High surface area a-alumina. I.

Applied CatalysisA: General 111 (1994) N-173

High surface area cu-alumina. I. Adsorption properties and heats of adsorption of carbon monoxide, carbon dioxide, and ethylene Ching-Feng Mao, M. Albert mm Vannice* Department of

ChemicalEngineering, The Pennsylvania State University, 158 Fen&e Laboratory, Shortlidge Road, University Park. PA 16802, USA

( Received 12 November

1993. revised manuscript

received 10 January 1994)

Abstract

High surface area (HSA) a-aluminas have now been produced from both ground (G) and unground (U) diaspore. The porosities of these (HSA) cY-aluminaswere.determined using nitrogen adsorption at 77 K and mercury porosimetry. Volumetric and calorimetric studies of CO, COZ, and CzH4 adsorption at 300 K were performed to examine the surface properties of these (HSA) a-aluminas as well as a commercial low surface area (LSA) a-alumina so that comparisons could be made. A pore size distribution with a maximum at a radius of around 60 8, was observed with both HSA samples, but the unground sample possessed only a small quantity of mesopores ( < 0.01 cm3 g- ‘) while the ground sample had an appreciable mesopore volume (ca. 0.07 cm3 g- ‘). Carbon monoxide and ethylene adsorption was completely reversible, and the corresponding heats of adsorption for carbon monoxide adsorption on the unground and ground samples were 6.1 and 15.4 kcal mol-‘, respectively. The ethylene heat of adsorption on these two (HSA) cr-aluminas was 3 kcal mol-‘, which is one-third the values reported for n-alumina. Significant amounts of irreversible carbon dioxide adsorption occurred on the (HSA) o-alumina samples, i.e., 106 and 101 pmol gg ’ for the unground and ground samples, and the respective heats of adsorption were 21.4 and 16.0 kcal mol - I. However, on a normalized basis, the respective carbon dioxide coverages on the two (HSA) (Yaluminasof 8.2.10” and5.8. lOI moleculemP2 were lower than the value of 9.8. lOI for the (LSA) a-alumina. When compared to heat of adsorption (Q,) values for transition aluminas, the low value for carbon monoxide on the unground sample indicates weaker Lewis acid sites while the higher value for the ground sample indicates acid strengths more similar to transition aluminas. The Qadvalues for carbon dioxide adsorption are quite consistent with previous heats of adsorption on transition aluminas, and the concentration of basic sites on aluminas appears to have the following relationship: (LSA) a-Al203 > (HSA) a-Al,Os(U) > (HSA) (~-Al203(G) 3 y- and v-Al2O3.

Key words: alumina surface area *Corresponding

(a-);

adsorption

of carbon monoxide,

carbon dioxide and ethylene;

author.

0926-860X/94/$07.00 0 1994 Elsevier Science B.V. All rights reserved Pzn,no,L_PLrlY ,az+,nnnno_r.

heats of adsorption;

152

C-F Mao, M.A. Vannice /Applied Catalysis A: General 11 I (1994) 151-I 73

1. Introduction A novel method for preparing high surface area (HSA) a-A&O, which involves the topotactic transformation of diaspore (a-AlOOH) has recently been developed by Alcoa. The conventional preparation of CX-A~~O~is usually done by thermal dehydration of aluminum oxides or hydroxides at 1500 K for several hours, which routinely leads to very low surface area materials ( ca. 1 m* g - ’ or less). Two samples of (HSA) (Y-A~~O~were prepared by using either ground or unground diaspore as the starting material, and the resulting BET surface areas are approximately one hundred times higher than those of commercial low surface area (LSA) a-A1203 and are closer to those of transition aluminas. In industrial catalytic processes, aluminas have been used extensively as catalyst supports for noble metals as well as for metal oxides; for example, platinum dispersed on activated aluminas is widely used in catalytic reforming processes to yield high-octane gasoline [ 11, and noble metals including platinum and palladium are dispersed on monolithic aluminas in automotive catalytic converters to reduce pollutants in exhaust gases. Low surface area (LSA) a-Al,O, is considered to be an “inert” support and is currently used as the support for silver partial oxidation catalysts. Because of its extensive usage as a support as well as its catalytic behavior in certain reactions such as the dehydration of alcohols to alkenes [ 11, many studies have been conducted to explore the surface structure and chemistry of transition aluminas, but few such studies have addressed a-Al,O, [ 21. Consequently, one major goal of the present work was to examine the surface properties of (HSA) cu-aluminas and compare them to transition aluminas and (LSA) a-alumina, when possible. The surface structure of transition aluminas has been reviewed by Knozinger and Ratnasamy [ 31, and this paper provides helpful information to better understand the surface properties of aluminas. A model of surface structure based on the major crystallographic planes of the defect spine1 lattices of y- and 7-A1203 has been proposed to explain the five O-H stretching bands observed in their infrared spectra. It states that aluminum ions may be located at either tetrahedral or octahedral sites, and the type of terminal hydroxyl group is then determined by the number and the nature of the coordinated aluminum ions. When aluminas are heated to a certain temperature, dehydration occurs and coordinatively unsaturated (cus) surface aluminum cations and oxygen anions are generated. The resulting cus tetrahedral aluminum ions are always considered to be stronger Lewis acid sites compared to cus octahedral aluminum ions. In contrast, a-alumina exhibits an ideal corundum structure, which has all the aluminum atoms situated at octahedral sites; thus, a-alumina is typically considered to be less active in an acid-catalyzed reaction than transition aluminas due to the absence of tetrahedral aluminum ions. However, an infrared study of pyridine adsorption on a-aluminas prepared by heating transition aluminas for different periods of time showed the existence of incomplete transformation of tetrahedral coordination to octahedral coordination even after a long calcination step [ 41. Therefore, one crucial question is whether (HSA) cu-Al,O,, once prepared, possesses any surface tetrahedral aluminum ions”. If it does, it would then be important to characterize its adsorption behavior “A recent paper appeared after submission of this manuscript which also describes a (HSA) CI-A&O~ (66.2 m* g-l), and it demonstrates that this cr-A1,03 has no activity for I-butene isomerization under conditions where a -y-A&O, with similar surface area is very active [ 351. This was attributed to the presence of only octahedral Al sites at the surface of the (HSA) (Y-AI~O~,

C-F Mao, M.A. Vannice /Applied

Catalysis A: General I I1 (1994) 151-I 73

153

and Lewis acidity and compare them to transition aluminas, which exhibit certain catalytic properties. The critical question is: “Can a thermally stable (HSA) cu-AlzOs support be prepared which retains the chemical inertness of (LSA) cu-Al,OJ but acquires the capability to give high dispersions of noble metals?” In the present study, the adsorption of CO, COP, and C,H., at 300 K on (HSA) cy-aluminas was used in an effort to probe their surface properties and to compare them with those of transition aluminas as well as (LSA) cu-alumina. Both the uptakes and the integral heats of adsorption of these probe molecules were measured and compared to those in the literature. In addition, physical properties of these (HSA) LTaluminas, such as surface area, porosity and pore volume, were also determined by nitrogen isotherms at 77 K and by mercury porosimetry. In a subsequent paper [ 51 it is shown that highly dispersed silver catalysts can indeed be prepared with these (HSA) cY-aluminas.

2. Experimental The (HSA) cu-aluminas used in this study were experimental samples provided by Alcoa and were manufactured by a topotactic transformation of d&pore ((u-AlOOH) at 723 K. Two types of the starting diaspore material were used -an unground ( < 1 m* g-i) sample and an alumina ball-milled (25 m* g-i) samplewhich led to two (HSA) (u-A1203 samples, denoted as (U) and (G), with BET surface areas of 78 and 104 m* g-i, respectively. A list of their physical properties is shown in Table 1. A (LSA) cu-Al,O, (Norton SA-5202, 99.6%, 0.92 m* gg ‘). denoted by (N), was used in the volumetric adsorption study for comparison. The HSA samples were in their original powder form, while the LSA sample was ground, sieved, and the 40-80 mesh cut was taken. All samples were heated at 823 K in flowing air (Linde, dry grade) at 500 seem for 2 h before any volumetric or calorimetric measurements were made. The calcined samples were then stored in a desiccator. X-ray diffraction (XRD) measurements were performed using a Rigaku Geigerflex diffractometer with a Cu Ka radiation source and settings of 40 kV and 25 mA. The powder samples were tightly pressed into a hollow aluminum holder and held in place by a glass window. They were then scanned between 28 values of 20 and 80” in a continuous mode at a rate of 4” min - ’ . Chemical analyses for impurities in these HSA samples were done at the Alcoa Technical Center, and the results are in Table 2. Both samples have a purity greater than 99.7%, and the major impurity is titania (approximately 0.2%). Table 1 Physical properties of (HSA) Sample

a-AIzOj

BET surface area

Pore volume (cm’/g)

Average pore size (4VIA)”

(m’/g)

cr-Al,O, ( U ) a-Al,O,( G) a-Al,O,( N)

78 104 0.92

(nm) N2 adsorption

Hg porosimetry

0.071 0.160 Porosity = 54-6W0

0.082 0.175

“V is the pore volume from nitrogen adsorption.

3.6 6.2

154

C-F Mao, M.A. Vannice /Applied

Catalysis A: General I1 l(I994)

151-l 73

Table 2 Chemical analysis for impurities in (HSA) a-AlzOOJ Sample

wAl,O,(U) LY-AllOS(G)

Impurity (%) SiOz

FezO,

TiO*

NazO CaO

MgG

ZnO

CuO

NiO

Ga,O,

MnO

Cr,03

0.008 0.016

0.022 0.027

0.190 0.210

0.001 0.012

0.004 0.004

0.002 0.003

0.002 0.003

0.008 0.019

0.003 0.004

0.000 0.001

0.001 0.003

0.001 0.002

Nitrogen isotherms at 77 K were obtained with a Quantachrome/Quantasorb system. Samples were loaded in U-shaped sample cells and heated at 423 K in 40 seem helium (Liquid Carbonic, 99.999%) on a separate Quantachrome Monotector gas flow system until no change in the carrier gas was detected by the thermal conductivity detector (TCD) . The cell was then transferred to the Quantasorb system, immersed in liquid nitrogen, and isotherm measurements were made by introducing a premixed gas of N2 (MG Indust., 99.999%) and He (Liquid Carbonic, 99.999%) at various compositions, whose flow-rates were regulated by micrometering valves (Hoke) and measured by mass flowmeters (Teledyne Hastings-Raydist) . The adsorption branch of each isotherm was obtained simply by measuring the amount of nitrogen desorbed after reaching equilibrium under the desired gas composition. However, the desorption branch was obtained by first introducing 98.9% Nz, waiting for equilibrium to be attained, and then repeating the same procedure used for the adsorption branch. A complete nitrogen isotherm at 77 K with both adsorption and desorption branches was thus established for each sample and its hysteresis behavior could be examined. The amount of nitrogen adsorption at 77 K under 98.9% NZ was interpreted as representing the pore volume of pores with radii less than ca. 500-1000 A [6]. The desorption branches were also utilized to calculate pore size distributions and cumulative pore volumes based on the Kelvin equation. The pore volumes given by mercury porosimetry were obtained at the Alcoa Technical Center. Carbon monoxide, carbon dioxide, and ethylene adsorption on these aluminas were determined volumetrically in a stainless steel adsorption system utilizing an Edwards Diffstak MK2 diffusion pump capable of a vacuum of low8 Torr ( 1 Torr = 133.3 Pa). Approximately one gram of each HSA sample was loaded into a Pyrex adsorption cell and attached to the system, whereas a large adsorption cell capable of holding eight grams of the (LSA) CY-A&O~was used to increase the sensitivity of the measurements for this sample. Isotherms were obtained using a Texas Instruments Model 145 pressure gauge. Further details of the adsorption system have been described elsewhere [ 71. Before the volumetric measurements for each probe molecule, samples were subjected to a standard pretreatment to remove residual adsorbed species which involved evacuation overnight at room temperature, calcination in 10% oxygen in helium (40 seem) at 773 K for 2 h, exposure to flowing hydrogen (20 seem) at 673 K for 2.5 h, then evacuation at 673 K for 10 min before cooling down. This procedure was used because it is identical to that employed for the Ag/cr-A&O3 catalysts we have studied [ 51. Once the temperature of interest was reached, volumetric measurements were conducted at pressures between 40 and 160 Torr. After evacuation for 20 min, a second isotherm was measured to investigate the reversibility of the adsorption. In general, the first pressure measurement in the isotherm was taken after a 30-min equili-

C-F Mao, MA. Vannice /Applied Catalysis A: Genemi I I1 (I 994) 151-I 73

155

bration period while subsequent pressure measurements stabilized in less than 10 min. The gases used -HZ (MG Indust., 99.999%), He (MG Indust., 99.999%). O2 (MG Indust., 99.999%), and CO (Matheson, 99.99%)were purified by passing through drying tubes (Supelco, 5A molecular sieve) before use, and the first two gases were also passed through Oxytraps (Alltech Asso.), while CO* (MG Indust., 99.995%), and CIH, (MG Indust., 99.5%) were used without further purification. The calorimetric measurements were performed with a modified Perkin-Elmer DSC-2C differential scanning calorimeter attached to a gas handling system. All gases used in the calorimetry system were of the same grade and purity as those used in the adsorption system and were subject to the same purification steps. Approximately 50 mg of the sample was placed in the sample cell and an equal weight of nonporous glass beads was placed in the reference cell. To maintain a stable baseline when the probe gas was introduced, the appropriate composition of the initial purge gas consisting of a mixture of helium and argon (MG Indust., 99.999%) was determined using empty sample pans. Additional details of the procedure have been given elsewhere [ 81, A pretreatment procedure similar to that used for the volumetric measurements was conducted before each calorimetric measurement except that the 20-min evacuation was replaced by a l-h purge by the carrier gas and the sample was exposed to 20% H2 in argon (40 seem) . The probe gas was then introduced into the system sample cavity for 30 min using a mixture of 7.9% probe gas in argon (38 seem), which gives a partial pressure of 60 Torr for the probe gas. Each exotherm was recorded in a computer for later analysis. A second exotherm was obtained after a 30-min purge with the carrier gas following completion of the first exotherm. Integral heats of adsorption were calculated based on the integration of the exotherm peaks, which was done by computer over the initial 3-min period because the exotherms usually reached the original baseline in this time period. The peak areas representing total energy changes were converted to integral heats of adsorption by dividing by the appropriate gas uptake. For reversible adsorption, the second exotherm was used and reversible integral heats of adsorption were obtained using the uptake at 60 Torr from Fig. 6. On the other hand, the irreversible energy change was obtained by subtracting the peak area of the second exotherm from that of the first one, and the heats of adsorption for irreversible carbon dioxide adsorption were calculated by dividing this value by the difference between the uptake of the initial point in the first isotherm and that in the second isotherm taken at 3 min. Because of the limitation in the size of the sample holders, heats of adsorption on the LSA a-A&O3 could not be obtained.

3. Results 3.1. Physical characterization The XRD patterns for the three unpretreated samples showed only the principal diffraction peaks of the corundum structure typical for a-aluminas, as shown in Fig. 1; however, two differences between the LSA and HSA samples were observed. First, the peaks in the XRD spectrum for the LSA sample were narrower than those for the HSA samples, implying that the average crystallite size of the LSA sample was larger than those of the HSA samples.

156

C-F Mao, M.A. VannicelApplied

20

I

Catalysis A: General 111(1994)

A

40

60

151-173

A 60

20

Fig. 1. XRD patterns of a-A&O, samples: bottom, CY-AI~O~ (U); middle, cr-Al,O,(G); top, a-A&O,(N)

Secondly, the relative intensity for each plane was different. Considering the relative intensity of the peak at 43.3”, which is usually the strongest peak in the corundum structure and associated with the ( 113) plane, the LSA sample had intensities of the two second strongest peaks at 35.2” and 57.5” which were tantamount to that of the 43.3” peak and relatively higher than those for the HSA samples. This discrepancy indicates that the HSA samples possessed a preferential orientation of crystallite planes different from that of the LSA sample. The physical properties of the HSA samples obtained by nitrogen adsorption and mercury porosimetry are listed in Table 1. The pore volumes determined by nitrogen adsorption at 0.98P. and mercury porosimetry were close, but the latter was somewhat larger than the former for each HSA sample. In addition, the values of the pore volume of the ground sample were about two times larger than those of the unground sample for both measurements. Due to this difference, the ground sample had an average pore size twice that of the unground sample. Pores are classified by IUPAC into three categories: macropores (d > 50 nm) , mesopores (2 < d < 50 nm), and micropores (d < 2 nm). The mesoporosity of aluminas can be estimated by examining their nitrogen isotherms at 77 K (Fig. 2)) and it can be noted that neither of them could be exactly classified as one of the five general isotherms, but the presence of hysteresis loops indicates the existence of mesoporosity. For example, the inset in Fig. 2 shows the enlargement of the minor type-E hysteresis of the unground sample in the range from 0.4 to 0.6Po, implying the existence of a small amount of mesopores, probably with an ink-bottle shape, i.e., larger pore volumes with constricted openings [ 91. On the other hand, the ground sample showed a significant type-B hysteresis extending from 0.4P. to the saturation pressure, which implies the presence of a wide distribution of slit-shaped mesopores [ 91. Also, the amount of nitrogen adsorption was similar for both samples below 0.4Po, but at higher relative pressures the unground sample featured an adsorption plateau whereas the ground sample showed a great increase of adsorption capability. The adsorption branch of the isotherm was also used to construct t-plots (Fig. 3) based on the table of statistical thickness t over nonporous aluminas given by Lippens et al. [ 101. Both showed that a straight line initially passing through the origin was formed, thus indicating the lack of microporosity. Between t = 5 and 7.5 A, corresponding to 0.3 and

C-Fhfoo, MA. Vonnice/Applied

0.0

0.2

Cotolysis A: General 111(19!M) 151-173

0.4

0.6

0.6

157

’

1.0

P/PO

Fig. 2. N2 isothermsfor (HSA) u-AIZ03 at 77 K: a-A&0,( U)-adsorption (O), desorption (0); a-A&O,(G)adsorption (El), desorption ( n ). Inset: hysteresis loop for a-AIZO,(U).

0.6Po, a small increase in the slope was observed for the ground sample, which is indicative of capillary condensation in mesopores. As the pressure increased, the slopes then decreased, indicating surface areas were no longer accessible for multilayer condensation. The mesopore size distributions in Fig. 4 were calculated from the desorption branch of the nitrogen isotherms using the Kelvin equation. To eliminate false peaks resulting from the scatter of data points or the interpolation of curves by a conventional method, the amounts of desorption against relative pressures were adjusted by a least squares method fitting to a polynomial function [ 11I. The polynomial-fitted curves were then divided into small intervals and a conventional calculation was followed to incorporate the thickness of the adsorbed layer and build cumulative pore volume curves [6 1, and pore size distributions were then obtained by differentiating the resulting cumulative pore volume curves. These distributions, as shown in Fig. 4, exhibited a sharp increase in pore radius near 20 A, which has been attributed by Gregg and Sing [ 61 to an artifact resulting from the breakdown of the Kelvin equation. The valid portion of these curves showed that the maximum mesopo-

t (4 Fig. 3. f-Plots for N, adsorption at 77 K on (HSA) a-A&O,: cx-A1203(U)(O), a-A&4(G)

(0).

158

C-F Mao, M.A. Vannice /Applied Catalysis A: General I I I (1994) 151-I 73

lo-

Pore radius rp (A) Fig. 4. Pore size distributions dotted curve.

for (HSA) a-AlzOs ( N, adsorption at 77 K) : a-A&O,(U)

-solid curve; a-A&O,(G)

-

rosity occurs with a radius near 60 A for both samples, with the ground sample exhibiting a higher concentration of these pores. The corresponding plot of cumulative mesopore volumes (Fig. 5) was constructed from pore radii 35 A and larger to avoid the invalidity of the Kelvin equation. The cumulative mesopore volume of the ground sample was greater than that of the unground sample in the range of radii from 35 to 300 A, which is in agreement with the pore volumes in Table 1; however, the mesopore volumes in pores with radii of 300 A or less were significantly smaller than those listed in Table 1. This discrepancy will be addressed in the discussion section. 3.2. Volumetric measurements of carbon monoxide, carbon dioxide and ethylene adsorption The adsorption isotherms for CO, CO* and C2H4 at 300 K on (HSA) cy-aluminas are shown in Fig. 6. There was no irreversible carbon monoxide and ethylene adsorption, and 0.06

5

u

0.00 10'

lo2

lo3

Pore radius rp (A) Fig. 5. Cumulative mesopore volume for (HSA) Al,O,( G)-dotted curve.

a-A&O3 (N2 adsorption at 77 K): wA1203( U)-solid curve; (Y-

C-F Mao, M.A. Vannice /Applied

Catalysis A: General I II (1994) 151-173

b

a

!

159

/

1

40

60 Pressure

120

0

160

40

60 Pressure

(torr)

120

160

(torr)

V.

0

40

120

60 Pressure

160

(torr)

Fig. 6. (a) Adsorption of carbon monoxide on (HSA) a-A&O, at 300 K: a-A&0,( U) : firstisotherm(A), second isotherm (A); a-A120s( G): first isotherm (A), second isotherm (0). (b) Adsorption of carbon dioxide on (HSA) a-A&O9 at 300 K: a-A1209(U): first isotherm (A), second isotherm (A); a-Al,O,(G): first isotherm (0). second isotherm (0). (c) Adsorption of ethylene on (HSA) a-A&O, at 300 K: a-Al,O,( U) : firstisotherm (A), second isotherm (A); a-Al,O,(G): first isotherm (0). second isothekm (0).

the second isotherms were usually slightly higher than the first isotherms (Fig. 6a and 6c), indicating the possibility of activated adsorption or perhaps surface reconstruction to generate more adsorption sites after initial adsorption followed by evacuation. In contrast, an appreciable amount of irreversible carbon dioxide uptake existed on both samples (Fig. 6b). thus showing the formation of strongly bonded species. The irreversible carbon dioxide uptakes were 106 and 101 ;CLmol/g for samples (U) and (G), respectively, while the corresponding surface fractional coverages were 0.057 and 0.041 (assuming a surface site density on the cu-A&O, (001) face of 24 pmol mP2, i.e., 1.5. 1Ol9 sites mm2). In general, based on sample weight the uptakes on the ground sample were somewhat larger than those on the unground sample but the isotherms had similar slopes. The adsorption behavior of the (LSA) cw-A1203,as shown in Fig. 7, was similar to the HSA samples - no irreversible carbon monoxide and ethylene adsorption occurred but appreciable irreversible carbon dioxide adsorption did. The uptakes for CO, CO2 and C2H, at 300 K on each sample at 60 Torr are summarized in Tables 3-5. A similar trend was

160

C-F Mao, M.A. Vannice/Applied

Catalysis A: General 11 I (1994) 151-173

40

0

60

120

160

Pressure (torr)

Fig. 7. Adsorption of carbon monoxide, carbon dioxide, and ethylene on (LSA) (r-A1203(N) at 300 K: first isotherm of CO2 (A), second isotherm of CO2 ( A.), second isotherm of CZH, ( 0). second isotherm of CO ( W)

Table 3 Carbon monoxide adsorption Sample

on a-A120s at 300 K

Gas uptake (molecule/m’~

(wmol/g)

a-AllOg ( U ) cr-AlzO,( G) wA1203( N)

IO-“)

Rev.”

Irrev.

Rev.”

Irrev.

11.7 12.9 0.052

0 0 0

0.90 0.12 0.34

0 0 0

Saturation coverage (I)’ Rev.

Qad (kcal/mol ) Rev.”

1.4 1.1

6.1 15.4 _

“At 60 Torr. “Calculated from the second exotherm. cBased on site density of 1.5. lOI molecule rn-’

Table 4 Carbon dioxide adsorption Sample

on wA1203 at 300 K

Gas uptake (molecule/m2.

(wmollg)

wAla03(U) a-A&O&G) a-A1203(N)

Saturation coverage (%)”

Qad ( kcall mol )

IO-“)

Rev.”

Irrev.

Rev.”

Irrev.

Rev.

Total

Rev?

IIIW.

48.6 59.4 1.61

106 101 1.49

3.1 3.4 10.9

8.2 5.8 9.8

8.9 1.2 18.2

14 11 24

10.8 8.4 -

21.4 16.0

“At 60 Torr. ‘Calculated from the second exotherm. ‘Based on site density of 1.5. IO’” molecule m-‘.

C-F Mao, M.A. Vannice /Applied

Catalysis A: General lll(1994)

151-173

161

Table 5 Ethylene adsorption on a-A&O9 at 300 K and 443 K Sample

T(K)

Gas uptake (moleculelm’~

(pmol/g) Rev.”

lrrev.

Rev.”

IITW.

lo-“)

Saturation coverage (96)’ Rev.

Q, (kcal/mol) Rev.’

U)

300 443

16.8 0.54

0 0

1.3 0.042

0 0

5.1

3.1

a-A120s(G)

300 443

20.7 0.35

0 0

1.2 0.020

0 0

4.1

2.9

a-A&O,(N)

300 443

0.65 0.061

0 0

4.2 0.40

0 0

8.5

a-Al,O,(

“At 60 Torr. *Calculated from the second exotherm. “Based on site density of 1.5. lOI molecule m-‘.

found for all the samples in that both the uptakes and the slopes of the isotherms varied in the following manner: CO;? > C,H, > CO. All the reversible isotherms fit the Langmuir equation very well except carbon monoxide adsorption on the LSA sample, but this was probably due to the fact that the experimental uncertainty was significant compared to the small carbon monoxide uptakes caused by the low surface area alumina. The saturation coverage from the Langmuir isotherms for only reversible adsorption for each gas also followed the aforementioned trend. For example, the saturation coverages for reversible adsorption on the unground sample were 166.7, 95.3, and 26.5 pm01 g-’ for CO*, C2H4 and CO, respectively, while the corresponding values on the ground sample were 179.3, 102.5, and 27.1 pmol gg ‘. Likewise the saturation amounts on the LSA sample exhibited the same trend with 4.02 pmol g-i for CO, and 1.88 pm01 gg ’ for C&. This represents fractional reversible carbon dioxide coverages of 0.089, 0:072, and 0.182 on (HSA) a-A1,03( U), (HSA) cr-Al,03( G), and (LSA) a-A1,03, respectively; however, total carbon dioxide coverages including irreversible carbon dioxide are 0.14, 0.11, and 0.24, respectively. The lowest saturation coverages were for carbon monoxide, and the fractional coverages were 0.014 and 0.011 for the (HSA) (Y-A&O~(U) and (G) samples, respectively. The comparable coverages for ethylene were 0.05 1, 0.041, and 0.08 for (HSA) a-A1,03( U), (HSA) a-Al,O,( G), and (LSA) a-A1203, respectively. To better compare uptakes on the HSA and LSA samples, the uptakes were normalized to the BET surface areas, as shown in Fig. 8 and Tables 3-5; for clarity, only the second isotherms for carbon monoxide and ethylene adsorption are present. In general, the reversible uptakes per unit surface area on both HSA samples were similar for each probe gas but the values on the unground sample were always somewhat higher. For the irreversible carbon dioxide uptakes per unit surface area, that on the unground sample was 40% higher than on the ground sample. In comparison, the normalized adsorption capacities on the LSA sample at 60 Torr deviated significantly from those on the HSA samples. For instance, the reversible carbon monoxide uptake per unit surface area on the LSA sample at 60 Torr was

162

C-F Mao, M.A. Vannice/Applied

I

Catalysis A: General lIl(l994)

151-173

a

40

0

60

Pressure

0

40

60

pressure

120

160

120

160

(torr)

(torr)

2c

0

40

60

Pressure Fig. 8. (a) CO ALO, (A ) second isotherm second isotherm A&O,(G) (0).

120

160

(torr)

adsorption per m2 at 300 K on different cy-aluminas: a-A1209(U) (0). a-AI?O,(G) (Cl), Q(b ) CO2 adsorption per m2 at 300 K on different waluminas: a-AlzO,( U) : first isotherm ( 0). (0); a-Al,O,(G): first isotherm ( n ). second isotherm (Cl); a-A120?(N): first isotherm (A), (A). (c) C,H,+ adsorption per mz at 300 K on different cr-aluminas: a-A&O,(U) (0), aa-A1203(N) (A).

C-FMao. M.A. Vannice/Applied

Catalysis A: General llI(I994)

151473

163

one half of those on the HSA samples (Table 3) whereas the reversible carbon dioxide and ethylene uptakes were about three times larger than those on the HSA samples (Tables 4 and 5). As for the irreversible carbon dioxide adsorption, that on the LSA sample was significantly higher compared to that on the HSA samples. A similar trend was also observed with the saturation coverages for reversible adsorption from the Langmuir isotherms. The adsorption of 02, Hz and C,H, was also measured at 443 K. On all three samples, the isotherms for oxygen and hydrogen yielded tiny reversible uptakes and no irreversible uptakes. The adsorption of ethylene was very low compared to that at 300 K, as shown in Table 5. 3.3. Calorimetric measurements Typical exotherms for CO, CO2 and C,H, adsorption on the unground sample are shown in Figs. 9-l 1. The tiny perturbation at the beginning of the exotherm in Fig. 9 is due to the switch to the carrier gas with the adsorbate. In Figs. 9 and 11, the first and second exotherms were almost identical and only one is shown. For the exotherms associated with reversible adsorption, most of the energy change occurred in one minute and they had short tails. On the other hand, the first exotherm for carbon dioxide adsorption possessed a long tail while the second exotherm returned to the baseline after about four minutes. The integral heats of adsorption ( Qad) for CO, CO2 and C,H, obtained from integration over these exotherms are listed in Tables 3-5. The Q, value for carbon monoxide adsorption on the unground sample was 6.1 kcal mol- ‘, which is reasonable for reversible adsorption; however, that on the ground sample was two and a half times larger. The exotherm for carbon monoxide adsorption on the ground sample also exhibited unusual behavior, as shown in Fig. 12, in that an extremely broad peak occurred on the first exotherm while the second exotherm had a significant tail, both of which were not observed with the unground sample. The exotherms for carbon dioxide and ethylene adsorption as well as the isotherms for carbon monoxide on the ground sample were normal and showed no unusual behavior.

0

1

2

3

4

Time (min)

Fig. 9. Calorimetric exothem for CO adsorption on (HSA) cr-Al,03( U) at 300 K.

164

C-F Mao, M.A. Vannice /Applied Catalysis A: General 1 I I (1994) 151-I 73

0

1

2

3

4

Time (min)

Fig. 10. Calorimetric exotherms for CO2 adsorption on (HSA) a-A&O,(U) tion, lower: reversible CO, adsorption.

at 300 K. Upper: initial CO2 adsorp-

The reason for these unusual exotherms is not clear. The large heat evolution shown in the first exotherm suggests that some type of reaction might occur when carbon monoxide was first introduced to the sample. After the following purge by the carrier gas, the ground sample then showed an energetically different surface from that of the initial ground sample. On the other hand, carbon dioxide adsorption had similar behavior on both HSA samples, with the irreversible Qad values always twice the reversible ones, but both the reversible and irreversible Qad values on the unground sample were somewhat higher than those on

0

1

2

3

4

Time (min)

Fig. 11. Calorimetric

exotherm for C,H., adsorption on (HSA) a-A1203( U) at 300 K.

C-F Mao, M.A. Vannice /Applied

0

1

Catalysis A: General 111(1994)

2

3

151-173

165

4

Time (min) Fig. 12. Calorimetric exotherms forC0 lower: reversible CO adsorption.

adsorption on (HSA) a-A&4(G)

at 300 K. Upper: initial CO adsorption,

the ground sample. As for ethylene adsorption, the heats of adsorption on both HSA samples were almost equal and were around 3.0 kcal mol - ‘. An estimation of the enthalpy of ethylene adsorption on the HSA samples was made based on the isotherms at 300 K and 443 K. The isotherm at 300 K was fitted to the Langmuir equation to obtain the saturation coverage and the adsorption equilibrium constant; however, uptakes at 443 K were very small and did not fit the Langmuir equation very well, so they were approximated using Henry’s law to give a slope that is equal to the product of the equilibrium constant and the saturation coverage. By assuming that the saturation coverage remains constant, the equilibrium adsorption constant at 443 K was obtained, and the heat of adsorption for the unground and ground samples was then estimated from the thermodynamic correlation between the equilibrium constant and reciprocal temperature to be 5.8 and 7.3 kcal mol-‘, respectively. Although somewhat larger, these values are still low and similar to those obtained in the calorimetric study, and they all indicate weak chemisorption.

4. Discussion 4.1. Nitrogen adsorption and porosity The internal pore structure and surface area of aluminas usually depends on the starting material and preparation methods, but has little dependency on the phase of alumina [ 12141. However, cy-aluminas prepared in a conventional way always lose surface area and porosity due to the high-temperature pretreatment used. Several investigations have been conducted on the pore structure of a variety of aluminas. De Boer and Lippens [ 121 have

166

C-F Mao, M.A. Vmnice /Applied Catalysis A: General 11 I (1994) 151-173

examined nitrogen isotherms at 77 K for boehmite and bayerite samples which had been subjected to different pretreating temperatures, and they reported type-A, B, andE hysteresis loops as well as a combination of them, indicating the presence of a variety of pore shapes for aluminas. Kotanigawa et al. [ 131 studied the effect of using divergent starting materials on the pore structure of y-Al,O, and they found particle shapes to be either plate-like or cubic or a mixture of the two by scanning electron microscopy, and hysteresis loops of the nitrogen isotherms at 77 K were either type B or E. The pore structure of aluminas was thus established to be strongly dependent on the starting materials. The corresponding pore size distributions were either unimodal, with a peak radius around 20 A, or bimodal; however, in retrospect, all the peaks around 20 A in the pore size distribution plots were probably due to the breakdown of the Kelvin equation [ 61. Gregg and Langford [ 141 have studied the effect of compacting on the porosity of an alumina powder with a spherical shape, and they found that the process of compaction produced a well-defined type-E hysteresis, indicating the generation of ink-bottle shaped pores. Since the literature data shows divergent pore structures for aluminas, a meaningful comparison of the pore structure between HSA a-aluminas and other aluminas does not seem possible. The two (HSA) cw-aluminas samples in this study possessed several common features related to the porosity given by the nitrogen adsorption experiments. First, a large difference existed between the total pore volume measured from nitrogen adsorption at 0.98Pc and the calculated cumulative mesopore volume; for example, the unground sample had a nitrogen pore volume of 0.07 1 cm3 g - ’ while the corresponding cumulative mesopore volume was less than 0.01 cm3 g- ’ . For nitrogen adsorption near the saturation pressure, the process is known to involve condensation in micropores and mesopores as well as multilayer formation on macropore and external surfaces. Since the microporosity appears to be insignificant from the r-plots (Fig. 3) and the mesopore volume is small, the major contribution to nitrogen adsorption is considered to be the combination of the last two processes. Second, the difference between the pore volumes in Table 1 obtained from nitrogen adsorption at 0.98Pa and those determined by mercury porosimetry suggests the existence of significant macroporosity in both samples since mercury porosimetry probes macropores and the upper range of mesopores. Third, the t-plots for both samples also exhibited similar characteristics: A straight line passed through the origin in the initial coverages while a decrease in the slope occurred as the thickness increased (Fig. 3). The former suggests the absence of microporosity, and the latter implies that the space is not accessible for nitrogen condensation at high relative pressures. This can be explained by the poor accessibility of the macropores and slit-shaped mesopores of these aluminas. The tiny hysteresis in the nitrogen isotherm for the unground sample suggests the absence of an appreciable amount of mesopores, which is further confirmed by the very broad pore size distribution and the extremely low cumulative mesopore volume. Furthermore, the shape of this hysteresis is close to that of type E, indicating that the mesoporosity may result from ink-bottle shaped pores or the voids between close-packed particles [ 91. As mentioned before, with (HSA) a-aluminas the pore volume measurements obtained from nitrogen adsorption at 0.98P, contain condensation in mesopores as well as adsorbed layer formation in macropores and on the external surface. Because of the shortage of mesoporosity on the unground sample, the adsorbed layers play a dominant role. However, the porosity in the unground sample is not completely accessible for nitrogen condensation, as indicated by

C-F Mao. M.A. Vannice /Applied

Catalysis A: General II l(1994)

151-I 73

167

the formation of the plateau in the isotherm in the region above O.SP, and the lower slope of the f-curve in the range of high relative pressure. Unlike the unground sample, the ground sample possessed an appreciable amount of mesopores, as shown by the much larger value in its cumulative mesopore volume plot and the evident hysteresis in the nitrogen isotherm. The type-B hysteresis further suggests that the mesoporosity is representative of slit-shaped pores with parallel walls [ 91. Moreover, the ground sample had significant contributions from both mesopore condensation (the value of cumulative mesopore volume at 250 A was 0.065 cm3 g- ‘) and multilayer formation to the total pore volume obtained by nitrogen adsorption (0.16 cm3 g-i). Another difference between these two HSA samples is that the ground sample possesses more space accessible to nitrogen condensation than the unground sample does, as indicated by the higher slope in the t-plots and the rise of the isotherm in the region above 0.5Pc,. This discrepancy can be explained by the large distribution of mesoporosity for the ground sample. Similarly, the larger surface area for the ground sample can also be explained by the presence of abundant mesoporosity on the ground sample as compared to the unground sample.

5. Carbon monoxide adsorption Infrared spectra for carbon monoxide adsorption at low temperature have been extensively used to characterize the coordination and valency states on metal oxide surfaces. Since carbon monoxide is a very weak base and more selective than other often used probe molecules like ammonia and pyridine, it has also been frequently employed to probe the surface acid/base properties of metal oxides, and numerous studies of its adsorption on aluminas either at low temperature [ 15,161 or above 273 K [ 17-201 can be found in the literature. The nature of the resulting carbonyl bonds on aluminas is generally believed to be the linkage of carbon monoxide to cus surface aluminum cations via a s-donor bond. In infrared spectra this always leads to a blue shift in the C-G stretching band (in the range of 2250 to 2150 cm - ’ ) relative to that in the gas phase (2 143 cm- ’ ) . The high frequency bands are usually assigned to carbon monoxide molecules s-bonded to strong Lewis sites, which have been shown to be cus tetrahedral aluminum cations, while the lower frequency bands represent carbon monoxide bonded to weaker Lewis sites. Hence carbon monoxide has been used successfully to examine the Lewis acidity of aluminas, and the concentrations and bond strengths have been demonstrated to depend strongly on the alumina phase and the activation temperature [ 15,181. Gattaet al. [ 181 have investigated carbon monoxide adsorption on several alumina phases using infrared spectroscopy and microcalorimetry. No carbon monoxide uptake was observed on their LSA CY-A&O3samples, which contradicts our results. Based on this evidence, they proposed that a-A1203 has no surface aluminum cations with tetrahedral coordination; however, Mot-terra et al. [ 41 investigated the same aluminas in a later paper using pyridine as a probe, and this time the results suggested the presence of surface aluminum tetrahedral coordination sites. Another interesting observation they reported was that a small portion of the surface carbonyl groups (5.4 and 1.8. lOi molecule rnp2 for rl_ and y-alumina, respectively) was not removable after a lo-min degassing period, and this

168

C-F Mao, M.A. Vannice/Applied

Table 6 Reported heats of adsorption

and uptakes for carbon monoxide on alumina surfaces

Alumina phase

n- and BAl,Os y-A1203 Commercial activated alumina Y-Al& y_AW,

Adsorption temperature

Activation temperature

(R)

(R)

309 309 343 343 315 1-l

1013 1013 943 123 923 1073

“Extrapolating isotherms at 60 Torr. *Integral values through 60 Torr. Values for the highest bands calculated residual

Catalysis A: General 111 (1994) 151473

Uptake at 60 Torr ( molecule/m2~ 1O- “)

Qti (kcallmol)

Reference

1.7” 2.0” 0.36

14and2

1181

12.1 9.5 12.6b ca. 12-14’

1191 Cl91 I201 cl61

0.56

1181

from the equation given by Paukshtis et al. [ 211

carbon monoxide was that with the highest C-0 stretching bands, which is associated with strong Lewis acid sites. The main difference between carbon monoxide adsorption on our a-alumina samples and their transition aluminas is that a-alumina not only has no irreversible uptake but also less reversible adsorption than the transition aluminas (Table 6). They reported two heat of adsorption values of 14 and 2 kcal mol- ‘, thus indicating the existence of at least two different carbon monoxide adsorption sites on transition aluminas. Several other investigators also reported heats of adsorption and uptakes for carbon monoxide adsorption on a variety of aluminas. In comparing their values to ours, the different adsorption temperatures and the divergent pretreatment procedures must be considered; consequently, these values are listed in Table 6 along with the adsorption and activation temperatures utilized. Two comments should be made here. First, none of these other studies found any irreversible carbon monoxide uptake, which is consistent with our results. Second, Zecchina et al. [ 161 calculated heats of carbon monoxide adsorption from the shift in the C-O stretching band using the correlation of AH (kJ mol-‘) = 10.5 +0.5 ( vco-2143) given by Paukshtis et al. [ 211. Based on this correlation, the C-O stretching frequencies can be easily converted to heats of adsorption; for example, the highest wavenumberreported by Gatta et al. [ 181, 2245 cm-‘, corresponds to a heat of adsorption of 14.6 kcal mol-‘, which is in good agreement with the measured heat of adsorption of 14 kcal mol-‘. The heats of adsorption listed in Table 6 fall in the range of 9 to 14 kcal mol- ’ except for the low value reported by Gatta et al. [ 181, which indicates that the nature of Lewis acid sites on aluminas are similar regardless of activation procedures; however, a large variation exists in adsorption capacity, which is presumably due to the extent of dehydration. When these values are compared to ours, it is noted that the unground sample has a Qad value of 6.1 kcal mol- ‘, which is much lower than the values reported for transition aluminas, and which indicates a weaker interaction with carbon monoxide compared to the transition aluminas, and leads to the conclusion that the unground a-alumina sample possesses a weaker Lewis acidity than transition aluminas. In contrast, the ground sample has a Q,, value of 15.4 kcal mol- ‘, which is somewhat higher than those for transition aluminas, and indicates Lewis acid sites with strengths similar to those on transition aluminas. Since carbon monoxide uptakes depend strongly on pretreatment procedures, a strict comparison

C-F Mao. M.A. Vannice/Applied

Catalysis A: General lll(1994)

151-173

169

cannot be made based on Table 6; however, in general the (HSA) cY-alumina samples have concentrations of strong Lewis acid sites, as probed by carbon monoxide adsorption, that are similar to or lower than those on transition aluminas, whereas the concentration on the (LSA) a-alumina may be slightly lower. 5.1. Carbon dioxide adsorption Viewed as an acidic anhydride, carbon dioxide has been utilized as a probe molecule to examine surface basicity on a variety of metal oxides [22]. Unlike carbon monoxide adsorption, several types of surface species such as bicarbonate, bidentate and monodentate carbonate, and bridging carbonate groups can be generated by carbon dioxide adsorption and interaction on metal oxide surfaces; consequently, more difficulties are encountered when interpreting results of carbon dioxide adsorption. Spectroscopic and calorimetric studies of carbon dioxide adsorption on aluminas have been reported for o-alumina [ 231 and transition aluminas [20,24-261. The most abundant species formed during carbon dioxide adsorption on aluminas have been found to be carbonate and bicarbonate groups which are strongly bound at low carbon dioxide coverages and are associated with the irreversible adsorption, whereas the reversible uptakes usually involve weakly bound carbonate and bicarbonate species as well as carbon dioxide molecules weakly adsorbed on Lewis acid sites. The above studies have demonstrated the surface heterogeneity that can exist on aluminas after carbon dioxide adsorption, and the relative amounts of these species are determined by pretreatment conditions while their nature mainly depends on surface coordination [ 261. Morterra et al. [23] have investigated two different cY-alumina samples which were activated at different temperatures, and they found that no, or relatively little, bicarbonate species existed after carbon dioxide adsorption on a-aluminas while the predominant species on transition aluminas were bicarbonate groups [ 261. The amount of bicarbonate groups relative to carbonate species on transition aluminas depends on the activation temperature, and at high temperatures this ratio decreases [ 271. Another feature of cy-aluminas is that subjection to high activation temperatures creates a lower surface basicity, as evidenced by the decreasing ratio of monodentate to bidentate carbonate groups. These carbonate species are suggested to be mobile at temperatures above 423 K [ 241. The heats of adsorption and uptakes for carbon dioxide adsorption on aluminas reported in the literature are listed in Table 7; however, few calorimetric studies of carbon dioxide adsorption on aluminas have been performed. Auroux and Gervasini [ 221 have reported a Qad value of 14.6 kcal mol-’ for total carbon dioxide adsorption on y-alumina, which lies between our values for reversible and irreversible adsorption. At high carbon dioxide coverages, Qad values from 4.5 to 12 kcal mol- ’ were measured by Rosynek [ 241, and since this represents the regime of reversible adsorption, these values are consistent with the reversible Qad values in Table 4. Since the reversible adsorption of carbon dioxide on aluminas can be attributed to the formation of different surface species, that is, a nonselective adsorption process occurs, any analysis of this adsorption behavior provides limited information about the surface properties of aluminas. On the other hand, the irreversible adsorption provides valuable information about surface basicity. The irreversible carbon dioxide adsorption as well as the reversible uptake on our (LSA) (Y-A&O~ is in extremely good

C-F Mao, M.A. Vannice /Applied Catalysis A: General I1 I (1994) 151-173

170

Table 7 Reported heats of adsorption Alumina phase

and uptakes for carbon dioxide on alumina surfaces

Adsorption temperature

Uptake at 60 Torr (molecule/m2~ IO-“)

Q,d

Rev.

Rev.

Reference

(kcallmol)

(R)

Y-AU%

296

y_AWh ~&0x

315 295

4.8

w&Q

370 300 300

2.2 5.ld 10.ld

r)-&Q a-Al,03 “Average *Integral “Isosteric dUptakes

Irrev.

lrrev.

[221

14.6“ 4.0 6.0 0.84 5.1 10.1

14.6’

19.6’ ca. 4%IT

[201 ~241

[251 [261

t261

value for the total isotherm. values through 60 Torr. heats of adsorption at high surface coverages. at 40 Torr.

agreement with that reported by Morterra et al. [ 261, while the irreversible adsorption on the (HSA) a-Al,O,(G) sample was similar to values reported for -y- and 77-A1203 but it was somewhat higher on the (HSA) cu-Al,Os( U) sample. These results imply the following relationship for the concentration of basic sites on alumina: (LSA) cr-A&O3 > (HSA) (YAl,O,( U) > (HSA) a-Al,O,( G) >, y- and rl_Al,O,. The Qad value for irreversible carbon dioxide adsorption on the unground sample may be somewhat greater than that for yalumina 1201, and it is also higher than that for the ground sample. This implies that the unground sample possesses more basic sites with a stronger interaction with carbon dioxide than the ground sample. Unfortunately, because of the scarcity of Q,., values for carbon dioxide adsorption on aluminas, a relationship between transition aluminas and HSA (Yaluminas cannot be established at this time. 5.2. Ethylene adsorption The nature of adsorbed ethylene species on aluminas has not yet been well established. Two groups have proposed different models to explain irreversible ethylene adsorption on q-alumina. Yates and coworkers [ 28,291 have proposed two adsorption species - an ethyl group on a single site and the associative adsorption of ethylene on two surface sites, SCH,CH,-S-based on the specific bands observed for CH3 groups and -CH*- groups in infrared spectra. In contrast, Schubart and Knozinger [ 251 have suggested that ethylene adsorbs on alumina to form a vinyl species via heterolytic cleavage of C-H bonds on acidbase pair sites that involve anion vacancies and OH groups coordinated to tetrahedral aluminum ions. Since ethylene adsorption yielded no irreversible adsorption on the (Yaluminas in this study, at least after our pretreatment, the possibility of adsorption on these sites can be eliminated. In regard to reversibly adsorbed species, little work has been done to elucidate its nature. Gordymova and Davydov [ 301 have examined propene adsorption on sodium-doped y-aluminas, and a weakly bound species that could be removed after

C-F Mao, M.A. Vannice/Applied

Catalysis A: General 111(19!44) 151473

171

evacuation was observed in the infrared spectra. This species was assigned to propene coordinated to surface cus aluminum ions via its m bond with a methyl group hydrogen bonded to surface hydroxyl groups in the immediate vicinity. The heat of adsorption for this species was approximately 2 kcal mol - ‘, which is very close to our measured values of 3 kcal mol- ’ for ethylene. Isosteric heats of ethylene adsorption on y-alumina of 9.5 and 8.8 kcal mol- ’ have been reported at a coverage of 10 pmol g - ‘, i.e., 0.4.10” molecule m-*, which included both irreversible and reversible adsorption [ 3 1,321. A previous study of ethylene adsorption on sintered v-alumina found only reversible ethylene adsorption occurred with a heat of adsorption of 10.1 kcal mol- ’ [ 331. All three values for transition aluminas are three times greater than those measured here on (HSA) a-alumina, thus indicating a much weaker interaction between ethylene and these a-alumina surfaces.

6. Summary Nitrogen adsorption at 77 K combined with mercury porosimetry was utilized to determine the porosity of two (HSA) a-alumina samples. With the unground sample macropores predominate and only small amounts of mesopores exist, as evidenced by a small hysteresis and the small cumulative mesopore volume. In contrast, the ground sample has considerable amounts of slit-shaped mesopores as well as macroporosity. Thus the starting material ground or unground diasporecan have an important effect on the mesoporosity of the (HSA) a-alumina formed. The Lewis acid site concentration on the surface of (HSA) and (LSA) cY-aluminas was probed by carbon monoxide adsorption at 300 K, which was completely reversible and was attributed to carbon monoxide molecules coordinated on cus surface aluminum ions. The uptakes for carbon monoxide adsorption at 300 K on the unground and ground samples at 60 Torr were 0.90 and 0.72.10” molecule m-* , respectively, which are less than or similar to those on transition aluminas but greater than that on (LSA) a-alumina. The unground sample had a Qa,, value of 6.1 kcal mol -’ for carbon monoxide adsorption, which is less than that for transition aluminas, and that for the ground sample was 15.4 kcal mol- ‘, which is comparable to that for transition alumina.% Irreversible carbon dioxide adsorption at 300 K was employed to probe surface basicity on these a-alumina as carbonate or bicarbonate groups have been previously found to be the major species after carbon dioxide adsorption. The irreversible carbon dioxide uptakes for samples (U), (G), and (N) were 8.2, 5.8, and 9.8 * 10” molecule m-*, respectively, thus the (LSA) a-alumina has the highest value among these three a-alumina samples, and the uptake on the unground sample is greater than on the ground sample, which is greater than or similar to that on transition aluminas. The corresponding Qti values for the unground and ground samples were 21.4 and 16.0 kcal mol- ‘, respectively, which implies the unground sample has a stronger interaction with carbon dioxide than the ground sample. These observations lead to a view that the concentration of basic sites on these cu-aluminas has the following relationship: (LSA) (~-A1,0~ > (HSA) a-Al,Os( U) > (HSA) crAl,O,(G) 2 transition aluminas, and the unground sample possesses basic sites with strengths greater than those on the ground sample.

172

C-F Mao, M.A. Vannice/Applied

Catalysis A: General llI(1994)

151-173

Ethylene adsorption provided little information about the surface properties of these (Yaluminas because no irreversible ethylene adsorption occurred. The heat of ethylene adsorption was 3 kcal mol-’ for (HSA) a-alumina, which is about one third that reported for v-aluminas, and this weakly adsorbed ethylene species is ascribed to ethylene molecules rr-bonded to cus surface aluminum cations. The weak interaction with ethylene allows the possibility that this a-A1,03 may be a satisfactory support for silver partial oxidation catalysts [ 341.

Acknowledgements This study was sponsored by the Aluminum Company of America. We would also like to thank Dr. A.J. Perrotta at Alcoa for providing characterization results and for useful comments.

References ] I] [2] [3] [4] [5] [6] [7]

[ 81 [9] [IO] [ II] [ 121 [ 13 ] [ 141 [ 151 [ 161 [ 171 [ 181 [ 191 [20] [21] [22] [23] [ 241 [25] [26] [27] [28] [29] [30]

C.L. Thomas, Catalytic Processes and Proven Catalysts, Academic Press, New York, 1970. V.E. Hemich, Rep. Prog. Phys., 48 (1985) 148. H. Knozingerand P. Ratnasamy, Catal. Rev. Sci. Eng., 17 (1978) 31. C. Morterra, S. Coluccia, A. Chiorino and F. Boccuzzi, J. Catal., 54 ( 1978) 348. C.-F. Mao and M.A. Vannice, in preparation. S.J. Gregg and K.S.W. Sing, Adsorption, Surface Area, and Porosity, Academic Press, London, 1982. S.R. Seyedmonir, D.E. Strohmayer, G.L. Geoffroy and M.A. Vannice, Adsorpt. Sci. Technol., 1 (1984) 253. M.A. Vannice, B. Sen and P. Chou, Rev. Sci. Instrum., 58 ( 1987) 647. J.H. De Boer, The Structure and Properties of Porous Materials, Butterworth, London, 1958. B.C. Lippens, B.G. Linsen and J.H. De Boer, J. Catal., 3 (1964) 32. A. Marcilla-Gomis, M. Molina-Sabio and F. Rodriguez-Reinoso. J. Catal., 99 (1986) 226. J.H. De Boer and B.C. Lippens, J. Catal., 3 (1964) 38. T. Kotanigawa, M. Yamamoto, M. Utiyama, H. Hattori and K. Tanabe, Appl. Catal., 1 ( 1981) 185. S.J. Gregg and J.F. Langford, J. Chem. Sot., Faraday Trans. I,73 (1977) 747. T.H. Ballinger and J.T. Yates, Jr., Langmuir, 7 (1991) 3041. A. Zecchina, E.E. Plater0 and CO. Arean, J. Catal., 107 ( 1987) 244. C. Morterra, S. Coluccia, E. Garrone and G. Ghiotti, J. Chem. Sot., Faraday Trans. I, 75 (1979) 289. G.D. Gatta, B. Fubini, C. Ghiotti and C. Mortcrra, J. Catal., 43 ( 1976) 90. S. Ishida, S. Imamura and Y. Fujimura, React. Kinet. Catal. L&t., 43 (1991) 447. M.C. Manchado, J.M. Guil, A.P. Masia, A.R. Paniego and J.M.T. Menayo, Langmuir, submitted for publication. E.A. Paukshtis, R.I. Soltanov and E.N. Yurchenko, React. Kinet. Catal. Lett., 16 (1981) 93. A. Auroux and A. Gervasini, J. Phys. Chem., 94 (1990) 6371. C. Morterra, S. Coluccia, G. Ghiotti and A. Zecchina, 2. Phys. Chem N.F., 104 ( 1977) 275. M.P. Rosynek, J. Phys. Chem., 79 (1975) 1280. W. Schubart and H. Knbzinger, Z. Phys. Chem. N.F., 144 (1985) 117. C. Morterra. A Zecchina, S. Coluccia and A. Chiorino, J. Chem. Sot., Faraday Trans. I,73 (1977) 1544. N.D. Parkyns, J. Phys. Chem., 75 (1971) 526. D.J.C. Yates and P.J. Lucchesi, J. Phys. Chem., 67 (1963) 1197. P.J. Lucchesi, J.L. Carter and D.J.C. Yates, J. Phys. Chem., 66 (1962) 1451. T.A. Gordymova and A.A. Davydov, Kinet. Katal., 20 (1979) 727.

C-F Mao, h4.A. Vannice / Applied Catalysis A: General 1 I1 (1994) 151-173

[31 I [32] [33] [34] [35]

173

J.D.D.C Ruiz, J.M.G. Pinto, A.P. Masia and A.R. Paniego, J. Chem. Tbermodyn., 18 (1986) 903. S. Cavallaro, P. Antonucci, N.V. Truong, N. Giordanoand J.C.J. Bart, Z. Phys. Chem., 262 (1981) 996. K.L. Anderson, J.K. Plischke and M.A. Vannice, J. Catal., 128 (1991) 148. P.T. Conner, S. Kovenklioglu and D.C. Shelly, Appl. Catal., 71 (1991) 247. T. Tsuchida, Appl. Catal, 105 (1993) Ll41.