Topics in Catalysis

, 54:1266

Li-doped MgO From Different Preparative Routes for the Oxidative Coupling of Methane

Authors

  • S. Arndt
    • Institut für ChemieTechnische Universität Berlin
  • U. Simon
    • Institut für WerkstoffwissenschaftenTechnische Universität Berlin
  • S. Heitz
    • Institut für ChemieTechnische Universität Berlin
  • A. Berthold
    • Institut für WerkstoffwissenschaftenTechnische Universität Berlin
  • B. Beck
    • Institut für ChemieTechnische Universität Berlin
  • O. Görke
    • Institut für WerkstoffwissenschaftenTechnische Universität Berlin
  • J. -D. Epping
    • Institut für ChemieTechnische Universität Berlin
  • T. Otremba
    • Institut für ChemieTechnische Universität Berlin
  • Y. Aksu
    • Institut für ChemieTechnische Universität Berlin
  • E. Irran
    • Institut für ChemieTechnische Universität Berlin
  • G. Laugel
    • Institut für ChemieTechnische Universität Berlin
  • M. Driess
    • Institut für ChemieTechnische Universität Berlin
  • H. Schubert
    • Institut für WerkstoffwissenschaftenTechnische Universität Berlin
    • Institut für ChemieTechnische Universität Berlin
Original Paper

DOI: 10.1007/s11244-011-9749-z

Cite this article as:
Arndt, S., Simon, U., Heitz, S. et al. Top Catal (2011) 54: 1266. doi:10.1007/s11244-011-9749-z

Abstract

Li-doped MgO was prepared on different preparative routes and with different loadings. The catalytic activity was found to decay for all catalysts for 40 h time on stream. A detailed structural analysis of 0.5 wt% Li-doped MgO showed heavy losses of Li, reduced surface area and grain growth. A correlation between these factors and the deactivation could not be found. The reaction temperature and the flow rate were found to be the main deactivation parameters.

Keywords

Li/MgO[Li+O]Oxidative coupling of methaneOCMWet impregnationPrecipitationSingle source precursorsMixed millingDeactivation

Abbreviations

AAS

Atomic absorption spectroscopy

BET

Brunauer Emmett Teller

ENDOR

Electron nuclear double resonance spectroscopy

EPR

Electron paramagnetic resonance

ESR

Electron spin resonance

eV

Electron volt

FID

Flame ionisation detector

FWHH

Full width at half height

MAS

Magic angle spinning

NMR

Nuclear magnetic resonance spectroscopy

OCM

Oxidative coupling of methane

PFTR

Plug flow tubular reactor

S

Selectivity

SEM

Scanning electron microscopy

TCD

Thermal conductivity detector

TEM

Transmission electron microscopy

TPPM

Two pulse phase modulation

X

Conversion

XRD

X-ray diffraction

1 Introduction

The known resources of natural gas are large [1]. The exact composition depends on the origin, but the main component of natural gas is always methane. However, methane is highly underutilized, due to two problems:
  1. 1.

    Conversion into value added products is difficult, because CH4 is the most stable hydrocarbon and therefore difficult to activate.

     
  2. 2.

    Difficulties in transporting natural gas from the source to the consumer.

     
The oxidative coupling of methane (OCM), shown in Eq. 1, could overcome these problems and therefore, it is a reaction of great industrial interest.
$$ CH_4 + O_2 \rightarrow C_{2}H_{6} \,\hbox {or} \, C_{2}H_{4} + H_{2}O $$
(1)

However, this reaction has still not been put into practical application due to some drawbacks. The main ones are the high temperature at which this process runs and connected problems like the lack of active and selective catalysts and the thermal stability of possible candidates.

Li-doped MgO is a catalyst for this reaction which has been the subject of intensiv research, nevertheless, many aspects remain unclear, e.g., the active center, structure activity relationship and the stability of the catalyst.

Lunsford et al. correlated electron paramagnetic resonance (EPR) signals with the methyl radical formation rate, concluding that a [Li+O] defect is the active center [24]. It has become a widely accepted fact, even so contradictory results exist in the literature. Mirodatos et al. found that a tight interface between Li2CO3 and MgO is necessary for a good catalytic performance but it only occurs after pretreatment at temperatures allowing the liquefaction of Li2CO3 [5, 6]. Later, Goodmann and co-workers provided evidence that F-centers are the active center and that Li is a promoter for the formation of F-centers [710].

The stability of the Li-doped MgO catalysts is often not adressed in the literature and/or large amounts of inert dilutent are used in the feed gas avoiding strong deactivation. However, Kimble and Kolts showed that Li is lost from the catalysts after calcination at 850 °C [11]. Moreover, Mirodatos and co-workers showed that Li-doped MgO suffers from severe deactivation due to sintering and loss of Li [5].

Li-doped MgO was found to deactivate rapidly by Ross et al. [12]. They also showed that, if Li is still available, the catalyst can be regenerated by treatment with CO2 under reaction conditions. The deactivation can be avoided if CO2 is added to the reaction mixture in low concentrations. They furthermore concluded that CO2 temporarily poisons the active site and simultaneously stabilises it against deactivation [12, 13]. The selectivity did not change so much in their long-term experiments, therefore the conclusion was that the nature of the active site has not changed much but the number decreased.

Another important finding was that the use of experimental equipment made of quartz glass is detrimental to the stability of Li-doped MgO [13, 14]. It was shown that the catalyst deactivates due to a loss of Li as the volatile LiOH or as Li2SiO3, due to the fact that many laboratory reactors are made of quartz glass. However, this is not limited to quartz devices, the Li caused problems in reactors made of alonized steel, Al2O3, Al2SiO5 and ZrO2 (stabilized with Ca) [15, 16]. Generally, it seems a problem that currently no material exist which is stable against the highly mobile Li.

Perrichon and Durupty investigated the thermal stability of Li, Na and K deposited on MgO, SiO2, Al2O3 and Cr2O3 [17]. It was found that on SiO2, Al2O3 and Cr2O3 the alkali metals are rather stable, however, on MgO a loss of alkali metal was observed at calcination temperatures higher than 500 °C. This effect increased from Li to K at 800 °C. A reason for this could be that MgO is not able to form stable compounds, unlike SiO2, Al2O3 and Cr2O3 which can form silicates, aluminates and chromates. Supports impregnated with alkali metals always had a lower surface area, and the temperature had no influence on this. That effect was explained with the formation of a conglomerate of alkali metal salt and the support. Furthermore, it was assumed that the alkali metal phase, which was present at the interface between the support particles, underwent transformations, especially fusion, during sintering. Therefore, it induced partial dissolution of the support and thus the growth of larger crystals.

A detailed discussion of the developement and the stability of Li-doped MgO can be found in Chapter 6 of reference [18]. In this reference the morphological aspects of Li-doped MgO and other catalysts and their influence on the OCM are discussed too (Chapter 10 of [18]).

Many different procedures for the synthesis of Li-doped MgO have been reported with different results regarding the catalytic performance, though, a comparison is usually difficult because the applied test conditions vary strongly. However, this is a common problem in research on OCM catalysts [19]. It is well known that the catalytic activity depends on the preparation procedure. Choudhary and co-workers studied the influence of the precursors for Li2O and MgO on the surface and catalytic properties. The applied preparation procedure was wet impregnation [20]. They reported that the catalytic activity and other characteristics of the catalysts strongly depended on the precursors used for the preparation, however, deactivation was not considered in their publication.

Up to our knowledge, no detailed comparison of Li-doped MgO prepared on different routes exists. Furthermore, detailed structural investigations after different times on stream for different catalysts has also not been reported so far.

In this publication, we present the results of a study of Li-doped MgO from different preparative routes and with Li loadings of 0, 0.5, 1, 2, 4, and 8 wt%. The applied preparative routes and the denomination of the catalysts are shown in Table 1. This comparison was performed in order to find a synthetic route that produces the most stable catalyst.
Table 1

The applied preparative routes and the denomination as used in this manuscript

No.

Preparation

Abbreviation

1

Single source precursors

Li@MgO

2

Wet impregnation

Li/MgO

3

Precipitation

Li−MgO

4

Mixed milling

Li+MgO

Since the Li-doped MgO system has been shown to be unstable [5, 1214], time on stream experiments were used to determine the developement of the catalytic activity. After the determination of the catalytic activity and the stability, one Li-loading from each preparation route was choosen and structural analysis at different times on stream were done to observe the changes of the catalysts during the course of reaction. Moreover, the impact factors on the catalyst stability, such as a the flow rate and temperature, were studied with time on stream experiments.

2 Experimental Part

2.1 Catalyst Preparation

In this publication, the loading of Li is calculated from Eq. 2.
$$ \hbox{Li (wt\%)} = \frac{\hbox{Mass\,of\,Li (g)}}{\hbox{Mass of Li (g) + Mass of MgO (g)}} \times 100 \% $$
(2)
The amount of the determined impurities is calculated with Eq.3.
$$ \hbox{Impurity (wt\%)} = \frac{\hbox{Mass\,of\,impurity (g)}}{\hbox{Mass of Li (g) + Mass of MgO (g)}} \times 100 \% $$
(3)

Four different preparation routes have been applied for the synthesis of Li-doped MgO. The routes and their denomination is shown in Table 1, disregarding of their actual Li content.

2.1.1 Li@MgO

The preparation of solid catalysts via the decomposition of single source precursors [21, 22] has recently been developed and already applied to Li-doped ZnO [2325]. This procedure allows a reproducible synthesis of solid materials with controlled stoichiometry.

The Li@MgO is prepared via the thermal decomposition of alkyl–Mg–alkoxides and Li–Mg–alkoxides. In Fig. 1 the reaction is shown schematically. The procedure for MgO and Li-doped MgO has been described in detail, including analytic data, by Heitz et al. [26, 27]. Since the prepared materials were a nano-powder, they were not crushed and sieved.
https://static-content.springer.com/image/art%3A10.1007%2Fs11244-011-9749-z/MediaObjects/11244_2011_9749_Fig1_HTML.gif
Fig. 1

The thermal decomposition of Li–Mg alkoxides cubanes leads to a very pure and reproducible preparable Li-doped MgO

2.1.2 Li/MgO

The preparation of Li/MgO via wet impregnation is a well-known and common procedure, which has also been used by Lunsford et al. [24]. However, it remains unclear if this procedure is suitable to insert Li into the MgO lattice and to produce reasonable amounts of [Li+O], which corresponds to the active center for some researchers. In the vast majority of the literature dealing with Li-doped MgO, wet impregnation has been used as preparation procedure. In favor of this simple route, all other synthetic routes have been neglected, even so they should be more suitable for the preparation of [Li+O].

MgO (Alfa Aesar, 99.99 %) and Li2CO3 (Fluka ≥99.0 %) were mixed in distilled water. The water was evaporated until only a thick paste remained. The samples were dried over night at 140 °C and then calcinated in air at 465 °C for 1 h. The materials were crushed and sieved, to obtain only particles ≤200 μm for catalytic studies.

2.1.3 Li–MgO

Aqueous solutions of Mg(NO3)2 were prepared by dissolving Mg(NO3)2 × 6H2O (p.A., Merck) in distilled H2O. The nitrate solution was slowly added to stirred ammonia solution while keeping the pH value above 11. The gelatinous precipitated magnesium hydroxide was rinsed with distilled H2O and mixed with aqueous LiOH solution (LiOH × H2O p.A., Riedel de Haen) with appropriate Li concentrations in a tubular mixer. Finally, the solution was quick-frozen using liquid N2. Afterwards, it was freeze–dried over 72 h using a freeze–dryer (Gamma 2–20 (Christ)). After calcination at 900 °C for 1 h in MgO crucibles, Li–MgO powders were produced. The samples with a Li content of 0 and 0.5 wt% were too fluffy, therefore, they were pressed at 100 bar for 10 min and subsequently crushed and sieved. All samples were sieved and only the fraction ≤200 μm was used for testing.

2.1.4 Li+MgO

We also chose mixed milling to prepare Li-doped MgO, as this route could be applied to produce this catalysts fast and easy in larger amounts, which is of interest with respect to an industrial application.

LiNO3 (Fluka, purity ≥99 %) and MgO (Sigma Aldrich, 325 mesh, purity 99+ %) were milled in a centrifugal ball mill (Retsch S 100) for 1 h at 400 rounds per min with alternating directions. The grinding jar and the grinding balls were made of stainless stell, material number 1.4034. The data sheet names the main components as Fe 82.925, Cr 14.500, Mn 1.000 and Si 1.000%. Afterwards, the prepared samples were calcinated at 400 °C for 3 h. The materials were crushed and sieved, only particles ≤200 μm were used.

2.2 Catalytic Tests

The catalytic experiments were carried out in a packed-bed, U-shaped, tubular reactor made of quartz glass. The outer diameter was 8 mm and the inner diameter 6 mm. For each catalytic run, 100 mg catalyst were diluted with approximately 1.5 mL quartz sand (quartz sand: purchased from Merck, it has already been washed with HCl and calcined, ca. 60% of the particles have the size 0.2–0.8 mm) for proper heat transfer. The reactor behaves like a plug flow tubular reactor (PFTR). The length of the catalytic bed was ca. 50–55 mm. Below and above the catalyst bed, a small amount of pure quartz sand was put to ensure proper heat transfer. The particle size of the catalyst was below 200 μm in each experiment to exclude internal mass transfer effects.

The gas was fed to the reactor using mass flow controllers of Brooks 5850 TR series. Methane and synthetic air, as oxygen source, were fed into the reactor with a flow rate of 60 mL/min and a feed gas composition of CH4:O2:N2 = 4:1:4. Before entering the reactor, the reactants passed a pre-heater, heated to 180 °C. The reactor was heated with an electric furnace (LK 1000-20-310-1 from HTM Reetz Berlin); isothermal conditions were insured. Under reactant gas flow, the reactor was heated to 750 °C with a heating rate of 20 K/min. 19 min after reaching the 750 °C, the first data point was measured. After reaching the reactor temperature of 750 °C, each catalyst was tested for at least 40 h time on stream, counting from the time when the reactor temperature was reached. The same test protocol was applied for every experiment.

To investigate the dependence on the residence time, the flow rates were set to 30, 60, 120 and 180 mL/min, respectively, and the catalytic activity was recorded for at least 18 h time on stream under otherwise identical experimental conditions. For the investigation of the temperature dependence, the experiments with a reaction temperature of 650, 700, 750 and 800 °C and a flow rate of 60 mL/min were conducted, with otherwise identical experimental conditions.

The analysis was performed with a gas chromatograph Agilent 7890 A, equipped with a flame ionisation detector (FID), a thermal conductivity detector (TCD), a HP-PLOT/Q and a HP Molsieve column. The analysed compounds were O2, N2, CO, CO2 via TCD, CH4, C2H4 and C2H6 via FID. N2 was used as internal standard. The reproducibility of conversion (X) and selectivity (S) is sufficient. The conversion and selectivity is calculated with a mass balance based on the inlet and outlet concentration of reactants and products, see Eqs. 4 and 5. The carbon balance was always well above 95%.
$$ \hbox{X} = \frac{\sum(\hbox{Reaction products})}{\sum(\hbox{Reaction products}) + \hbox{Unconverted reactant}} $$
(4)
$$ S = \frac{\hbox{Product}}{\sum(\hbox{Reaction products})} $$
(5)

The desired reaction products of the oxidative coupling of methane are C2H4 and C2H6. The selectivity of these two products is discussed as a sum, the C2 selectivity. Higher reaction products, especially C3H8 and C3H6, have not been detected, although the applied method was suitable for their detection.

In order to determine the degree of thermal conversion of CH4, blank experiments with pure quartz sand, with quartz balls with a particle size of 0.4 ± 0.1 mm (Quarzglas QCS, Maintal, Germany) and with an empty reactor were conducted under the above described reaction conditions. The importance of the homogeneous gas phase reaction in the OCM is significant [28], therefore, the obtained results for CH4 conversions and the according selectivities are shown in detail in Table 2. At 700 °C and below the contribution of the thermal reaction is negligible in the applied experimental set-up. At 750 and 800 °C, a contribution of the thermal reaction is observed. However, since the CH4 conversion was rather small the influence of the thermal reaction has been neglected. The degree of contribution of the gas phase reaction strongly depends on several different factors, such as partial pressures of all compounds, temperature, residence time, free reactor volume et cetera [28]. Therefore, it is not surprising that different CH4 conversions are observed in a reactor which is empty, filled with quartz sand or quartz balls.
Table 2

Contribution of the homogenous gas phase reaction to the oxidative coupling of methane, in reactors with different inert fillings. At 750 °C and below the contribution of the thermal reaction is small and therefore negligible

Reaction conditions

Quartz balls

Quartz sand

Empty reactor

Temperature (°C )

Flow (mL/min)

\(\hbox{X}_{O_2}\)(%)

\(\hbox{X}_{\hbox{CH}_4}\) (%)

\(\hbox{S}_{\hbox{C}_{2}}\) (%)

\(\hbox{X}_{\hbox{O}_2}\) (%)

\(\hbox{X}_{\hbox{CH}_4}\) (%)

\(\hbox{S}_{\hbox{C}_{2}}\) (%)

\(\hbox{X}_{\hbox{O}_2}\) (%)

\(\hbox{X}_{\hbox{CH}_4}\) (%)

\(\hbox{S}_{\hbox{C}_{2}}\) (%)

700

30

2.7

0.1

100.0

700

60

700

90

700

120

700

180

750

30

0.9

0.1

100.0

3.5

0.4

100.0

2.3

0.3

100.0

750

60

1.4

0.2

100.0

0.9

0.1

100.0

750

90

1.1

0.1

100.0

0.5

0.1

100.0

750

120

1.0

0.1

100.0

750

180

0.8

0.1

100.0

800

30

3.5

0.6

100.0

8.6

2.5

52.8

9.0

2.8

54.5

800

60

1.3

0.2

100.0

4.3

0.7

100.0

3.3

0.6

100.0

800

90

0.5

0.1

100.0

3.2

0.5

100.0

1.9

0.3

100.0

800

120

0.5

0.1

100.0

2.5

0.4

100.0

1.2

0.2

100.0

800

180

0.5

0.1

100.0

2.3

0.3

100.0

0.8

0.1

100.0

2.3 Sample Preparation for Structural Analysis

For a structural analysis after the reaction it is necessary to retrieve the catalysts. With the applied quartz sand this is not possible, as its particle size distribution is too large and it overlaps with the particle size of the catalysts. Thus, quartz balls with a particle size of 0.4 ± 0.1 mm (Quarzglas QCS, Maintal, Germany) and a very narrow particle size distribution were used instead of quartz sand as inert dilutent, enabling a separation of catalysts and inert material after the reaction by sieving. The reaction conditions for the sample preparation are identical to those described in Sect. 2.2. For the separation of catalyst and dilutent, a 200 μm sieve was used.

The contribution of the gas phase reaction is different when quartz sand or quartz balls are used, as shown in Table 2. Therefore, it could be that the catalytic behavior and the deactivation process of the reaction with quartz sand and quartz balls is not absolutely identical. However, the general trends should still be observable in a reliable manner.

2.4 Catalyst Characterization

2.4.1 Atomic Absorption Spectroscopy

The Li content of the different samples was quantified via atomic absorption spectroscopy (AAS), using a AAS NovAA 400 G device from Analytik Jena via flame. Furthermore, Fe, Cr, Mn, Ca and Cu were determined as possible impurities.

2.4.2 BET

The specific surface area was determined by a Micromeritics Gemini III 2375 Surface Area Analyzer, using N2 adsorption at −196 °C. Before measuring, the samples were degassed at 300 °C and 0.15 mbar at least for 30 min. The surface areas were calculated by the method of Brunauer, Emmett and Teller (BET).

2.4.3 X-Ray Diffraction

Powder X-ray diffractograms (XRD) were obtained (CuKα1 radiation wavelength 0.154 nm) using a Bruker AXS D8 ADVANCE X-ray diffractometer. The diffractorgramms were analysed with the programm STOE WinXPOW. The lattice parameter were determined with the algorithm of Werner and the particle size with the Scherrer equation.

2.4.4 Solid State NMR

The solid state MAS (magic angle spinning) NMR (nuclear magnetic resonance) measurements were carried out at a Bruker Avance 400 spectrometer operating at 155.5 MHz for 7Li using a Bruker 4 mm double-resonance probehead operating at a MAS spinning rate of 12 kHz. TPPM (two pulse phase modulation) proton decoupling was applied during the acquisition. The 7Li spectra were referenced to a 1 M solution of LiCl in water using solid LiCl as a secondary reference.

2.4.5 Scanning Electron Microscopy

Scanning electron microscopic (SEM) studies were applied to characterize surface morphologies by using a Cross Beam Microscope (ESB 1540, Zeiss, Germany) with integrated energy dispersive X-ray Spectroscopy (Thermo Fisher Scientific, Germany). The specimen were bonded on conducting carbon pads and finally covered with a thin carbon layer via evaporating to avoid charging.

2.4.6 Transmission Electron Microscopy

Transmission electron microscopic (TEM) images have been recorded on a Tecnai G2 20 S-TWIN (operating at 10 keV) with an energy dispersive X-ray spectrometer (EDAX, r-TEM SUTW) located at the ZELMI, Technische Universität Berlin.

3 Results

3.1 General Characterization

The Li-content for all catalysts is shown in Table 3. No Li was detected for pure MgO. For most catalysts there is a discrepancy between the target loading of Li and the actual loading. This is due to a loss during the calcination process, which has already been described in the literature [5, 11]. The Li-loading of 0.04 wt% for the 0.5 wt% Li–MgO can be attributed to the very high calcination temperature of 900 °C. The general trend is, that with higher loadings of Li, the difference between the target and the actual loadings increases. The target loading of Li is used for denomination, to avoid any confusion.
Table 3

Lithium contents in wt% of the differently prepared Li-doped MgO materials

Loading

Li@MgO

Li/MgO

Li–MgO

Li+MgO

0.0

0.00

0.00

0.00

0.00

0.5

0.41

0.49

0.04

0.51

1.0

0.61

0.92

0.84

0.97

2.0

1.33

2.09

1.71

1.69

4.0

2.24

4.04

2.76

2.48

8.0

5.99

7.43

6.15

3.66

The results for the BET experiments for all prepared catalysts are shown in Table 4. The specific surface area depends on the preparation procedure and the applied calcination temperature, albeit the highest BET values are observed for pure MgO. Upon Li-addition, the specific surface area decreases and with increasing loadings of Li, the BET surface area decreases. The strongest reduction of the specific surface area upon Li-addition is observed for Li/MgO and Li–MgO.
Table 4

The table shows the BET surface area in m2/g for all preparation routes and Li-loadings

wt%

Li@MgO

Li/MgO

Li–MgO

Li+MgO

0.0

99.3

120.7

34.4

67.8

0.5

81.7

6.3

27.3

23.2

1.0

73.4

5.4

 ≤1

16.7

2.0

26.3

5.6

 ≤1

20.6

4.0

7.5

3.4

 ≤1

24.0

8.0

 ≤1

5.1

 ≤1

14.5

Li can not be detected via XRD, SEM and TEM. Therefore, these results are not described here. However, the characterization of the catalysts after different times on stream still leads to useful data, which will be discussed in the according paragraphs.

3.2 Time on Stream Experiments

3.2.1 Li@MgO

Figure 2 shows the developement of the CH4 conversion and the C2 selectivity during 40 h time on stream for the Li@MgO catalyst with different loadings of Li. In the first 3–5 h, a strong deactivation is observed. With higher Li loadings, the deactivation rate increases. For 0.5 and 1 wt% Li@MgO, the strong deactivation is followed by a reactivation. For 0, 0.5, and 1 wt% Li@MgO, the CH4 conversion seems to approach the same residual value. The strong scattering of the trajectories for 4 and 8 wt% is due to the small conversion. They are too small for an exact determination of the reaction products and therefore, for an exact calculation conversion and selectivity. The gap at 6.7 h in the trajectory of 8 wt% Li/MgO can be explained that at low conversions the concentration of one of the reaction products (usually CO2) falls below the detection limit of the TCD detector. The selectivity of these two catalysts is therefore not being discussed, because the conversion and the selectivity can not exactly be determined. The C2 selectivity decreases with time on stream and approaches a residual value. However, for 2 wt% Li, this decrease is much slower than for lower Li-loadings. The catalytic performance of 0.5 and 1 wt% is similar. Compared to materials prepared on other synthetic routes, the 0 wt% Li@MgO has a rather high C2 selectivity.
https://static-content.springer.com/image/art%3A10.1007%2Fs11244-011-9749-z/MediaObjects/11244_2011_9749_Fig2_HTML.gif
Fig. 2

CH4 conversion (left) and C2 selectivity (right) of Li@MgO as function of time on stream

3.2.2 Li/MgO

Figure 3 shows the developement of the CH4 conversion and the C2 selectivity during 40 h time on stream for the Li/MgO catalyst with different loadings of Li. A strong and constant decrease in the CH4 conversion can be observed. The deactivation rate increases with higher Li-loadings, 0 wt% Li/MgO exhibits the slowest deactivation. The high initial CH4 conversion falls below 4% after 40 h time on stream. The gap in the trajectory at 7 and 8 h for 4 and 8 wt% Li/MgO, respectively, occurs due to the same reasons like for Li@MgO. The selectivity is therefore not being discussed. There is no significant difference for the C2 selectivity for 0.5, 1, and 2 wt% Li/MgO. However, as already reported by other researchers, the catalysts without Li shows a very low C2 selectivity.
https://static-content.springer.com/image/art%3A10.1007%2Fs11244-011-9749-z/MediaObjects/11244_2011_9749_Fig3_HTML.gif
Fig. 3

CH4 conversion (left) and C2 selectivity (right) of Li/MgO as function of time on stream

3.2.3 Li–MgO

Figure 4 shows the developement of the CH4 conversion and the C2 selectivity during 40 h time on stream for the Li–MgO catalyst with different loadings of Li. All materials with a Li-loading higher than 0.5 wt% drastically deactivate within 5 h time on stream. However, the loss in activity for 0 wt% Li–MgO and 0.5 wt% Li–MgO is rather low compared to other materials. For 0.5 wt% Li–MgO, it might very well be due to the rather low actual Li content compared to the other preparation routes. The gap in the trajectory at 3 and 6.3 h for 4 and 8 wt% Li–MgO, respectively, can be explained by a low conversion and therefore a low concentration of one of the reaction products (usually CO2), which then falls below the detection limit of the TCD detector. The selectivity of these two catalysts is also not being discussed. The C2 selectivity remains constant for 0 wt% MgO and it is constant after an initial deactivation time for 0.5 wt% Li–MgO with time on stream. It is evident, that 0.5 wt% Li–MgO has the higher CH4 conversion as well as the higher C2 selectivity.
https://static-content.springer.com/image/art%3A10.1007%2Fs11244-011-9749-z/MediaObjects/11244_2011_9749_Fig4_HTML.gif
Fig. 4

CH4 conversion (left) and C2 selectivity (right) of Li–MgO as function of time on stream

3.2.4 Li+MgO

Figure 5 shows the developement of the CH4 conversion and the C2 selectivity during 40 h time on stream for the Li+MgO catalyst with different loadings of Li. The catalysts, with higher loadings of Li, deactivate quickly and their residual activity is rather low. The difference between 0.5 wt% Li+MgO (red) and 1 wt% Li+MgO (green) is very distinct. However, the difference between 0 wt% MgO (milled) and the unmilled commercial MgO is remarkable. The unmilled material has a residual CH4 conversion of 6.0% and a C2 selectivity of 22.3%. When the material is milled, the residual CH4 conversion is 15.6% and the C2 selectivity is 29.0%. Moreover, the deactivation of the milled MgO seems to be retarded compared to unmilled commercial MgO. The C2 selectivity decreases for all materials at the beginning of the experiment, however, for some materials the C2 selectivity becomes stable, while for others the selectivity reduction proceeds, i.e., for commercial MgO. The commercial MgO and the 8 wt% Li+MgO catalyst exhibit a very poor C2 selectivity. All other catalysts show a C2 selectivity which is between 30 and 35%.
https://static-content.springer.com/image/art%3A10.1007%2Fs11244-011-9749-z/MediaObjects/11244_2011_9749_Fig5_HTML.gif
Fig. 5

CH4 conversion (left) and C2 selectivity (right) of Li+MgO as function of time on stream

3.2.5 Summary of Time on Stream Experiments

The data for the residual activity for all catalysts is summarized in Table 5. Generally, the activity is rather low, disregarding of the preparation procedure and the Li-loading.
Table 5

Summary of the residual activity (O2 and CH4 conversion, C2 selectivity and the ratios of C2H6 to C2H4 and CO2 to CO after 40 h time on stream for the prepared catalysts. If the conversion was too low and the selectiviy was not discussed, then the according values are replaced with a “-”

Single source precursor

Catalyst

0 wt% Li@MgO

0.5 wt% Li@MgO

1 wt% Li@MgO

2 wt% Li@MgO

4 wt% Li@MgO

8.7 wt% Li@MgO

\(\hbox{X}_{\hbox{O}_{2}}\)

26.0%

24.9%

24.6%

10.7%

2.6%

2.1%

\(\hbox{X}_{\hbox{CH}_{4}}\)

7.0%

6.7%

6.7%

3.2%

0.7%

1.0%

\(\hbox{S}_{\hbox{C}_{2}}\)

28.9%

25.8%

26.1%

31.4%

47.9%

45.8%

\({\hbox{C}_{2}\hbox{H}_{6}}/{\hbox{C}_{2}\hbox{H}_{4}}\)

3.0

3.5

3.3

5.4

8.3

CO2/CO

0.3

0.3

0.3

0.2

Wet impregnation

Catalyst

0 wt% Li/MgO

0.5 wt% Li/MgO

1 wt% Li/MgO

2 wt% Li/MgO

4 wt% Li/MgO

8 wt% Li/MgO

\(\hbox{X}_{\hbox{O}_{2}}\)

13.7%

9.5%

9.5%

9.0%

6.6%

4.1%

\(\hbox{X}_{\hbox{CH}_{4}}\)

3.5%

2.6%

2.2%

2.4%

1.6%

1.0%

\(\hbox{S}_{\hbox{C}_{2}}\)

22.3%

30.5%

32.0%

32.9%

43.8%

59.3%

\({\hbox{C}_{2}\hbox{H}_{6}}/{\hbox{C}_{2}\hbox{H}_{4}}\)

5.2

5.5

6.9

5.8

60.1

9.0

CO2/CO

0.3

0.3

0.4

0.3

0.0

Precipitation

Catalyst

0 wt% Li–MgO

0.5 wt% Li–MgO

1 wt% Li–MgO

2 wt% Li–MgO

4 wt% Li–MgO

8 wt% Li–MgO

\(\hbox{X}_{\hbox{O}_{2}}\)

15.8%

25.5%

1.1%

1.3%

1.7%

2.9%

\(\hbox{X}_{\hbox{CH}_{4}}\)

3.8%

6.3%

0.2%

0.2%

0.2%

0.4%

\(\hbox{S}_{\hbox{C}_{2}}\)

15.0%

22.7%

\({\hbox{C}_{2}\hbox{H}_{6}}/{\hbox{C}_{2}\hbox{H}_{4}}\)

4.6

3.3

7.5

CO2/CO

0.3

0.3

Mixed milling

Catalyst

Unmilled MgO

Milled MgO

0.5 wt% Li+MgO

1 wt% Li+MgO

2 wt% Li+MgO

4 wt% Li+MgO

8 wt% Li+MgO

\(\hbox{X}_{\hbox{O}_{2}}\)

25.4%

69.8%

52.0%

29.2%

15.2%

8.9%

10.2%

\(\hbox{X}_{\hbox{CH}_{4}}\)

6.0%

15.6%

12.8%

7.4%

4.2%

2.9%

2.2%

\(\hbox{S}_{\hbox{C}_{2}}\)

22.3%

29.0%

32.2%

30.1%

30.7%

33.0%

25.1%

\({\hbox{C}_{2}\hbox{H}_{6}}/{\hbox{C}_{2}\hbox{H}_{4}}\)

3.3

1.4

1.6

2.6

4.0

5.5

6.1

CO2/CO

0.7

1.8

1.1

0.6

0.4

0.3

0.8

The data in this table is also shown in a converion selectivity diagram in Fig. 6

Figure 6 is a conversion selectivity diagramm, containing all catalysts with the initial activity (small data points) and the residual activity (large data points). The inital activity shows a strong scattering for the different materials. However, after 40 h time on stream, the activity is significantly reduced. Moreover, the catalytic data for the differently prepared Li-doped MgO catalysts do not differ so strong anymore. Since none of the catalysts reached a stable state, it can be expected that this scattering will be further reduced when the time on stream is extended, disregarding of the preparation route.
https://static-content.springer.com/image/art%3A10.1007%2Fs11244-011-9749-z/MediaObjects/11244_2011_9749_Fig6_HTML.gif
Fig. 6

The initial activity (small data points) and residual activity (large data points) for all catalysts of Table 5 is shown as a conversion selectivity diagram. While the inital activity scatters, the residual activity and the scattering is strongly reduced, irrespective of the Li-loading or of the preparation procedure

3.3 Structural Analysis for Different Times on Stream

From each preparation procedure the samples with a loading of Li of 0.5 wt% were chosen. Samples were taken from the reactor after 1, 3, 7, 14, and 24 h time on stream and structural analysis was performed. The obtained catalytic results, such as O2 conversion, CH4 conversion, C2 selectivity, C2 yield, and the ratios of C2H6 to C2H4 and CO2 to CO of these four materials are shown in detail in Fig. 7 for 24 h time on stream.
https://static-content.springer.com/image/art%3A10.1007%2Fs11244-011-9749-z/MediaObjects/11244_2011_9749_Fig7_HTML.gif
Fig. 7

Comparison of the O2 conversion, CH4 conversion, C2 selectivity, C2 yield, and the ratios of C2H6 to C2H4 and CO2 to CO for the differently prepared 0.5 wt% Li-doped MgO samples

3.3.1 AAS

Figure 8 shows the Li content after different times on stream for all 0.5 wt% Li-doped MgO samples. The most significant reduction is observed within the first hour under reaction conditions, irrespective of the preparation procedure. Afterwards, the further reduction is still signifcant, but small. After approximately 7 h time on stream, the amount of Li in the catalyst is not significantly reduced anymore. Disregarding of the exact initial Li-loading and the preparation procedure, the residual Li-content approaches the same, or at least similar value, which lies between 0.01 and 0.03 wt%.
https://static-content.springer.com/image/art%3A10.1007%2Fs11244-011-9749-z/MediaObjects/11244_2011_9749_Fig8_HTML.gif
Fig. 8

Decrease of the Li content of the differently prepared Li-doped MgO catalysts as a function of time on stream

Moreover, possible impurities such as, Fe, Cr, Mn, Ca and Cu were determined via AAS before the reaction and after 24 h time on stream. The results are shown in Table 6. Considerable amounts of Fe and Ca were found. Cr, Mn and Cu were also determined, though in smaller amounts. For Li/MgO, Mn was found, even so in rather small amounts. Li+MgO contains the highest amount of impurities, which is probably due to the milling. The measured amounts of impurities did not change significantly after the reaction. The strong increase in Ca for Li/MgO is probably an outlier. The amounts of impurities seem to be relatively stable under reaction conditions.
Table 6

Amount of the determined impurities in wt% before and after the reaction for the different 0.5 wt% Li-doped MgO

Catalyst

0 h

24 h

 

Fe

Li@MgO

0.014

0.011

Li/MgO

0.011

0.014

Li–MgO

0.012

0.009

Li+MgO

0.130

0.122

 

Cr

Li@MgO

0.006

0.007

Li/MgO

0.006

0.007

Li–MgO

0.007

0.006

Li+MgO

0.006

0.008

 

Mn

Li@MgO

0.001

0.001

Li/MgO

0.000

0.000

Li–MgO

0.001

0.000

Li+MgO

0.006

0.005

 

Ca

Li@MgO

0.049

0.014

Li/MgO

0.041

0.942

Li–MgO

0.017

0.014

Li+MgO

0.939

0.452

 

Cu

Li@MgO

0.002

0.001

Li/MgO

0.002

0.005

Li–MgO

0.002

0.001

Li+MgO

0.002

0.002

3.3.2 BET

The change of the BET surface area with time on stream is shown in Table 7 and the relative change is shown in Fig. 9. It is evident, that the Li@MgO, prepared via the decomposition of single source precursors, had the highest BET surface area at the beginning of the experiment, but it also showed the strongest reduction. The Li/MgO, prepared via wet impregnation, did not have a high BET surface at all, but after 1 h time on stream, the BET surface area fell below the detection limit of the used Micromeritics Gemini III 2375 Surface Area Analyzer. Therefore, after 3 h time on stream, the specific surface has not been determined anymore. The Li–MgO, prepared via precipiation, suffered from a loss of BET surface area only within the 1 h time on stream. After that no change was observed anymore. For Li+MgO, a strong decrease was observed after 1 h time on stream and after that only a comparatively small reduction was observed. From 7 h time on stream until the end of the experiment, no change was observed.
Table 7

The table shows the change of the BET surface area in m2/g with time on stream of the different 0.5 wt% Li-doped MgO

Time (h)

Li@MgO

Li/MgO

Li–MgO

Li+MgO

0

81.7

6.3

27.3

23.2

1

3.1

 ≤1

20.9

14.5

3

3.8

 ≤1

20.0

14.9

7

5.2

n.d.

22.8

10.0

14

4.3

n.d.

21.4

11.1

24

2.6

n.d.

20.5

11.0

n.d. not determined

https://static-content.springer.com/image/art%3A10.1007%2Fs11244-011-9749-z/MediaObjects/11244_2011_9749_Fig9_HTML.gif
Fig. 9

Relative loss of BET surface area with time on stream. The data of Table 7 is plotted in this figure. A strong reduction is observed for Li@MgO and Li/MgO, while Li–MgO and Li+MgO remain constant after an initial reduction period

3.3.3 XRD

The XRD diffractograms show the reflexes for MgO. Small reflexes for Li2CO3 were found for the samples of Li/MgO and Li@MgO, before reaction. Li2O was not found in any sample. The Li@MgO samples showed small reflexes which are probably due to carbon containing materials, shown in Fig. 10. Moreover the reflexes of Li@MgO were rather broad, indicating that they either have a rather low degree of crystallinity and/or have a small grain size. With time on stream the width of the reflexes decreased, indicating a higher regularity or crystallinity of the particles.
https://static-content.springer.com/image/art%3A10.1007%2Fs11244-011-9749-z/MediaObjects/11244_2011_9749_Fig10_HTML.gif
Fig. 10

The blue graph (below) represents the XRD pattern of Li@MgO recorded before reaction. The reflexes are realtively broad and at angles of approximately 18, 38 and 58 θ, they are most probably due to carbon contamination from the preparation. The green graph (above) represents the XRD of Li@MgO recorded after 24 h time on stream. The reflexes for the carbon containing compounds disappeared and the MgO reflexes have become narrower, indicating a higher degree of crystallinity and/or a larger particle size

The lattice parameters were determined before reaction and after 24 h time on stream. The values are in the range of the theoretical lattice parameter for MgO of 4.217 Å. The differences are insignificant and do not exhibit an evident trend.

However, the change in grain size was more significant. The change is shown in Fig. 11. The freshly prepared materials had a relatively low grain size. All materials, except Li–MgO, showed a drastic increase in the grain size within the first 3 h time on stream. After that, no significant change was observed. However, Li@MgO showed an increase in the grain size after 14 h time on stream, which was concomitant to a small increase in the catalytic activity. The Li–MgO sample did not show any signifcant change in the particle size with time on stream.
https://static-content.springer.com/image/art%3A10.1007%2Fs11244-011-9749-z/MediaObjects/11244_2011_9749_Fig11_HTML.gif
Fig. 11

The change of the grain sizes for the different Li-doped MgO catalysts. The Li–MgO did not show an increase in particle size, while the other samples exihibited a drastic increase in grain size within the first 3 h time on stream

3.3.4 Solid State NMR

Solid state 7Li MAS NMR measurements have been carried out for all four Li-doped MgO catalysts before reaction and after 1, 3, 7, 14 and 24 h time on stream. All samples before reaction show a single resonance at 0 ppm, with the Li/MgO sample showing a slight shoulder at 1.4 ppm. It has to be noted that no signals at 3.2 ppm which would be indicative of Li2O were found in any of the samples. The 7Li resonance peaks of the four samples before reaction exhibit different linewidths (full width at half height—FWHH) of 380, 285, 200 and 190 Hz for Li/MgO, Li+MgO, Li@MgO and Li–MgO, respectively. The narrow lines in the latter samples indicate a higher degree of cristallinity and uniformity in the latter samples.

In the samples measured after different reaction times no significant shift in the 7Li signal is observed for any of the samples, indicating that the chemical environments of the 7Li nuclei are not changed dramatically during the reaction. In the Li–MgO material a small resonance at 1.5 ppm is emerging after 1 h of reaction time and slightly increasing up to 24 h which can be explained by the formation of lithium silicates, but the dominant 7Li signal (90% of Li signal intensity) remains at 0 ppm.

However, in all four materials the 7Li NMR linewidth is significantly decreased after reaction yielding resonances as narrow as 250, 70, 85 and 100 Hz (FWHH) after 24 h time on stream for Li/MgO, Li+MgO, Li@MgO and Li–MgO, respectively, indicating a significantly higher degree of order (i.e., a higher crystallinity) after the reaction. For example, Fig. 12 shows the 7Li MAS NMR spectra of material Li+MgO before reaction and after 1, 3, 7, 14 and 24 h on stream. Interestingly, the linewidth has already narrowed to 85 Hz (FWHH) after 1 h on stream and no further narrowing is observed after 3 h of reaction time. Similar results are observed for the other three materials. In summary, the 7Li MAS NMR measurements of the materials indicate that while no significant change in the chemical environment of 7Li nuclei can be observed during the catalytic reactions, a significant increase in the cristallinity of the 7Li environments is observed.
https://static-content.springer.com/image/art%3A10.1007%2Fs11244-011-9749-z/MediaObjects/11244_2011_9749_Fig12_HTML.gif
Fig. 12

7Li MAS NMR spectra of 0.5 wt% Li+MgO recorded before reaction and after 1, 3, 7, 14, and 24 h time on stream

The Li-signal at 24 h time on stream is smaller than at 0 hours time on stream. It can not be excluded, that the signal which is present at 24 h is already present at 0 h, but being overlapped due to the broadness of that peak.

3.3.5 SEM

The SEM images for Li@MgO are shown in Fig. 13. After 24 h time on stream, restructuring is observed. Moreover, particle coarsening up to 500 nm and sintering effects such as neck growth and the loss of sharp edges are found.
https://static-content.springer.com/image/art%3A10.1007%2Fs11244-011-9749-z/MediaObjects/11244_2011_9749_Fig13_HTML.jpg
Fig. 13

SEM imgages for Li@MgO, before reaction (left) and after 24 h time on stream (right)

The images for Li/MgO are shown in Fig. 14. Similar to Li@MgO restructuring, extensive inhomogeneous particle coarsening up to 1 μm and sintering effects such as neck growth and the loss of sharp edges are discerned.
https://static-content.springer.com/image/art%3A10.1007%2Fs11244-011-9749-z/MediaObjects/11244_2011_9749_Fig14_HTML.jpg
Fig. 14

SEM imgages for Li/MgO, before reaction (left) and after 24 h time on stream (right)

For Li–MgO, see Fig. 15, the agglomerates have a fine and acicular grain structure, with a grain size ≤100 nm, which is in accord with grain size calculated via XRD. After 24 h time on stream, homogeneous grain coarsening and sintering effects such as neck growth and the loss of sharp edges are discerned.
https://static-content.springer.com/image/art%3A10.1007%2Fs11244-011-9749-z/MediaObjects/11244_2011_9749_Fig15_HTML.jpg
Fig. 15

SEM imgages for Li–MgO, before reaction (left) and after 24 h time on stream (right)

The images for Li+MgO are shown in Fig. 16. The primary grain size is ca. 200 nm. After the reaction, alike for the other materials, homogenous grain coarsening and sintering effects such as neck growth and the loss of sharp edges are noted.
https://static-content.springer.com/image/art%3A10.1007%2Fs11244-011-9749-z/MediaObjects/11244_2011_9749_Fig16_HTML.jpg
Fig. 16

SEM imgages for Li+MgO, before reaction (left) and after 24 h time on stream (right)

3.3.6 TEM

The TEM images before the reaction and after 24 h time on stream for Li@MgO are shown in Fig. 17, Li/MgO in Fig. 18, Li–MgO in Fig. 19 and Li+MgO in Fig. 20. Generally, the agglomerates consist of irregularly shaped nanocrystals, which is in accord with the results obtained by XRD and SEM. The Li@MgO sample has the smallest particle size, this coincides with the observations of the XRD, as shown in Fig. 11. However, after 24 h time on stream crystal growth is observed for all Li-doped MgO materials.
https://static-content.springer.com/image/art%3A10.1007%2Fs11244-011-9749-z/MediaObjects/11244_2011_9749_Fig17_HTML.jpg
Fig. 17

TEM imgages for Li@MgO, before reaction (left) and after 24 h time on stream (right)

https://static-content.springer.com/image/art%3A10.1007%2Fs11244-011-9749-z/MediaObjects/11244_2011_9749_Fig18_HTML.jpg
Fig. 18

TEM imgages for Li/MgO, before reaction (left) and after 24 h time on stream (right)

https://static-content.springer.com/image/art%3A10.1007%2Fs11244-011-9749-z/MediaObjects/11244_2011_9749_Fig19_HTML.jpg
Fig. 19

TEM imgages for Li–MgO, before reaction (left) and after 24 h time on stream (right)

https://static-content.springer.com/image/art%3A10.1007%2Fs11244-011-9749-z/MediaObjects/11244_2011_9749_Fig20_HTML.jpg
Fig. 20

TEM imgages for Li+MgO, before reaction (left) and after 24 h time on stream (right)

3.4 Main Deactivation Parameters

These parameters were also determined for the 0.5 wt% Li-doped MgO prepared via the four described preparation procedures. For all four Li-doped MgO samples the same trends were observed, however, Li+MgO exhibited the most distinct behavior. Therefore, it has been chosen to show the influence of the different deactivation parameters: temperature and flow rate.

A direct comparison of the deactivation at different flow rates and temperatures is not possible. Therefore, the relative CH4 conversion is used. The first recorded data point is used for normalization, so that the relative decay can be compared for different reaction conditions.

3.4.1 Flow Rate

Figure 21 shows the absolute and the relative CH4 conversion (first data point equal to 100%) and the C2 selectivity for different flow rates as a function of time on stream for the catalyst Li+MgO, recorded at 750 °C. The strong deactivation correlates with the flow rate, it is distinct that with increasing flow rates the deactivation rate increases. At a flow rate of 30 mL/min, the catalyst loses only 20% activity within 18 h time on stream. The C2 selectivity also decreases with time on stream, but it remains approximately the same for 60, 120 and 180 mL/min. For the flow rate of 30 mL/min, a much lower C2 selectivity is observed, which seems to increase towards the end of the experiment. The deactivation is accelerated at higher flow rates. This is remarkable, as usually the opposite is the case. An explanation has to be searched in the interaction of different process parameters.
https://static-content.springer.com/image/art%3A10.1007%2Fs11244-011-9749-z/MediaObjects/11244_2011_9749_Fig21_HTML.gif
Fig. 21

The absolute CH4 conversion (above), the relative CH4 conversion (middle) and the C2 selectivity (below) of Li+MgO as function of the flow rate and time on stream, recorded at 750 °C. The deactivation increases with increaseing flow rates, the differences are distinct. The C2 selectivity increases with time on stream and the C2 selectivity for the flow rate of 30 mL/min is much lower than for the other flow rates

The Li content was determined after the reaction, the results are shown in Table 8. The residual Li-loading after the experiment is independent of the applied flow-rate. In fact, the result is similar to the residual loadings of Li, obtained in Sect. 3.3.1.
Table 8

Li content in wt% after the end of the reaction, conducted with different flow rates at 750 °C

30 (mL/min)

60 (mL/min)

120 (mL/min)

180 (mL/min)

0.02 (wt%)

0.03 (wt%)

0.03 (wt%)

0.03 (wt%)

3.4.2 Temperature

In Fig. 22, the CH4 conversion, the relative CH4 conversion and the C2 selectivity are shown. It has been reported in detail, that the CH4 conversion and the C2 selectivity increase with increasing reaction temperature. This was also the case for Li+MgO. However, the relative CH4 conversion shows that the reduction of CH4 conversion was more severe at lower temperatures. At 650 °C, the catalysts showed only 30% of its inital CH4 conversion, meaning a loss of 70%. At higher temperatures the activity loss was reduced, for 800 °C the loss accounted only to 20%.
https://static-content.springer.com/image/art%3A10.1007%2Fs11244-011-9749-z/MediaObjects/11244_2011_9749_Fig22_HTML.gif
Fig. 22

The absolute CH4 conversion (above), the relative CH4 conversion (middle) and the C2 selectivity (below) of Li+MgO at 60 mL/min as function of the reactor temperature and time on stream. The CH4 conversion and the C2 selectivity increase with increasing reaction temperature, which is according to the literature. The deactivation decreases at increasing temperature, the differences are distinct. The highest temperature exhibited the lowest deactivation and vice versa

Like in Sect. 3.4.1, the Li content was determined at the end of the experiment. The results are shown in Table 9. At 650 °C, a higher residual Li content was found. Above this temperature, the amount of Li was found to be 0.03 wt%, irrespective of the applied reaction temperature.
Table 9

Li content after the end of the reaction, conducted at different reaction temperatures and a flow rate of 60 mL/min

650 (°C)

700 (°C)

750 (°C)

800 (°C)

0.07 (wt%)

0.03 (wt%)

0.03 (wt%)

0.03 (wt%)

3.4.3 Temperature and Flow Rate

Due to the unexpected behavior, experiments were conducted with different flow rates at different temperatures. Like in the Sects. 3.4.1 and 3.4.2, all tested materials show the same behavior. The results of 0.5 wt% Li+MgO are presented, since this material showed the most distinct trends. In Fig. 23, the CH4 conversion, the relative CH4 conversion and the C2 selectivity are shown. The CH4 conversion depended on the flow rate, but the C2 selectivity did not show changes at different flow rates.
https://static-content.springer.com/image/art%3A10.1007%2Fs11244-011-9749-z/MediaObjects/11244_2011_9749_Fig23_HTML.gif
Fig. 23

The absolute CH4 conversion (above), the relative CH4 conversion (middle) and the C2 selectivity (below) of Li+MgO as function of the reator temperature, the flow rate and time on stream. The decrease in CH4 conversion was independed of the flow rate at 650 °C. However, at 750 °C, the reduction depended on the flow rate

After 18 h time on stream, at 650 °C the relative CH4 conversion showed the same decrease at 60 mL/min and at 180 mL/min. For 750 °C, this was not the case. Here, a reduction of CH4 conversion of 30% for 60 mL/min was observed and of 50% for 180 mL/min. The degree of reduction depended on the flow rate, but only at high temperatures.

The Li content after the experiment was determined and the results are shown in Table 10. At 650 °C, the residual loading of Li was determined to be 0.07 wt% at a flow rate of 60 mL/min and 0.08 wt% at a flow rate of 180 mL/min. However, for 750 °C, the same Li content was found, irrespective of the flow rate.
Table 10

Li content after the end of the reaction, conducted with two different reaction temperatures and flow rates

 

650 (°C)

750 (°C)

60 (mL/min)

0.07 (wt%)

0.03 (wt%)

180 (mL/min)

0.08 (wt%)

0.03 (wt%)

4 Discussion

The stability of the catalytic performance for different Li-doped MgO catalysts has been tested for 40 h time on stream. Until now, these kind of life time tests are not reported in the literature of Li-doped MgO. None of the prepared Li-doped MgO or pure MgO catalysts was found to be stable. The catalysts did not reach a stable range within 40 h time on stream. Furthermore, the impression of stability at long times on stream are only due to the scaling of the axis. As shown in Fig. 5, the residual activity does not scatter strongly anymore after 40 h time on stream. For even longer times on stream, it can be expected that the scattering is further reduced, approaching a common activity. Therefore, it can be concluded that the preparation procedure has an influence on the initial activity, but not on the activity in a steady state, being reached after several hours time on stream.

Li@MgO and Li/MgO exhibit the trend that with increasing loading of Li, the residual conversion is reduced but the residual selectivity is increased. However, on a consecutive reaction pathway, the selectivity decreases with increasing conversion following a single trajectory. Figure 24 shows a conversion selectivity diagram for the residual values of Li@MgO and Li/MgO. It is obvious that the points for each catalyst lie on a separate trajectory, indicating different catalytic behavior within the OCM reaction network.

It has to be criticized that in many publications concerning Li-doped MgO, the stability is not reported or considered and only catalytic results which are obtained within a few hours time on stream are reported. But this kind of catalytic data can only be a snapshot on the deactivation trajectory of the Li-doped MgO catalysts.

All catalysts lose the majority of their initial Li content within the first hour time on stream, as shown in Fig. 8. For higher Li-loadings, the decrease in Li content might be retarded. The Li content of 0.01 to 0.03 wt% at 24 h time on stream, is independent of the preparation procedure. Furthermore, this value is in accordance with the estimation of Anderson and Norby for the maximum solubility of Li in MgO [29].

A signifcant loss of BET surface and an increase in grain size and crystallinity are also observed. The changes of these parameters seem to depend strongly on the applied preparation method. However, it can not be excluded that impurities play a significant role for the change of surface area and grain size.

In order to find a correlation between deactivation and physical properties a comparison of the relative changes of CH4 conversion, C2 yield, Li content, BET surface area and grain size (with inverse values for better comparability) for the four different Li-doped MgO samples is done in Fig. 25. The loss of BET surface area, increase in grain size and loss of Li content correlate, this is well in agreement with the results of Mirodatos et al. [5, 6]. It becomes evident that the decline of CH4 conversion and C2 yield does not correlate with any of the other determined features. This fact gains further support, when the deactivation as a function of flow rate and temperature is considered. A high flow rate and a low temperature resulted in a high deactivation rate, while under a low flow rate and high temperatures the CH4 conversion does not decline so strongly. Moreover, the reduction of the CH4 conversion is the same at 650 °C, irrespective of the flow rate, but not at 750 °C. This observation indicates two different sintering phenomena, which is consistent with the results of Mirodatos et al. [5]. The Li content differed only with different temperatures, not with a different flow rate. Therefore, it can be concluded, that the influence of the Li content on the catalyst activity is much lower than previously thought.
https://static-content.springer.com/image/art%3A10.1007%2Fs11244-011-9749-z/MediaObjects/11244_2011_9749_Fig24_HTML.gif
Fig. 24

Conversion selectivity diagram for the Li@MgO and Li/MgO. For each material the data points lie on one single trajectory

https://static-content.springer.com/image/art%3A10.1007%2Fs11244-011-9749-z/MediaObjects/11244_2011_9749_Fig25_HTML.gif
Fig. 25

Relative changes of CH4 conversion, C2 yield, Li content, BET surface area and grain size for the four different Li-doped MgO samples. The values for the gain size are inverse to improve the comparability

The apparent activation energy was calculated for 0.3 and 24 h time on stream. This was not possible for Li/MgO due to the fast deactivation. The calculated apparent activation energies are reported in Table 11 and the Arrhenius plot for Li+MgO is presented in Fig. 26. There is a strong increase in the apparent activation energie with time on stream. The determined values for the apparent activation energy after 24 h time on stream are within the range of experimentally determined values of 90 kJ/mol [30] and 235 kJ/mol [31], reported by other research groups, while at 0.3 h time on stream, they are considerably lower. This observation indicates a substantial change in the catalytic performance of the material, rather than only a depletion of active sites.
Table 11

Apparent activation energies for Li@MgO, Li–MgO and Li+MgO in kJ/mol

Catalyst

0.3 (h)

24 (h)

Li@MgO

40

160

Li–MgO

68

89

Li+MgO

47

98

https://static-content.springer.com/image/art%3A10.1007%2Fs11244-011-9749-z/MediaObjects/11244_2011_9749_Fig26_HTML.gif
Fig. 26

Arrhenius plots for Li+MgO for the determination of the apparent activation energy

The impurities that were determined seem to be stable under reaction conditions, their concentration did not change significantly. Li+MgO has the highest is the catalyst with the highest initial and residual activity and it shows a relatively slow deactivation, however, it also has the highest concentration of impurities. The determined list of impurities is not complete, there could be even other impurities of considerable amounts in the Li-doped MgO catalysts.

We would also like to comment on the debate on the active center of Li-doped MgO. Lunsford et al. suggested [Li+O] to be the active center [4]. This proposal is based on the following experiments: EPR spectra of the prepared Li/MgO were recorded, but they were not measured in situ, giving rise to signal at g = 2.054. The relative intensity of this signal was reported to increase with Li-loading of the catalyst. Moreover, the formation of methyl radicals was reported to increase with an increasing Li-loading. These two trajectories exhibited a very good correlation. The group of Abraham and Schirmer also found that signal with EPR and electron spin resonance (ESR) when studying Li-doped MgO single crystals [32, 33]. However, because this signal is not univocal, Abraham et al. and Rius et al. measured ENDOR spectra (electron nuclear double resonance spectroscopy), proving the existence of [Li+O −] [3436]. In the works of Lunsford, this prove is missing. Furthermore, the problem of stability, based on loss of Li and the effect of impurities are not considered. A recent study by Myrach et al. could not detect [Li+O], though the capability of their equipiment was demonstrated with Li-doped MgO single crystals [37]. Moreover, the finding of methyl radicals at the outlet of the reactor, could also be an effect of gas phase reactions; as described by Yates and Zlotin, who found a signifcant conversion and a medium C2 selectivity in blank reactor studies [38].

To question the contribution of either Li, transition metals or further candidates for the active site 1 wt% Li/MgO was subjected to a pretreatment with H2S, a well-known poison for transition metals. The pretreatment consisted in submitting the catalyst for 2 h in 100% H2S flow at 450 °C. The Li content was determined to be the same before and after the pretreatment. The 150 mg catalysts were tested for 15 hours time on stream at 700 °C. The other experimental conditions are as described in Sect. 2.2. The result is shown in Fig. 27. It is outstanding that with H2S pretreatment, the catalyst does not show any OCM activity anymore. Moreover, the activity could not be retained via calcination of the pretreated catalysts at 700 °C in air. The poisoning effect of H2S on the activity of Li/MgO is a strong indication that not the Li is the essential component of the active site, because it does not form stable sulfur components.
https://static-content.springer.com/image/art%3A10.1007%2Fs11244-011-9749-z/MediaObjects/11244_2011_9749_Fig27_HTML.gif
Fig. 27

Poisoning effect of H2S on 1 wt% Li/MgO. After pretreatment (for 2 h in 100% H2S flow at 450 °C), the 1 wt% Li/MgO catalyst does not show any activity anymore

5 Summary and Conclusion

It has been shown that the Li-doped MgO is unstable, irrespective of the preparation procedure. The different catalysts are unstable over the tested time period of 40 h, without the prospect of a range with stable activity. The most drastic changes occur during the first hour time on stream and the measured Li content reaches the proposed limit of Li solubility in MgO after approximately 14 h time on stream.

Since no correlation between the catalytic activity and the Li-loading, specific surface area and grain size was found, it was concluded that these are not the main deactivation factors. The deactivation is most probably due to a very complicated process, influenced by several different factors. However, at lower temperatures (650 °C) and at elevated temperatures (750 °C), different deactivation mechanism are predominant.

Due to the strong deactivation and the loss of Li and its effect on reactor materials, we do not see Li-doped MgO as a potential candidate for an industrial application of the oxidative coupling of methane.

Depending on our own results and the results published in the literature, it is concluded that there is no proof for [Li+O] being the active center. This is further supported by the poisoning effect H2S has on Li/MgO. Since Li does not form stable sulfides, it is most probably not part of the active center.

Acknowledgements

We would like to thank the Deutsche Forschungsgemeinschaft for funding the Excellence Cluster “Unicat” (Unifying Concepts in Catalysis) and the IMPRS (International Max Planck Research School) of the Fritz Haber Institute of the Max Planck Society for financial support. We are also obliged to Mr. Axel Schiele and the workshop for their support with the equipment. We thank Dr. Traugott Scheytt and his team for the performance of the AAS analysis and Dr. Nakhal for the H2S pretreatment of the Li/MgO catalyst. We are indepted to Dr. Thomas Risse and Dr. Raimund Horn for their valuable advice. We also thank our apprentices Mrs. Anna Paliszewska and Mr. Domenic Jelinski, for their support with the sample preparation and analysation. We would like to thank Prof. Dr. Arne Thomas for the permission to use his multi-sampling XRD machine and Dr. Kamalakannan Kailasam for the XRD experiments. We thank the Helmholtz Center, Berlin for the permission to use their electron microscope. We thank also the ZELMI (Zentrales Laboratorium für Elektronenmikroskopie, TU, Berlin) for TEM analyses.

Copyright information

© Springer Science+Business Media, LLC 2011