Cordierite

Feb 25, 2010 - Ljubiša Gavrilović , Jan Brandin , Anders Holmen , Hilde J. Venvik , R. Myrstad , and Edd A. Blekkan. Industrial & Engineering Chemis...
0 downloads 0 Views 3MB Size
3176

Ind. Eng. Chem. Res. 2010, 49, 3176–3183

Effect of Alkali Vapor Exposure on Ni-MgO/γ-Al2O3/Cordierite Monolithic Catalyst for Biomass Fuel Gas Reforming Y. P. Li, T. J. Wang, C. Z. Wu,* Y. Gao, X. H. Zhang, C. G. Wang, M. Y. Ding, and L. L. Ma Key Laboratory of Renewable Energy and Gas Hydrate, Guangzhou Institute of Energy ConVersion, Chinese Academy of Sciences, Guangzhou, Guangdong 510640, P.R .China

Fly ash compounds, such as alkali salts, in the raw biomass fuel gas can contaminate and deposit on traditional granular Ni-based catalysts, which resulted in catalyst deactivation and pressure increase of the downstream reformer. The impact of alkali salt exposure (KCl, K2SO4, K2CO3, by evaporation at about 7.8 mg/L for 6 h) on dry CH4/CO2 reforming of model biomass fuel gas (H2/CO/C2H4/CH4/CO2/ N2 ) 15.8/12.1/2.51/ 15.0/22.1/32.6 vol %) over Ni-MgO/γ-Al2O3/cordierite monolithic catalyst (MC) was investigated and studied. The results showed that CH4 and CO2 conversions and CO and H2 yields increased at 700-850 °C for undeposited and deposited MC. Compared with undeposited MC, the deposited catalysts show lower CH4 conversion but higher CO2 conversion and CO yield at 750-830 °C. The stability tests also show that CH4 conversion and H2 content in the tail gas decreased dramatically from 87.2% to 32.0% and from 35.1% to 26.7%, respectively, after 17 h time on stream (TOS) for the deposited MC, while CH4 conversion kept steady of above 90% after 60 h TOS for undeposited MC at 750 °C. Characterization by N2-physisorption, XRD, ICP-AES, SEM-EDS, and XPS of MC indicate that alkali salt aerosol covering the catalyst surface or blocking mesopore channels was the main reason for the decreased reforming performance and MC deactivation, which occurred mainly at the top part of monolithic catalyst (K ) 1.39 wt % by EDS), vicinal to the alkali source. The reforming of real biomass fuel gas (H2/CO/C2H4/CH4/CO2/N2 ) 10.2/16.8/0.5/6.4/15.2/51.0 vol %) from air gasification of pine sawdust in the pilot plant (200-250 kg/h) by the reformer packed with MCP, larger in size than MC, exhibits pressure drop of less than 700 Pa, CH4 conversion of about 84%, and tar content from 4.8-5.3 g/m3 to 0.12-0.14 g/m3 during 60 h TOS at 600 °C. The porosity structure of MCP catalytic bed and relatively low alkali (K, Na ) 0.03-0.07 wt %) deposition by fly ash from real biomass fuel gas were the main reasons for the excellent reformer performance. 1. Introduction As one of the most promising technology to substitute fossil fuels, second-generation biofuel production has been studied widely,1 such as the Hynol project in the United States, BioFuels projects in Sweden, Sun-diesel in Choren Company, Germany.2 Yet thermochemical gasification of biomass is inevitably accompanied by the formation of tar and inorganic impurities(Si, Ca,K, Na, S, Cl, Mg, Zn, Fe, Mn) and the obtained raw fuel gas is usually CO2-rich with low H2/CO ratio.3 To fulfill the stoichiometric requirement for syngas conversion to liquid fuels, reforming of CH4 and tar to increase H2/CO ratio has been carried out extensively in the past decades.4 Although Nibased catalysts showed high efficiency in reforming, their deactivation was obvious due to carbon deposition during short operation period.5,6 And the deterioration of inorganic impurities like alkali compounds in fly ash is another adverse factor on reforming, which has not received considerable attention until recently.7,8 During biomass gasification, some alkali and alkali earth compounds, abundant in the feedstocks, are apt to form sulphates and chlorides and cause heterogeneous nucleation and agglomeration in the gasifier.9,10 And the alkali particles may exist in the fly ash or form aerosol particles in fuel gas, which often foul the downstream gas turbine and even inactivate reforming catalyst ultimately.11,12 Several techniques have been applied to reduce the alkali vapor emissions, such as alkali absorbent in the gasifier, hot-gas filter and gasification temperature decrease, et al.9,13 These procedures, however, are * To whom correspondence should be addressed. E-mail: wucz@ ms.giec.ac.cn. Tel.: +86-20-87057688. Fax: +86-20-87057737.

insufficient to drastically eliminate alkali content in the raw fuel gas. It was reported that CH4 conversion of Pt-Rh/MgAl(O) catalyst was decreased to below 25% after exposed to the generated K2SO4 aerosol particles and to the alkali aerosol particles from the water-soluble part of biomass fly ash during steam reforming in lab-scale.14 Few concrete experiments have been focused on other catalysts, especially for monolithic catalysts,15 which have advantages as high catalytic performance per unit mass of active metal, high operating stability and low pressure drop of catalyst bed over fix-bed reactor or slurry reactor.16 Preliminary studies from our group have achieved remarkable CH4 and tar reforming performance from biomass fuel gas over monolithic Ni-based catalyst.17 In the present work, the effect of alkali salt vapor exposure (KCl, K2SO4, and K2CO3) on NiMgO/γ-Al2O3/cordierite monolithic catalyst(MC) for model biomass fuel gas reforming, as well as the stability of undeposited and deposited catalyst were investigated to evaluate the impact of inorganic salts (expected to be present in the biomass fuel gas) on a catalytic reformer unit. And the typical reforming results of real biomass fuel gas from fluidized bed gasification of pine sawdust (200-250 kg/h) in the pilot plant were also discussed. 2. Experimental Section 2.1. Catalyst Preparation. Two types of ceramic cordierite monolithic substrates (Yixing Nonmetallic Chemical Factory, Jiangsu Province, China) with the density, porosity, BET, cell density, specification of 1.71 g/cm3, 60%, 10.9 m2/g, 60 cell

10.1021/ie901370w  2010 American Chemical Society Published on Web 02/25/2010

Ind. Eng. Chem. Res., Vol. 49, No. 7, 2010

3177

XCO2 ) (FinXin,CO2 - FoutXout,CO2)/(FinXin,CO2) YH2 ) (FoutXout,H2 - FinXin,H2)/(2FinXin,C2H4 + 2FinXin,CH4) YCO ) (FoutXout,CO - FinXin,CO)/(2FinXin,C2H4 + FinXin,CO2 + FinXin,CH4)

Figure 1. Schematic diagram of the experimental apparatus.

s/cm2, φ4.0*5.0 cm and 1.03 g/cm3, 90%, 10.4 m2/g, 1.5cells/ cm2, φ9.0*10.0 cm were used to prepare catalysts for reforming model biomass fuel gas and real biomass fuel gas respectively. At first, the bare monoliths were immersed into 30 wt % oxalic acid solution for 0.5 h, and then dipped into the ethanol suspension of ultrafine γ-Al2O3 powder under vacuum condition to increase pore volume and surface area. Then γ-Al2O3-coated monoliths were impregnated by 1N aqueous magnesium nitrate (Mg(NO3)2.6H2O) solution, dried overnight at 110 °C and calcined at 900 °C. After that, MgO-loaded monolith was impregnated again by 1 N aqueous nickel nitrate (Ni(NO3)2.6H2O) solution, dried overnight at 110 °C, and calcined at 850 °C. The two types of well-prepared catalysts were denoted as MC and MCP, respectively. 2.2. Alkali Salt Vapor Deposition on MC and Model Biomass Fuel Gas Reforming Procedure. The deposition of alkali salt vapor on MC was carried out in a horizontal reactor with two thermocouples, one was for alkali salt vaporization at 750 °C and the other was for reforming temperature controlling in Figure 1. In each run, MC was reduced in situ in 300 mL/ min of 5% H2/N2(vol) flow at 750 °C for 2 h, which was very uniformly black, and then exposed to the 300 mL/min N2 flow containing alkali salt vapor(KCl, K2SO4 and K2CO3), settled in a porcelain boat before MC bed. The alkali content in the carrier gas was determined by the weight loss of the boat after 6 h deposition and the effluent gas was introduced into water to absorb surplus alkali salts. The dry CH4/CO2 reforming activity of model biomass fuel gas (H2/CO/C2H4/CH4/CO2/N2) 15.8/12.1/2.51/15.0/22.1/32.6 vol %) over undeposited and deposited MC was carried out at atmospheric pressure, 300 mL/min of gas flow rate, and gas hour space velocity (GHSV) of 541.8 h-1. The general reaction time for the reaction was kept constant of 3-5 h to make the reforming results comparable. The tail gas during this period passed through water absorber, and then was sampled into gas bag every 15 min for off-line gas chromatogram analysis. And the derivations were usually below 5%. CO, N2, CH4, and CO2 were analyzed by carbon-sieve column with TCD detector; gaseous hydrocarbons (CH4, C2H4) were analyzed by Porapak Q column with FID detector. N2 in the gas was used as internal standard to calculate CH4 and CO2 conversion. CO2 or CH4 conversion and CO or H2 yield were calculated as the following equations: XCH4 ) (FinXin,CH4 - FoutXout,CH4)/(FinXin,CH4)

where Xin,p is inlet gas composition (vol%), Xout,p is outlet gas composition (%), Fin is total inlet gas flow rate(mL/min), Fout is total outlet gas flow rate (mL/min), X is conversion (%), and Y is yield(%). 2.3. Biomass gasification-reforming procedure. As shown in Figure 2, the system was made up by biomass feeder, gasifier, downer reactor, reformer, water bath, gas tank. Pine sawdust of 200-250 kg/h was air-gasified. The raw fuel gas passed through a purification unit to separate most fly ash particles and then through a multitube(49 columns) reformer ((cube-shaped, 1 m*1 m*1 m) heated with a burner to supply reforming heat from diesel combustion. The 49 tubes (ID: 93 mm, Length: 1.2 m) was arranged to be 7 columns and 7 lines in the reformer. Ten pieces of MCP were packed in one tubular tube and each piece of MCP was separated by 1 cm of inert Rashig rings. After that, the gas was washed by water to further eliminate particles and other impurities and pumped by roots blower to gas tank for dimethyl ether synthesis. 2.4. Catalyst Characterization. The elemental compositions of the catalysts were determined by inductively coupled plasmaatomic emission spectroscopy (ICP-AES) on a PEPLASMA2000 instrument. The textural properties of undeposited and deposited MC were measured by N2 physisorption at -196 °C on Micromeritics SORBFLOWIII-3210 apparatus and pore distribution was described by BJH model. X-ray diffraction (XRD) patterns were obtained on a BDX3300 using CuKa´ radiation at 40 kV and 30 mA. Ni Metal crystallite size was determined from the peak at 2θ ) 44.3°, which is assigned to the diffraction of Ni(1 1 1) plane in XRD patterns [30], using Scherrer equation.18 The morphology of MC was obtained by scanning electron microscopy(SEM) on a JSM-6330F equipment with energy diffusive spectroscopy detector (EDS). The elemental compositions of the catalyst surface were identified by NP-1 X-ray Photoelectron Spectrometer (XPS) equipped with Mg-KR radiation. The binding energies were referenced to the C1s band at 284.6 eV. 3. Results and Discussion 3.1. Textural Properties of MC. The textural properties of cordierite substrate and MC were summarized in Table 1. It was commonly accepted that alkali metal elements could be converted from organic form into inorganic form as sulphates and chlorides after gasification or combustion and the principal alkali metal was K, followed by Na.19,20 In this paper, KCl was used as model alkali salt in fly ash deposited on MC. The K vapor content in the carrier gas was 7.8 mg/L and K content in MC was 0.85 wt % after 6 h exposure by ICP-AES, which was higher than that from fly ash deposition during practical biomass gasification.21 It confirmed that the alkali salt vaporization operation was applicable for simulating alkali salt deposition in a catalytic reforming bed. While K content after reforming reaction decreased to 0.57 wt %, which mainly resulted from metal loss during high temperature reforming reaction. It is clear that BET and BJH surface area increase remarkably from 10.9 m2/g to 40.0 m2/g and from 7.75-9.52 m2/g to 47.0-48.1 m2/g, respectively, after oxalic acid immersion and

3178

Ind. Eng. Chem. Res., Vol. 49, No. 7, 2010

Figure 2. Schematic diagram of biomass gasification-reforming system (1, pine sawdust feeder; 2, fluidized bed gasifier; 3, Downer reactor; 4, power generation; 5, low BTU gas for direct combustion; 6, gas purification and upgrading; 7, burner; 8, reformer; 9, chimney; 10, MCP packed in the reformer; 11, water bath; 12, roots blower; 13, gas tank; 14, cross-section image of the reformer and analysis positions). Table 1. Textural Properties of MCa items

cordierite

undeposited MC

reduced MC

BET specific surface area (m2/g) BJH cumulative surface area (1.7-300 nm) (m2/g) BJH cumulative volume (1.7-300 nm) (cm3/g) average pore diameter (nm) K concentration (wt %) Ni particle size (nm)

10.9 7.75-9.52 0.01 4.10

40.0 47.0-48.1 0.094 8.14

39.6 49.2-49.3 0.09 10.11

a

42.2

KCl deposited MC

KCl deposited MC after reaction

15.1 16.9-17.5 0.05 10.3 0.85 42.3

11.2 9.77-12.0 0.03 12.2 0.57

Chemical composition of untreated undeposited MC: Al2O3/MgO/NiO/Cordierite ) 6.6/3.3/5.5/84.6 wt %.

superfine γ-Al2O3 deposition (BET surface area ) 350 m2/g) in Table 1. The pore volume and the average pore diameter also increased to 0.094 cm3/g and 8.14 nm, respectively, which attributed to mesopore amount increase by pore expansion treatment on cordierite substrate. Reduction in H2/N2 flow did not change textural properties of the catalyst, because relatively low Ni and Mg impregnation amounts of 5.5 wt % and 3.3 wt % in Table 1. While BET area of KCl deposited MC decreased evidently to 15.1 m2/g, similarly with 10.9 m2/g of the original cordierite substrate. Pore volume decreased nearly by half to 0.05 cm3/g, but pore diameter kept constant of about 10 nm.11 The decrease of mesopore volume, BET might be caused by blockage of some mesopore channels and coverage of surface area by alkali salt aerosol deposition, discussed in section 3.3. After reforming reaction, BET area and pore volume of KCl deposited MC decrease further to 11.2 m2/g and 0.03 cm3/g, respectively. According to BJH pore distribution results, mesopore fraction of the used deposited MC decreased with increased macropore fraction and increased pore diameter of 12.2 nm. 3.2. Structure Properties of MC. Figure 3 illustrates the XRD patterns of undeposited, reduced, and KCl deposited MC. No obvious nickel diffraction peaks were detected for undeposited MC, but it exhibited obvious presence of NiO

Figure 3. XRD patterns of MC catalyst samples.

(JCPDS 34-0396, 37.1°, 43.3°, 62.8°). According to the literature,22 three major peaks at 44.3°, 51.8°, 76.5° and at 36.8°, 44.9°, 65.3° of the reduced MC were identified as Ni0 and NiAl2O4 (or NiMgO2), respectively. Because of the overlapping of main diffraction peaks of NiAl2O4 and NiMgO2 phase, it was difficult to identify the two compositions by XRD solely. The weak diffraction intensity corresponding to Ni0 for the reduced MC and KCl deposited MC indicates small Ni crystallite size and high dispersion of Ni particles on the catalyst surface. It might also be attributed to the formation of NiAl2O4 (or NiMgO2) spinel and the interaction between active Ni metal and high surface area of γ-Al2O3 support, which is difficult to be reduced. And the existence of NiO diffraction lines for the reduced MC and KCl deposited MC was the result of the reoxidation of metal Ni under atmosphere condition during XRD analysis. The similar XRD patterns also indicated that no new phase was formed during KCl deposition on MC or K content was too low to be identified by XRD or K existed as a noncrystal form.23 Calculated by Scherrer equation, Ni(111) crystal size for reduced MC and KCl deposited MC were 42.2 and 42.3 nm, respectively, in Table 1, which illustrated that no obvious sintering happened during KCl deposition process. It could also be concluded preliminarily that the decreased catalyst surface area should be attributed to KCl exposure. Such results are in disagreement with the results from Moradi,24 who found BET area slightly increased over KCl deposited Pt-based catalyst at 300 °C, but are correspondent with the work of Einvall,21 which also showed the BET decrease over Ni-based catalyst. The difference may be resulted from the various deposition process and distinctly different deactivation mechanism of these catalysts. 3.3. SEM and XPS Analysis. To evaluate the change of surface morphology after KCl exposure, MC samples, perpendicular to biomass fuel gas flow were obtained from different positions of MC bed. The photos of sampling positions were shown in Figure 4 and the corresponding SEM images were shown in Figure 5.

Ind. Eng. Chem. Res., Vol. 49, No. 7, 2010

3179

Figure 4. Axial photos of MC samples: (A) reduced, (B) KCl deposited, and (C) KCl deposited after reaction (1 or 2 refers to SEM-EDS and XPS analysis position in Figure 5 and Table 2).

Figure 5. SEM images of MC samples: (A) reduced, (B) KCl deposited at position 1 in Figure 4B, (C) KCl deposited at position 2 in Figure 4B, and (D) KCl deposited after reaction at position 2 in Figure 4C (Inset is the close-up micrograph of filament domain morphology).

The surface micrograph of the reduced MC revealed that the monolithic surface was covered by amorphous massive materials, mainly NiAl2O4 spinel structures, which are adhered by polydispersed spherical particles of Ni phase according to the EDS analysis. This is in good accordance to XRD characterization in Figure 3, where there are existences of NiAl2O4 and Ni phase. The particle size measured by SEM dates was larger than that by XRD patterns because the particle observed by SEM was the agglomeration result of numerous supported Ni crystalline by XRD. As shown in Figure 5B, some catalyst surface of the top part of KCl deposited MC(position 1 in Figure 4B) is covered by irregularly shaped sheet microstructure with round edges, mainly K aerosol with the content of 1.39 wt % by EDS. In the same time, the K/Ni ratio at this position was much higher than that at the middle part of MC(position 2 in Figure 4B) in Table 2, about 17-fold by EDS. And the SEM image at the middle part of KCl deposited MC exhibits that small spherical particles are attached to the surface of the amorphous surface, similar to the freshly reduced MC in Figure 5A. It could be concluded that alkali salt deposition mainly happened on the positions vicinal to the alkali salt source. It thus appears that the majority of the alkali aerosols were retained at the top part of MC. So the application of monolithic catalyst would reduce the alkali salt deposition tendency down the length of catalytic bed.

The K/Ni atom ratio difference by EDS and XPS for the top part of KCl deposited MC in Table 2 was due to the variation of layer thickness analyzed by EDS and XPS, about 5 µm by EDS and 5 nm by XPS which means that XPS analyses only few layer of the samples. About 52-fold of K/Ni atom ratio by EDS and XPS, which was 5.74 and 0.11 respectively, indicated that K not only deposited on the MC surface, but permeated in the mesopores of MC and decreased active metal surface in section 3.1. According to the literature,25 binding energy at 294.7 and 292.7 eV was assigned to K2p and K2p+, respectively. The XPS results of binding energy at 293.0 eV for KCl deposited MC might be Kσ+ with σ varying from 0.5 to 0, which means an indirect electrostatic interaction of K with the active Ni metal.26 For the top part of reduced MC, XPS pattern shows two peaks at ∼855.8 and 862.0 eV, respectively, indicated that Ni was in two different conditions and corresponded to NiAl2O4 and NiO respectively with relative intensities of 70.6% and 29.4%. These results also suggest the strong interaction between Ni and γ-Al2O3. And the oxidation state of Ni for the reduced MC by XPS was due to the reoxidation of Ni by O2 in air condition before analysis, similar with the oxide phase of NiO by XRD in Figure 3. As shown in Figure 5D, the filamentous carbon was found on the surface of KCl deposited MC after dry CH4/CO2 reforming reaction. Closer inspection from inset of Figure 5D shows that these filaments were entangled with catalyst particles. EDS spectra of the filamentous domain also shows peaks corresponding to catalytic Ni element, cordierite elements and carbon. The spent KCl deposited MC was also examined for the amount of carbon formed by combustion with air flow over the sample at 800 °C for 2 h in a furnace. The weight change of 1.94 wt % was primarily caused by carbon deposition on the sample since the weight change due to Ni phase was relatively low. Nevertheless Ni-based catalysts are more prone to carbon deposition than precious metal catalysts during reforming reaction, particularly due to the formation of filamentous carbon, which was found by many other researchers.5 While the carbon deposition degree in this paper was relatively low, there still exists small domains of spherical particles of active phase on MC surface. This would be the reason why the catalyst could maintain its activity after few hours’ reaction in section 3.5. 3.4. CH4/CO2 Reforming Activities for Model Biomass Fuel Gas over MC. Both reduced MC and KCl, K2SO4, and K2CO3 deposited MC were evaluated for dry CH4/CO2 reforming of model biomass fuel gas mentioned in section 2.2 at 700-850 °C. The results are listed in Table 3. C2H4 in the feed gas was converted completely. For undeposited MC and

3180

Ind. Eng. Chem. Res., Vol. 49, No. 7, 2010

Table 2. SEM-EDS and XPS Analysis Results of MC K/Ni atomic ratio by EDS analysis (%)

position (Figure 4)

catalyst reduced catalyst

1

deposited catalyst

1 2

K/Ni atomic ratio by XPS analysis (%)

Ni2p3/2

K2p

0.11

855.8 (70.648%) 862.0 (29.352%) 856.0

293.0

5.74 0.33

binding energy (eV)

Table 3. Effect of Potassium Salt Vapor Deposition on MC Reforming Activitiesa catalyst

undeposited catalyst

KCl deposited catalyst

K2SO4 deposited catalyst

K2CO3 deposited catalyst

a

X (%)

Y (%)

Xout,p (%)

temperature (°C)

Fout/Fin

CH4

CO2

H2

CO

CH4

CO2

H2

CO

nH2/nCO

850 830 780 750 700 850 830 780 750 700 850 830 780 750 700 850 830 780 750 700

1.21 1.16 1.08 1.08 1.09 1.22 1.21 1.11 1.07 1.08 1.29 1.19 1.17 1.06 1.01 1.24 1.18 1.15 1.06 1.03

95.45 96.37 91.20 88.86 78.99 92.19 90.23 88.26 89.29 76.43 90.39 78.65 50.65 32.95 14.84 91.24 88.65 85.35 82.36 74.68

90.03 75.03 68.78 64.18 50.78 88.81 79.76 75.43 67.62 56.26 91.37 90.07 80.74 65.40 61.17 81.57 80.14 75.23 65.54 60.38

74.59 66.54 56.84 58.53 58.97 71.37 72.28 60.79 60.25 57.45 78.54 60.53 49.70 24.90 15.10 72.23 70.32 59.97 54.01 55.10

75.82 63.32 46.31 43.34 33.93 78.13 70.58 53.68 45.22 36.08 90.07 77.47 66.38 45.81 33.70 80.17 70.47 56.38 45.13 34.70

0.56 0.46 1.19 1.66 3.01 0.96 1.21 1.04 1.46 3.11 1.12 2.67 6.30 9.48 12.61 1.01 1.26 2.30 2.78 3.81

1.74 4.64 8.26 9.28 11.59 2.02 3.67 4.85 6.52 8.47 1.46 1.83 3.61 7.17 8.43 2.06 3.63 4.96 6.17 8.53

34.47 33.44 33.21 33.46 33.26 33.96 34.03 33.78 33.93 33.18 33.67 31.09 38.44 23.29 21.02 33.47 32.32 30.01 30.15 29.78

36.20 33.14 27.51 25.31 22.78 36.76 34.31 30.99 28.25 23.57 38.58 37.23 33.97 29.44 25.83 37.78 36.53 32.79 28.34 24.29

0.95 1.00 1.13 1.32 1.38 0.91 0.99 1.07 1.19 1.39 0.87 0.83 0.84 0.79 0.81 0.89 0.88 0.92 1.06 1.23

Note: nH2 is outlet H2 content (%, vol), and nCO is outlet CO content (%, vol).

deposited MC, both CH4 and CO2 conversion increased with the increased temperature. This can be explained by the following chemical equations, mainly occurring in the catalytic bed:27,28 CH4 + CO2 ) 2CO + 2H2

(1)

C2H4 + 2CO2 ) 4CO + 2H2

(2)

CO2 + H2 ) CO + H2O

(3)

2CO ) C + CO2

(4)

CH4 ) C + 2H2

(5)

The endothermic dry reforming reaction between CH4 and CO2 was enhanced by temperature increase to favor CH4 and CO2 conversions, which also resulted in the increased H2 and CO yields. While high reaction temperature also promoted the endothermic reverse water gas shift reaction, which consumed H2 and produced CO. So CO yield increase is higher than that of H2, which resulted in decreased H2/CO ratio in tail gas, accordingly from 700 to 850 °C for undeposited and deposited MC in Table 3. Compared with undeposited MC, most potassium salt vapor deposited MC exhibited lower CH4 conversion, but a little higher CO2 conversion and CO yield under same reaction condition at 750-830 °C. It could be attributed to the presence of alkali aerosol deposition on MC surface, which caused the decrease of BET area, pore volume, Ni metal surface, and CH4 adsorption amount. Praliaud described that alkali metal acted as a promoter, existing in the form of alkali silicate or nitrate, and mainly associated with an anion.26 However, K vapor deposited on Ni(111) were in the form of Kσ+ with σ varying from 0.5 to 0 as the alkali aerosol coverage increase by XPS results in this

paper. It could be assumed that an indirect electrostatic interaction occurred between the deposited alkali metal and the active Ni metal. Alkali metal would migrate from the top part of the monolithic catalyst (vicinal to the alkali source) to the whole reforming bed.24 Besides, the deposition of alkali salt vapor may decrease catalyst surface acidity and change the adsorption characteristics of MC, which promoted the carbon elimination reaction, shown in eq 4. And the electron enrichment of metallic phase by alkali metal might increase the backdonation of metal electron to the 2π* antibonding of CO2,29 so CO2 molecules were adsorbed preferentially on MC surface. That promoted the reversed reaction of eq 4, which increased CO2 conversion and CO yield, but decreased the H2/CO ratio in the tail gas. Although K2SO4 has the highest melt temperature of 1067 °C, K2SO4 deposited MC showed most serious decrease of CH4 conversion among three potassium salts of KCl, K2SO4 and K2CO3. And the deactivation results in Table 3 were similar to that of powdered Pt/Rh catalysts, where BET order after salt deposition was: NiKCl > NiZnCl2 > NiK2SO4 and BET decrease degree was in linear relationship with molecular weight of the deposited salt.11 Besides, KCl was in the melting state, while K2SO4 and K2CO3 still kept in solid state under reforming reaction of 750 °C in this paper, so it might result in different catalytic reforming mechanism on these three potassium salt vapor deposited MC. In the same time, the deactivation of K2SO4 deposited MC may also be caused by sulfide poisoning besides the active Ni phase decrease mentioned above.14 It also should be pointed that the irregular CO2 conversion results at high temperature of 850 °C and H2 yield might also result from the different reforming mechanism and interactions of the complex reforming system,30 where the main reactions were accelerated at different temperature respectively according to their thermodynamic properties.

Ind. Eng. Chem. Res., Vol. 49, No. 7, 2010

3181

Figure 6. CH4 conversion (A) and tail gas composition (B) with time on stream over undeposited MC(close symbols) and KCl deposited MC(open symbols).

Figure 7. Typical reforming results of real biomass fuel gas over MCP with time on stream: (A) pressure drop of the reformer, (B) CH4, CO2, and tar conversion, and (C) tail gas composition.

3.5. Stability of KCl Deposited MC. It is known that Nibased catalyst deactivation due to carbon deposition via Boudouard and methane decomposition reactions was serious during CH4 or liquid fuel reforming.31 In this paper, it was mainly focused on the effect of alkali impurities on biomass fuel gas reforming system, which also contributed to catalyst. The stability tests were carried out at 750 °C. For undeposited MC, CH4 conversion maintained about 90% after 60 h time on stream (TOS) and tail gas compositions changed slightly in Figure 6: CH4 content increased from 1.03 to 1.38 vol % and H2 content was almost the same of 32.0%. While after KCl vapor deposition, CH4 conversion decreased from 87.2% to 32.0% and H2 content in tail gas decreased from 35.1 to 26.7 vol % after 17 h TOS. According to SEM-EDS results, it could be conferred that the deactivation by KCl deposition was probably caused by a weak interaction between alkali and Ni and MgO, resulting in rapid migration of alkali metal through the whole catalyst bed in 3.3. In the beginning, the alkali was primarily retained at the top part of MC bed (position 1 in Figure 5B). Subsequently, extra alkali salt vapor spread the bed and poisoned the whole catalyst, which might be the main reason for the decreased CH4 conversion due to the coverage of Ni phases by alkali vapor deposition. 3.6. Reforming of Real Practical Biomass Fuel Gas over MCP. The quick deactivation of KCl deposited MC for dry CH4/CO2 reforming of model biomass fuel gas in section 3.5 was due to the low Ni loading amount(5.5 wt %, NiO) and high KCl deposition amount(7.8 mg/L KCl in carrier gas for 6 h, K content of 0.85 wt % and 1.39 wt % at the top part of deposited MC by ICP-AES and EDS respectively), which is far higher than K amount by emission of atomized alkali salt solution for 2 h(37 mg/m3)11 and real K content in fly ash during biomass gasification.24 To investigate the real performance of cordierite monolithic catalyst of MCP, larger in size than MC,

the reforming procedure of our pilot-scale integrated dimethyl ether synthesis system from air gasification of pine sawdust (200-250 kg/h) was carried out and the results are shown in Figure 7 and Table 4. It can be seen that the pressure drop of the reformer fluctuated over the first 2 h TOS, and the highest pressure drop was up to 1200 Pa (Figure 7A) because of the unstable gasification process. But after about 2 h TOS, the pressure drop kept stable of about 700 Pa and that would decrease the power requirement for roots blower, which ventilated biomass fuel gas to the gas tank. The reforming effect of real raw biomass fuel gas over MCP was obvious: CH4 conversion of raw biomass fuel gas (initial compositions, H2/CO/C2H4/CH4/CO2/N2 ) 10.2/16.8/0.5/6.4/ 15.2/51.0 vol % in Figure 7C) was above 84% during 60 h TOS at 600 °C in Figure 7B. CO2 conversion was relatively low to below 40% because of the low reforming temperature in the reformer, resulting from inadequate external heat transfer of the burner by combusting diesel. And tar content decreased from 4.8-5.3 to 0.12-0.14 g/m3 after catalytic reforming, which resulted in more than 40 vol % content of H2 and CO with H2/ CO ratio of around 1, higher than the original content of 27 vol %. The particle size of fly ash was 0.15-0.85 mm in this gasification process, confirmed by SEM. After 60 h operation, the K and Na content in MCP at different positions of reformer (as shown in Figure 2) was similar of 0.03-0.07 wt % except at the reformer inlet (0.26 and 0.11 wt %, respectively, in the intermediate tube) in Table 4. It was because of the relatively slow fuel gas flow in the up-draft reformer and vicinity to the alkali source, where K and Na in the fly ash were inclined to deposit. And this was accordance with high K content at the top part of MC during model experiments in section 3.3, which approved the application possibility of model alkali salt evaporation method to investigate fly ash impurities on reform-

0.06 0.06

Conclusion The deposition of alkali salt vapor (KCl, K2SO4 and K2CO3) on the Ni-MgO/γ-Al2O3/cordierite monolithic catalyst showed that the high content of alkali was mainly existence at the top part of monolithic catalyst(vicinal to alkali source), which caused quick catalyst deactivation for dry CH4/CO2 reforming performance of model biomass fuel gas(H2/CO/C2H4/CH4/CO2/N2) 15.8/12.1/2.51/ 15.0/22.1/32.6 vol %). The extent of catalyst deactivation increased with alkali salt agglomeration on the catalyst surface and the potassium compound, in the form of Kσ+ (Naσ+) with σ varying from 0.5 to 0, which resulted in an indirect electrostatic interaction between the oxygen atom of adsorbed CO2 and the active Ni metal. Compared with the high conversion of CH4 (above 80%) and tar (above 90%) and high H2, CO content (> 40 vol %) in tail gas after 60 h real biomass fuel gas reforming operation in the pilot-scale integrated gasification/reforming system, it could be concluded that far lower K and Na content in monolithic catalyst in the reformer, caused by real fly ash, was the main reason. And the reforming process in the pilot-scale plant was tested to be feasible for produce biomass synthesis gas for the following liquid fuel synthesis in pilot-scale or large-scale and the utilization of monolithic catalyst was applicable because of its low pressure drop of the reformer and excellent performance. In the same time, a longer reforming operation and direct dimethyl ether synthesis process are ongoing by our group. The synthesis reaction heat and tail gas combustion will be applied to generate heat for the reformer as much as possible.

0.07 0.07

0.07

0.11

0.06

0.07

0.07

0.07

0.06 0.06

0.04 0.03 0.04 0.04 0.06 0.07 0.05 0.05 0.26 0.06 0.07

0.07

bottom middle top bottom middle top top middle bottom middle bottom

top

ing performance. Because of the low alkali content in practical reforming process, CH4 conversion, H2, and CO yields kept high after 60 h operation in Figure 7, which is different from the quick deactivation of deposited MC by high K amount in Figure 6.

0.05 0.05

This work was supported by the National Key Basic Research Program 973 Project founded by MOST of China (Project No. 2007CB210207) and the National High Technology Research and Development Program of China via 863 plan (Project No. 2007AA05Z416) and the Natural Science Foundation of Guangdong Province China (Project No. 9451007006004086). Literature Cited

The positions were shown in Figure 2. a

0.07 0.07

0.07 0.08

0.08

0.08 0.06

0.05 0.05 0.06 0.06

0.06 0.07

0.06

0.05 0.05

top bottom middle top

bottom middle top

bottom

middle

Acknowledgment

mcp packed position in the tube K content (wt %) Na content (wt %)

tube position in the reformer

left A

left A

left A

left B

left B

left intermediate intermediate intermediate intermediate intermediate intermediate intermediate intermediate intermediate right B A A A B B B C C C A

right right right A A B

right right B B

Ind. Eng. Chem. Res., Vol. 49, No. 7, 2010

Table 4. K and Na Content in MCP Packed at Different Positions of the Reformera

3182

(1) Hamelinck, C. N.; Faaij, A. P. C. Outlook for advanced biofuels. Energy Policy 2006, 17, 3268. (2) Li, Y.; Wang, T.; Yin, X.; Wu, C.; Ma, L.; Li, H.; Sun, L. Design and operation of integrated pilot-scale dimethyl ether synthesis system via pyrolysis/gasification of corncob. Fuel 2009, 11, 2181. (3) Corella, J.; Caballero, M. A.; Aznar, M.-P.; Brage, C. Two advanced models for the kinetics of the variation of the tar composition in its catalytic elimination in biomass gasification. Ind. Eng. Chem. Res. 2003, 13, 3001. (4) Miyazawa, T.; Kimura, T.; Nishikawa, J.; Kado, S.; Kunimori, K.; Tomishige, K. Catalytic performance of supported Ni catalysts in partial oxidation and steam reforming of tar derived from the pyrolysis of wood biomass. Catal. Today 2006, 1-4, 254. (5) Chen, X.; Gould, B. D.; Schwank, J. W. n-Dodecane reforming over monolith-based Ni catalysts: SEM study of axial carbon distribution profile. Appl. Catal., A 2009, 2, 137. (6) Zhang, W. D.; Liu, B. S.; Zhan, Y. P.; Tian, Y. L. Syngas production via CO2 reforming of methane over Sm2O3-La2O3-supported Ni catalyst. Ind. Eng. Chem. Res. 2009, 16, 7498. (7) Liao, C.; Wu, C.; Yan, Y. The characteristics of inorganic elements in ashes from a 1 MW CFB biomass gasification power generation plant. Fuel Process. Technol. 2007, 2, 149.

Ind. Eng. Chem. Res., Vol. 49, No. 7, 2010 (8) Olsson, J. G.; Pettersson, J. B. C.; Padban, N.; Bjerle, I. Alkali metal emission from filter ash and fluidized bed material from PFB gasification of biomass. Energy Fuels 1998, 3, 626. (9) Uberoi, M.; Shadman, F. Simultaneous condensation and reaction of metal compound vapors in porous solids. Ind. Eng. Chem. Res. 2002, 4, 624. (10) Christensen, K. A.; Stenholm, M.; Livbjerg, H. The formation of submicron aerosol particles, HCl and SO2 in straw-fired boilers. J. Aerosol Sci. 1998, 4, 421. (11) Einvall, J.; Albertazzi, S.; Hulteberg, C.; Malik, A.; Basile, F.; Larsson, A.-C.; Brandin, J.; Sanati, M. Investigation of reforming catalyst deactivation by exposure to fly ash from biomass gasification in laboratory scale. Energy Fuels 2007, 5, 2481. ¨ hman, M.; Kjellstro¨m, B. Alkali retention/ (12) Gabra, M.; Nordin, A.; O separation during bagasse gasification: a comparison between a fluidised bed and a cyclone gasifier. Biomass Bioenergy 2001, 6, 461. (13) Yang, H. Q.; Xu, Z. H.; Fan, M. H.; Bland, A. E.; Judkins, R. R. Adsorbents for capturing mercury in coal-fired boiler flue gas. J. Hazard. Mater. 2007, 1-2, 1. (14) Albertazzi, S.; Basile, F.; Brandin, J.; Einvall, J.; Fornasari, G.; Hulteberg, C.; Sanati, M.; Trifiro, F.; Vaccari, A. Pt-Rh/MgAl(O) catalyst for the upgrading of biomass-generated synthesis gases. Energy Fuels 2008, 1, 573. (15) Toledo, J. M.; Corella, J.; Molina, G. Catalytic hot gas cleaning with monoliths in biomass gasification in fluidized beds. 4. Performance of an advanced, second-generation, two-layers-based monolithic reactor. Ind. Eng. Chem. Res. 2006, 4, 1389. (16) Corella, J.; Toledo, J. M.; Padilla, R. Catalytic hot gas cleaning with monoliths in biomass gasification in fluidized beds. 2. Modeling of the monolithic reactor. Ind. Eng. Chem. Res. 2004, 26, 8207. (17) Wang, C. G.; Wang, T. J.; Ma, L. L.; Gao, Y.; Wu, C. Z. Partial oxidation reforming of biomass fuel gas over nickel-based monolithic catalyst with naphthalene as model compound. Korean J. Chem. Eng. 2008, 4, 738. (18) JCPDS Powder Diffraction File, International Centre for Diffraction Data. , 2003. (19) Keown, D. M.; Hayashi, J.-i.; Li, C.-Z. Effects of volatile-char interactions on the volatilisation of alkali and alkaline earth metallic species during the pyrolysis of biomass. Fuel 2008, 7, 1187.

3183

(20) Salo, K.; Mojtahedi, W. Fate of alkali and trace metals in biomass gasification. Biomass Bioenergy 1998, 3, 263. (21) Larsson, A.-C.; Einvall, J.; Andersson, A.; Sanati, M. Targeting by comparison with laboratory experiments the SCR catalyst deactivation process by potassium and zinc salts in a large-scale biomass combustion boiler. Energy Fuels 2006, 4, 1398. (22) Devi, T. G.; Kannan, M. P. X-ray diffraction (XRD) studies on the chemical states of some metal species in cellulosic chars and the Ellingham diagrams. Energy Fuels 2007, 2, 596. (23) Boman, C.; Nordin, A.; Bostrom, D.; Ohman, M. Characterization of inorganic particulate matter from residential combustion of pelletized biomass fuels. Energy Fuels 2003, 2, 338. (24) Moradi, F.; Brandin, J.; Sohrabi, M.; Faghihi, M.; Sanati, M. Deactivation of oxidation and SCR catalysts used in flue gas cleaning by exposure to aerosols of high- and low melting point salts, potassium salts and zinc chloride. Appl. Catal., B 2003, 1, 65. (25) Pitchon, V.; Gallezot, P.; Nicot, C.; Praliaud, H. Study by analytical electron microscopy of the potassium distribution on silica-supported nickel and palladium catalysts. Appl. Catal. 1989, 2, 357. (26) Praliaud, H.; Tardy, B.; Bertolini, J. C.; Martin, G. A.; Claudio Morterra, A. Z. a. G. C. Comparison of alkali promoters on silica-supported Ni and Ni (111): Chemical state, localization, and range of the promoter effect. Stud. Surf. Sci. Catal. 1989, 749. (27) Rostrup-Nielsen, J. R. Sulfur-passivated nickel catalysts for carbonfree steam reforming of methane. J. Catal. 1984, 1, 31. (28) Glassman, I. Chemical thermodynamics and flame temperatures. In Combustion, 3rd ed.; Academic Press: San Diego,CA, 1996; pp 1. (29) Diaz, A.; Acosta, D. R.; Odriozola, J. A.; Montes, M. Characterization of alkali-doped Ni/SiO2 catalysts. J. Phys. Chem. 1997, 10, 1782. (30) Guo, J.; Lou, H.; Zhao, H.; Chai, D.; Zheng, X. Dry reforming of methane over nickel catalysts supported on magnesium aluminate spinels. Appl. Catal., A 2004, 1-2, 75. (31) Wang, T.; Chang, J.; Cui, X.; Zhang, Q.; Fu, Y. Reforming of raw fuel gas from biomass gasification to syngas over highly stable nickelmagnesium solid solution catalysts. Fuel Process. Technol. 2006, 5, 421.

ReceiVed for reView August 31, 2009 ReVised manuscript receiVed January 22, 2010 Accepted February 3, 2010 IE901370W