Raman spectroscopic and EPR investigations of oxygen species on SrCl2-promoted Ln2O3 (Ln=Sm and Nd) catalysts for ethane-selective oxidation to ethene

Raman spectroscopic and EPR investigations of oxygen species on SrCl2-promoted Ln2O3 (Ln=Sm and Nd) catalysts for ethane-selective oxidation to ethene

Applied Catalysis A: General 202 (2000) 1–15 Raman spectroscopic and EPR investigations of oxygen species on SrCl2 -promoted Ln2 O3 (Ln=Sm and Nd) ca...

361KB Sizes 0 Downloads 19 Views

Applied Catalysis A: General 202 (2000) 1–15

Raman spectroscopic and EPR investigations of oxygen species on SrCl2 -promoted Ln2 O3 (Ln=Sm and Nd) catalysts for ethane-selective oxidation to ethene H.X. Dai, C.F. Ng, C.T. Au∗ Department of Chemistry and Center for Surface Analysis and Research, Hong Kong Baptist University, Kowloon Tong, Hong Kong, China Received 24 September 1999; received in revised form 20 December 1999; accepted 6 January 2000

Abstract The SrCl2 -promoted Ln2 O3 (Ln=Sm and Nd) catalysts have been investigated for the oxidative dehydrogenation of ethane (ODE) to ethene. With the doping of SrCl2 into Ln2 O3 , the C2 H4 selectivity and C2 H6 conversion were enhanced considerably. We also found that the addition of SrCl2 to Ln2 O3 could markedly reduce the deep oxidation of C2 H4 . The 40 mol% SrCl2 /Ln2 O3 catalysts were stable for 60 h of on-stream ODE reaction. Under the reaction conditions of temperature=640◦ C and space velocity=6000 ml h−1 g−1 , 40 mol% SrCl2 /Sm2 O3 showed 80.3% C2 H6 conversion, 70.9% C2 H4 selectivity, and 56.9% C2 H4 yield while 40 mol% SrCl2 /Nd2 O3 gave 63.8% C2 H6 conversion, 74.3% C2 H4 selectivity, and 47.4% C2 H4 yield. X-ray photoelectron spectroscopic and chemical analysis of chloride indicated that the Cl− anions were evenly distributed in the 40 mol% SrCl2 /Ln2 O3 catalysts. We observed that Cl leaching was insignificant. The results of temperature-programmed desorption of oxygen and temperature-programmed reduction studies demonstrated that the addition of SrCl2 to Ln2 O3 enhanced the activation of oxygen molecules. We believe that such improvement is closely associated with the defects formed during the exchanges of ions between the SrCl2 and Ln2 O3 phases. X-ray powder diffraction results revealed that the Ln2 O3 lattices were enlarged, whereas the SrCl2 lattices contracted in the 40 mol% SrCl2 /Ln2 O3 catalysts. In situ Raman results indicated that there were dioxygen adspecies such as O2 2− , O2 n − (1
1. Introduction Ethane is the second largest component of natural gas and is a main product of the oxidative coupling of ∗ Corresponding author. Tel.: +852-2339-7067; fax: +852-2339-7348. E-mail address: [email protected] (C.T. Au)

methane (OCM) reaction. It has been used as a starting material for the production of ethene via the oxidative dehydrogenation of ethane (ODE) process. In the past decades, compounds such as transition metal oxides, alkali metal oxides, alkaline earth oxides, rare earth oxides, and perovskite-type mixed oxides have been investigated as OCM and ODE catalysts. Rare earth oxides have been reported to be active for the two

0926-860X/00/$ – see front matter © 2000 Elsevier Science B.V. All rights reserved. PII: S 0 9 2 6 - 8 6 0 X ( 0 0 ) 0 0 4 5 3 - 1

2

H.X. Dai et al. / Applied Catalysis A: General 202 (2000) 1–15

reactions e.g., [1–9]. Otsuka et al. [1] pointed out that the catalytic performance for the OCM reaction of rare earth oxides increased in the order of CeO2 , TbOx , PrOx
The catalytic performance was tested with 0.5 g of the catalyst in a fixed-bed quartz micro-reactor (i.d.=4 mm) at atmospheric pressure using a gaseous mixture of ethane (14.8 ml min−1 ) and air (35.2 ml min−1 ), corresponding to a C2 H6 /O2 /N2 molar ratio of 2/1/3.7 and a space velocity of 6000 ml h−1 g−1 . A thermocouple was placed in the middle of the catalyst bed to determine the reaction temperature. The product mixtures (C2 H6 , C2 H4 , CH4 , CO, CO2 , and O2 ) were analyzed on-line by a Shimadzu 8A TCD gas chromatography with Porapak Q and 5A molecular sieve columns. For the variation of space velocity, the catalyst mass was varied at a fixed reactant gas flow rate of 50 ml min−1 . Ethane conversion and ethene, methane, carbon monoxide, and carbon dioxide selectivities were calculated based on the balance of carbon [19]. The balances of carbon and oxygen were estimated to be 100±2 and 100±3%, respectively, for every run over the catalysts. The crystal phases of the catalysts were determined by X-ray powder diffraction (XRD, D-MAX, Rigaku Rotaflex) operating at 40 kV and 40 mA using Cu K␣ radiation (λ=1.542 Å) filtered with a nickel monochromator. X-ray photoelectron spectroscopy (XPS, Leybold Heraeus-Shengyang SKL-12) was used to characterize the catalyst surfaces. The contents of surface halides were calculated according to the procedure described in [14]. The specific surface areas of the catalysts were measured by a NOVA 1200 instrument and estimated according to the BET method. The continuous flow chromatographic technique was adopted, with helium as the carrier gas and nitrogen as the adsorbate. In situ Raman experiments were performed on a Nicolet 560 FT laser Raman spectrometer. The samples were treated in O2 , C2 H6 , and C2 H6 /O2 /N2 (molar ratio=2/1/3.7), respectively, at different temperatures. After various treatments, the samples were monitored at 25◦ C without being exposed to air. The electron paramagnetic resonance (EPR) experiments were carried out on a JES-TE100 spectrometer. Samples (ca. 1.5 g) treated under various conditions were examined in the X band at 25◦ C. The EPR system was equipped with a quartz-tube reactor which was easily transferable between the sample chamber and an oven. The sample in the quartz reactor could be heated to 900◦ C and exposed to gas(es) without being exposed to air.

H.X. Dai et al. / Applied Catalysis A: General 202 (2000) 1–15

For the studies of temperature-programmed desorption (TPD) of oxygen, the samples (0.2 g) were placed in the middle of a quartz micro-reactor (i.d.=4 mm). The outlet gases were analyzed on-line by mass spectrometry (HP G1800A). The heating rate was 10◦ C min−1 and the temperature range was from room temperature to 800◦ C. Before one performed the O2 -TPD experiments, the samples were calcined in situ at 800◦ C for 30 min in an oxygen flow of 20 ml min−1 , followed by cooling in oxygen to room temperature and then purging with helium (flow rate, 20 ml min−1 ) for 30 min. The quantification of O2 desorbed from the catalysts was made by calibrating the peak areas against a standard pulse of O2 . Temperature-programmed reduction (TPR) was conducted by using a 7% H2 -93% N2 (v/v) mixture. The flow rate of the carrier gas was 50 ml min−1 and a thermal conductivity detector was employed. The amount of the sample used was 0.2 g and the heating rate was 10◦ C min−1 . Before one performed the TPR experiments, the sample was first calcined in situ at 800◦ C for 30 min in an oxygen flow of 15 ml min−1 followed by cooling in oxygen to room temperature. The titration method of chlorine content has been described in one of our previous reports [15]. 3. Results 3.1. Catalytic performance Table 1 shows the catalytic performances at 640◦ C of Ln2 O3 (Ln=Sm and Nd), SrCl2 , 40 mol% SrCl2 /Ln2 O3 , and quartz sand. Clearly, quartz sand and SrCl2 exhibited poor catalytic activities. Both

3

Sm2 O3 and Nd2 O3 were active for the ODE reaction; the catalytic activity of the latter was better than that of the former. With the addition of SrCl2 to the host oxides, the performances increased significantly. Under the reaction conditions of temperature=640◦ C and space velocity=6000 ml h−1 g−1 , we achieved 80.3% C2 H6 conversion, 70.9% C2 H4 selectivity, and 56.9% C2 H4 yield over 40 mol% SrCl2 /Sm2 O3 , and 63.8% C2 H6 conversion, 74.3% C2 H4 selectivity, and 47.4% C2 H4 yield over 40 mol% SrCl2 /Nd2 O3 . The catalytic performances of Ln2 O3 (Ln=Sm and Nd) and 40 mol% SrCl2 /Ln2 O3 as related to reaction temperature at a space velocity of 6000 ml h−1 g−1 are shown in Table 2. One can observe that, over the undoped Ln2 O3 catalysts, with the rise in temperature from 540 to 680◦ C, C2 H6 conversions, C2 H4 selectivities, and C2 H4 yields increased, whereas COx (i.e., CO+CO2 ) selectivities decreased; O2 conversion remained at high values. After the addition of SrCl2 to Ln2 O3 , there were significant improvements in catalytic performance. With the increase in temperature, C2 H6 and O2 conversions as well as CH4 selectivity increased, whereas C2 H4 selectivity and C2 H4 yield reached a maximum value at 640◦ C. Table 3 shows the results of the oxidation of C2 H4 and C2 H6 , respectively, over Ln2 O3 and 40 mol% SrCl2 /Ln2 O3 at 640◦ C and 6000 ml h−1 g−1 . With the addition of SrCl2 to Ln2 O3 , C2 H4 selectivity increased, whereas C2 H4 conversion decreased markedly and the CO/CO2 ratios in the products increased considerably in the oxidation of C2 H4 . Such results indicate that, with the introduction of SrCl2 to Ln2 O3 , the deep oxidation of C2 H4 was reduced significantly.

Table 1 The catalytic performances after an on-stream time of 1 h over Ln2 O3 (Ln=Sm and Nd) and 40 mol% SrCl2 /Ln2 O3 at 640◦ C and 6000 ml h−1 g−1 Catalyst

sandb

Quartz Sm2 O3 Nd2 O3 SrCl2 SrCl2 /Sm2 O3 SrCl2 /Nd2 O3 a b

COx =CO+CO2 . Tested at 660◦ C.

Conversion (%)

Selectivity (%)

Yield (%)

C2 H 6

O2

COx a

CH4

C2 H 4

C2 H 4

5.0 41.4 55.0 2.0 80.3 63.8

– 99.8 99.9 11.1 93.0 89.9

12.0 54.8 49.4 11.3 25.8 22.3

0 1.7 2.8 0.1 3.3 3.4

88.0 43.6 47.8 88.6 70.9 74.3

4.4 18.0 26.3 1.8 56.9 47.4

4

H.X. Dai et al. / Applied Catalysis A: General 202 (2000) 1–15

Table 2 Catalytic performances of the undoped Ln2 O3 (Ln=Sm and Nd), and 40 mol% SrC12 /Ln2 O3 catalysts as related to reaction temperature at 6000 ml h−1 g−1 Catalyst

Temp. (◦ C)

Conversion (%)

Selectivity (%)

Yield (%)

C2 H 6

O2

COx

CH4

C2 H 4

C2 H 4

Sm2 O3

540 580 600 620 640 660 680

34.6 36.9 38.1 40.6 41.4 42.8 44.4

99.9 99.9 99.9 99.9 99.9 99.8 99.8

67.3 62.1 59.7 57.1 54.8 52.1 50.6

1.6 1.6 1.7 1.7 1.7 1.8 1.7

31.2 36.3 38.7 41.2 43.6 46.1 47.7

10.8 13.4 14.7 16.7 18.0 19.7 21.2

Nd2 O3

540 580 600 620 640 660 680

40.3 47.4 49.1 52.9 55.0 56.6 57.6

88.5 93.2 96.4 97.8 99.9 99.9 99.9

59.0 56.9 54.6 52.1 49.4 46.2 44.7

2.1 2.2 2.3 2.5 2.8 3.0 3.1

38.9 40.9 43.1 45.4 47.8 50.8 52.2

15.7 19.4 21.1 24.0 26.3 28.8 30.1

SrCl2 /Sm2 O3

580 600 620 640 660 680

57.8 66.7 73.4 80.3 82.8 84.2

84.3 87.8 89.9 93.0 95.7 98.2

41.9 35.5 30.3 25.8 28.4 29.1

1.6 1.9 2.5 3.3 3.8 4.5

56.5 62.6 67.2 70.9 67.8 66.4

32.7 41.7 49.3 56.9 56.1 55.9

SrCl2 /Nd2 O3

540 580 600 620 640 660 680

30.4 48.8 54.8 58.3 63.8 64.4 65.0

73.8 79.9 84.3 86.6 89.9 91.1 93.8

45.5 32.3 28.5 25.6 22.3 23.3 24.6

2.5 2.7 2.8 3.2 3.4 4.3 4.8

52.0 65.0 68.7 71.2 74.3 72.4 70.6

16.5 31.7 37.6 41.5 47.4 46.6 45.9

In order to investigate the effect of SrCl2 loading on the catalytic performance, we prepared SrCl2 /Ln2 O3 catalysts with different SrCl2 loadings and monitored their activities; the results are plotted in Fig. 1. With the increase in SrCl2 content, C2 H4 selectivities in-

creased while O2 conversions decreased; C2 H6 conversions and C2 H4 yields reached the highest values at a SrCl2 loading of 40 mol%. For lifetime studies, one can see from Fig. 2 that within 60 h of on-stream reaction, the activities of the two 40 mol% SrCl2 /Ln2 O3

Table 3 The catalytic performances of Ln2 O3 and 40 mol% SrCl2 /Ln2 O3 in the oxidation of ethane and ethene at temperature=640◦ C and space velocity=6000 ml h−1 g−1 Catalyst

Sm2 O3 Nd2 O3 SrC12 /Sm2 O3 SrCl2 /Nd2 O3 a

Oxidation of C2 H4 a

Oxidation of C2 H6

C2 H4 Conv. (%)

CO/CO2 (ratio)

C2 H6 Conv. (%)

C2 H4 Sel. (%)

27.6 31.2 15.3 13.8

1/13.7 1/12.8 1/2.1 1/1.6

41.4 55.0 80.3 63.8

43.6 47.8 70.9 74.3

At C2 H4 /O2 /N2 molar ratio=2/1/3.7.

H.X. Dai et al. / Applied Catalysis A: General 202 (2000) 1–15

Fig. 1. The catalytic performance of (a) SrCl2 /Sm2 O3 and (b) SrCl2 /Nd2 O3 at 640◦ C and 6000 ml h−1 g−1 as related to SrCl2 loading (䊏) C2 H6 conversion, (䉬) C2 H4 selectivity, (䉱) C2 H4 yield, and (䊊) O2 , conversion.

catalysts were stable. With the rise in space velocity from 4000 to 10,000 ml h−1 g−1 , there were a rise in C2 H4 selectivity and drops in C2 H6 conversion and COx selectivity; a maximum C2 H4 yield was observed at 6000 ml h−1 g−1 (Fig. 3). Similar results were observed when the two catalysts were well dispersed in quartz sand (0.5 g catalyst/5.0 g quartz sand), indicating that the problem of hot spots was insignificant. 3.2. BET, XRD, and XPS studies Table 4 shows the phase compositions and surface areas of Ln2 O3 (Ln=Sm and Nd) and 40 mol% SrCl2 /Ln2 O3 measured before and after 60 h of ODE on-stream reaction at 640◦ C and 6000 ml h−1 g−1 . After 8 h, although no new phases were detected, the surface areas of undoped Sm2 O3 and Nd2 O3 were reduced by ca. 26 and 21%, respectively. For the 40 mol% SrCl2 /Sm2 O3 and 40 mol% SrCl2 /Nd2 O3 catalysts, we observed very weak signals of SrCO3

5

Fig. 2. Lifetime study of (a) 40 mol% SrCl2 /Sm2 O3 and (b) 40% SrCl2 /Nd2 O3 during 60 h of on-stream ODE reaction at 640◦ C and 6000 ml h−1 g−1 . (䊏) C2 H6 conversion, (䉬) C2 H4 selectivity, (䉱) C2 H4 yield, and (䊊) O2 conversion.

after 60 h; the surface areas of the catalysts decreased slightly, by only ca. 1%. The chloride compositions on the surface and in the bulk of the fresh and used (after 60 h of on-stream reaction) 40 mol% SrCl2 /Ln2 O3 catalysts were calculated based on the results of XPS and chemical analyses (Table 5). One can observe that both surface and bulk chloride concentrations were rather similar for each of the two SrCl2 -doped catalysts. In other words, chloride ions were distributed rather evenly throughout the two catalysts. The similarity in chloride contents in the fresh and used catalysts indicates that the problem of Cl leaching was insignificant; i.e., the two SrCl2 -modified catalysts were stable within a period of 60 h ODE reaction. Table 6 shows the lattice parameters of the crystal phases of Ln2 O3 and SrCl2 in 40 mol% SrCl2 /Ln2 O3 estimated according to the d values of XRD lines. One can realize that the Sm2 O3 and Nd2 O3 lattices were enlarged and the SrCl2 lattices were contracted. Also, the extent of SrCl2 lattice contraction

6

H.X. Dai et al. / Applied Catalysis A: General 202 (2000) 1–15

Fig. 4. The O2 -TPD profiles of (a) fresh Sm2 O3 , (b) fresh and (b0 ) used (60 h of on-stream reaction) 40% SrCl2 /Sm2 O3 , (c) fresh Nd2 O3 , and (d) fresh and (d0 ) used 40 mol% SrCl2 /Nd2 O3 catalysts. Fig. 3. The catalytic performances of (a) 40 mol% SrCl2 /Sm2 O3 and (b) 40 mol% SrCl2 /Nd2 O3 at 640◦ C as a function of space velocity. (䊏) C2 H6 conversion, (䉬) C2 H4 selectivity, (䉱) C2 H4 yield, and (䊉) COx selectivity.

in 40 mol% SrCl2 /Nd2 O3 was larger than that in 40 mol% SrCl2 /Sm2 O3 . 3.3. O2 -TPD and TPR studies Fig. 4 shows the O2 -TPD profiles observed over fresh Ln2 O3 and fresh and used (60 h of on-stream ODE reaction) 40 mol% SrCl2 /Ln2 O3 . For fresh

Sm2 O3 and 40 mol% SrCl2 /Sm2 O3 , there were a large desorption peak centered at ca. 572◦ C and a broad one centered at ca. 720◦ C (totally amount to 1.66 ␮mol g−1 ) over the former and a large peak centered at ca. 606◦ C and a broad one centered at ca. 699◦ C (totally amount to 3.12 ␮mol g−1 ) over the latter. However, for the undoped Nd2 O3 and SrCl2 -doped Nd2 O3 samples, there was only one desorption peak centered at ca. 588◦ C (amount to 1.02 ␮mol g−1 ) over the former and at 616◦ C (amount to 2.29 ␮mol g−1 ) over the latter. From Fig. 4, one can see that the addition of SrCl2 into Ln2 O3 enhanced the extent of O2 desorption. In addition, the desorption behaviors

Table 4 The crystal phases and surface areas of Ln2 O3 (Ln=Sm and Nd), and 40 mol% SrCl2 /Ln2 O3 catalysts measured before and after ODE reaction at 640◦ C for 60 h Catalysts

Sm2 O3 Nd2 O3 SrCl2 /Sm2 O3 SrCl2 /Nd2 O3 a b

Surface area (m2 g−1 )

Crystal phasea Before

After

Before

After

Sm2 O3 (s) Nd2 O3 (s) Sm2 O3 (s), SrCl2 ·6H2 O (m) Nd2 O3 (s), SrCl2 ·6H2 O (m)

Sm2 O3 (s) Nd2 O3 (s) Sm2 O3 (s) SrCO3 (vw), SrCl2 ·6H2 O (m) Nd2 CO3 (s) SrCO3 (vw), SrCl2 ·6H2 O (m)

4.6 7.4 3.7 4.8

3.4b 5.8b 3.4 4.3

Nd2 O3 , is hexagonal. Sm2 O3 is cubic. SrCO3 is orthorhombic. SrCl2 ·6H2 O is hexagonal. s, strong; m, medium; vw, very weak. 8 h of on-stream ODE reaction.

H.X. Dai et al. / Applied Catalysis A: General 202 (2000) 1–15

7

Table 5 The surface and bulk halide compositions of 40 mol% SrCl2 /Ln2 O3 (Ln=Sm and Nd) catalysts estimated from the results of XPS and chemical analyses, respectively Catalyst

SrCl2 /Sm2 O3 SrCl2 /Nd2 O3 a

Surface composition (wt.%)

Bulk composition (wt.%)

Before

Aftera

Before

Aftera

10.11 10.46

10.20 10.56

10.30 10.55

10.22 10.36

After 60 h of ODE reaction at 640◦ C.

Table 6 The lattice parameters of Ln2 O3 and SrCl2 , phases in the 40 mol% SrCl2 /Ln2 O3 catalysts calculated according to the d values of XRD patterns Catalyst

40 mol% SrCl2 /Sm2 O3

Phase

Sm2 O3 SrCl2

40 mol% SrCl2 /Nd2 O3

Nd2 O3 SrCl2

a

Lattice parameter (Å) a

b

c

10.9478 (10.9306)a 7.9419

10.9478 (10.9306) 7.9419

10.9478 (10.9306) 4.1122

3.8322 (3.8297) 7.9361

3.8322 (3.8297) 7.9361

6.0109 (5.9987) 4.1072

The values in brackets are the PDF-2 data of the pure compounds.

of the fresh and used 40 mol% SrCl2 /Ln2 O3 samples were rather similar, except for the slight shift in peak positions towards higher temperature(s). It indicated that the two SrCl2 -doped Ln2 O3 catalysts were basically intact after 60 h of on-stream reaction. Fig. 5 shows the TPR profiles of Ln2 O3 and 40 mol% SrCl2 /Ln2 O3 . For Sm2 O3 , there were a broad TPR band stretching from 450 to 610◦ C and

a band centred at ca. 700◦ C. For Nd2 O3 , two TPR bands were observed at ca. 480 and 694◦ C, with the latter being much bigger than the former in intensity. With the addition of SrCl2 to Sm2 O3 and Nd2 O3 , however, the onset temperatures for reduction were lowered from ca. 450 to 425◦ C and from ca. 480 to 430◦ C, respectively; two TPR bands were also observed at ca. 644 and 672◦ C over the former and at ca. 636 and 655◦ C over the latter, with the first band being much bigger than the second one. From Fig. 5, one can realize that, aside from the decrease in reduction temperature, the doping of SrCl2 into Ln2 O3 caused the reduction bands to increase in intensity. In other words, the modification of Ln2 O3 with SrCl2 enhanced the amount of oxygen that was reducible. 3.4. In situ laser Raman studies

Fig. 5. The TPR profiles of (a) Sm2 O3 , (b) Nd2 O3 , (C) 40 mol% SrCl2 /Sm2 O3 , and (d) 40 mol% SrCl2 /Nd2 O3 catalysts.

Fig. 6 shows the Raman spectra of Sm2 O3 and Nd2 O3 treated in O2 at 25 and 700◦ C for 30 min, respectively. At 25◦ C, we observed three very weak peaks at ca. 811, 927, and 964 cm−1 over the Sm2 O3 sample; a rise in treatment temperature to 700◦ C caused these signals to increase in intensity. We assign these three signals to O2 2− and/or O2 n − (1
8

H.X. Dai et al. / Applied Catalysis A: General 202 (2000) 1–15

Fig. 6. In situ Raman spectra of (a) Sm2 O3 and (b) Nd2 O3 when the samples were treated in O2 at 25 and 700◦ C for 30 min, respectively.

[25–27]. There was no peak observed over a Nd2 O3 sample at 25◦ C; however, with the rise in treatment temperature to 700◦ C, five weak peaks were observed at ca. 900, 978, 1082, 1233, and 1316 cm−1 . The former three peaks could be ascribed to O2 2− and/or O2 n − (1
Fig. 7. In situ Raman spectra of 40 mol% SrCl2 /Sm2 O3 when it was treated: (a) in O2 at 25◦ C for 15 min, (b) then in O2 at 500◦ C for 15 min, (c) then in O2 at 700◦ C for 15 min, (d) then in H2 at 500◦ C for 15 min, (e) then in H2 at 640◦ C for 15 min, (f) then in H2 at 700◦ C for 15 min, and (g) then in C2 H6 /O2 /N2 (molar ratio=2/1/3.7) for 30 min.

1054 cm−1 , a shoulder at ca 878 cm−1 , and a broad peak centered at ca. 1366 cm−1 were observed over the sample treated in O2 at 700◦ C for 15 min. We assign the signal at ca. 878 cm−1 to O2 2− [25,26], those at ca. 988 and 1054 cm−1 to O2 n − (1
H.X. Dai et al. / Applied Catalysis A: General 202 (2000) 1–15

9

Fig. 8. In situ Raman spectra of 40 mol% SrCl2 /Nd2 O3 when it was treated: (a) in O2 at 25◦ C for 15 min, (b) then in O2 at 500◦ C for 15 min, (c) then in O2 at 700◦ C for 15 min, (d) then in H2 at 500◦ C for 15 min, (e) then in H2 at 640◦ C for 15 min, (f) then in H2 at 700◦ C for 15 min, and (g) then in C2 H6 /O2 /N2 (molar ratio=2/1/3.7) for 30 min.

O2 for 30 min at temperatures which varied from 25 to 700◦ C, the intensities of the signals with g components of 2.036, 2.008, and 1.975 increased significantly (Fig. 9a, b and a0 , b0 ). The signal at 2.036 could be assigned to O2 − and the doublet superhyperfine structure at 2.008 and 1.975 to O− [19]. Further treatment in C2 H6 at 700◦ C for 30 min would result in the detection of a signal at 2.007 ascribable to trapped electrons [16] (Fig. 9c and c0 ). After treatment in O2 at 700◦ C for 30 min, the signal due to trapped electrons disappeared and the signals due to O2 − and O− species reappeared (Fig. 9d and d0 ). Fig. 10 shows the EPR spectra of 40 mol% SrCl2 /Sm2 O3 treated under different conditions. When the sample was kept in O2 at 25◦ C for 30 min, only O− species was detected with g tensors of 2.016 and 1.975 (Fig. 10a). By treating the sample in O2 at 700◦ C for 30 min, signals at 2.034, 2.016, and

Fig. 9. EPR spectra of Sm2 O3 (a–d) an Nd2 O3 (a0 –d0 ) when the samples were treated: (a, a0 ) in O2 at 25◦ C for 30 min, (b, b0 ) then in O2 at 700◦ C for 30 min, (c, c0 ) in H2 at 700◦ C for 30 min, and (d, d0 ) then in C2 H6 /O2 /N2 (molar ratio=2/1/3.7) at 640◦ C for 30 min.

1.975 were observed (Fig. 10b). At the same treatment temperature, the introduction of a flow of H2 (20 ml min−1 ) to the sample for 30 min would result in the generation of trapped electrons with a g value of 2.006 (Fig. 10c). When the sample treated in O2 at 700◦ C for 30 min was exposed to C2 H6 for 30 min at temperatures which varied from 500 to 640◦ C, the signal at 2.016 disappeared and that at 1.975 decreased in intensity (Fig. 10d and e). Further treatment in C2 H6 at 700◦ C for 30 min would give rise to

10

H.X. Dai et al. / Applied Catalysis A: General 202 (2000) 1–15

O2 at 700◦ C for 30 min, except for the decrease in intensities of these signals (Fig. 10g). The EPR spectra of 40 mol% SrCl2 /Nd2 O3 recorded under different treatment conditions are shown in Fig. 11. After treatment in O2 at 25◦ C for 30 min, signals at 2.015 and 1.977 due to O− species were observed (Fig. 11a). When the sample was treated in O2 at

Fig. 10. EPR spectra of 40 mol% SrCl2 /Sm2 O3 when the sample was treated: (a) in O2 at 25◦ C for 30 min, (b) then in O2 at 700◦ C for 30 min, (c) then in H2 at 700◦ C for 30 min, (d) first in O2 at 700◦ C for 30 min and then in C2 H6 at 500◦ C for 30 min, (e) then in C2 H6 at 640◦ C for 30 min, (f) then in C2 H6 at 700◦ C for 30 min, and (g) then in C2 H6 /O2 /N2 (molar ratio=2/1/3.7) at 640◦ C for 30 min.

the detection of the signal due to trapped electrons at g=2.006 (Fig. 10f). After treatment in a C2 H6 /O2 /N2 (molar ratio=2/1/3.7) flow (20 ml min−1 ) at 700◦ C for 30 min, the EPR spectrum recorded was similar to that obtained after the sample was treated in

Fig. 11. EPR spectra of 40 mol% SrCl2 /Nd2 O3 when the sample was treated:(a) in O2 at 25◦ C for 30 min, (b) then in O2 at 700◦ C for 30 min, (c) then in H2 700◦ C for 30 min, (d) first in O2 at 700◦ C for 30 min and then in C2 H6 at 500◦ C for 30 min, (e) then in C2 H6 at 640◦ C for 30 min, (f) then in C2 H6 at 700◦ C for 30 min, and (g) then in C2 H6 /O2 /N2 (molar ratio=2/1/3.7) at 640◦ C for 30 min.

H.X. Dai et al. / Applied Catalysis A: General 202 (2000) 1–15

700◦ C for 30 min, signals at 2.035, 2.015, and 1.977 were detected (Fig. 11b). When the sample was further treated in H2 at 700◦ C for 30 min, the signal of trapped electrons with a g value of 2.007 was detected (Fig. 11c). By exposing the sample treated in O2 at 700◦ C to C2 H6 for 30 min at temperatures which varied from 500 to 640◦ C, we found that the signal at g=2.015 disappeared and the one at 1.977 decreased in intensity (Fig. 11d and e). Further treatment in C2 H6 at 700◦ C for 30 min would lead to the detection of the signal of trapped electrons (g=2.007) (Fig. 11f). When the sample was further treated in a C2 H6 /O2 /N2 (molar ratio=2/1/3.7) flow for 30 min, the signal due to trapped electrons disappeared and all the signals due to O2 − and O− species reappeared (Fig. 11g).

4. Discussion 4.1. Catalytic performance A number of rare earth oxides have been tested as OCM or ODE catalysts [1–9]. Most of them showed moderate catalytic activity. Adding promoters such as alkali metal oxides [33], MO (M=alkaline earth metals) [13,17–19,33], and MX2 (X=F, Cl, and Br) [15–19] to the lanthanide oxides could improve their catalytic abilities. For the ODE reaction, as shown in Table 1, a single-component Sm2 O3 or Nd2 O3 catalyst performed moderately. However, the doping of SrCl2 into the two oxides enhanced the C2 H6 conversion and C2 H4 selectivity significantly. From Fig. 1, one can observe that, at a 40 mol% loading of SrCl2 in the SrCl2 -promoted Sm2 O3 and Nd2 O3 catalysts, the C2 H4 yield was at its maximum. During the 60 h of on-stream reaction, the 40 mol% SrCl2 /Sm2 O3 and 40 mol% SrCl2 /Nd2 O3 catalysts showed stable activities (Fig. 2a and b). In the space velocity studies (Fig. 3a and b), with the rise in space velocity, the C2 H6 conversion and COx selectivity decreased, while the C2 H4 selectivity increased. Similar results were observed when the two SrCl2 -modified Ln2 O3 catalysts were well dispersed in quartz sand (0.5 g catalyst/5.0 g quartz sand). This implies that the obvious enhancement in catalytic performance is a result of catalytic action rather than a result of hot spot generation. From the activity data in Table 2, one can realize that, besides the improved catalytic behaviors,

11

the introduction of SrCl2 to Ln2 O3 caused O2 conversions to drop at high C2 H6 conversions and C2 H4 selectivities. This is understandable, since an enhancement in C2 H4 selectivity signals the dominance of the ODE reaction, in which a smaller amount of O2 can convert relatively more C2 H6 , as compared to the deep oxidation reactions. It should be noted that, when the reaction temperature was at or above 660◦ C, the C2 H4 selectivity decreased, whereas CH4 and COx selectivities increased over the SrCl2 -doped catalysts. The formation of CH4 requires the cleavage of a C–C bond. As suggested by Kennedy and Cant [8], methane generation follows two possible routes: i.e., ethane decomposition in the gas phase and a heterogeneous pathway involving an ethylperoxy intermediate. Ethylperoxy reacted with surface oxygen species to form CH4 and HCO2 ; the latter was further oxidized to COx . In C2 H6 oxidation, O2 conversion, C2 H4 selectivity, and C2 H6 conversion can increase simultaneously if deep oxidation is reduced. Usually, a comparison of C2 H4 selectivities between various catalysts should be based on a similar level of C2 H6 conversion. It is difficult to make such a comparison because, at a particular C2 H6 conversion, the reaction temperature and O2 conversion varied significantly from catalyst to catalyst. Hence, we chose to compare the C2 H4 selectivities and C2 H6 conversions of the catalysts under similar reaction conditions (for example, at 640◦ C and at ca. 90% O2 conversion). We conclude that catalytic performance increased in the order: Sm2 O3
12

H.X. Dai et al. / Applied Catalysis A: General 202 (2000) 1–15

C2 H4 conversions in the C2 H4 oxidation reaction and much higher C2 H4 selectivities in the ODE reaction. Similar effects were observed over the BaX2 -doped Ho2 O3 [17] and Y2 O3 [18] catalysts. Therefore, we conclude that the addition of SrCl2 can reduce the deep oxidation of C2 H4 formed in the ODE reaction and, thus, enhance significantly the C2 H4 selectivity. 4.2. Defective structure induced by SrCl2 modification From the XRD results (Table 4), one can observe that the crystal structures of Sm2 O3 and Nd2 O3 in the undoped catalysts remained unaltered after 8 or 60 h of on-stream reaction. As for the SrCl2 -doped ones, we detected weak signals of orthorhombic SrCO3 phase after 60 h. The surface areas of the undoped catalysts reduced by more than 21%, whereas those of the SrCl2 -promoted ones decreased only slightly after 60 h of ODE reaction. It indicates that the addition of SrCl2 could be beneficial for the stabilization of surface area of the rare earth oxides. From Table 5, it can be seen that the Cl compositions on the surface and in the bulk of each of the SrCl2 -modified catalysts were rather similar, indicating a uniform distribution of Cl− ions in the catalysts; furthermore, the small difference in Cl− concentration in the two catalysts before and after 60 h of reaction means that chlorine leaching was insignificant. The sustainable catalytic performances of 40 mol% SrCl2 /Ln2 O3 are pieces of supporting evidence for this implication. It has been generally accepted that ionic exchanges or substitutions take place between rare earth oxides and alkaline earth oxides or halides. Osada et al. [10] reported the occurrence of Y2 O3 lattice distortions induced by the infiltration of Ca2+ into Y2 O3 in the binary oxide Y2 O3 -CaO catalyst. Erarslanoglu et al. [36] pointed out that the incorporation of Sr2+ cations into the Y2 O3 lattice led to the formation of defective structure. Filkova et al. [13] ascribed the improved catalytic performance of the SrO-doped Nd2 O3 catalyst to the incorporation of Sr into the Nd2 O3 lattice. Previously, we proposed that the presence of SrF2 in SmOF [19] and of BaX2 (X=F, Cl, and Br) in Nd2 O3 [15] or Y2 O3 [18] caused the rare earth oxyfluoride or oxide lattices to enlarge and the halide lattice to contract, due to partial ionic exchanges between phases. The size of Sr2+ ion (radius, 1.18 Å) is larger than

those of Sm3+ (radius, 0.96 Å) and Nd3+ (radius, 0.98 Å) whereas a O2− ion (radius, 1.40 Å) is smaller than a Cl− ion (radius, 1.81 Å). The ionic exchanges between SrCl2 and Sm2 O3 or Nd2 O3 would result in the enlargement of the Sm2 O3 or Nd2 O3 lattice and the contraction of the SrCl2 lattice in 40 mol% SrCl2 /Ln2 O3 . It should be noted that, in addition to ionic radii, lattice distortion is also closely associated with the temperature adopted for calcination. As pointed out by West [37], solid reactions between two powders can be sped up in the presence of a small amount of liquid phase that acts as a medium for the transport of matter. According to this viewpoint, we chose 950◦ C (which is near the melting point of SrCl2 ) for the calcination of the catalysts. The extent of ionic exchanges would determine the defect density and hence the catalytic activity. The addition of SrCl2 modified the surface and bulk natures of Sm2 O3 and Nd2 O3 as well as those of SrCl2 itself. As shown in Table 6, the lattices of Sm2 O3 and Nd2 O3 in 40 mol% SrCl2 /Ln2 O3 were enlarged, whereas those of SrCl2 in 40 mol% SrCl2 /Ln2 O3 contracted. Obviously, the extent of Sm2 O3 lattice expansion in 40 mol% SrCl2 /Sm2 O3 was larger than that of the Nd2 O3 lattice in 40 mol% SrCl2 /Nd2 O3 , indicating that the amount of SrCl2 which entered into the Sm2 O3 lattice was more than that which entered into the Nd2 O3 lattice. On the contrary, the extent of SrCl2 lattice contraction in 40 mol% SrCl2 /Nd2 O3 was larger than that in 40 mol% SrCl2 /Sm2 O3 , indicating that the infiltration of Sm2 O3 into the SrCl2 lattice was less than that of Nd2 O3 . By comparing the activity data in Tables 1 and 2, one can realize that the 40 mol% SrCl2 /Sm2 O3 catalyst (which contained more SrCl2 in Sm2 O3 and less Sm2 O3 in SrCl2 ) performed better than the 40 mol% SrCl2 /Nd2 O3 catalyst (which contained less SrCl2 in Nd2 O3 and more Nd2 O3 in SrCl2 ). The difference in C2 H6 conversion and C2 H4 selectivity might be associated with the distortions of the SrCl2 and Ln2 O3 lattices. 4.3. Activation of oxygen and reactivity of di- and mono-oxygen adspecies towards ethane As illustrated in the O2 -TPD studies (Fig. 4), with the addition of SrCl2 to Ln2 O3 , the amount of O2 desorbed from the catalysts increased markedly and followed the order of 40 mol% SrCl2 /Sm2 O3 >40 mol%

H.X. Dai et al. / Applied Catalysis A: General 202 (2000) 1–15

SrCl2 /Nd2 O3 >Nd2 O3 >Sm2 O3 , coinciding with the sequence of their catalytic performances. The used 40 mol% SrCl2 /Ln2 O3 samples showed O2 desorption peaks (Fig. 4b0 and d0 ) rather similar to those of the fresh samples (Fig. 4b and d), indicating that the 40 mol% SrCl2 /Ln2 O3 catalysts were quite intact during the 60 h of on-stream reaction. According to the results of TPR studies (Fig. 5), the amount of oxygen in these catalysts reduced by H2 increased with the introduction of SrCl2 , and the order is the same as that of O2 desorption. It is clear that the addition of SrCl2 to Ln2 O3 could enhance significantly the extent of O2 adsorption. In other words, the activation of O2 was enhanced via the modification of Sm2 O3 or Nd2 O3 by SrCl2 . In the past decades, oxygen adspecies on catalysts have been investigated intensively. Eysel and Thym [26] suggested that the vibrational frequency for O2 2− lies in the 730–950 cm−1 region. Valentine [27] considered that the perturbed intermediate oxygen species O2 n − (1O2 − >O2− . Comparing the amounts of O2 2− , O2 n − , O2 − , and O2 δ− adspecies on the two SrCl2 -doped catalysts and their decreased reactivity towards C2 H6 , one can deduce that 40 mol% SrCl2 /Sm2 O3 should show higher C2 H6 conversion and lower C2 H4 selectivity than

13

40 mol% SrCl2 /Nd2 O3 under similar reaction conditions. Such a deduction is confirmed by the catalytic activities listed in Tables 1 and 2. By treating the two SrCl2 -modified samples (which had been calcined in O2 at 700◦ C for 15 min) in C2 H6 at temperatures which varied from 500 to 640 and then to 700o C, one finds that all the Raman bands of dioxygen adspecies decreased in intensity (Figs. 7d, e and 8d, e) and disappeared at 700◦ C (Figs. 7f and 8f), indicating that these dioxygen adspecies had reacted with C2 H6 . When a flow of reactant mixture was introduced at 640◦ C to the treated samples, all the Raman bands reappeared (Figs. 7g and 8g), indicating that the O2 in the reactant mixture had adsorbed on the catalysts. Based on the results of in situ Raman studies, we conclude that dioxygen such as O2 2− , O2 n − (1
14

H.X. Dai et al. / Applied Catalysis A: General 202 (2000) 1–15

signals due to O− species were observed over the samples treated in O2 at 25◦ C (Figs. 10a and 11a). When the O2 − treatment temperature was raised to 700◦ C, the signals due to O2 − species increased, whereas those due to O− species decreased in intensity (Figs. 10b and 11b), indicating that a certain amount of O− species had transformed into dioxygen. By comparing the EPR signal intensities of O− and O2 − species observed over 40 mol% SrCl2 /Nd2 O3 and 40 mol% SrCl2 /Sm2 O3 , one can see that there were more O2 − but less O− on the former catalyst. These results are in agreement with the in situ Raman observation that the amount of O2 − species on the former was larger than that on the latter (Figs. 7 and 8). Further treatment in H2 at 700◦ C brought about the formation of trapped electrons (Figs. 10c and 11c). With the rise in C2 H6 -treatment temperature from 500 to 700◦ C, all the signals due to O− and O2 − disappeared and trapped electrons were generated (Figs. 10d–f and 11d–f), indicating that the O− and O2 − species had reacted with C2 H6 . After passing a flow of reactant mixture to the treated samples, all the signals of O− and O2 − species reappeared (Figs. 10g and 11g), an indication of the occurrence of oxygen adsorption. By comparing the Raman and EPR signal intensities of O2 2− , O2 n − (140 mol% SrCl2 /Nd2 O3 >Nd2 O3 >Sm2 O3 (Figs. 6–8) and those of O2 n − and O2 − species decreased in the order of 40 mol% SrCl2 /Nd2 O3 >40 mol% SrCl2 /Sm2 O3 >Sm2 O3 >Nd2 O3 (Figs. 6–8), whereas that of O− species increased in the sequence of 40 mol% SrCl2 /Sm2 O3 <40 mol% SrCl2 /Nd2 O3 < Nd2 O3 40 mol% SrCl2 /Sm2 O3 >Nd2 O3 >Sm2 O3 (Tables 1 and 2). Taking into account the C2 H6 conversion as well, the performance of the catalysts followed the sequence of 40 mol% SrCl2 /Sm2 O3 >40 mol% SrCl2 /Nd2 O3 >Nd2 O3 >Sm2 O3 (Tables 1 and 2). Therefore, we conclude that the O2 2− , O2 n − , O2 − , and O2 δ− species are more selective for the oxidation of C2 H6 to C2 H4 , whereas the O− species are relatively more active for the deep oxidation of C2 H6 .

5. Conclusions The adding of SrCl2 to Ln2 O3 could significantly improve the catalytic performance and could considerably reduce C2 H4 deep oxidation. The 40 mol% SrCl2 /Ln2 O3 catalysts performed well and were durable in 60 h of on-stream reaction. Based on the results of XPS and chemical analyses of chloride, we considered that the Cl− anions were uniformly distributed on the surface and in the bulk of the 40 mol% SrCl2 /Ln2 O3 catalysts. We observed that the leaching of Cl was not serious under the reaction conditions adopted. XRD results revealed that the Ln2 O3 lattices were expanded while the SrCl2 lattices were contracted in the 40 mol% SrCl2 /Ln2 O3 catalysts. The results of O2 -TPD and TPR studies revealed that the addition of SrCl2 to Ln2 O3 could enhance the activation of gaseous oxygen molecules. We propose that such a effect is closely related to the lattice defects generated via ionic exchanges between the SrCl2 and Ln2 O3 phases. In situ Raman results indicated that there were O2 2− , O2 n − (1
Acknowledgements The project described above was fully supported by a grant from the Research Grants Council of the Hong Kong Special Administration Region, China (Project No. HKBU 2050/97P). H.X. Dai thanks the HKBU for a Ph.D. studentship. References [1] K. Otsuka, S. Yokoyama, A. Morikawa, Chem. Lett., 1985, 319. [2] K. Otsuka, K. Jinno, A. Morikawa, Chem. Lett., 1985, 499. [3] K. Otsuka, K. Jinno, A. Morikawa, J. Catal. 100 (1986) 353. [4] K. Otsuka, T. Komatsu, Chem. Lett., 1987, 483.

H.X. Dai et al. / Applied Catalysis A: General 202 (2000) 1–15 [5] S.J. Korf, J.A. Roos, J.W.H.C. Derksen, J.A. Vreeman, J.G. van Ommen, J.R.H. Ross, Appl. Catal. 59 (1990) 259. [6] V.D. Sokolovski, O.V. Buyevskaya, L.M. Plyasova, G.S. Litvak, N.Ph. Uvarov, Catal. Today 6 (1990) 489. [7] A.M. Maitra, Appl. Catal. 104 (1993) 11. [8] E.M. Kennedy, N.W. Cant, Appl. Catal. 75 (1991) 321. [9] S. Sugiyama, K. Sogabe, T. Miyamoto, H. Yayashi, J.B. Moffat, Catal. Lett. 42 (1996) 127. [10] Y. Osada, S. Koike, T. Fukushima, S. Ogasawara, T. Shikada, Appl. Catal. 59 (1990) 59. [11] J.M. Deboy, R.F. Hicks, J. Chem. Soc. Chem. Commun., 1988, 982. [12] J.M. Deboy, R.F. Hicks, J. Catal. 113 (1988) 517. [13] D. Filkova, D. Wolf, G. Gayko, M. Baerns, L. Petrov, Appl. Catal., A 159 (1997) 33. [14] C.T. Au, Y.W. Liu, C.F. Ng, J. Catal. 171 (1997) 231. [15] C.T. Au, Y.W. Liu, C.F. Ng, J. Catal. 176 (1998) 365. [16] C.T. Au, X.P. Zhou, Y.W. Liu, W.J. Ji, C.F. Ng, J. Catal. 174 (1998) 153. [17] C.T. Au, K.D. Chen, H.X. Dai, Y.W. Liu, J.Z. Luo, C.F. Ng, J. Catal. 179 (1998) 300. [18] H.X. Dai, Y.W. Liu, C.F. Ng, C.T. Au, J. Catal. 187 (1999) 59. [19] C.T. Au, X.P. Zhou, J. Chem. Soc. Faraday Trans. 93 (3) (1997) 485. [20] T. Ito, J.H. Lunsford, Nature 314 (1985) 721. [21] D. Dissanayake, J.H. Lunsford, M.P. Rosynek, J. Catal. 143 (1993) 286. [22] G. Mestl, H. Knözinger, J.H. Lunsford, Ber. Bunsenges. Phys. Chem. 97 (1993) 319. [23] J.H. Lunsford, X. Yang, K. Haller, J. Laane, G. Mestl, H. Knözinger, J. Phys. Chem. 97 (1993) 13810. [24] G.J. Hutchings, M.S. Scurrell, J.R. Woodhous, Catal. Today 4 (1989) 255.

15

[25] C.T. Au, X.P. Zhou, H.L. Wan, Catal. Lett. 40 (1996) 101. [26] H.H. Eysel, S. Thym, Z. Anorg. Allg. Chem. 411 (1975) 97. [27] J.S. Valentine, Chem. Rev. 73 (1973) 237. [28] C. Li, K. Domen, K. Manuya, T. Onishi, J. Am. Chem. Soc. 111 (1989) 356. [29] A. Metcalfe, S. Udesgankar, J. Chem. Soc. Faraday Trans. I 76 (1980) 630. [30] J.L. Gland, B.A. Sexton, G.B. Fisher, Surf. Sci. 95 (1980) 587. [31] M. Che, J.E. Tench, Adv. Catal. 33 (1983) 1. [32] A. Zecchina, G. Spoto, S. Coluccia, J. Mol. Catal. 14 (1982) 351. [33] L. Ji, J.S. Liu, J. Chem. Soc. Chem. Commun., 1996, 1203. [34] R. Burch, E.M. Crabb, G.D. Squire, S.C. Tsang, Catal. Lett. 2 (1989) 249. [35] C. Shi, M.P. Rosynek, J.H. Lunsford, J. Phys. Chem. 98 (1994) 8371. [36] Y. Erarslanoglu, I. Onal, T. Dogu, S. Senkan, Appl. Catal. A 145 (1996) 75. [37] A.R. West, Solid State Chemistry and its Applications, Courier International, Great Britain, 1990. [38] K. Aika, J.H. Lunsford, J. Phys. Chem. 81 (1977) 1393. [39] K. Aika, J.H. Lunsford, J. Phys. Chem. 82 (1978) 1794. [40] Y. Takita, J.H. Lunsford, J. Phys. Chem. 83 (1979) 683. [41] M. Iwamoto, J.H. Lunsford, J. Phys. Chem. 84 (1980) 1710. [42] Y. Takita, M. Iwamoto, J.H. Lunsford, J. Phys. Chem. 84 (1980) 3079. [43] A. Bielañski, J. Haber, Oxygen in Catalysis, Marcel Dekker, New York, 1991. [44] J.X. Wang, J.H. Lunsford, J. Phys. Chem. 90 (1986) 5883. [45] C.T. Au, X.P. Zhou, J. Chem. Soc. Faraday Trans. 92 (10) (1996) 1793.