Catalyst loss and retention during alkali-catalysed carbon gasification in CO2

Carbon Vol.2Y. No. 7, pp Printedin GreatBritain.

929-941.

0008-622319 I $3.00+ .11(1 Copyright0 1991PergamonPress plc

1991

CATALYST LOSS AND RETENTION DURING ALKALI-CATALYSED CARBON GASIFICATION IN CO2 Department

R. MEIJER, M. WEEDA, F. KAPTEIJN, and J. A. MOULIJN of Chemical Engineering, University of Amsterdam, Nieuwe Achtergracht 1018 WV Amsterdam, The Netherlands

166,

(Received 2 July 1990; accepted in revised form 5 November 1990) Abstract-Characterization of the catalyst/carbon samples by outgassing yielded reproducible and consistent results, provided the amounts of CO1 and CO released are related to the amount of alkali metal actually present. The catalytically active cluster is anchored to the carbon by phenolate groups and is capable of chemisorbing one CO, molecule per three to five alkali metal atoms. Between KIC=0.02 to 0.04 the alkali cluster size increases, but the specific gasification rate, rK, remains unaffected. During potassium catalysed CO, gasification initially 20% to 40% of the potassium was lost; the rest remained present throughout the gasification process. The amount of catalyst that can be stabilised on the carbon increases in the order Na < K < Cs and depends on the amount of oxygen in the carbon. The latter is reduced by heat treatment of the applied carbon, and hence a smaller amount of catalyst can be stabilised. The differences between the alkali metals arise from their ability to migrate into the carbon matrix and their interaction with lattice oxygen at which they are trapped. This migration of Cs and K into the carbon matrix explains the burn-off profiles of catalyst/carbon samples after TPD treatment. Initially only part of the catalyst is active for gasification, but gradually more catalyst becomes available, resulting in an increasing gasification rate. This behaviour is absent for sodium. Key Words-Alkali-catalysed,

carbon gasification, carbon reactivity

1. INTRODUCTION It is well known that addition of alkali metals to carbonaceousmaterials hasacatalyticeffectontheirgasification. A generally postulated mechanism[l-51 for catalytic gasification consists of an oxidation-reduction cycle in which oxygen is transferred to the carbon matrix through the catalytically active alkali species (l),(2), followed by a rate determining decomposition of the oxidised carbon site (3), producing CO. In this process, [*] and [0-*] represent the “empty” and oxygen containing alkali species active for oxygen transfer to a free carbon site C, resulting in a surface oxygen intermediate C(0). Kinetic measurements have shown that addition of alkali metal enhances the steady state concentration of C(0) complexes, rather than altering the reaction pathway[3,6-81. Furthermore, several studies[2,4,6,9121 have shown that under (sub)gasification conditions the active alkali species can contain considerable amounts of chemisorbed CO, and are capable of catalysing oxygen exchange reactions. In most studies published rates of gasification and oxygen exchange and results of temperature-programmed desorption (TPD) experiments are related to the amount of alkali metal initially present on the sample. However, at high temperatures, during TPD or gasification, catalyst can be lost. It is obvious that the amount of alkali metal actually present on the carbon surface is therefore an important parameter in the interpretation of experimental data related to CAR

catalysed gasification. co,

+ * +

o-*

+ c, + C(0) *

I

co

2

+ o-*

C(0) co

+ *

(I) (2) (3)

The aim of this study was to examine if, how, and to what extent catalyst is lost during alkali-carbonate (Na,K,Cs) catalysed gasification in CO, and heat treatment of a model carbon, and how this affects the activity and stoichiometry of the remaining catalytic sites. This was achieved by determination of the alkali metal content after reactivity, CO, chemisorption and temperature-programmed desorption (TPD) experiments.

2. EXPERIMENTAL 2.1. Sample preparation The model carbon used in this study is Norit RX1 Extra, an acid washed, steam activated peat char with a high specific surface area (1100 m2 g-’ (CO1 adsorption at 273 K), 1500 mz . g-’ (N, adsorption at 77 K)). Catalyst was added to the carbon (d, = 0.25-0.6 mm) by pore volume impregnation with an aqueous solution of alkali carbonate. In Table 1 the preparation data are listed of the catalyst/carbon

29:7-E 929

930 Table

R. MEIJERet al. 1. Preparation data of the samples investigated

catalyst/carbon

experiments simultaneous GC (FID) analysis was performed. The sample mass curve is corrected for changes in flow and buoyancy forces at all experimental stages. 2.3. Experimental procedure In this study two experimental schemes were used (Fig. l), resulting in the following seven different procedures (stages between brackets are optional, SSG refers to Steady State Gasification and TPD to temperature-programmed desorption):

type

that have been studied. The initial catalyst loading is expressed as the atomic alkali metal to carbon (M/C), ratio. The Na and K content were determined (in duplo) by inductive coupled plasma atomic emission spectroscopy (ICP-AES), the Cs content by atomic absorption spectroscopy (AAS). The alkali metal could be quantitatively recovered from the catalyst/carbon residue by leaching it with an aqueous 2% HN03 solution. Before impregnation, the properties of the carbon have been varied by heat treatment, in an inert atmosphere (Ar), at temperatures up to 2000 K. samples

2.2. Apparatus The data obtained in this study were generated in a thermobalance (TGA) and a fixed-bed flow reactor (FBR). Basically the FBR apparatus, described in more detail elsewhere[l3], consists of two gas-mixing sections in which the desired flow mixture is generated, an oven (T,, = 1273 K) containing a quartz glass reactor (I.D. = 3-5 mm, P,,, = 0.15 MPa) with the carbon sample and a gas chromatograph (GC, He carrier, TCD) and mass spectrometer (MS, Leybold-Heraeus Q200) for gas analysis. All gases used are of HP or UHP grade and are purified (0, and/or H,O removal) before they are fed to the reactor. Mass loss curves (TG) and the derivative signals (DTG) of the catalyst/carbon samples during the different experimental schemes were obtained in a thermobalance (Setaram TG 85)[14] under conditions similar to those in the fixed-bed reactor. The TGA flow system consists of two gas inlets to the reactor: the main inlet through which the carrier gas (Ar) is supplied and a side inlet, just above the catalyst/carbon sample, through which CO1 is added. This set-up was chosen to ensure a fast response after addition or removal of CO, to or from the gas phase. The catalyst/carbon sample (typically 50 mg) was held in a ceramic cup placed in the center of the oven (T,,,, = 1700 K). The sample mass and oven temperature were monitored by a PLC (umac 5000) through which data were stored for further quantitative analysis. During TPD and CO* exposure

la lb 1C Id 2a 2b 2c

treatment SSG SSGSSGSSGTPD TPDTPD-

TPDCO2 exposure[CO1 exposure]-

SSG TPD TPD-

CO2 exposure-TPD

SSGCO2 exposure-

[TPD] TPD-

[SSG]

In the FBR gasification rates and CO and CO* desorption patterns during TPD were obtained. Experiments in the TGA were performed to monitor changes in the sample mass during CO, exposure and temperature-programmed desorption (TPD) experiments. At the end of each FBR and TG experiment the sample was cooled to room temperature and the alkali metal content of the residue was determined.

He j

1200

CO,

/ He

I

;CO,:

,

HB

p0.j

H.3

-

,oocl-

Fig. 1. Experimental procedures: type 1 (upper part) and type 2 (lower part). SSG and TPD refer to Steady State Gasification and Temperature Programmed Desorption. The arrows indicate the points in the experiments at which alkali metal content measurements were performed.

931

Catalysed carbon gasification in CO, 2.4. Fixed-bed reactor

The first step in each experiment was drying of the catalyst/carbon sample (typically 50 to 100 mg) in situ (T = 473 K, He). Partial (approximately 25% burn-off) gasification of the catalyst/carbon sample was performed in 20 kmol CO2 . s-r (T = 1000 K; pC0, = 0.12 MPa). The steady state gasification (SSG) rate in the FBR is based on either the actual amount of alkali metal (rM) or on the initial amount of carbon (r,) present, and expressed as (mol carbon gasified) . (mol M, or C,))’ . s-l. The gasification rate as a function of the percentage of carbon burnoff is referred to as the “burn-off profile” of the catalyst/carbon sample. Gasification rates of catalyst/carbon samples that have been heat treated according to one of the experimental schemes are referred to as “used” samples. TPD patterns in the FBR are obtained in a helium flow of 20 bmol . s-l with a heating rate of 10 K . min’ up to 1200 K, followed by an isothermal period of 30 minutes at 1200 K. The rate of CO and CO2 release is plotted as a function of temperature and expressed as (mol CO, desorbed) . (mol M, present))’ s-l. After partial (approximately 25% burn-off) gasification in CO, the sample was exposed to helium at 1000 K and either cooled to room temperature for alkali metal analysis or cooled to 673 K followed by either TPD and gasification of CO, exposure and TPD. In some cases a second CO, exposure and TPD were performed. After TPD of a fresh sample the sample was cooled to either room temperature for alkali metal analysis or to 673, 873, or 1000 K for CO* exposure, followed by a second TPD.

2.5. Thermobalance Partial gasification of the catalyst/carbon sample was performed in a CO*, Ar flow of 150 ml . min’ (STP, 33 ~01% CO,). TPD patterns, with the same temperature program as applied in the FBR, were obtained in an argon flow of 100 ml . min’. The effect on the sample mass of CO* exposure at 673 K was studied by introduction and removal of CO, from the argon gas phase. 3. RESULTS

3.1. Gasification reactivity vs. catalyst loading In Fig. 2 the steady state gasification rates (pCOz = 0.12 MPa, T = 1000 K) as a function of burn-off (“burn-off profiles”) are shown for fresh catalyst/carbon samples. During the first stages of burn-off due to removal of oxygen present on the sample, maximum gasification rate is observed. All samples exhibit a constant and reproducible gasification rate between approximately 20% and 50% burn-off. At higher burn-off due to pore plugging and catalyst loss a strong decrease in gasification rate is observed. Comparison of steady state gasification rates of fresh and used catalyst/carbon samples and TPD experiments are therefore performed in the 20 to 50% burn-off region. The steady state gasification rate for fresh and used catalyst/carbon samples at different stages in the experimental schemes is shown in Fig. 3. For fresh samples (open symbols) the observed gasification rate rK at 25% burn-off shows a constant value between (K/C), = 0.02 to 0.04, indicating that over this K/C range the activity, per potassium atom present. is constant. After different treatments (solid

3.0 I\ I I

Ice *

\

’

2.0

b r \

1 .o L’

0.0 40

%

60

80

Burn-off

Fig. 2. Steady state gasification rate in CO, of fresh M,COJNorit RX1 samples (7’ = 1000 K, pCQ = 0.12 MPa) is a function of burn-off(%). --------- (Na/C), = 0.019; -(K/C), = 0.019; --(WC), = 0.019); - - - (K/C), = 0.043).

R. MEIJERet al.

932

80

60

7 co :

40

z \

20

L’

(K/C), Fig. 3. Steady state CO, gasification rates per initial amount of carbon (r.) and actual amount of potassium (rK) of fresh (open symbols) and used (solid symbols) K,COJNorit RX1 samples (T = 1000

K, pC0, = 0.12 MPa) as a function of the actual K/C ratio. 0 (K/C), = 0.019; v (K/C), = 0.031; A (K/C), = 0.043.

symbols), nearly independent of the initial loading, the observed gasification rate r, corresponds well with rates of fresh samples. 3.2. Burn-off behaviour of fresh and heat-treated samples

In Fig. 4 burn-off profiles are compared of fresh and used catalyst/carbon samples that have experienced different treatment. In the case of Na/Ci = 0.019 (1,B) and K/C, = 0.043 (III, B and C) the steady state gasification rate has decreased after TPD up to 1200 K. After TPD K/C, = 0.019 (II, B and C) shows a gasification rate initially similar to that of a fresh sample and shows a small increase in rate at higher burn-off. For Cs/Ci = 0.019 after TPD (IV, B and C) a different behaviour is observed. Initially the gasification rate is low, but with increasing burn-off a slowly increasing rate to a level similar to that of a fresh cesium catalyst/carbon sample (IV,A) is observed. A fresh catalyst/carbon sample (V,A), of which the carbon has experienced a heat treatment in argon (T = 1650 K) before impregnation, shows a continuously decreasing gasification rate, whereas after TPD a lower but more constant rate is observed (V,B). 3.3. Catalyst loss Figure 5 gives the fraction of potassium actually present as a function of burn-off for samples with different initial potassium loading. It was found that already at low burn-off about 20% to 40% of the potassium is lost, independent of the initial potassium loading. This amount remains further fairly constant over the burn-off range investigated (0% to 60%).

k--_-A

I m

,---____________-----------cy

II

Ill

IV

V

0

20

40

60

80

% Burn-off Fig. 4. Burn-off profiles during gasification in CO, of fresh and used M&OJNorit RX1 samples (T = 1000 K, pC0, = 0.12 MPa) after different treatment (SSG underlined is shown). I. (Na/C) = 0.019; II. (K/C), = 0.019; III. (K/C), = 0.043; IV. (Cs/C), = 0.019; V. (K/C), = 0.019 carbon treated at 1650 K. A. SSG; B. TPD - CO2 @ 673 K - TPD - XSG; C. TPD - SSG - TPD - SSG; D. TPD - SSG TPD - SSG.

Catalysed

1.0

-

0.8

-____-_______“;-__ A

carbon

gasification

933

in CO?

+

ii\ Yrn

____ +_ _--____ +-- _-_-

0.6

A A -.__!_.__.____________________________

0.4

-

0.2

I

I

I

20

40

60

0.0 0

i+ l .

%

80

Burn-off

Fig. 5. Fraction of potassium actually present (K,IK,) as a function of burn-off during alkali catalysed gasification in CO, (T = 100 K, pC0, = 0.12 MPa). + (K/C), = 0.019; A various loadings (K/C), = 0.0054 - 0.043.

In Fig. 6 the K/C ratio based on the amount of potassium actually present (ICP-AES values) is plotted for (K/C), = 0.019, 0.031, and 0.043 samples after different treatment. Clearly, during TPD extensive potassium loss takes place on catalyst/carbon samples with initially high potassium loadings. After a second TPD up to 1200 K the (K/C), ratios, independent of the initial K/C ratio, are in the range of 0.018 to 0.022. After TPD of samples prepared with the heat-treated carbons (not shown), lower (K/C), values are obtained (0.007-0.010).

0.06 0.04 m G 2 0.02

6. Actual (K/C) 0.043 samples after TPD (type shaded

ratio

(K/C),

TPD (type lD, 2B the data scatter.

and 2C). The

3.4. Outgassing patterns In Table the total amounts of CO and CO1 released in experiments after different are listed. TPD patterns of fresh catalyst/carbon samples 7~) are and are fairly similar all samples investigated. They exhibit K) CO, desorption, by a two-peak CO TPD patterns of of which the carbon, has been heat-treated show CO desorption of the carbon, the peak maximum to higher temperature. TPD patterns obtained after in CO2 at 1000 K and subsequent cooling in helium, for potassium catalyst/carbon samples with different initial loading, show only one CO peak and no significant COz desorption (Fig. 8~). In general, the maximum CO desorption shifts to lower temperature with increasing potassium loading and the CO/K, desorption ratio decreases in the same order (Table 2). In Fig. 8B TPD patterns are shown after SSG and cooling in helium followed by CO2 exposure at 673 K. The low-temperature CO? desorption is associated with this CO2 exposure, whereas one or two CO desorption peaks are observed at high temperature, depending on the initial potassium loading. If Fig. 8B is compared with TPD directly after SSG and cooling in helium (Fig. 8~), it is clear that the first CO desorption peak is induced by CO, exposure at 673 K, whereas the second CO desorption peak is not affected. TPD patterns after TPD and CO, exposure at 673 K also show low-temperature CO2 desorption from the alkali cluster (Fig. 8c, note different scale). The

934

R. MEIJER et

al.

Table 2. Total amount of CO, and CO released per alkali metal actually present (mol/mol) in TPD experiments after

A

different treatment. Underlined is the actual TPD, the rest indicates the sample history

B

C

amount of CO or CO, desorbed during this second TPD has been corrected for potassium loss during the first TPD. From Fig. 8c it is clear that the TPD patterns have become comparable, independent of the initial potassium loading. However, the amount of CO desorbed per K atom present during this TPD is considerably lower than in a TPD after gasification (Table 2).

1.2

A 0.8 v) Y”

0.4

‘ij

E . ”

0.0 1.2

II

I

8

z

0.8 -

E

400

1200

800

T/K Fig. 7. TPD patterns obtained in the FBR of fresh (A) CO). and heat treated (B) samples (-------- = COz; -= A. (M/C), = 0.019 + Na; 0 K; n Cs B. (K/C), = 0.019 heat treated at 1650 K (A) and 2000 K (V.

T/K of Fig. 8. TPD patterns after different treatment K,COJNorit RX1 samples with different initial potassium loading. After SSG the catalyst/carbon sample is cooled in helium. (------ = CO1; -= CO; 0 (K/C), = 0.019; V (K/C), = 0.031; A (K/C), = 0.043). A. SSG - TPD; B. SSG - CO, @ 673 K - TPD; C. SSG - CO2 @ 673 K TPD - CO2 @ 673 K - TPD.

In Fig. 9 outgassing patterns of differently treated potassium catalyst/carbon samples ((K/C), = 0.019) are compared. A fresh sample shows extensive lowtemperature CO2 desorption followed by a two-peak CO desorption at high temperature. If CO, exposure is performed at 873 K, the total amount of CO/K, desorbed increases significantly compared to CO, exposure at 673 K, approaching the CO/K, ratio desorbed (CO/K, = 0.92) after partial gasification (25% burn-off) at 1000 K and cooling in helium (Table 2). If after TPD and SSG a second TPD is performed, the same CO desorption pattern is obtained as by TPD after only SSG. However, during TPD after SSG and cooling in CO, (Fig. 9), besides lowtemperature CO, desorption (not shown), a twopeak CO desorption similar as for a fresh (K/C), = 0.019 sample is observed. For the differently treated cesium and sodium catalyst/carbon samples that have been investigated ((M/C), = 0.019) a similar desorption behaviour (Fig. 10) is observed as for potassium ((K/C), = 0.019). For (Cs/C), = 0.019 the CO desorption pattern of a fresh sample is quite comparable with the CO desorption pattern obtained after TPD and partial gasification. TPD after direct SSG for (CS/C)~ = 0.019 shows a higher CO desorption compared with TPD after TPD and SSG.

Catalysed carbon gasification in CO,

935

T/K Fig. 9. TPD patterns for K,COJNorit RX1 Extra (K/C), = 0.019 samples (------- = CO,; -= CO). Sample history: l fresh sample; A TPD - CO, @ 673 K - TPD; V TPD - CO2 @ 873 K - TPD; 0 SSG(CO,, @ 1000 K, 25% B.0) cool in He - TPD; A SSG(CO,, @ 1000 K, 25% B.O.) cool in CO, - TPD.

3.5. Thermobalance 3.5 . 1. Heat trearment.

1.0

0.5

0.0

400

1200

800

T/K Fig. 10. TPD patterns for Na&O,/and Cs&O,/Norit RX1 Extra (M/C), = 0.019 samples (-------- = CO,; -= CO). Sample history: 0 fresh sample; A TPD CO, @ 673 K - TPD; 0 TPD - SSG(CO,, @ 1000 K, 25% B.O.) cool in CO, - TPD; A SSG(CO,, @ 1000 K, 25% B.O.) cool in CO* - TPD.

Thermogravimetric measurements show that the amount of oxygen present in Norit can be decreased by increasing the heat treatment temperature in an inert (Ar) atmosphere. Norit that has been heat treated at 2000 K in argon, shows no high-temperature mass loss (CO desorption) on reheating up to 1700 K in the thermobalante, even after being exposed to air at room temperature. The carbon sample heat-treated at 1373 K and a fresh sample both show a high temperature (T < 1400 K) mass loss. The difference in mass loss with the sample heat-treated at 2000 K corresponds to 36 pmol CO desorption (CO/C = 8.6 * 10m3). Carbon samples heat-treated up to 16.50 and 2000 K still have high surface areas of 1016 and 752 m* . g-’ (CO, adsorption at 273 K). 3.5.2. CO2 exposure. Experiments performed in the thermobalance with catalyst/carbon samples that have experienced different treatment show an instantaneous mass increase on CO, exposure at 673 K, followed by a very slowly increasing sample mass (Fig. 11). Independent of the initial loading and alkali metal, an increase in mass is observed, corresponding to 14-16 gram per mol alkali metal actually present (Fig. 11~). Figure 11~ shows the difference in mass gain, on introduction of CO2 at 673 K, before (I) and after (II) TPD up to 1200 K of a (K/C), = 0.043 sample. Upon removal of CO, all samples exhibit a slow loss of sample mass due to CO2 desorption. 3.5.3. TPD. Mass loss patterns obtained during heat treatment (TPD) in the thermobalance with simultaneous GC analysis provide additional information on the amount of alkali metal lost from the

R. MEIJER et al.

936 20

The values of the percentage and rate of potassium loss during alkali-catalysed gasification or heat treat7 =a 16 ment in an inert atmosphere reported in the literature differ enormously[l0,15-191. The amount of alkali metal that can be stabilised on the carbon is zE 12 dependent on the characteristics of the applied carbon. The oxygen content of the carbon, which is CD 6 strongly correlated with the heat treatment temperature, is an important parameter. The rate at which ; 4 alkali metal is lost is, however, strongly dependent a on the experimental conditions (apparatus, T, P, 0 flow rate, gas phase) and the method of impregna4 B tion[ 151. Ar Ar i CO, / The results show that the amount of potassium 3that can be stabilised on the carbon (K/C = 0.0180.022) during TPD up to 1200 K is nearly indepenI F dent of the initial potassium loading. In agreement 2with results reported in the literature[l7,20], sodium shows less interaction with the carbon matrix com; -----------___J_!______. *__--pared to potassium, resulting in a smaller amount of a I,' : sodium ((Na/C), = 0.008-0.010) that can be stabilised on the carbon. In contrast with potassium and I OL---! sodium, no cesium loss is observed after heat treat1800 0 600 1200 ment (TPD) of a cesium catalyst/carbon sample ((Cs/C) = 0.019). t / s Furthermore, our data show that after heat treatment (1650-2000 K) of the carbon sample the Fig. 11. Mass changes upon CO2 exposure at 673 K after different treatment of M,CO,/Norit RX1 samples. amount of potassium that can be stabilised on the A. SSG - TPD - CO, exp. --TPD. I. (Na/C), = 6.019; II. carbon at 1200 K is considerably lower than for un(K/0; = 0.019: III. (K/Cl. = 0.043. treated carbon samples (K/C = 0.007 to 0.010 vs. B. SSG -‘kO,‘kxp. (I) i TPD‘ - C’b, exp. (II) - TPD. 0.018 to 0.022). (K/C), = 0.043. Removal of lattice oxygen by heat treatment of the carbon shows a strong correlation with the amount of potassium that can be stabilised on the catalyst/carbon sample through evaporation. In Fig. 12 the mass loss curves during TPD are shown, as carbon after TPD up to 1200 K. The total amount of oxygen removed (-36 pmol) during heat treatcalculated from TG and GC analysis, after partial ment of the carbon up to 2000 K is on the same order gasification (25% burn-off) of a (K/C), = 0.043 sample at 1000 K. Above 1100 K extensive mass loss as the difference in absolute amount of potassium that can be stabilised during TPD between “fresh” takes place that can not be attributed solely to CO, and heat-treated (K/C), = 0.019 samples (K,,,, - 40 or CO desorption, but must be associated with evappmol) . oration of alkali metal (oxide). In Table 3 the total Loss of alkali metal during heat treatment can be mass loss during TPD after SSG, calculated from due to evaporation of alkali metal or an oxidic form. TG and GC analysis, is shown for all samples investigated. From the difference in total mass loss The results in Table 3 suggest that potassium is lost in the metallic state. between TG and GC analysis the amount of alkali Figure 12 shows that during TPD of an (K/C), = metal loss (pmol) is calculated assuming alkali evap0.043 sample, alkali metal loss is observed at high oration as M, MO, or M,O. This is compared with temperature. The amount of alkali metal loss, alkali metal loss values as determined by ICP and observed by TG analysis, is strongly dependent on AAS. the alkali metal and alkali metal loading (Table 3) and in good agreement with measurements of 4. DISCUSSION alkali metal content. Table 3 shows that for (CslC), = 0.019 and (K/C), = 0.019 the absolute 4.1. Catalyst loss amount of alkali metal loss is negligible compared During alkali-caltalysed gasification in CO*, using to (Na/C), = 0.019 and (K/C)i = 0.043. pore volume impregnated catalyst/carbon samples, Our results concerning the amount of alkali metal substantial catalyst loss (Fig. 5) at low burn-off is that can ultimately be stabilised by the carbon surobserved. This was also found by Sulimma et al. [15] face are consistent with results reported by Shadman in the case of steam gasification. They furthermore et a1.[17], but higher by a factor of ten. They observed a substantial increase in catalyst loss with reported K/C = 0.002 and Na/C = 0.0008 after increasing burn-off. co*

Ar

A

931

Catalysed carbon gasification in CO,

-6.0

400

I

I

800

1200 T/K

Fig. 12. Mass loss curves during TPD of a partially (25% B.O.) gasified K#ZO,/Norit RX1 (K/C), = 0.043 sample. (TG; -------- calculated from GC analysis).

complete reduction in nitrogen at 1073 K for an alkali-carbonate/Carbopack B sample (graphitized carbon, S, = 100 m2 . g-l). The fact that potassium and sodium show interac-

tion with the carbon, resulting in an amount that can be stabilised on the carbon at high temperature was already observed by Tromp and Cordfuncke[20]. They also concluded that potassium shows more interaction than sodium and that pretreatment of a catalyst/activated carbon sample at higher temperature resulted in a decrease in alkali carbon interaction. Saber et a1.[18] concluded that after heat treatment in an inert atmosphere at 1350 K the amount of potassium loss by volatilization increases due to a decreasing oxygen content of the carbon. Therefore, values reported in literature on the amount of alkali metal that can be retained on the carbon differ due to the different oxygen content of the applied carbons or chars. During heat treatment (TPD), no cesium loss is observed with the (Cs/C), = 0.019 sample. Evidently a saturation level is not reached with this loading. The fact that cesium shows a better interaction with the carbon compared to potassium or sodium can also be seen from the outgassing pattern

Table 3. Comparison of mass loss during TPD of MCOJ carbon samples after SSG (CO,, T = 1000 K) to approximately 25% burn-off (see text)

obtained with a fresh cesium catalyst/carbon sample. This is comparable to that obtained after TPD and SSG[21], indicating that cesium, during this first TPD, has sufficient interaction with the carbon to prevent evaporation. As for cesium, no significant potassium loss is observed for the (K/C), = 0.019 sample, during TPD up to 1200 K. For the same M/C ratio, sodium loss during TPD clearly is considerable, as for potassium at higher initial loadings ((K/C), = 0.031 and 0.043). This decrease in interaction with the carbon in the range Cs > K > Na corresponds well with results reported in the literature for Na and K[17,20] and is consistent with saturation levels of alkali metal loadings for gasification in COZ[lO]. It is suggested that stabilisation of the alkali metal during heat treatment is obtained through diffusion into the carbon matrix, where it interacts with stable lattice oxygen. This migration might be a diffusion of metallic species between the graphitic planes in the carbon, as in the case of intercalation, until it is trapped at an oxygen site. The differences in alkali metal interaction observed in this study correspond with the fact that Na forms only lower stage intercalation compounds (C,Na) compared to Cs and K (C,K, C,Cs)[22,23]. Hence, the amount of Na that can enter the carbon matrix is much lower, compared to K and Cs, thus Na mainly interacts with the easily accessible oxygen present on the reactive surface. The interaction of the alkali element with the carbon oxygen is a second factor that determines the amount of alkali metal that can be stabilised. Lang[24] showed that alkali phenolate formation is most favourable for Cs, followed by K and Na, in that order, in agreement with the results presented here.

938

R. MEIJERet al.

4.2. Gasification

Burn-off patterns of potassium and sodium catalyst/carbon samples, which have experienced substantial catalyst loss, all exhibit a burn-off behaviour similar to freshly gasified catalyst/carbon samples. From Fig. 3 it can be seen that in the K/C range investigated (0.02-0.04) the gasification activity, per potassium atom present, of fresh samples is constant. Gasification of used potassium catalyst/carbon samples that have experienced considerable catalyst loss, shows that the observed gasification rate per pmol potassium present (rK) is similar to that of fresh samples with the same (K/C), ratio. Sams et al.[19] concluded that independent of initial loading and heat treatment, the gasification rate is a unique function of the (K/C), ratio with the restriction that the K/C value is below the saturation level. At lower loadings (especially at K/C < 0.01) the reactivity per K atom is lower because the high state of dispersion of potassium on the carbon surface does not result in formation of alkali clusters, which represent the active centres in gasification[9,10]. Therefore, more isolated phenolate groups, with a higher stability, are formed[25]. At loadings above (K/C) = 0.043 a decrease in reactivity per K atom is observed, due to formation of inactive stable (bulk K2C03) crystalline material[25], For an alkali-carbonate/carbon black system Matsukata et al.[26-281 already observed Cs migration into the carbon bulk at 700 K and concluded that the ability of the alkali metal to sink into the bulk is associated with the stability of the alkali metal oxide and can only take place after the alkali metal is in the metallic form. The burn-off profile of heattreated (TPD) cesium catalyst/carbon samples supports the fact that during heat treatment all available cesium can be stabilised on the carbon (no cesium loss is observed) against evaporation, due to the presence of lattice oxygen (Fig. 4). The low initial gasification rate (after TPD) and its increase with increasing burn-off can be explained by the fact that at the start of gasification only part of the Cs is available for gasification. The rest is trapped in the carbon matrix and is inaccessible for the gasifying agent. During gasification, due to carbon consumption, gradually all Cs initially present becomes available, resulting in an increase in gasification rate up to the original level of a fresh cesium catalyst/carbon sample (Fig. 4-IV). The difference in initial gasification rate between fresh (2*10e4 s-l) and heat treated (1*10e4 s-l) (Cs/C), = 0.019 samples is similar as observed by Kapteijn et a1.[21]. However, they only gasified the sample up to low burn-off levels and therefore did not observe rate enhancement during gasification after TPD. Although less pronounced, this increase in gasification rate with increasing burn-off is also observed for potassium-containing samples. For a (K/C), = 0.019 ratio the sample that was first outgassed in a TPD even surpasses in activity that of a directly gasified sample at higher burn-off levels (Fig. 4-11).

This should be ascribed to the initial potassium loss in the latter case (Fig. 5), whereas by TPD treatment of a (K/C), = 0.019 sample all potassium is stored in the carbon matrix, which in a subsequent gasification will contribute to the activity at higher burnoff levels. This phenomenon of alkali metal migrating into the carbon matrix and gradually becoming available for catalytic action with increasing burnoff can explain numerous burn-off profiles that show a minimum in gasification activity at lower burn-off and a maximum at higher burn-off levels[29-311. 4.3. Outgassing patterns TPD patterns of fresh and partially gasified alkalicatalyst/carbon samples are in good agreement with those reported by Kapteijn et a1.[21]. The first CO desorption peak is due to secondary reaction(s) of desorbing CO, with the catalyst/carbon system, while the second is ascribed to decomposition of phenolate groups [9, 16,21,32,33], the obtained (CO/K), ratio approaching unity. After TPD it is assumed that the alkali cluster is completely reduced. TPD patterns in Fig. 8~ show that CO, exposure at 673 K after partial gasification and cooling in helium results in the same high-temperature CO desorption peak as with direct outgassing after gasification (Fig. 8~). However, after CO2 exposure at 673 K, during TPD a first CO desorption peak appears that can be attributed to secondary reaction(s) of desorbing CO,. After SSG and CO, exposure at 673 K, samples with different initial catalyst loading still show distinct differences in outgassing pattern (Fig. 8~). A second outgassing after CO, exposure at 673 K with the same sample shows, as expected from the alkali metal measurements (Fig. 6), CO2 and CO desorption patterns that have become nearly independent of the initial catalyst loading (Fig. 8~). TPD after TPD and CO, exposure shows a distinct difference in CO desorption pattern with increasing chemisorption temperature (Figs. 9 and 10). From Table 2 it is clear that the amount of CO desorbed after CO, exposure at 673 K (CO/K, = 0.3-0.4) is considerably lower than after gasification at 1000 K or CO* exposure at 873 K (CO/K, = 0.9). From this small amount of high temperature CO desorption it is concluded that the phenolate groups are only partially reformed at low CO, exposure temperature (673 K), whereas after TPD followed by CO2 exposure at 1000 K (SSG) these phenolate groups can be completely reformed. So, with CO, exposure at 673 K, only a small amount of oxygen is built in with direct interaction with the carbon. Ratcliffe[32] showed that on outgassing of a KCOJSpherocarb sample after ‘CO, chemisorption at 333 K, “CO desorption was detected between 700 and 800 K as a result of catalyst and carbon oxidation. This 13C0 desorption was succeeded at higher temperature by WO release, due to desorption of the oxidized carbon sites, with a maximum at 1030 K. This was also

939

Catalysed carbon gasification in CO,

alkali cluster is justified, because with pure carbon no significant mass change is observed. From Table 4, in which the corrected mass loss during TPD (CO, + 0 from CO) is compared with the mass gain during CO, exposure at 673 K, it can be seen that this balance is fairly good for all samples investigated. The suggestion that during the initial gasification stages less Cs is available after a TPD treatment is supported by the results of Fig. 10. After gasification of a fresh sample all Cs is present as phenolate groups (Table 2), resulting in a maximum high temperature CO desorption during TPD. After an initial TPD, part of the Cs is not yet available for gasification and not present in a phenolate form. The resulting CO desorption in the following TPD is consequently lower (Fig. 10, Table 2).

Table 4. Mass gain during COz-exposure at 673 K after SSG and (TPD) of M,CO,/Norit RX1 samples compared with the corrected mass loss in a subsequent TPD (see text) treatment

reported by Kelemen and Freund[6], and corresponds to the outgassing patterns in Figs. 8 to 10, in which clearly two ‘CO desorption peaks can be observed. The first is due to secondary reaction of desorbing CO, with the still partially reduced catalyst/carbon sample. The second CO desorption peak can be ascribed to two processes: desorption of oxidised carbon sites (C(0) + CO) and decomposition of phenolate groups (C-O-K -+ CO + K). From other experiments[l3] it was concluded that the decomposition of phenolate groups contributes most (>90%) to this high temperature CO desorption. TPD patterns obtained after TPD and CO, exposure at 673 to 1000 K show that more phenolate groups are restored with increasing CO,-exposure temperature. The amount of CO, and CO desorption observed during TPD must correlate with the mass gain observed during a previous CO, exposure at 673 K. Due to the fact that gasification takes place, this comparison can not be made for CO, exposure at 873 or 1000 K and a subsequent TPD. During CO, exposure at 673 K, only CO, chemisorption and some catalyst oxidation takes place. The total mass gain observed in TGA must therefore correlate with the total amount of CO2 and all oxygen from CO, desorbed in a subsequent TPD. The assumption that the change in sample mass can be attributed to (partial) catalyst oxidation and CO, chemisorption in the

4.4. Model of the alkali metal cluster at different experimental conditions The total amount of CO, and CO desorbed during TPD after partial gasification in CO, (Table 2) indicates that the catalytically active alkali cluster contains on the average one CO, molecule on every three to five potassium atoms present. These values are in good agreement with CO, chemisorption results reported in the literature[4,9,33,34]. Furthermore, the CO/K, ratio decreases with increasing potassium loading. Thus with increasing alkali metal loading the alkali cluster contains more potassium and chemisorbed CO,. Nevertheless, the specific gasification rate, rK, remains fairly constant with a K/C ratio between 0.02 and 0.04. The fact that the interaction between the alkali metal and the carbon oxygen, leading to the formation of phenolate groups, decreases in the order CS > K > Na[24], is reflected by a decrease in gasification activity, for samples with the same (M/C), ratio, in the same order[lO]. Additionally, the saturation level for phenolate formation increases from Na to Cs, which is in good agreement with results presented here. Dependent on the alkali metal. no co, K

KI

K

K

K I

K

K/C, Fig. 13. Schematic model for the active alkali cluster initially present and after heat treatment (TPD) as a function of the actual potassium loading. The dashed lines enclose the (K/C), range to which all catalyst/carbon samples with a higher (K/C), value are reduced after repeated TPD up to 1200 K.

R. MEIJER et al.

940

673

K CO,

1000

V K CO,

1200

K He

Fig. 14. Schematic model for the M$O,/carbon

additional phenolate groups are formed above a certain loading, but the alkali cluster size increases. Based on these and earlier results[lO], a schematic representation of the catalytically active alkali cluster, as a function of the potassium loading, is given in Fig. 13. This picture also accounts for the decreasing amount and size of the alkali clusters due to potassium loss after heat treatment (TPD) of potassium catalyst/carbon samples with high initial potassium loading. For potassium catalyst/carbon this will result in a (K/C), ratio of 0.018 to 0.022 independent of the initial catalyst loading, whereas for sodium a lower (0.008-0.01) and for cesium a higher value will be obtained. Finally, Fig. 14 is a schematic representation of the catalyst/carbon system under the different experimental conditions in this study. Under gasification conditions (1000 K, CO,) the catalyst/carbon system is fully oxidised. However, upon exposure to helium at 1000 K, CO2 will desorb from the alkali cluster and carbon oxygen complexes will decompose, producing CO[10,13]. A subsequent CO, exposure at 673 K restores the CO, chemisorption in the alkali cluster. This chemisorbed CO, is easily exchangeable with (labelled) gas phase CO,[9], but is stable in an inert gas at 673 K[9,14]. During a TPD in helium up to 1200 K the catalyst/carbon system is reduced and at high temperatures alkali metal loss can occur. This reduced catalyst/carbon sample can be (partially) reoxidised upon CO, exposure at (sub)gasification temperatures (673-1000 K). The amount of phenolate groups that are reformed is

673

system under the different experimental

K CO,

conditions.

strongly dependent on the CO, exposure temperature. Reoxidised alkali clusters show the same gasification activity and stoichiometry as fresh catalyst/carbon samples at similar loadings. 5. CONCLUSIONS In this study it has been shown that, provided the gasification activity and amounts of CO* and CO desorbed during outgassing are related to the amount of alkali metal actually present, reproducible and consistent results are obtained. The catalytically active alkali cluster is anchored to the carbon by phenolate groups and is capable of chemisorbing one CO, molecule per three to five alkali metal atoms. Between K/C = 0.02-0.04 the cluster size increases, but the specific rate, rK, remains constant. Upon outgassing, the chemisorbed CO, is released and the phenolate groups are decomposed. Exposure of this reduced sample to COZ at and above 1000 K completely restores the active alkali cluster. Below 1000 K the phenolate groups are only partly restored. During potassium-catalysed CO, gasification catalyst is lost in the initial stages to an extent of 20% to 40% of the initial amount. By heat treatment in an inert atmosphere (TPD up to 1200 K), depending on the alkali metal and loading, more catalyst can be lost, probably in the metallic state. The amount lost decreases in the order Na > K > Cs (- 0). The amount of catalyst that can be stabilised on the carbon during heat treatment (TPD) increases in the order

Na < K < Cs and depends

on the amount

of

Catalysed carbon gasification in CO? oxygen in the carbon. The latter is reduced by heat treatment of the applied carbon, and hence a lower amount of catalyst can be stabilised. The differences between the alkali elements arise from their ability to interact with lattice oxygen at which they are trapped. This migration of cesium and potassium into the carbon matrix explains the burn-off profiles of catalyst/carbon samples after a TPD treatment. Initially only part of the catalyst is active for gasification, but gradually more catalyst becomes available resulting in an increasing gasification rate. This behaviour is absent for sodium.

Acknowledgement-B.

van der Linden is gratefully acknowledged for performing the thermobalance experiments and the laboratory of Analytical Chemistry for their kind permission to use the ICP and AAS apparatus. These investigations have been executed within the framework of the Dutch National Coal Research Programme fNOK). which is managed by the Project Office-for Energy Research (NOVEM) and financed by the Ministry of Economic Affairs.

REFERENCES

1 F. Kapteijn and J. A. Moulijn, Fuel 62, 221 (1983). 2 M. B. Cerfontain, R. Meijer, and J. A. Mouhjn, Proc. Inl. Conf. Coal Science, Sydney, Australia, (1985), p.

277. 3. F. Kapteijn, 0. Peer, and J. A. Moulijn, Fuel65, 4. 5. 6. 7. 8.

9.

10.

1371 (1986). C. A. Mims and J. K. Pabst, J. Cafal. 107,209 (1987). J. A. Moulijn, M. B. Cerfontain, and F. Kapteijn, Fuel 63, 1043 (1984). S. R. Kelemen and H. Freund,/. Catal. 102,BO (1986). H. Freund. Fuel 64. 657 (1985). M. B. Cerfontain and J. ‘A. Moulijn, In Advances in Coal Chemistry (Edited by N. P. Vasilakos), pp. 233263 Theophrastus Publ. Co., Athens, Greece;‘(1988). M. B. Cerfontain. R. Meiier. F. Kaoteiin. and J. A. Mouhjn, J. Catal. 107, 173 (1987). ’ ’ M. B. Cerfontain, Ph.D. Thesis, Univ. ofAmsterdam, The Netherlands (1986).

941

11. K. J. Htittinger, B. Masling, and R. Minges, Fuel 65, 932 (1986).

12. C. A. Mims and J. K. Pabst, Prepr. Amer. Chem. Sot. Div. Fuel Chem. 25, 258 and 263 (1980). 13. R. Meijer, S. C. van Eck, F. Kapteijn, and J. A. Moulijn, in preparation. 14. R. Meijer, B. van der Linden, F. Kapteijn, and J. A. Moulijn, Fuel 70, 205 (1991). 1.5. A. Sulimma, H.-J. Miihlen, and K. H. van Heek, Proc. Inf. Conf. Coal Science, Maastricht, The Netherlands, (1987) p. 551. 16. C. A. Mims and J. K. Pabst. Fuel 62. 176 (1983). 17. F. Shadman, D. A. Sams, and W. A. Punjak, Fuel 66, 1658 (1987).

18. J. M. Saber, J. L. Falconer, and L. F. Brown, Fuel65, 1356 (1986). 19. D. A. Sams, T. Talverdian, and F. Shadman, Fuel 64, 1208 (1985). 20. P. J. J. Tromp and E. H. P. Cordfunke, Thermochim. Acfa 81,113 (1984). 21. F. Kapteijn, G. Abbel, and J. A. Mouhjn. Fuel 63, 1036 (1984). 22. M. A. M. Boersma, Cat. Rev. Sci. Eng. 10,243-280 (1974). 23. N. N. Greenwood and A. Earnshaw, Chemistry of the Elements, Pergamon Press, London (1984). R. J. Lang, Proc. Fund. Cat. Coal and Carbon Gas. Rolduc, The Netherlands (1986), p, 73. 25. F. Kapteijn, G. Abbel, and J. A. Moulijn, Ext. Abstr., 17th Bienn. Conf on Carbon, Lexington. KY, 1985, p. 781. 26. M. Matsukata, T. Fujikawa, E. Kikuchi, and Y. Morita, Energy & Fuels 2, 750 (1989). 27. M. Matsukata, T. Fujikawa, E. Kikuchi, and Y. Morita, Energy & Fuels j, 336 (1989). 28. M. Matsukata, Ph.D. Thesis, University of Waseda, Tokyo, Japan (1989). 29. R. T. Hamilton, D. A. Sams, and F. Shadman, Fuel, 63, 1008 (1984). 30. R. Meijer, H.-J. Miihlen, F. Kapteijn, and J. A. Moulijn, accepted for publication. 31. T. Wigmans. A. Hoogland, P. Tromp, and J. A. Moulijn, Carbon 21, 13 (1983). 32. C. T. Ratcliffe, Prepr. Amer. Chem. Sot. Div. Fuel Chem. 30, 330 (1985).

33. C. T. Ratcliffe, and S. N. Vaughn, Prepr. Amer. Chem. Sot. Div. Fuel Chem. 30, 304 (1985). 34. P. C. Koenig, R. G. Squires, and N. M. Laurendeau, J. Catal. 100, 228 (1986).