A quantitative diffuse reflectance spectroscopy study of supported

Working Surface Science Model for the Phillips Ethylene Polymerization Catalyst: Preparation and Testing. P. C. Thüne, C. P. J. Verhagen, M. J. G. va...
1 downloads 0 Views 1MB Size
4156

J. Phys. Chem. 1993,97, 4156-4163

A Quantitative Diffuse Reflectance Spectroscopy Study of Supported Chromium Catalysts Bert M. Weckbuysen,' Leo M. De Ridder, and Robert A. Schoonheydt Centrum voor Opperulaktechemie en Katalyse, K.U. Leuven, Kardinaal Mercierlaan 92, B-3001 Heverlee, Belgium Received: December 4, 1992; In Final Form: February 2, 1993

The diffuse reflectance spectra of chromium supported on alumina, on silica, and on silica-alumina have been investigated after calcination, C O reduction, and recalcination. A method for quantifying the Cr6+, Cr3+,and Cr2+concentrations was developed. After calcination at 550 OC,the chr0mate:dichromate ratio is m, 0.56, and 2.18 on respectively alumina, silica, and silica-alumina. Reduction results in the formation of three new species: octahedral Cr3+, octahedral Cr2+, and pseudotetrahedral Cr2+. On silica and silica-alumina, the relative proportions of these three species depend on the reduction temperature. On alumina, only octahedral Cr3+ and small amounts of octahedral Cr2+ were observed after reduction. Recalcination reestablishes the chromate: dichromate ratio; however, this regeneration is complete only on silica. On alumina and silica-alumina, about one-third and two-thirds respectively of the total Cr content remain as Cr3+. The sensitivity of Cr6+to reduction is greater on silica than on alumina and silica-alumina. This is indicated by the increase of the Cr2+:Cr3+ratio from alumina over silica-alumina to silica.

Introduction Chromium on silica (Cr/SiOz) is the famous Phillips catalyst for the polymerizationof ethyleneunder relatively low pressures.] This catalyst is the basis for the Phillips particle-form process in the production of high-density polyethylene,2 and the use has been extendedto the production of linear low-density polyethylene3 and the copolymerizationof ethylene and butene- 1 .4 Chromium on silica-alumina (Cr/Si02*A1203)catalyses the polymerization of propylene5 and chromium on alumina (Cr/A1203)catalysts are active in the dehydrogenation of alkanes, in dehydrocyclization, and in catalytic reforming.6 The Phillips catalyst is one of the most studied and yet most controversial systems,' while the studies of Cr/A1203and Cr/ Si02.Al203 are more limited. All these studies deal with the formation and nature of the chromium species on the catalyst surface or with the associated problem of identifying the valence state of the species which are responsible for catalysis. Different techniques,such as infrared spectroscopy(IR)? diffuse reflectance spectroscopy(DRS)? X-ray photoelectron spectroscopy (XPS),lo Raman spectroscopy,Iland electronspin resonance (ESR),12were used. There is a general consensus that the detectable oxidation states are Cr6+,Cr5+,Cr3+,and Cr2+. However, the coordination geometry of these surface complexes and the degree of polymerization of Cr on the surface are a matter of debate. After calcination, most of the Cr is stabilized as Cr6+on the Si02 and Alz03surfaces;only the exact mechanism of stabilization and the molecular structure of the stabilized Cr6+species are points of discussion. According to Hogan and McDaniel, only chromate is formed on the Si02surface,I3J4while other research groups favor the formation of dichromate.15-19 Raman studies on Cr/Si02 give evidence for mono-, di-, tri-, and tetramers under hydrated conditionsZoand highly distorted monochromate under dehydrated conditions.21 On A1203,mostly chromate is formed after calcination.17v22 However, Raman studies reveal the presence of monomers and dimers under hydrated conditions20 and polymeric oxide species under dehydrated condition^.^^ To our knowledge, the molecular structure of Cr on Si0yA1203 is not yet reported in the literature. Reduction of the calcined catalysts results in the formation of traces of Cr5+,Cr3+,and Cr2+,but the relative proportion of these species depends on the reduction temperature and on the

* To whom correspondence should be addressed.

support. The coordination of the reduced species is unclear. Cr3+is mostly thought of as octahedral, but according to Myers and Lunsford, a small part can be in a lower symmetry.24 Cr2+ can be unsaturated, tetrahedrally coordinated, and squareplanar, 18,19.26.29 Table I give a survey of DRS absorption bands and proposed assignments. All these DRS studies are qualitative, describing the different oxidation states formed after calcination and reduction. However, the data of different groups cannot be easily compared because of differencesin pretreatments and Cr loadings. Therefore, the purpose of this work is a quantitative study of chromium-supported catalysts by DRS. This technique allows the quantitative determination of Cr2+,Cr3+, and Cr6+ after calcination, reduction, and recalcination at least at small loadings. The redox behavior of Cr on Al203, S O 2 , and Si02.A1203will be discussed. Experimental Section Sample Preparation and Characterization. 1. Preparation. S i 0 2and SiO2.A1203 were home-made. Si02 was prepared by mixing 2 vol parts H20 at pH 2 (adding HCl) and 1 vol part tetraethyl orthosilicate (TEOS) during 5 h at room temperature. The mixture was titrated under stirring to pH 6 with a NH40H solution of pH 9.5. After 16 h of gelation, the gel was dried at 130 O C for 72 h and calcined at 250 and 550 OC for respectively 3 and 16 h. The so-obtained cake was crushed. Si02.A1203 with 40 wt 5% Si02was prepared following a modified method of Chen et al.33 TEOS (35.5 g) and aluminum triisoprop'xide (65.36 g) were mixed in 128 mL of ethanol during 30 min at room temperature. After adding 35 mL of 1 M HCl, the hydrolysis started and the solution was mixed for 1 h. Acid hydrolysis resulted in a sol and a flocculated phase, suggesting a heterogeneous SiOyA1203. The gel was dried at 60 and 100 OC for 8 h for both and calcined at 550 O C for 16 h. The Si02.A1203cake was crushed. A1203was delivered by Rhbne Poulenc (reference number EC 21/581/88). The Cr catalysts were prepared by the incipient-wetnessmethod with chromium(V1)oxide (Cr03).The chromium loadings were 0.1, 0.2,0.4, and 0.8 wt %. 2. Characterization. The characteristics of the supports were measured by dynamic N2 adsorption on an Omnisorp 100 (Coulter), after pretreatment in vacuo at 200 O C for 8 h. Si02 had a BET surface area of 735 m2/g, a pore volume of 0.8 mL/g, and a pore size of 1-3 nm. SiOyA1203 had a BET surface area

0022-3654/93/2097-4156$04.00/0 0 1993 American Chemical Society

Spectroscopy Study of Supported Cr Catalysts

The Journal of Physical Chemistry, Vol. 97, No. 18, 1993 4151

TABLE I: Literature Survey of DRS Absororption Bands of Cr Catalysts and Their Assignments sample

sample pretreatment

absorption bands, cm-I

assignments'

ref

0.3 wt % Cr/SiO2

CO: 3OOOC

02: 6OOOC CO: 35OOC CO: 35OOC 0 2 : 55OOC CO: 35OOC 0 2 : 6OOOC 02: 4OOOC CO: 35OOC 02: 65OOC CO: 3OOOC CO: 5OOOC 0 2 : 5OOOC C O 44OOC 0 2 : 700 OC 0 2 : 6OOOC 02: 4OOOC 02: 4OOOC

Cr2+,Cr3+, Cr6+ Cr2+tslr Cr5+,; dichr Cr2+, Cr3+ Cr2+ dichr (chrom) Cr2+Una Cr6+mpl Cr6+ Cr2+ chrom Cr2+,,, Cr2+,n, dichr Cr2+,,l Cr3+, Cr6+ Cr6+1,1 Cr6+ Cr6+

26

0.5 wt % Cr/SiOz

12 500,13 000,15 000,21 500,25 000,26 000 13 000 18 000,21 700,26 000,28 500,39 000 7500,lO 000,12 500,15 000,31000,38 500 8000,12 200,28 000,31 000 21 000,29 000,41000 8000,12 200,28 000,30 000 16 700,21 500,27 800,34 500 21 700,27 000 13 700 22 200,27 000 7800,12 000,28 000 8000,13 000,32 000 20 5000,28 000,38 200,47 000,5 1 000 14 500,20 700 16 700,21 600,27 000,36 000 23 300,27 800,36 400 17 250,23 250,27 000,37 000 16 700,23 250,27 000,37 000

0.5 wt % Cr/Si02 0.5 wt % Cr/SiOz 0.5 wt % Cr/Si02 1.O wt % Cr/SiO2 1.O wt % Cr/Si02 1.O wt % Cr/Si02 2.0 wt % Cr/Si02 9.0 wt 0.5 wt 1.5 wt 1.8 wt

% Cr/Si02 % Cr/A1203 76 Cr/A1203 % Cr/A1203

co: m0c

15 27 28 25 29 14 30 19 30? 25 31 32

mpol, more polymerized; Ipol, less polymerized; uns, unsaturated; chrom, chromate; dichrom, dichromate; sqpl, square-planar.

of 364 m2/g, a pore volume of 0.92 mL/g, and a pore size of 1.5-7 nm with a bimodal distribution (4 and 6 nm). A1203 had a BET surface area of 3 16 m2/g, a pore volume of 0.35 mL/g, and a pore size of 1.5-4 nm. Pretreatmentand Spectroscopy. 1. Pretreatment. The samples were dried at 50 OC for 8 h and granulated. The size fraction of 0.16-0.25 mm was loaded in a quartz flow cell with Suprasil window for DRS and a side arm for ESR. With this cell, it was possible to take DRS and ESR spectra after the same pretreatment. The ESR spectra will be discussed in a following paper. The samples were subsequentlydried at 90 OC during 16 h followed by calcination at 550 OC during 6 h in an oxygen stream. DRS spectra were recorded on these calcined samples. The samples were then reduced with CO at 200,300,400, and 600 OC during 30 min. After each reduction step, DRS spectra were taken. In the last step, the samples were recalcined in oxygen during 6 h at 550 OC and DRSspectra were taken again. The same procedure was followed for the supports without chromium. Their spectra were used as reference. An oxygen flow of 3600 mL/h and a CO flow of 1800 mL/h were used for all the treatments. 2. DiffuseReflectanceSpectroscopy. DRS spectra were taken on a Varian Cary 5 UV-vis-NIR spectrophotometer at room temperature. The spectra were recorded against a halon white reflectance standard in the range 2200-200 nm. The computer processing of the spectra consisted of the following steps: (1) subtraction of the baseline; (2) conversion to wavenumber; (3) calculationof the Kubelka-Munk (KM) funtion; (4) subtraction of the spectrum of the parent support from that of the chromiumsupported catalysts measured after the same treatment. The spectra were deconvoluted with the Spectra Calc program of Galactic Industries Corp. into Gaussian bands. The values of the KM function at the band maxima were used for quantitative DRS. Spectra of the solutions of K2Cr04,K2Cr207,and Cr(NO3)3-9H20 were taken at room temperature on a Cary 17, in the range of 800-210 nm. The spectra of the solids (K2Cr04, K2Cr207, Cr(N03)3.9H20,and Cr203) were taken on a Varian Cary 5 UV-vis-NIR at room temperature in the range 2200200 nm. Results The absorption maxima of the spectra and the color of the reference compounds are shown in Table 11. The deconvoluted spectra of catalysts pretreated in an identical way contain the same number of bands. The positions and widths are independent of the Cr loading. The position of the maxima and widths are summarized in Table 111. The consistency of the deconvolution is an indication that the proposd deconvolution has physical

TABLE II: Band Maxima of Spectra of Reference Compounds compound K2Cr04 (soh) K2Cr04 (solid salt) K~Cr207(soln) K2Cr207 (solid salt) Cr(N03)3-9H20 (soln) Cr(NO3)3*9H20 (solid) amorphous Cr2O3

absorption bands.' cm-I 22 700 (sh, vw), 27 000 (sl. 36 350 (s) . 21 800 (s), 29 400 (s), 37 700 (sl. 43 600 (sl 22 700 (Wj, 28 400 (S j, ' 39 200 (s) 19 000 (s, br), 30 100 (s), 38 200 (s), 43 600 (s) 17 400 (s), 24 400 (s), 33 000 (s) 17 400 (s), 24 400 (s), 32 900 (s), 38 000 (sh) 14 000 (sh), 15 500 (sh), 16 800 (s), 21 700 (s), 28 500 (s), 36 500 (s)

color vellow yellow orange orange-red green green green

s, strong; m, medium; w, weak; vw, very weak; sh, shoulder; and br,

broad.

meaning. The fact that a consistent interpretation can be given is another strong point of the deconvolution. 1. Calcined Samples. The color of the samples was yellow for Cr/A1203 and Cr/SiO**A1203and yellow-orange for Cr/Si02. The spectra of the calcined 0.2 wt ?6 loading are shown in Figure 1, together with the proposed deconvolution in Gaussian bands. The spectra of Cr/A1203 are dominatedby two bands at 27 500 and 41 000 cm-I. By deconvolution,two low-frequencyshoulders at 22 500 and 34 400 cm-l are resolved. Such a spectrum is typical for metallochr~mates.~~ The bands are 0 Cr6+chargetransfer bands with the followingassignment? the band at 22 500 cm-I is the symmetry-forbiddentransition (1tl 2e), while the otherbandsat 27 000(lt]-2e),34 400(lt,-7t2),and41 400 cm-I (6t2 2e) are symmetry-allowed. The first transition is partially allowed in solid salts, as is concluded by comparing the spectra of K2Cr04in solution and as a solid (Table 11). For Cr/SiOz, four bands at 15 500,22 000,30 500, and 40 500 cm-I and a shoulder at 27 000 cm-I are visible by the eye. After deconvolution, 11bands are resolved. Three weak bands at 15 500, 21 500, and 33 900 cm-I with equal width and intensity are ascribed to octahedral Cr3+,in accordance with the literature.36 The remaining eight bands are due to chromate (21 000,27 000, 34 000, and 37 300cm-I) anddichromate (22 000,31 000,35 500, and 40 700 cm-'). The dichromate bands are systematically shifted t o the blue with respect to the chromate bands. This in accordance with the solution spectra in Table I1 and literature data.37138The assignments are the same as previously given for Cr/A1203.

-

-

4758 The Journal of Physical Chemistry, Vol. 97, No. 18. 1993

Weckhuysen et a].

TABLE IIk Survey of the Positions and Widths of the Bands in the Deconvoluted Spetra of the Cr Catalysts Cr catalyst

pretreatment calcination 550 OC reduction 200 OC reduction 300 "C reduction 400 OC reduction 600 OC recalcination 550 OC calcination 550 OC reduction 200 "C reduction 300 OC reduction 400 "C reduction 600 OC recalcination 550 OC calcination 550 OC reduction 200 OC reduction 300 OC reduction 400 "C reduction 600 OC recalcination 550 "C

absorption bands, cm-I

27 000 (5500),22 500 (4900).34 400 (8000),41 400 (9200) 27 000 (5500),22 500 (4900),34 700 (8700),41 400 (8500) 12 500 (6500),17 000 (8000),23 OOO (8000),23 000 (8000),33 000 (8000), 22 600 (3000),27 000 (6000),34 400 (5600),41 300 (1 1 000) 12 500 (8700),15 300 (SOW), 26 100 (8000),34 700 (8000),40 600 (9600) 12 500 (6700),16 500 (SOW), 23 500 (6500),33 000 (6000),39 100 (7700) 17 000 (8000),23 000 (8000),33 000 (8000),22 500 (4000),27 000 (5800), 34 400 (7500),41 400 (9000) 15 500 (6000),21 500 (6000),33 900 (6000),21 000 (3500),22 000 (3500), 27 000 (SOOO),31 000 (5500),34 000 (4000),35 500 (4000), 37 300 (SOOO), 40 7000 (6000) 15 500 (6000),21 500 (6000),33 900 (6000),21 000 (3500);22 000 (3500), 27 000 (SOOO), 31 000 (5400),34 000 (4000);35 500 (4000).37 300 (SOOO),40 700 (6300) 10 000 (SOOO), 12 500 (SOOO), 15 500 (6100),21 000 (6000),33 400 (6000), 32 300 (8500),39 500 (8800) 10 000 (SOOO), 12 5000 (SOOO),15 5000 (6000),20 000 (6000),32 800 (6000), 32 700 (9000), 39 000 (8600) 10 000 (SOW), 13 000 (5000), 15 500 (6000),19 5000 (6100),32 500 (6000), 32 000 (8000),39 700 (9400) 15 000 (6000),21 000 (6000),33 000 (6000),21 100 (3500),22 000 (3500), 27 300 (SOOO), 31 000 (SOW), 34 100 (4000),35 400 (4100),37 3000 (6500),40 800 (7600) 22 200 (SSOO), 23 100 (5500), 27 300 (4900),31 000 (4600), 34 100 (4000), 35 500 (4000),39 200 (10000),41 500 (10000) 22 200 (5600),23 000 (5500).27 300 (4900),31 000 (4700),34 00 (4000), 36 600 (4000),39 200 (10 200),41 200 (10000) 10 000 (5000),13 1000 (6000), 17 100 (7900),23 100 (8100),33 000 (8100), 22 000 (5500),23 100 (5500),27 300 (5200),31 500 (4500),34 100 (4000),35 600 (4000), 39 300 (10600),41 500 (10 100) 10 000 (5000),13 100 (6000),17 100 (7900), 23 100 (8000);36 000 (8000), 28 200 (6500),39 100 (7900),33 700 (7900),41 000 (7000) 10 000 (5100),13 000 (5800),16 100 (5500). 22 200 (5100),35 300 (5500), 32 100 (8000),41 100 (6000) 15 100 (8100),23 100 (8000), 33 100 (8100),22 200 (5300),23 100 (5500), 27 300 (4700),31 000 (4200),34 100 (4000),35 600 (4200),39 200 (10000),41 800 (10300)

The DRS spectra of Cr/Si02.A1203 consist of two main bands at 27 000 and 41 000 cm-' and one shoulder at 22 000 cm-I. After deconvolution, eight bands are resolved. The bands at 22 200, 27 300, 34 100, and 39 200 cm-I are due to chromate, while the bands at 23 100, 31 000,35 500, and 41 500 cm-' are those of dichromate. The ratio of the intensitiesof the 27 000-cm-I band of chromate and the 3 1 OOO-cm-' band of dichromate is taken as a measure of the chr0mate:dichromate ratio on the different supports. This ratio of intensities is -, 0.62, and 2.40 for Cr/A1203, Cr/SiO2, and Cr/Si02.A1203, respectively, with 0.2 wt % and independent of the loading. If it is assumed that the ratio of the extinction coefficients of chromate and dichromate is the same in solution and on the the surface, then formula 1 allows the calculation of

-

Z(27 000 cm-I) - %hromateCchromate 1(31 000 Cm-l) 'dichromateCdichromate

-

Cchromate

(1)

Cdichromate

thechr0mate:dichromate ratios on the surface with K = 1.1 (from solution spectra). These ratios are -, 0.56, and 2.18 for Cr/ A1203, Cr/Si02, and Cr/SiO2.A1203, respectively. All the deconvoluted spectra have a broad and intense shoulder near 50 000 cm-I. It is more intense on Cr/Si02 than on Cr/AI2O3 and Cr/SiO*-A1203. It can only result from the interaction of Cr with the support. However, the band occurs in the limit of the working range of the integration sphere. As a consequence, both the intensity and the position are somewhat dubious, and we will not discuss it. 2. ReducedSamples. In Figures 2-4, the deconvolutedspectra of the reduced samples with 0.2 wt % Cr loading are shown. The spectra for the other loadings are identical, except for the intensity of the bands. In each case, reduction starts above 200 O C and is accompanied with typical color changes. 2.1. Cr/A1203. After reduction at 300 O C , the color changes to yellow-green and two new bands appear at 12 500 and 17 000 cm-I. The band at 17 000 cm-I is due to Cr3+ in octahedral

200

100

0

N

b r

*

500

2 v

LL

C

400

20c

C 4b0g

ZCCCC

cm" Figure 1. Deconvolutionsof the DRS spectra of 0.2 wt %I Cr/A1203 (A), Cr/SiO2 (B), and Cr/Si02.A1203 (C) after calcination at 550 OC.

Spectroscopy Study of Supported Cr Catalysts

The Journal of Physical Chemistry, Vol. 97, No. 18, 1993 4759

a 200

100

0

100

N

50

b r

*

z 0

*

2

v

LL

I

6oc

v

LL

LOO 50

2co

0

0

D

50

0 1

20000

40000

cm"

.?coo0

40000

cm"

Figure 2. Deconvoluted spectra of 0.2 wt % Cr/A1203 after reduction at 200 O C (A), 300 O C (B),400 OC (C), and 600 OC (D).

Figure 3. Deconvoluted spectra of 0.2 wt W Cr/SiOl after reduction at 200 O C (A), 300 O C (B),400 ' C (C), and 600 O C (D).

coordination,while the weak 12 500-cm-I band can be ascribed to octahedral Cr2+. An octahedral complex of a high-spin 3d4 ion such as CrZ+would be expected to show only a single spinallowed d-d transition, 5E, 5 T ~ gin, the visible region of the spectrum.39 However, the band may be broadened due to a JahnTeller effe~t.3~ After reduction at 400 OC, the samples turn white and the spectra are dominated by the three d-d transitions of octahedra Cr3+ at 15 300, 26 100, and 34 700 cm-I. The first band shifts by 1200 cm-I to higher wavenumbers after reduction at 600 O C . An intense band at 39 000 cm-' is always present. The Cr2+ band is weak at all reduction temperatures. 2.2. Cr/SiOz. The color of the sample turns to green-blue after reduction at 300 OC, and the DRS spectrum changes drastically. The Cr6+ bands completely disappear. In the d-d region, two new bands at 10 000 and 12 500 cm-l appear. The intensities of these bands vary independently with the reduction temperature, suggesting the presence of two different species. The 12 500-cm-I band is ascribed to Cr2+ in octahedral coordination, while the band at IO 000 cm-' suggests the presence of a Cr2+ ion in a lower symmetry. Similar band maxima are reported for distorted tetrahedral complexes of CrZ+complexes like CrC142-and CrBrd2-i0ns.~6Besides Cr2+,the threed-d bands

of octahedral Cr3+are also found. Finally, two new bands appear in the UV at 32 300 and 39 500 cm-1. After reduction at 400 OC,the sample was blue-violet and became blue after reduction at 600 OC. The CrZ+band intensities increase with the reduction temperature at the expense of the Cr3+bands. The intensities of the bands at 32 000 and 39 500 cm-I change independently with reduction temperature. The 32 000-cm-l band decreases and the 39 500-cm-I band increases with increasing reduction temperature. 2.3. Cr/Si02AlP3. Increasing the reduction temperature resultsinaslowdecreaseoftheintensityofCfl+bands. Chromate and/or dichromate bands disappear completely only after reduction at 600 O C . After reduction at 300 OC, the color turns to yellow-green, while after reduction at 400 and 600 OC the samples are light blue and blue-gray, respectively. Three new bands are observed in the DRS spectrum after reduction at 300 OC: a band at 10 000 cm-I typically for pseudotetrahedral Cr2+, a band located at 13 100cm-l due to octahedral Cr2+,and a band of octahedral Cr3+at 17 100 cm-I. Two bands appear in the UV at 32 100 and 41 100 cm-I. The intensity of the 32 100-cm-1 band decreases with increasing reduction temperature but that of the 41 lOO-cm-' band remains more or less the same.

-

4160

The Journal of Physical Chemistry, Vol. 97, No. 18, 1993

Weckhuysen et al.

400

2CO

0

200

'C0 (Y

b r *

0

C

2cc -

I

2co

2cccc

I

'CCOC

cm-' Figure 5. Deconvolutions of the DRS spectra of 0.2 wt 5% Cr/A1203 (A), Cr/SiOz (B), and Cr/Si02.A1203 (C) after calcination at 550 "C.

^^-^^

-CC?C

LMl_i

cm"

four chromate bands (22 200,27 300,34 100, and 39 200 cm-I), and four dichromate bands (23 100,31 000,35 600, and 41 800 cm-I). However, only about one-third of the chromate and dichromate bands can be regenerated by recalcination. 4. Quantitative Diffuse Reflectance Spectroscopy. 4.1. Calibration. The KM function allows a quantitative determination of absorbing species according to

Figure4. Deconvolutedspectra of 0.2wt ?'& Cr/Si02.A1203 after reduction at 200 OC (A), 300 OC (B), 400 O C (C), and 600 O C (D).

2.4. Chemiluminescence of Reduced Samples. Cr/Si02 has a spectacular chemiluminescence, first described by Hogan after exposure of the reduced samples to oxygen.I3 The intensity of the yellow-orange light flash decreases with increasing reduction temperature. The chemiluminescence was more intense on Cr/ SiOz than on Cr/Si02.A1203, while no chemiluminescence was observed on Cr/A1203. 3. Recalcined Samples. In Figure 5 , the deconvoluted spectra of the recalcined samples are shown. On Cr/A1203, the chromate bands (22 500,27 000,34 400, and 41 400 cm-I) reappear in the spectrum, but the Cr3+ bands located at 17 000, 23 000, and 33 000 cm-' do not completely disappear. As a consequence, the intensities of the chromate bands are lower than after the initial calcination, and the color of the recalcined samples was yellowgreen. On Cr/SiO2, the spectrum after recalcination was almost the same as before the reduction cycle and the sample color was orange, indicating complete oxidation. The slightly higher intensities for all the absorption bands in the spectrum are due to differences in light reflectance properties of the recalcined samples. Recalcination of Cr/Si02.A1203gives the samples a green-yellow color. The DRS spectrum reveals three bands (15 100, 23 100, and 33 100 cm-I) typical for octahedral Cr3+,

where R, is the ratio of the light intensity reflected from the sample and the light reflected from the standard.40 K and S are respectively the KM absorption and scattering coefficients. When at a given wavelength S is constant, eq 1 gives a linear relation between F(R,) and the Cr concentration, Cc,. Conditions, inherent to the use of the KM function are ( 1 ) infinitely thick sample, (2) random distribution of particles which are much smaller than the layer thickness, (3) isotropic scattering and diffuse irradiation, and (4) samples with low absorption.41 The first, second, and third conditions are fulfilled by using powdered samples, while the third condition suggests the use of low Cr loadings. It is also assumed that S does not change with loading and treatment. In the following procedure, calibration lines are obtained for Cr3+and Cr6+. On Cr/A1203, Cr6+is present as chromate, and the Cr6+ calibration line is obtained by plotting the intensity of the 27 000-cm-1 band of chromate as Cr loading. From the ratio chr0mate:dichromate = 2.40, it is possible to obtain the Cr6+ calibration curve for CrlSiO2aA1203from the sum of the intensities of the 27 000-cm-I band (chromate) and 31 000-cm-1 (dichromate) bands. For Cr/SiO2, the same procedure was followed

The Journal of Physical Chemistry, Vol. 97, No. 18, 1993 4761.

Spectroscopy Study of Supported Cr Catalysts

100

1000 A 900 800

* roo r a c 600 a -.cI 500 400 300 t 200 -

c.

0

.-0

Y

A

0 0.00

J

t

0

lo",

80

E

CI

. 0.16

.. , . . . ., 0.32

0.48

. . . , . . . .

1

60 40

20 0

0.80

0.64

calc.

200

300

400

600 recalc.

Pretreatments

wt% C P 100 L .

60

1

80

E

I I

le

CI

a 8

60

5

J

t

40 20 0

calc.

200

400

300

600 recalc.

Pretreatments

0.00

0.08

0.16

0.24

0.32

0.40

wt% C P

Figure 6. Calibration lines for Cr6+ (A) and Cr?+ (B): intensity of vs Cr6+13+concentration on absorption band of Cr6+13+(in F(R,) Cr/A1203 (A), Cr/SiOz (V), and Cr/SiO2-A1203 (0).

after correction for small amounts of Cr3+. This correction was obtained from the Cr3+ calibration line, discussed below. The Cr3+-linefor Cr/A1203 and Cr/SiO2*Al203was based on the spectra of the recalcined samples. These contain Cr3+and Cr6+. The latter is determined with the Cr6+calibration lines. The Cr3+loadings are then easily obtained by subtracting the Cr6+loadings from the total Cr loadings. The Cr3+calibration lines for Cr/A1203and Cr/SiO2.Al2O3were obtained by plotting respectively the intensity of the 17 000-cm-I band and of the 15 100-cm-1 band vs the calculated Cr3+ concentrations. The Cr3+calibration line for Si02 was obtained by relating the intensity of the 15 500-cm-I band with the Cr3+ concentration (36%of the total Cr loading) in the calcined samples of different chromium loadings. Thecalibration lines for Cr6+and Cr3+on thedifferent supports are shown in Figure 6. The calibration line for Cr6+on A1203 is almost linear, while that of Cfi+on Si02 and Si02-A1203deviates from linearity above 0.16 and 0.4 wt %, respectively. The same holds for the Cr3+calibration lines, but that of the Cr3+ on Si02 cannot be called linear. Therefore, we restrict a quantitative analysis to samples with 0.1 wt % Cr. One can apply the method for high loadings too in the case of Cr/A1203 and Cr/Si02-A1203 but not for Cr/SiOz. In any case, the slopes of the linear portions of the calibration lines increase from Cr/A1203 over Cr/Si02=A1203to Cr/SiOz. This is due to Sin eq 2 which follow the order A1203> Si02.Al203 > Si02. 4.2. Results. Theconcentrationsof Cr2+,Cr3+,and Cfi+ after different pretreatments are shown in Figure 7. Cr2+ is always determined by thedifference [Cr2+]= Crlolal-([Cr6+] [Cr3+]). On Cr/A1203, the following observations are made: (a) The chromateloading decreaseswith increasing reduction temperature to reach zero after reduction at 400 OC. (2) About 96%of the reduced Cr species is Cr3+; the remaining is Cr2+. (3) The concentration of Cr2+is low, and the variations are within the limits of experimental accuracy. (4) After recalcination, only 58% of the total Cr content shows up as chromate; the remaining 42%is Cr3+.

+

calc.

200

300

400

600 recalc.

Pretreatments

Figure7. Distribution of Cr6+(black), Cr3+(dark gray), and Cr2+(light gray) on Cr/A1203 (A), Cr/SiO2 (B), and Cr/SiO2-Alz03 (C) after different pretreatments.

TABLE I V Cr2+:Cr3+Ratios on 0.1 wt % Cr Catalysts Determined by DRS

Cr2+/Cr3+ratio after reduction at Cr catalvsts

300 OC

400 OC

600 OC

Cr/A1203 Cr/SiOz Cr/SiOyAl203

0.2 1.30 0.86

0.07 1.34 1.10

0.050 2.85 0.76

On Cr/SiOz, the method is less accurate because there are more species present on the surface. Nevertheless, the following trends are observed: (1) The Cr6+concentrationis quantitatively converted to Cr2+ and Cr3+after reduction at 300 OC. (2) The CrZ+:Cr3+ratio increases with increasing reduction temperature. (3) The maximum Cr2+concentration of 74%was obtained after reduction at 600 OC. (4) After recalcination, the same chromate: dichromate ratio was established as after the initial calcination. On Cr/Si02*Al2O3,the following observations are made: (1) The Cfi+concentrationdecreasesslowly with increasing reduction temperature and was zero only after 600 OC reduction. (2) Reduced Cr was Cr2+and Cr3+. The amount of Cr3+increases with increasing reduction temperature, while the Cr2+ concentration has a maximum of 44%after reduction at 400 O C . (3) After recalcination, only 37% of the total chromium was regenerated as chromate or dichromate. In Table IV, the CrZ+:Cr3+ratio is shown for A1203,Si02, and Si02-Al203 after reduction at 300,400, and 600 OC. This ratio increases from Cr/A1203 over Cr/SiO2*Al203 to Cr/SiO2, whatever the reduction temperature. Furthermore, while for Cr/ Si02 the Cr2+:Cr3+ratio increases with reduction temperature,

Weckhuysen et al.

4762 The Journal of Physical Chemistry, Vol. 97, No. 18, 1993

a maximum is found for Cr/Si02.A1203. On Cr/A1203, the variation of the Cr2+:Cr3+ ratio is insignificant because of the large errors on the Cr2+ concentrations. Discus5i0n 1. EduationoftheMetbod. It is possible toquantify supported Cr2+,Cr3+,and Cr6+ provided the followingis fulfilled: (1) a low Cr content; (2) the absence of Cr3+is calcined samples; and (3) the presence of Cr3+in recalcined samples. All these factors are fulfilled for Cr/A1203 and Cr/Si02-A1203, while the presence of Cr3+ complicates the quantification of the spectra of calcined Cr/SiOz. We have presented here the data of 0.1 wt % Cr. Higher loadings are possible but with less accuracy. The best results were obtained for Cr/A1203because there are only two species: chromate and Cr3+. Thus, the calibration lines (Figure 6 ) are straight lines up to 0.4 wt % Cr, and our analysis is valid between 0 and 0.4 wt % Cr. Cr/Si02 has a multitude of species: chromate, dichromate, and Cr3+on calcined samples and cr3+,Cr2+(Oh),Cr2+(Td)on reduced samples. Cr2+could only be determined by difference, and Cr2+(Oh) and Cr2+(Td) cannot be determined separately in a quantitative way. Cr/ Si02.Al203 takes a position intermediate between Cr/A1203 and Cr/Si02. On the basis of Figure 6 , quantitative determinations of Cr species are possible between 0 and 0.1 wt % for Cr/SiO2 and in the range 0-0.2 wt % for Cr/Si02.A1203. 2. Spectroscopy of the Chemistry of Supported Cr. With DRS spectroscopy in the UV-vis-NIR, we probe the charge-transfer and d-d transitions of Cr. The charge-transfer transitions are of the type 0 Cr and are typical for Cr6+in chromate and dichromate form. Cr2+ and Cr3+ have characteristic d-d transitions, which are much weaker than the charge transfers. In any case, both charge transfers and d-d transitions are characteristicfor theoxidation stateof Cr andofthenature andgeometry of the atoms in the coordination sphere, in our case oxygens. Long-range effects, for instance the distinction between monomericchromate and polymeric chromate, cannot be madedirectly. So our results do not allow us to make firm conclusions on the degree of polymerization of Cr on the various supports. 2.1. Calcined Samples. The chr0mate:dichromate ratio depends on the type of support, both before and after calcination. Two explanations can be advanced: (1) The chr0mate:dichromate ratio only depends on the surface properties of the supports. (2) The chromate:dichromate ratio expresses the degree of polymerization of Cr: the more dichromate, the more polymerized Cr is present on the surface. Both hypotheseshave their merits. Indeed, the surface of A1203 (pHzc = 7-8)42 is a neutral-to-basic surface, and only chromate is stable on this surface. Dichromate occurs on Si02 with ~ H Z C = 1-2,42 Le., an acidic surface. Our Si02nA1203 is an interesting case. The chr0mate:dichromate ratio can be explained within 5% of the true value (2.18) if it is assumed that it consists of 40% of the Si02 regions with a chr0mate:dichromate ratio of pure Si02 (0.56) and of 60% of the A1203 regions which carry chromate only. This means that the SiOz.AlzO3 region can be visualized by a zoned material. The second hypothesis is indicative of the fact that Cr6+is better dispersed on A l 2 0 3 than on Si02.25 2.2. Reduced Samples. (i) Formation and Coordination of Cr3+ and Cr2+. The reduction of Cr with CO starts above 200 OC and leads to the formation of Cr2+and Cr3+, with relative amounts depending on reduction temperature and support. The Cr3+species have characteristic octahedral spectra, and the energy of the first transition is equal to 10 Dq, the ligand field strength.36 This is at 15 500 cm-I for Cr/SiO2, at 17 100 cm-I for Cr/Si02.A1203, and initially at 17 000 cm-1 for Cr/A1203, decreasing to 15 300 cm-' at intermediate reduction temperatures and increasing again to 17 000 cm-1 after reduction. The 15 500-cm-I value for octahedral Cr3+ on silica must be ascribed to the surface phaseof Cr3+. On one hand, it is strongly interacting

-

with the support, because a-Cr203is not formed, yet it is reducible to Cr3+. Cr3+ on A1203 and on SiO~A1203is-at the small loadings of the present investigations-substituting for octahedral A13+in the surface layers of the supports. It cannot be reduced to Cr2+ to an appreciable extent for two reasons: (1) some Cr3+ migrates into the support, where it becomes unaccessible for reduction; (2) the majority of Cr3+, however, forms a-CrzO3 particles at high reduction temperatures, which cannot be reduced and only slowly oxidized. The Cr ions in the A1203regions of Si02~A1203behave as these of A1203; those in the Si02 regions behave as those of 5 3 0 2 . Octahedral Cr2+ is a surface species on all supports. The tetrahedral Cr2+,always more difficult to make and always at lower concentration,may be Cr2+in the surfaceof S O 2 ,occupying empty sites Si4+. Because of the size difference of Si4+and Cr2+, this can only be a distorted tetrahedral site. Summarizing, Cr is a surface species on Si02, fully available for reduction by CO and reoxidation with 0 2 . On A1203, part of Cr3+migrates in the support and part forms a-Cr203. Both types of Cr3+ cannot be reduced to Cr2+ and are only slowly reoxidizable to chromate. In this chemistry, the properties of the supports certainly play a role. In terms of hardness and softness principles, first introduced by Pears0n,4~one can state that A1203 is harder than Si02;4u6 Le., A1203is less polarizable or less susceptible to a change in number of electrons than on S O 2 .We suggest that is also one of the reasons for the difference in reducibility of Cr on alumina and silica. It implies that surface oxygens are actively involved in the reduction and reoxidation reactions. (ii) Interpretation of the Bands in the UV. In the UV region, two bands at 32 000 and 39 500 cm-I on reduced Cr/SiO2 were observed, while on reduced Cr/Si02.A1203 these bands were shifted to 32 500 and 41 000 cm-I. On reduced Cr/A1203, only the 39 lOO-cm-' band was found. Similar bands were observed on reduced Cr/Si02 by Zecchina et al.ls and Fubini et al.l* These bands may conveniently be explained as 0 Cr2+,Cr3+ CT bands. Indeed, using theconcept of optical electronegativity,4? the CT frequency can be calculated as

-

dCT(cm-') = 30 000 (xop,(02-)- xOpt(Cr2+,Cr3+))(3) where x,,,(02-) and xopt(Cr2+,Cr3+) are respectively the optical electronegativity of the surface oxygen and of Cr2+ and Cr3+. With xopt(Cr3+)= and xopt(02-)= 3.5 (Si02)49 or 3.3 (A1203),49 one has v c ~ ( 0 Cr3+) = 45 000 and 39 000 cm-l for Si02 and A1203,respectively. The latter is in agreement with the 39 500-41 000-cm-l band. Weascribethen the 32 500-cm-I band to an 0 Cr2+(Td) transition on silica. The shift to lower frequencies is in the right direction because tetrahedral species have smaller ligand field strengths than octahedral c ~ m p l e x e s . ~ ~ This interpretation follows that of Zecchina et al.15 and Fubini et However, it is also clear that the situation is more complicated because three bands have been reported and that each band may encompass several components. However, our results do not allow a more detailed discussion. (iii) Chemiluminescence of Reduced Cr. The chemiluminescence of reduced Cr/SiOz is described by several authors in the literat~re.~3~'**~~~SO Our observations suggest that only the Cr2+ species, corresponding to the absorption band at 10 000 cm-I, is responsible for the light flash. Octahedral Cr2+ is excluded, because it was observed on Cr/A1203 in small amounts, while no chemiluminescence was observed on these samples. The more intense chemiluminescence of Cr/SiO2 than of Cr/Si02.A1203 can be explained by the higher intensity of the 10 000-cm-l band. Further, the intensity of chemiluminescence is decreasing with decreasing intensity of the CTI,red. on Cr/Si02 and Cr/Si02.A1203, while no chemiluminescenceoccurred when the C T I ,was ~ ~absent, as for reduced Cr/A1203.

-

-

Spectroscopy Study of Supported Cr Catalysts

Conclusions The deconvolution of the DRS spectra of Cr catalysts allows a quantitative estimation of the amount of Cr6+, the chromate: dichromate ratio, and the Cr3+and Cr2+concentrations at any treatment temperature provided the loading is small, typically 0.2 wt % for alumina and silica-alumina and 0.1 wt % for silica. The chr0mate:dichromateratioincreases from silica over silicaalumina and alumina, and this is due to the surface properties of the supports and differences in dispersion. On alumina, CO reduction gives only Cr3+with traces of Crz+,and the amount of Cr2+increasesover silica-alumina to silica. In the latter case, (pseud0)tetrahedral and (pseud0)octahedral Cr2+are formed. Theintensechemiluminescenceisdueto(pseud0)tetrahedralCr2+. On silica, Cr2+/3+is completely reoxidizable to Cr6+. This is only partially the case for A1203and Si02.Al203, because formed cr-CrzO3 is only slowly oxidizable. The supports take an active part in the redox chemistry of Cr: the harder A1203only allows reduction to Cr3+, the softer Si02 to Cr2+.

Acknowledgment. B.M.W. acknowledges the I.W.O.N.L. (Instituut vmr Wetenschappelijk Onderzoek in Nijverheid en Landbouw) for a research grant. We thank J.B. and M.G. Uytterhoeven for the use of the Spectra Calc Program.

References and Notes (1) Hogan, J. P.; Banks, R. L. Belg. Pat. 530617, 1955. (2) Hogan, J. P.; Norwood, D. D.; Ayres, C. A. J . Appl. Polym. Sci. 1981, 36, 49. (3) Hogan, J. P. Applied Industrial Catalysis; Academic Press: New York, 1983; Vol. 1, p 149. (4) Clarck, A. Addition and Condensation Polymerization Processes; ACS Symposium Series; American Chemical Society: Washington, DC, 1968; p 387. (5) Charcosset, H.; Revillon, A.; Guyot, A. J . Catal. 1967, 8, 334. (6) Grunert, W.; Saffert, W.; Feldhaus, R.; Anders, K. J. Catal. 1986, 99, 149. Knozinger, H.; Ratnasamy, P. Catal. Rev.-Sci. Eng. 1978, 17, 31. (7) McDaniel, M. P. Ado. Catal. 1985, 33, 47. (8) Ghiotti, G.; Garrone, E.; Zecchina, A. J. Molec. Catal. 1988,46,61. Kim, C. S.;Woo, S. I. J. Molec. Catal. 1992, 73, 249. (9) See references in Table I. (10) Best, S.A.; Squires, R. G.; Walton, R. A. J . Catal. 1977, 47, 292. Merryfield, R.; McDaniel, M.; Parks, G. J . Catal. 1982, 77, 348. (1 1) Vuurman, M. A.;Stufkens, D. J.;Oskam, A.; Moulijn, J. A.;Kapteijn, F. J . Molec. Catal. 1990, 60, 83. (12) Ellison, A. J. Chem.Soc.,Faraday Trans.1 1984,80,2581. Kazanski, V. B.; Turkevich, J. J . Catal. 1967, 8, 231. (13) Hogan, J. P. J . Polym. Sci. 1970, 8, 2637. (14) McDaniel, M. P. J . Catal. 1982, 76, 17. (15) Zecchina, A.; Garrone, E.; Ghiotti, G.; Morterra, C.; Borello, E. J . Phys. Chem. 1975, 79, 966.

The Journal of Physical Chemistry, Vol. 97, No. 18, 1993 4763 (16) Krauss, H. L. Proceedings of the 5th International Congres on Catalysis, Palm Beach, 1972; North-Holland Amsterdam, 1973; Vol. 1, p 207. (17) Rebenstorf, B.; Larsson, R. Z . Anorg. Allg. Chem. 1981,478, 119. (18) Fubini, B.; Ghiotti, G.; Stradella, L.; Garrone, E.; Morterra, C. J. Catal. 1980, 66, 200. (19) Groeneveld, C.; Wittgen, P. P. M. M.; van Kersbergen, A. M.; Mestrom, P. L. M.; Nuijten, C. E.; Schuit, G. C. A. J . Catal. 1979,59, 153. (20) Hardcastle, F. D.; Wachs, I. E. J . Molec. Caral. 1988, 46, 173. (21) Kim, D. S.;Tatibouet, J. M.; Wachs, I. E. J. Catal. 1992,136, 209. (22) Iwasawa, Y.;Sasaki, Y.;Ogasawara, S. J. Chem. SOC.,Chem. Commun. 1981, 140. (23) Vuurman, M. A.; Wachs, I. E. J. Phys. Chem. 1992, 96, 5008. (24) Myers, D. L.; Lunsford, J. H. J . Caral. 1985, 92, 260. (25) Fouad, N. E.; Knozinger, H.; Zaki, M. I.; Mansour, S.A. A. Z . Phys. Chem. 1991, 171, 75. (26) Przhevalskaya, L. K.; Shvets, V. A.; Kazanskii, V. B. Kinet. Katal. 1970,11, 1310. (27) Ghiotti, G.; Garrone, E.; Della Gatta, G.; Fubini, B.; Giamello, E. J . Catal. 1983, 80, 249. (28) Krauss, H. L.; Stach, H. Z . Anorg. Allg. Chem. 1969, 366, 34. (29) Krauss, H. L.; Rebenstorf, B.; Westphal, U. 2.Anorg. Allg. Chem. 1975, 414, 97. (30) Cimino, A.; De Angelis, B. A.; Luchetti, A,; Minelli, G. J . Cural. 1976, 45, 316. (31) Ellison, A.; Oubridge, J. 0. V.; Sing, K. S.W. J . Chem. SOC.1970, 66, 1004. (32) Ellison, A.; Sing, K. S.W. J. Chem. Soc., Faraday Trans. 1 1978, 74, 2017. (33) Chen, K. C.; Tsuchiya, T.; Mackenzie, J. D. J . Non-Cryst. Solids 1986, 81, 227. (34) Szabo, 2. G.; Kamaras, K.; Szebeni, S.;Ruff, I. Spectrochim. Acta 1978, 34a, 607. (35) Hillier, J. H.; Saunders, V. R. Chem. Phys. Lett. 1971, 9, 219. (36) Lever, A. B. P. Inorganic Electronic Spectroscopy, 2nd ed.; Elsevier: Amsterdam, 1984. (37) Miskowski, V.; Gray, H. B.; Ballhausen, C. J. J. Molec. Phys. 1974, 28, 729. (38) Sieslak-Golonka, M. Coord. Chem. Rev. 1991, 109, 223. (39) Fackler, J. P.; Holah, D. G. Inorg. Chem. 1975, 4, 954. (40) Schoonheydt, R. A. In Characterizationof Heterogeneous Catalysts; Delannay, F., Ed.; Marcel Dekker: New York and Basel, 1984; Chapter 4. (4 1) Kellerman, R. In Spectroscopy in Heterogeneous Catalysis; Delgass, W. N., Haller, G. L., Kellerman, R., Lunsford, J. H., Eds.; Academic Press: New York, 1979; Chapter 4. (42) Parks, G. A. Chem. Rev. 1965,65, 177. (43) Pearson, R. G. J. Am. Chem. SOC.1963,85, 3533. (44) Langenaeker, W.; De Decker, M.; Geerlings, P.; Raymaekers, P. THEOCHEM 1990, 207, 115. Langenaeker, W. Graduate Thesis, V.U. Brussel, Belgium, 1989. (45) Beran, S. J. Phys. Chem. 1981, 85, 1956. (46) Corma, A.; Sastre, G.; Viruela, R.; Zicovich-Wilson, C. J . Catal. 1992, 136, 521. (47) Duffy, J. A. J . Phys. C: Solid State Phys. 1980, 13, 2979. (48) Duffy, J. A. J. Chem. Soc., Dalton Trans. 1983, 1475. (49) Duffy, J. A. J . Solid Stare Chem. 1986,61, 3428. (50) O’Reilly, D. E.; Santiago, F. D.; Squires, R. G. J . Phys. Chem. 1969, 73, 3172.