Kinetics and mechanism of the oxidative ... - Springer Link

101

Topics in Catalysis Vol. 46, Nos. 1–2, September 2007 ( 2007) DOI: 10.1007/s11244-007-0320-x

Kinetics and mechanism of the oxidative dehydrogenation of ethane over Li/Dy/Mg/O/(Cl) mixed oxide catalysts Stefan Gaaba, Josef Finda,b, Thomas E. Mu¨llera, and Johannes A. Lerchera,* a

Department Chemie, Lehrstuhl II fu¨r Technische Chemie, Technische Universita¨t Mu¨nchen, Lichtenbergstraße 4, 85747 Garching, Germany b Present Address: GWP, Georg-Wimmer-Ring 25, 85604 Zorneding, Germany

The microkinetic reaction network of the oxidative dehydrogenation of ethane to ethene over Li/Dy/Mg/O and Li/Dy/Mg/O/Cl catalysts was investigated. With Li/Dy/Mg/O catalysts, the reaction kinetics is compatible with a heterogeneous-homogeneous radical based reaction mechanism. The formation of ethyl radicals on the surface is concluded to be the rate-determining step. In contrast, the reaction kinetics for Li/Dy/Mg/O/Cl is in line with a purely surface catalyzed reaction mechanism. However, also in this case, alkane activation is rate determining. KEY WORDS: oxidative dehydrogenation; ethane; ethene; mixed oxide.

1. Introduction Alternative and novel routes to ethene and propene (two of the most important base chemicals in modern petrochemical industry) are of high interest, as the well-proven processes of (naphtha) steam cracking and fluid catalytic cracking of vacuum gas oil provide steadily decreasing margins. One of the problems lies in the production of by-products, such as aromatic molecules, which are valuable gasoline blending agents at present, but will be reduced in scale in the target products of the future. Thus, more selective routes to ethene and propene are expected to have a high economic potential. Conceptually, catalytic oxidative dehydrogenation of ethane and propane is an interesting option. The endothermic dehydrogenation reaction is coupled with the exothermic oxidation of hydrogen making the overall process exothermic. Furthermore, the presence of oxygen limits coke formation and extends the catalyst lifetime, which is a major advantage compared to the currently available processes of steam- and fluid catalytic cracking [1–6]. Li2O/Dy2O3/MgO [7–13] mixed oxides are excellently suited for catalyzing the oxidative dehydrogenation of alkanes. However, the catalytic activity of these materials can be considerably improved by incorporation of chloride, when using the proper synthesis procedure. The catalytic activity is directly correlated with the concentration of LiCl in these materials. In this study, the reaction kinetics of the oxidative ethane dehydrogenation over Li/Dy/Mg/O and Li/Dy/Mg/O/Cl catalysts was explored. Although the reaction kinetics is quite similar for the two catalysts, marked differences in the transient * To whom correspondence should be addressed. E-mail: [email protected]

kinetics were observed. The different pathways for the activation of ethane are discussed in detail.

2. Experimental 2.1. Catalyst preparation For preparation of Li/Dy/Mg/O, MgO (6.64 g, 0.165 mol) and Dy2O3 (0.60 g, 1.6 Æ 10)3 mol) were suspended in 90 mL de-ionized water and LiNO3 (2.76 g, 0.040 mol) was added to the suspension. For preparation of Li/Dy/Mg/O/Cl, MgO (6.64 g, 0.165 mol) and Dy2O3 (0.60 g, 1.6.10)3 mol) were suspended in 90 mL de-ionized water. LiNO3 (2.78 g, 0.040 mol) and an aqueous solution (10 mL) of HCl (37%, 1.6 mL, 0.019 mol) and NH4Cl (1.04 g, 0.019 mol) were added. The suspensions were stirred for 2 h at 353 K. The water was then removed under reduced pressure at 353 K and the residues dried for 12 h at 353 K. Subsequently, the precursors were calcined for 12 h at 973 K in a flow of synthetic air (100 NmL/min). The powder was finally sieved to 160 to 224 lm particle size. 2.2. Physicochemical characterization Physicochemical data of the catalysts are summarized in table 1. The specific surface area was determined by N2 adsorption at 77 K (BET method) on a PMI Automated BET Sorptometer. XRD phase analyses were performed at RT on a Philips X’Pert instrument in Bragg-Brentano geometry in the range 2h = 10 to 80 in steps of 0.04/2s. In-situ high temperature XRD profiles were recorded in a Parr chamber HTK 1200 in an atmosphere of synthetic air. In-situ Raman experiments were performed on a Renishaw Serie 1000 1022-5528/07/0900-0101/0 2007 Springer Science+Business Media, LLC

102

S. Gaab et al./Kinetics and mechanism of the oxidative dehydrogenation of ethane

Catalyst

Mg2+ Dy3+ Li+ Cl) BET area XRD phases [wt%] [wt%] [wt%] [wt%] [m2/g] at room temperature

Li/Dy/Mg/O 51.1 Li/Dy/Mg/O/Cl 45.0

6.7 5.9

3.5 3.1

– 15.4

6.7 11.3

MgO; LiDyO2 MgO; Dy2O3

laser Raman micro-spectrometer in the temperature range of 293 to 873 K. The Raman measurements were conducted using the 514 nm line of a 20 mW Ar laser for excitation. 2.3. Pulse experiments Pulse experiments were performed in a quartz reactor at ambient pressure. Approximately 350 mg of catalysts were activated in 50 NmL/min He flow by heating from room temperature to 873 K at 10 K/min. The samples were kept at 873 K for 1 h. The temperature was then adjusted to 773 K, 838 K, or 878 K. Oxygen and/or ethane was introduced in pulses of 1.8Æ10)7 to 1.8Æ10)6 mol. Argon was used as internal standard. The partial pressure of the probe molecules was either 10 mbar or 100 mbar. A quadrupole mass spectrometer (Pfeiffer Vacuum QME 200) was used for product analysis. The total amount of adsorbed or reacted probe molecules was calculated by integrating the corresponding peak areas. 2.4. Catalytic experiments

3.1. Catalyst characterization A chloride free and a chloride containing catalyst was prepared by wet impregnation of a mixture of MgO and Dy2O3 with an aqueous solution of LiNO3 (Li/Dy/Mg/O) or LiNO3/HCl/NH4Cl (Li/Dy/Mg/O/Cl). The phase composition of the two catalysts was determined by XRD. In both catalysts, MgO was the prevailing crystalline phase. Crystalline LiDyO2 was observed for

H2Orelease + CO2 release

CO2 release

LiClmelt

Cl-- free Cl-- containing 420

520

endothermic

3. Results

exothermic

Kinetic measurements were performed in a tubular fixed bed quartz reactor. Prior to each run, the sample was heated in He to 873 K for one hour to remove adsorbed water followed by a stability test at 853 K for at least two hours. Reaction rates were determined under differential conditions with a total flow of 42.8 NmL/min at 853 K. Ethane conversions were less than 15% and oxygen conversions were lower than 25% in all experiments unless noted otherwise. The composition of the products was analyzed with a Hewlett Packard 6890 Series gas chromatograph equipped with FID and TCD detector. A Pora Plot Q and a Molsieve column were used for product separation.

Li/Dy/Mg/O, while Dy2O3 was observed for Li/Dy/Mg/ O/Cl. It is interesting to note, that neither crystalline Li2O or LiCl, nor any other Li phases were observed for Li/Dy/ Mg/O/Cl. These phases were apparently either non crystalline or finely dispersed. Alternatively, LiCl might have been solvated forming a liquid film of a concentrated aqueous solution and, therefore, X-ray amorphous. For the chloride containing catalyst, the formation of a molten LiCl phase has been suggested for temperatures above 573 K [14]. To gain better understanding of phase transformations, which occur during heating to reaction temperature, TG-DSC analyses were performed (figure 1). Both catalysts showed broad endothermic signals between 540 and 730 K, which are attributed to release of water and decomposition of MgCO3 (pure MgCO3 has a decomposition temperature of 623 K [15]). Additionally, a weak signal was observed at 869 K for the Cl) free catalyst, which results from CO2 release by decomposition of Li2CO3. This suggests that MgCO3 and Li2CO3 were formed during storing of the catalyst at ambient conditions. It is most noteworthy that the Cl) containing catalyst showed a strong endothermic signal with a maximum at 877 K, which was not associated with a change in the catalyst mass. This is in perfect agreement with transformation of solid LiCl into a melt (the melting point of pure LiCl is 878 K [15]). It is interesting to note that finely dispersed LiCl starts melting at temperatures of 800 K. Temperature resolved XRD spectra of Li/Dy/Mg/O/ Cl (figure 2) revealed that MgO was present over the entire temperature range (373–873 K). LiCl, which was not observed at room temperature, appeared as a crystalline phase at 373 K and persisted as such up to 573 K. With increasing temperature, the XRD lines of LiCl gradually shifted to higher d-values. The deduced value of the thermal expansion coefficient a of 44.8 · 10)6 K)1 is in good agreement with literature values of 46.0 · 10)6 K)1 for pure LiCl [16,17]. The disappearance of the LiCl diffraction peaks at 573 K is consistent with transformation of LiClsolid into a melt.

DSC signal [a.u.]

Table 1 Chemical composition of the catalysts and data of the physicochemical characterization

620

720

820

920

Temperature [K] Figure 1. TG-DSC analysis of Li/Dy/Mg/O (grey line) and Li/Dy/ Mg/O/Cl (black line).

103


3.2. Oxidative dehydrogenation of ethane – effect of oxygen and ethane partial pressure

MgO

Dy2O3

Dy2O3

LiCl

The catalysts Li/Dy/Mg/O and Li/Dy/Mg/O/Cl were tested in the oxidative dehydrogenation of ethane to ethene. The main side products were CO and CO2 (equation 1 to 3).

LiCl 923 K

373 K

25

30

35

40

45

50

0:5 O2 þ 1 C2 H6 ! 1 C2 H4 þ 1 H2 O

ð1Þ

2:5 O2 þ 1 C2 H6 ! 2 CO þ 3 H2 O

ð2Þ

3:5 O2 þ 1 C2 H6 ! 2 CO2 þ 3 H2 O

ð3Þ

55

2 Theta [°] Figure 2. Temperature resolved XRD diffractograms of Li/Dy/Mg/O/ Cl (temperature steps of 50 K, synthetic air).

To obtain further information about the structure of Li/Dy/Mg/O/Cl, the catalyst was investigated with in situ Raman spectroscopy. Spectra were recorded at room temperature after (a) treatment in nitrogen, (b) followed by oxygen treatment, (c) subsequent reduction in hydrogen and (d) re-oxidation in oxygen. Three sharp bands were observed at 587, 463, and 373 cm)1 (figure 3) and are attributed to the lattice vibrations of Dy2O3. The assignment was confirmed by comparison with the spectrum of pure Dy2O3. No additional bands were observed for this catalyst after oxidation or reduction. This suggests that Dy2O3 was not affected by the pretreatment and that lithium was present mainly as LiCl, which is not Raman active. Note that the formation of LiOCl [18] and other highly active Li species in low concentrations during the oxygen treatment cannot be excluded (vide infra).

3.2.1. Chloride free catalyst (Li/Dy/Mg/O) The rate of product formation on Li/Dy/Mg/O as a function of ethane partial pressure is shown in figure 4. For all products, i.e., ethene, CO and CO2, the rate increased linearly up to 71 mbar ethane partial pressure. The reaction order was 0.94 for ethene formation, while the reaction orders for CO and CO2 formation were lower (i.e., 0.80 and 0.75, respectively). In the range 71–213 mbar ethane partial pressure, the rates of product formation increased further, but the pressure dependence was lower. The corresponding reaction orders for product formation were 0.71, 0.65 and 0.52, respectively; see table 2). The influence of the oxygen partial pressure on the rate of product formation is shown in figure 5. The rate of ethene, CO2 and CO formation increased sharply at low oxygen partial pressures (0 to 15 mbar). It is interesting to note that ethene formation was observed in the absence of oxygen, although the rate was about two orders of magnitude lower than at 15 mbar oxygen partial pressure. This suggests that, in the absence of oxygen, ethene is formed via dehydrogenation or that oxygen from the catalyst was used. Further increase in oxygen partial pressure resulted in a continuous

373

12 12

463

rate [10-1 µ mol/(g•s)]

Intensity [a.u.]

587 d) 1 h O2

c) 1 h H2 b) 1 h O2 a) 1 h N2

99

C2H4

66 CO2

3x

33 3x CO

700

600

500

400

Raman shift

300

200

[cm-1]

Figure 3. In-situ Raman spectra of Li/Dy/Mg/O/Cl recorded at room temperature after treatment for 1 h at 873 K in: (a) nitrogen, (b) followed by oxygen treatment, (c) subsequent reduction in hydrogen, and (d) re-oxidation in oxygen.

00

00

50 50

100 100

150 150

200 200

250 250

p(C2H6) [mbar] Figure 4. Rate of product formation versus ethane partial pressure over Li/Dy/Mg/O (p(O2) = 71 mbar; T = 853 K; total flow 42.8 NmL/min; mcat = 0.42 g).

104


increase in the rates. The highest apparent reaction order in oxygen partial pressure (0.50) was observed for CO2 formation (table 2). For ethene and CO formation the reaction orders had approximately half of the value found for CO2 (0.25 and 0.27, respectively).

3.3. Kinetic studies under non-steady state conditions 3.3.1. Discontinuous admission of oxygen in pulses In order to investigate, how oxygen interacts with the catalyst, oxygen was pulsed on pre-activated Li/Dy/Mg/ O and Li/Dy/Mg/O/Cl. Time resolved MS signals during oxygen pulsing for the chloride free catalyst are

rate [10-1 µ mol/(g•s)]

C2H4

66

44 CO2 3 x

22 CO

3x

00

0 0

25 25

50 50

75 75

100 100

p(O2) [mbar] Figure 5. Rate of product formation versus oxygen partial pressure over Li/Dy/Mg/O (p(C2H6) = 71 mbar; T = 853 K; total flow 42.8 NmL/min; mcat = 0.42 g).

12 12

rate [µmol/(g•s)]

3.2.2. Chloride containing catalyst (Li/Dy/Mg/O/Cl) The rate of product formation on Li/Dy/Mg/O/Cl was one order of magnitude higher than on Li/Dy/Mg/ O. The effect of ethane partial pressure is shown in figure 6. The rate of ethene and CO2 formation increased with the ethane partial pressure. However, the increase was less pronounced at higher partial pressures of ethane. The rate of ethene formation was about 20 and 10 times higher than the rate of CO and CO2 formation, respectively. The reaction orders determined for ethene, CO and CO2 formation were 0.63, 0.58 and 0.53, respectively (table 2). Note that the apparent reaction orders in ethane were similar for the chloride free and chloride containing catalyst at ethane partial pressures above 106 mbar, while significantly higher reaction orders were observed for the chloride free catalyst at low ethane partial pressures. The dependence of the rate of ethene, CO2 and CO formation on the oxygen partial pressure is depicted in figure 7. Similar to the chloride free catalyst, the rate increased sharply at low oxygen partial pressures. Interestingly, the rate of ethene formation in the absence of oxygen was about 70 times higher than the corresponding rate for the chloride free catalyst. This shows that Li/Dy/Mg/O/Cl also catalyses the direct dehydrogenation of ethene. With increasing oxygen partial pressure the rate of product formation increased moderately ( ‡ 15 mbar oxygen partial pressure). It is interesting to note that the apparent reaction order for ethene and CO formation was similar for the two catalysts, while the reaction order for CO2 formation was about 2 times lower for Li/Dy/Mg/O/Cl, than for the chloride free catalyst.

88

99

C2H4

66

CO2

33

3x 3x

CO

00

00

50 50

100 100

150 150

200 200

250 250

p(C2H6) [mbar] Figure 6. Rate of product formation versus ethane partial pressure over Li/Dy/Mg/O/Cl (p(O2) = 71 mbar; T = 853 K; total flow 42.8 NmL/min; mcat = 0.05 g).

shown in figure 8. At 773 K, the oxygen uptake was insignificant. At higher temperatures (838 K and 878 K), the maximum intensity of the oxygen signal was significantly reduced, the peaks were broadened compared to the internal standard (Ar) and the baseline was raised with temperature. This clearly indicates that oxygen adsorbs reversibly on the catalyst and that oxygen desorption was not complete in the time period between two consecutive pulses.

Table 2 Reaction orders for product formation with respect to ethane and oxygen for Li/Dy/Mg/O and Li/Dy/Mg/O/Cl in the range 14 mbar £ p(C2H6) £ 213 mbar and 15 £ p(O2) £ 90 mbar Li/Dy/Mg/O p(C2H6) £ 71 mbar r(C2H4) a [C2H6] 0.94 [O2] 0.25 r(CO) a [C2H6] 0.80 [O2] 0.27 r(CO2) a [C2H6] 0.75 [O2] 0.50

Li/Dy/Mg/O/Cl p(C2H6) ‡ 106 mbar r(C2H4) a [C2H6] 0.71 [O2] 0.25 r(CO) a [C2H6] 0.65 [O2] 0.27 r(CO2) a [C2H6] 0.52 [O2] 0.50

r(C2H4) a [C2H6] 0.63 [O2] 0.25 r(CO) a [C2H6] 0.58 [O2] 0.35 r(CO2) a [C2H6] 0.53 [O2] 0.23

S. Gaab et al./Kinetics and mechanism of the oxidative dehydrogenation of ethane 88

rate [ µ mol/(g•s)]

C2H4

66

44

CO2 3 x

22

CO

00

0 0

25 25

50 50

75 75

3x

100 100

p(O2) [mbar] Figure 7. Rate of product formation versus oxygen partial pressure over Li/Dy/Mg/O/Cl (p(C2H6) = 71 mbar; T = 853 K; total flow 42.8 NmL/min; mcat = 0.05 g).

105

3.3.2. Joint admission of ethane & oxygen pulses The MS signals for reactants and products during simultaneous pulsing of oxygen and ethane on the chloride containing catalyst are shown in figure 10. Ethene and CO2 were the main products in this experiment. It is interesting to note that oxygen had almost the same residence time as the Argon standard, whereas ethane and ethene were observed later (figure 11). This is attributed to adsorption of the hydrocarbons on the catalyst surface. Note that ethane and ethene were detected at the same time. This indicates that ethene desorption from the catalyst surface occurred immediately after ethene formation. The highest shift in time (relative to the Argon residence time standard) and, thus, strongest interaction was found for CO2. The CO2 signals were also broadened in comparison to all other components.

4. Discussion In contrast to the chloride free sample, it is interesting to note, that the amount of oxygen adsorbed on Li/Dy/Mg/O/Cl increased with temperature (figure 9). Approximately 0.3 and 1.4 lmol/gcat of oxygen were consumed at 838 K and 878 K, respectively. An increase in the oxygen baseline and prolonged residence time were not observed. To test, if oxygen was reversibly or irreversibly adsorbed, oxygen was pulsed on oxygen-saturated catalyst. Under these circumstances no further oxygen uptake was observed (not shown). This suggests that the catalyst was partly oxidized by molecular oxygen.

4.1. Chloride free catalyst (Li/Dy/Mg/O) It has been shown previously that the conversion of ethane on Li-doped magnesia proceeds via a heterogeneous-homogeneous reaction mechanism, in which the alkyl radicals are formed at the catalyst surface [19–24]. A detailed kinetic study on the oxidative dehydrogenation of propane on Li/Dy/Mg mixed oxides was performed by Leveles et al. [25,26]. Propane was shown to take part in the rate-determining step, i.e., the formation of propyl radicals on the catalyst surface. The propyl radicals react in the gas phase with

Figure 8. MS signal intensity during oxygen pulsing (1.8 Æ 10)7 mol/pulse) on pre-treated Li/Dy/Mg/O catalyst at 773, 838 and 878 K (pre-treatment in He for 1 h at 873 K). The magnification for the O2 signal is 1 : 1.6 : 2.4, respectively.


I n tensit y o f m /z = 32 [ a .u. ]

106

∆tt = 2 min

878 K 12 10 838 K 8 6 773 K

4 2

0

5

10

15

20

25

30

Time [min] Figure 9. MS signal intensity during oxygen pulsing (1.8 Æ 10)7 mol/pulse) of pretreated Li/Dy/Mg/O/Cl catalyst at 773, 838 and 878 K (pretreatment in He for 1 h at 873 K).

5 Ar (m/z 40)

Intensity [a.u.]

4 CO2 (m/z 44)

3 O2 (m/z 32)

2 C2H4 (m/z 28)

1 C2H6 (m/z 30)

0

10

20

30 Time [min]

40

50

Figure 10. Intensity of the MS signal during simultaneous pulsing of oxygen and ethane on Li/Dy/Mg/O/Cl (1.8 Æ 10)6 mol/pulse; p(O2) and p(C2H6) 100 mbar, 878 K).

4.2. Ethane activation The active sites for ethane activation are proposed to be [Li+O)], which are generated by the substitution of Mg2+ by Li+ within the MgO. Similar sites are also active for the oxidative coupling of methane [27]. In the rate-determining step, the [Li+O)] centers abstract hydrogen homolytically from ethane, forming [Li+OH)] and ethyl radicals according to equation 4. It is speculated that ethane is adsorbed in a weak precursor state on the surface prior to this reaction. Such a weakly

adsorbed precursor would be perfectly compatible with the observed reaction order of 0.94 in ethane (table 2). Note that the formation of surface OH) groups is supported by in situ DSC and DRIFTS studies on Li2O/ MgO catalysts under the conditions of methane oxida-

∆tt

∆t´ t´

Ar O2 C2H6

1

C2H4

Intensity [a.u.]

oxygen. Water and propene are formed in this process. For the present case, let us speculate that the oxidative dehydrogenation of ethane proceeds along this reaction mechanism. The apparent reaction order of 0.94 for ethene formation with respect to ethane for the chloride free catalyst is in line with the postulate that ethane activation is rate-determining. The decrease in reaction order in ethane at higher ethane partial pressures ( > 71 mbar) (table 2 and figure 4) is consistent with a gradual saturation of the ethane activation sites (Langmuir Hinshelwood type mechanism).

.8

CO2

.6 .4 .2 0 620

640

660

680

700

720

740

Time [sec]

Figure 11. Comparison of the MS signals for reactants and products during the first oxygen and ethane pulse according to figure 9.

107


tive coupling [28,29]. The sharp increase in the rates, as the oxygen partial pressure is increased to 15 mbar (figure 5), suggests that there is no pathway to regenerate the active [Li+O)] sites from surface OH) groups in the absence of oxygen [26,30]. Ethyl radical formation (r.d.s.): C2 H6 þ ½Liþ O s ! C2 H5 þ ½Liþ OH s

ð4Þ

which are able to regenerate two more active sites by hydrogen abstraction from [Li+OH)]. Thereby water is formed (equation 8). The overall equation shows that one oxygen molecule is able to regenerate four active sites (equation 5 to 8). If we assume that the reaction steps 5 to 8 are in equilibrium, the equilibrium constant Kp can be expressed by equation 9. Thus, first order dependence in ethane partial pressure and an order of 0.25 in oxygen partial pressure is expected. This is in perfect line with our observations (table 2).

4.3. Gas phase reactions Once the ethyl radical is formed, it desorbs from the catalyst surface into the gas phase. A second hydrogen atom is abstracted by oxygen in the gas phase to give a hydroperoxy radical and ethene (figure 12). Note that the hydroperoxy radical can remove a hydrogen atom from another ethane molecule to form an ethyl radical and hydrogen peroxide. The hydrogen peroxide might decompose to oxygen and water on the walls of the reactor and the catalyst surface. However, depending on the reaction conditions, the hydrogen peroxide can also be cleaved homolytically into two hydroxyl radicals, which can initiate a chain mechanism via reaction with ethane molecules (not shown). A relatively high partial pressure in ethane is necessary to initiate chain reactions. 4.4. Regeneration of the active site After the initiation by hydrogen abstraction according to equation 4, the catalytic active site has to be regenerated. For the oxidative coupling of methane on Li2O/ MgO catalysts dehydroxylation of the [Li+OH)] species was proposed [27]. However, this step requires the energetically unfavorable removal of lattice oxygen and might be operative only at high temperatures [973 K in ref. 31]. Under the reaction conditions used in this study, regeneration of the active site is speculated to take place according to the mechanism proposed by Sinev et al. [32,33]. In the first step one active site is regenerated by the reaction of [Li+OH)] with oxygen (equation 5). The peroxy radical formed can react with further [Li+OH)] to form [Li+O)] and hydrogen peroxide (equation 6). The decomposition of the hydrogen peroxide results in the formation of two hydroxyl radicals (equation 7), C2H4

O2

Initiated by catalytic reaction

C2H5•

Gas-phase

HO2•

Regeneration of the active site: ½Liþ OH s þ O2 ! ½Liþ O s þ HO2

ð5Þ

½Liþ OH s þ HO2 ! ½Liþ O s þ H2 O2

ð6Þ

H2 O2 ! 2HO

ð7Þ

2½Liþ OH s þ 2HO ! 2½Liþ O s þ 2H2 O

ð8Þ

4½Liþ OH s þ O2 ! 4½Liþ O s þ 2H2 O

ð5 8Þ

4 4 Kp ¼ Liþ O ½H2 O2 = Liþ OH ½O2 ð9Þ 1 ) Liþ O ¼ Kp Liþ OH ½O2 0:25 =½H2 O0:5 The overall process (equation 12) is described by the surface reactions 4 and 5–8 (equation 10) and the subsequent processes in the gas phase (equation 11). Note that the reaction order is determined by the rate determining step (surface reaction) and equilibria, which determine the concentrations in the rate determining step. Overall process: 4 C2 H6 þ O2 ! 4 C2 H5 þ 2 H2 O

ð10Þ

4 C2 H5 þ O2 ! 4 C2 H4 þ 2 H2 O

ð11Þ

4 C2 H6 þ 2 O2 ! 4 C2 H4 þ 4 H2 O

ð12Þ

4.5. Chloride containing catalyst (Li/Dy/Mg/O/Cl) The incorporation of chloride dramatically improved the catalytic performance of Li/Dy/Mg/O catalysts . The presence of molten LiCl is believed to open a new pathway for the ethane activation as will be discussed next. Note that there are similarities between the chloride free and the chloride containing Li/Dy/Mg mixed oxides with respect to the reaction orders for ethene formation, although the transient kinetics is totally different. 4.6. Formation of the active site

H2O + 0.5 O2

H2O2

C2H6

Figure 12. Gas phase reaction mechanism proposed.

The active site is proposed to be [LiOCl], which is formed in the LiCl melt by oxidation of LiCl with

108


dissolved oxygen according to equation 13. LiOCl is stable with respect to formation from the elements (DfG ) 90.7 kJ/mol at RT and ) 263.90 kJ/mol at 900 K, gas phase data [18]). However, LiOCl is not stable with respect to decomposition to LiCl and oxygen (DrG 137.22 kJ/mol at 900 K). The concentration of LiOCl in the molten LiCl phase is, therefore, low. As the decomposition is exothermal, the concentration of LiOCl increases with temperature. Formation of the active site: LiCl þ 1=2O2 ! ½Liþ OCl

ð13Þ

4.7. Proposed reaction mechanism Based on our observations a simplified reaction mechanism is proposed, which qualitatively explains the behavior of Li/Dy/Mg/O/Cl (equation 14 to 21). The mechanism distinguishes between three types of molecules, i.e., molecules in the gas-phase, molecules adsorbed on the surface of the melt (symbolized by h) and molecules dissolved in the melt (symbolized by brackets). Ethene formation by reaction of ethane with surface hypochlorite is assumed to be rate determining. k1 pO2 ! ½O2 k1 k2

½O2 ! 2½O k2

k3

½LiOCl ½O þ ½LiCl ! k3

K1 ¼ ½O2 =pO2

ð14Þ

K2 ¼ ½O2 =½O2

ð15Þ

K3 ¼ ½LiOCl=½O ½LiCl ð16Þ

k4

H0 þ ½LiOCl ! HLiOCl k4

K4 ¼ HLiOCl =½LiOCl H0 ð17Þ

k5

pC2 H6 þ H0 ! HC2 H6

K5 ¼ HC2 H6 =H0 pc2 H6 ð18Þ

k5

k6

HC2 H6 þ HLiOCl ! HLiOH þ HHCl þ pC2 H4 r:d:s

ð19Þ

k7

HLiOH þ HHCl ! HH2 O þ H0 þ LiCl k7

K7 ¼ HH2 O H0 ½LiCl=HLiOH HHCl ð20Þ k8

HH2 O ! pH2 O þ H0 k8

K8 ¼ pH2 O H0 =HH2 O ð21Þ

1 ¼ H0 þ HLiOCl þ HC2 H6 þ HLiOH þ HHCl þ HH2 O ð22Þ

In the first step oxygen is dissolved in the LiCl melt according to Eq. 14. In this respect improved oxygen solubility has been reported for alkali carbonate/ chloride mixtures after melting [34,35]. In the second step oxygen dissociates in the melt thereby generating two oxygen atoms (equation 15). The oxygen atoms react with molten LiCl to [LiOCl] (equation 16). The reaction is believed to take place at the interface of the support and the LiCl melt, since it is known [11,13,36] that the support influences catalyst activity and selectivity even at temperatures above the melting point of LiCl. Once the hypochlorite is formed it diffuses to the surface of the melt. The equilibrium of the hypochlorite concentration in the melt and its surface concentration is expressed by equation 17 (free surface places are symbolized by h 0). This equilibrium may account for the undetectable ethane uptake on the oxygen-saturated catalyst. However, the data of figure 11 provide evidence that ethane is adsorbed. Therefore, it is concluded that hydrocarbon activation occurs on the surface of molten LiCl. Ethane adsorbed at the surface is in equilibrium with ethane in the gas phase (equation 18). During reactant pulsing on Li/Dy/Mg/O/Cl, the shift in residence time was equal for ethane and ethene (figure 11). This suggests that ethene is immediately desorbed from the catalyst surface after its formation. Ethene is formed via the reaction of lithium hypochlorite and ethane at the surface of the molten LiCl phase (equation 19). Thereby LiOH and HCl are formed, which react to form adsorbed water, LiCl and an unoccupied adsorption site (equation 20). Desorption of water from the catalyst surface is described by equation 21. The active site is regenerated by reaction of LiCl with oxygen according to steps 14 to 16. As already mentioned, the formation of ethene is postulated to be the rate-determining step within the reaction sequence. According to equaion 23, the rate of ethene formation is dependent on the surface concentration of LiOCl and the concentration of adsorbed ethane. The surface coverage of ethane (hC2H6) and LiOCl (hLiOCl) is determined by equation 18 and 14 to 17, respectively. The concentration of empty surface sites can be derived from equation 20 to 22. As the partial pressure of water in the gas phase was very small, adsorption of water can be neglected. The overall reaction rate for ethene formation for the proposed reaction mechanism is expressed by equation 25. A reaction order of 0.5 in oxygen and 1 in ethane is derived. The lower reaction order in oxygen observed indicates that the sites for LiOCl adsorption are more saturated than the sites for ethane adsorption, i.e. Kb K5. rðC2 H4 Þ ¼ k6 HC2 H6 HLiOCl ðr.d.s.Þ

ð23Þ


One might ask, how the heterogeneous reaction mechanism can explain the increase in ethene selectivity with temperature, since ethene, CO and CO2 are primary products on Li/Dy/Mg/O/Cl. Concerning this question, one must take into account, that the LiCl phase transforms into a melt. Spreading of the chloride phase was shown to start below the melting point of pure LiCl [14]. Therefore, we suggest that non-selective sites are covered by or – more likely – dissolved in the molten LiCl phase, thereby leading to increased ethene selectivity.

r(C2H4) calculated [µmol/g]

12 y = 1.0053x R2 = 0.9826

10

109

8 6 4 2 0 0

2

4

6

8

10

12

5. Conclusions

r(C2H4) measured [µmol/g] Figure 13. Parity plot for the rate of ethene formation (model according to equation 22).

rðC2 H4 Þ k6 Ka p0:5 O2 pC2 H6 ¼ 2 1 0:5 1 þ Kb p0:5 O2 þ K5 pC2 H6 þ Kc pH2 O þK8 pH2 O ð24Þ with 0:5 Ka ¼K0:5 1 K2 K30 K4 K5 0:5 Kb ¼K0:5 1 K2 K30 K4

Kc ¼2 K07 K0:5 8 K30 ¼K3 ½LiCl ½LiCl0:5 K07 ¼K0:5 7 rðC2 H4 Þ ¼

k6 Ka p0:5 O2 pC2 H6 1 þ Kb p0:5 O2 þ K5 pC2 H6

2

ð25Þ

Calculated and measured rates are in perfect agreement (figure 13), which supports the heterogeneous reaction mechanism presented here. Note that the apparent activation energy for ethene formation was ca. 240 kJ/ mol. This is in line with the activation energy determined for the kinetic controlled regime at low temperatures for catalysts having high chloride loadings [14]. So far, we have dealt with a heterogeneous-homogeneous and a purely heterogeneous reaction mechanism for the oxidative ethane dehydrogenation on Li/Dy/Mg/O and Li/Dy/Mg/O/Cl catalysts, respectively. However, the reaction orders for the rate of ethene formation in respect of ethane and oxygen were similar for Li/Dy/Mg/O and Li/Dy/Mg/O/Cl (table 2). On this point one might tend to believe that the reaction mechanism for hydrocarbon conversion is the same for both catalysts, and most of the researchers believe that a heterogeneous-homogeneous based reaction mechanism is operative ([24] and references therein). However, it has clearly been demonstrated that the interaction of oxygen with the catalyst (figure 8 and 9) differs remarkably on the two samples.

Li/Dy/Mg/O and Li/Dy/Mg/O/Cl are effective catalysts for the oxidative dehydrogenation of ethane. The rate of reaction was ten times higher for the chlorine containing catalyst than for the chloride free catalyst. Although the observed reaction orders in oxygen and ethane were similar for both catalysts, fundamental differences in the mechanism of ethane activation exist, as indicated by transient kinetic experiments. Oxygen was reversibly adsorbed on the surface of Li/Dy/Mg/O. The active site is postulated to be [Li+O)], where homolytic abstraction of hydrogen from ethane provides [Li+OH)] and ethyl radicals. The ethyl radicals react in the gas phase with oxygen to selectively form ethene and water. In contrast, a molten LiCl phase was formed on the Li/Dy/Mg/O/Cl catalyst under reaction conditions. We speculate that [LiOCl] is formed at the interphase of melt and support. The [OCl)] ions diffuse to the catalyst surface, where they react with adsorbed ethane to ethene. For both pathways, ethane activation is rate determining. However, the reaction kinetics for Li/Dy/ Mg/O is compatible with a heterogeneous-homogeneous reaction mechanism, whereas the reaction kinetics for Li/Dy/Mg/O/Cl is in general agreement with a purely heterogeneous reaction mechanism.

Acknowledgments The author is grateful to Dipl.-Ing. X. Hecht and Dipl.-Ing. M. Neukamm for their help in GC chromatography and physicochemical analyses. Discussions on this topic within the framework of IDECAT, WP 5 are gratefully acknowledged. The project has been partially supported by a grant from the ’’Verband der Chemischen Industrie’’.

References [1] T. Blasco and J.M. Lopez Nieto, Appl. Catal. A: Gen. 157 (1997) 117. [2] R. Burch and E.M. Crabb, Appl. Catal. A: Gen. 97 (1993) 49. [3] M.A. Banãres, Catal. Today 51 (1999) 319. [4] H.H. Kung and M.C. Kung, Appl. Catal. A: Gen. 157 (1997) 105. [5] E.A. Mamedov and V. Corte´s Corberań, Appl. Catal. A: Gen. 127 (1995) 1.

110


[6] K. Ruth, R. Burch and R. Kieffer, J. Catal. 175 (1998) 27. [7] S.J. Conway, D.J. Wang and J.H. Lunsford, Appl. Catal. A: Gen. 79 (1991) L1. [8] S. Fuchs, L. Leveles, K. Seshan, L. Lefferts, A. Lemonidou and J.A. Lercher, Top. Catal. 15 (2001) 169. [9] M.V. Landau, M.L. Kaliya, A. Gutman, L.O. Kogan, M. Herskowitz and P.F. van den Oosterkamp, Stud. Surf. Sci. Catal. 110 (1997) 315. [10] M.V. Landau, M.L. Kaliya, M. Herskowitz, P.F. van den Oosterkamp and P.S.G. Bocque´, CHEMTECH 26 (1996) 24. [11] M. Herskowitz, M.V. Landau and M.L. Kaliya, German Patent DE 19502747C1, 1997. [12] L. Leveles, Oxidative dehydrogenation of lower alkanes to olefins, Print Partners Ipskamp, Enschede, 2002. [13] L. Leveles, S. Fuchs, K. Seshan, J.A. Lercher and L. Lefferts, Appl. Catal. A: Gen. 227 (2002) 287. [14] S. Gaab, M. Machli, J. Find, R.K. Grasselli and J.A. Lercher, Top. Catal. 23 (2003) 151. [15] R.C. Weast, M.J. Astle and W.H. Beyer, Handbook of Chemistry and Physics, 66 ed., CRC Press, Florida, 1986. [16] A. Levins, M. Straumanis and K. Karlsons, Zeitschr. Phys. Chem. B40 (1938) 146. [17] M. Straumanis, A. Levins and K. Karlsons, Zeitschr. Anorg. Allg. Chem. 238 (1938) 175. [18] Data on the thermodynamic stability of LiOCl are found in NIST Standard Reference Database 69, June 2005 Release: NIST Chemistry WebBook. [19] S.J. Conway and J.H. Lunsford, J. Catal. 131 (1991) 513. [20] J.A. Roos, S.J. Korf, R.H.J. Veehof, J.G. van Ommen and J.R.H. Ross, Catal. Today 4 (1989) 441.

[21] J.A. Roos, S.J. Korf, R.H.J. Veehof, J.G. van Ommen and J.R.H. Ross, Appl. Catal. A: Gen. 52 (1989) 147. [22] E. Morales and J.H. Lunsford, J. Catal. 118 (1989) 255. [23] Y. Ng Lee, F. Spina, E. Martinez, J.V. Folgado and V. Corte´s Corberań, Stud. Surf. Sci. Catal. 110 (1997) 747. [24] F. Cavani and F. Trifiro`, Catal. Today 24 (1995) 307. [25] L. Leveles, K. Seshan, J.A. Lercher and L. Lefferts, J. Catal. 218 (2003) 307. [26] L. Leveles, K. Seshan, J.A. Lercher and L. Lefferts, J. Catal. 218 (2003) 296. [27] T. Ito and J.-X. Wang, J. Am. Chem. Soc. 107 (1985) 5062. [28] S.C. Bhumkar and L.L. Lobban, Ind. Eng. Chem. Res. 31 (1992) 1856. [29] V.Y. Bychkov, M.Y. Sinev, V.N. Korchak, E.L. Aptekar and O.V. Krylov, Kinet. Catal. 30 (1989) 989. [30] E. Morales and J.H. Lunsford, J. Catal. 118 (1989) 255. [31] V.R. Choudhary, V.H. Rane and S.T. Chaudhari, React. Kinet. Catal. Lett. 63 (1998) 371. [32] M.Y. Sinev, V.Y. Bychkov, V.N. Korchak, E.L. Aptekar and O.V. Krylov, Kinet. Catal. 30 (1989) 1236. [33] M.Y. Sinev and V.Y. Bychkov, Kinet. Catal. 34 (1993) 309. [34] V.A. Volkovich, T.R. Griffiths, D.J. Fray and R. Thied, J. Nucl. Mater. 282 (2000) 152. [35] V.A. Volkovich, T.R. Griffiths, D.J. Fray and M. Fields, J. Chem. Soc., Faraday Trans. 93 (1997) 3819. [36] M.V. Landau, A. Gutman, M. Herskowitz, R. Shuker, Y. Bitton and D. Mogilyansky, J. Mol. Catal. A: Chem. 176 (2001) 127.