Chemical Characterization of Catalysts. 11 ... - ACS Publications

BETWEEN WATER AND CRACKING CATALYSTS. 5549 ... the velocity of the initiation step is dependent ... cracking catalysts were investigated by tracing th...
0 downloads 0 Views 757KB Size
OXYGENEXCHANGE BETWEEN WATERAND CRACKING CATALYSTS

Dec., 1950

dence i t can be concluded that: (1) without the presence of oxygen no decomposition of formic acid nor monomer polymerization occurs a t room temperature; (2) when oxygen is present in not too large amounts decomposition occurs and polymerization sets in. An induction period is present, and increases with increasing amount of oxygen in the solution; ( 3 ) in the presence of large excess of oxygen the decomposition reaction occurs but no ready polymerization can be detected. Therefore, there must be an optimum concentration of oxygen, which gives rise to polymerization with the shortest induction period. This can be explained in the following way. A t constant formic acid and monomer concentration the velocity of the initiation step is dependent on the amount of oxygen present on the surface = k: [Otlsurfsce (3) Under similar conditions the velocity of saturation of polymer radicals by molecular oxygen is

5549

of oxygen in the solution leads however to values of

vp which are too low to be measured. A know1edgment.-I wish to express my sin-

'it

cere hanks t o Professor H. S. Taylor for his help, advice and criticism during the course of this investigation. E. I. du Pont de Nemours and Company assisted this work through a grant-in-aid to Princeton University, for which assistance best thanks are also extended.

Summary

1. The decomposition of formic acid solutions a t the surface of platinum sol in the presence of oxygen and of methyl methacrylate initiates polymerization of the monomer. 2. Acrylonitrile is not polymerized under the same conditions since it poisons the surface for formic acid decomposition. 3. The rate of polymerization is proportional to the amount of platinum employed and to formic acid and monomer concentrations. 4. The average degree of polymerization is ut = k:C* [ 0 2 l l i q u i d phase (4 1 directly proportional to monomer concentration, Using steady state approximation the velocity inversely proportional to formic acid concentration and independent of the amount of platinum of propagation of polymer chains is employed. 5. Kinetic mechanisms to interpret the data op = k k' [MI [Otlaurface (5) 'k6 :02]liquid phaae have been derived. 6. Only a very small fraction of the decomFrom equation ( 5 ) i t can be seen that, for low values Of [02]liquid phase, the metal Surface not posed formic acid initiates chain polymerization of being saturated with oxygen, polymerization can the monomer. 7. The effect of oxygen on both formic acid deoccur a t a measurable rate. hS [ O ~ l l i q u i dphase increases, the surface saturation is reached, [O2Isurface composition and polymerization is discussed. becomes constant. The increased concentration PRINCETON, NEWJERSEY RECEIVED JUNE 22, 1950 VI

[CONTRIBUTION FROM

HOUDRY PROCESS CORPORATION, MARCUS HOOK,P A . ]

Chemical Characterization of Catalysts. 11. Oxygen Exchange between Water and Cracking Catalysts B Y G. A.

MILLSAND s. G. HINDIN

Introduction The reactions occurring between water and oxides used in petroleuin cracking catalysts are of special importance because (1) water is the medium from which oxide catalysts are prepared in hydrogel form, (2) water is normally responsible for loss of activity during catalyst use, and ( 3 ) water is believed to be fundamental to the mechanism of the cracking process.lS2 The reactions between water and the surfaces of cracking catalysts were investigated by tracing the reaction with isotopic oxygen. These exchange data have been correlated with ignition loss and surface area determinations, and the results interpreted in terms of the atomic structures of the surfaces involved. (1) Hansford, Ind. En#.Chcm., 89, 849 (1947). Milliken, Mills and Oblad, Trans. Far. Sac., Symposium on Heterogeneous Catalysis, 1950. (2)

While the bulk structure of many solids are known in detail, little experimental ,evidence is available to establish the surface structure. The measurement of physical properties, although giving valuable information, does not define the chemical nature of the surface, nor does i t allow conclusions concerning reactions that may be occurring there. These highly specific surface chemicaI properties have been widely recognized in colloidal phenomena and particularly in heterogeneous catalysis, and the first paper of this seriesS has discussed a chemical characterization of petroleum cracking catalysts through their reaction with basic substances. Although isotopic tracer methods have been used to elucidate the mechanisms of catalyzed hydrocarbon reactions, most investigators have been concerned with atomic rearrangements in (3) Mills, Boedeker and Oblad, THIS JOURNAL, 79, 1554 (1950).

3. Silica-Alumina.-This material was standard IIouthe hydrocarbon inolecules ; few hdve investigated Type S catalyst (12.5 wt. 70alumina, 87.5 wt. 96 the catalyst itself. In the case of oxide catalysts, dry silica) which had been dried a t 730" for twelve hours in the i t is possible to measure critical catalyst properties presence of air containing 8 7 , steam. Before use in exby following the transference of oxygen between change experiments it was recalcined for one hour at 540' catalyst and reactant molecules. Titani, Mo- under vacuum. 4. Clays.-(a) The kaolin clay, from Florida, is rita and their co-workers have made an extensive commercially as Edgar Company plastic kaolin. study of the exchange between metal oxides and known It was mixed with water, extruded and dried. Before usc molecular oxygen, and current publications indi- it was dried for four hours a t 100" under vacuum. (b) The bentonite clay is a fairly pure montmorillonite ~~,~ cate new interest in this e x ~ h d n g e . ~Oxygen exchange between inorganic oxygen-containing from Ash Meadows, Nevada. I t was mixed with water, extruded arid dried. Before use it was dried for four substances and water has also been investigated to hours a t 100" under vacuum. sonie extent and, for soluble inorganic oxy-anioiis, (c) Samples of activated bentonite-1 (Filtrol TCC) exchange was shown to proceed, it1 general, and -2 (Ash Meadows): Both clays were sulfuric acidthrough reversible anhydride f ~ r r n a t i o n . ~Ex, ~ ~ ~activated bentonites, water-washed free of sulfate ion and dried a t 105'. They were then calcined a t 565" for two change between water and metallic oxides has also hours in dry air. Before use they were redried for one been measured by 22oritd and Titdni This hour a t 460" under vacuum. worL has been revienwl hx ( h h '' Physical Properties.-~T h e moisture content wa.; ob-

Experimental Procedure Samples of the oxides were calciiieti at different temperatures, cooled, and then allomcd to react with water enriched in the heavy isotope of oxygen (H20'*)for various periods of time (fifteen minutes to one month) at 100" or 105". In some experirncnts the H301*was made acid or alkaline with hydrochloric acid or d i u m hydrozide. A few exchange experiments nere alio made d t 565 . After reaction, the water wab sepdrated and the reduction in isotopic concentration of the H,Ois was determined, using the mass spectrometei. From thii and the weights of materials used, the extent of ouvgcn exchange was calculated for the reaction

+ I M ~ .2-f?ir,o,

xzoV'b

%

t

HP

where M = Si or Al. A few experiments were also made using DzO, and the hydrogen content of the oxide calculated from the H-D exrlidngc.. Preparation of the Oxides 1. Silica.-Two silica prepardtions nere used (a) The first sample, SiOz-l, was prepared from Philadelphia Quartz Company N-Brand sodium silicate and sulfuric acid. The resuiting hydrogel wali washed sulfate-free with water and dried a t 70". The dry material was calcined for six hours a t 760" iii dry air. Before use in ehrecdkiried one hour a t 5 2 0 " change experiments, i t under vacuum. (b) The second sample, SiOz-5, vas prepared by tredt ing N-Brand sodium silicate with ammonium sulfate The hydrogel was washed with 10% ammonium sulfate solution, and then water-washed sulfate-free. Two separate portions of the hydrogel nere dried a t temper +tures of 100" and 450" a t about 1 mm. pressure. 2. Alumina.-The alumina mas prepared according to IIeard.11 Aluminum was dmalgamated with mercury, treated with acetic acid, and precipitated with ammonium hydroxide The hydrogel \ $ a i water-washed, though not to complete removal of ,trninonium and acetate ions (to prevent peptization) arid filtered. The dmmoninni and acetate ion contents of the mdteriak \$ere not determined. T w o separate portions of the hydrogel !\ere dried af 100 and 450" a t about 1 mm. preisure ( I ) Allen and Lauder, "Vafurc, 164. 142 (1949) ( 5 ) Houghton and Winter, rbid., 164, 1130 (1949) (6) Karpacheva and Rozen, Dokiady A k a d VVauk

S S S. R , 68, 1053 (1949) 17) Titalu and Goto, Btdl Chcna Soc J a p a n , 14, 77 (1939) (8) Mills, THISJOURNAL, 62, 2838 11940) (9) Winter and Briscor J C h o n Soc , 631 (1948) (IO) Geib, "Handbuch der Katalyse," edited by C: M bclirtdb L 01 6, Springer Verlag, Virnria, 1q43, pp 8fi e1 J c q (11) Heard, U S P 2 271 6 3 4 1042'

tained by blast lamp ignilioti in piatilium to constant weight. Cotnp1et.e nitrogen adsorption isotherms were measured, the surface areas calculated by the method of Rrunauer, Emmett and Teller and the pore size distribiition estimated by use of the Kelvin equation.'* Powdered aatnples w r r also cxainiried for their S - r a y diffraction patterns uiirig :I Sorelco machine with a copper target aiiri a nickel filter. Exchange Procedure.--A weighed aniouut. (~cri.0.3 9.) of water ciiriched with R20'8 was added to a w-eighed sample ( ( a . 0.6 9.) of the oside in a Pyrex tube. Thc mixture wai frozen to liquid nitrogen temperature, the air pumped off, and the tube sealed while under vacuuni, The tube was shaken vigorously and was then held at 100 ' or 105' for a rhosen time. For exchange reaction times up to eight hours, the tube was placed in a vigorously boiling water-bath; for longer reaction times, the tube was kept in a drying oven at 105 j: 9 '. After the exchange period the tube was broken open and the water distilled over into a receiver, under vacuum, using liquid nitrogen :LS coolant. Approximately one-half cc. of carbon dioside was added arid the water-carbon dioxidc mixture equilibrated for forty-eight hours a t room teinperature (about 2 7 " ) . The carbon dioxide lR/'OIB ratio determined by 11 a cracking catalyst and the silica dricti a t 760' were also exchanged at 100" for periods up to eight hours with H2OI9 solutions, tenth-normal with resprcf t o

hvtlrochloric acid or to sodiuni hydroxide. f3lank determinations were made on the water ciiriche(1 sealed in Pyres arid inaintained a t 105 * 2 O for with N201S prriotii up to four days, and on the 0.1 N alkaline solution fci- periods of time tip to eight hours a t 100". In neither ('asc did measurable exchmige with osygen from the glass iiirfaw occur. ' T h i h tuajnr sotirce of error involved iti this procedure from the incomplete removal of all sorbed water the oxidc surface after reaction, prior to equilibration with c:irbon dioxide and, a i :t result, thc exchange values lieved minimal. \\'ith the same oxidc ion ih limited priinarily by the accuracy otnetric atialysii, aiiii is of the ordcr of --tXg.b ;ilisolutc. \ \ l t h different prcparations of some o s ides, vari;Lt.iotis arc of iiiuch greater magnitude. The exrhnnge data are calculated in terms of the fraction of the oxygen in the oxide sample which has equilibrated with the water. This value is obtained from the ' of oxygen which has relationship: M8/,1f2X 100 = % Ml(1.53) = a(M1 equilibrated, where: M3(0.20) M,) and M I , M z and Ma are, respectively, the weights of oxygen in the added mater, in the oxide sample, and in the oxide sample which has equilibrated; a is the atom per cent. 0 ' 8 in the water after eschange (1.53 atom yo - .l__.l_

+

+

(12) Holmes and Emmett, . I . I'hn's. Colloid C h c v . , 61, 1280

c i94ii.

OXYGENEXCHANGE BETWEEN WATERAND CRACKING CATALYSTS

Dec., 1950

originally). This calculation is believed t o give a truer picture of the exchange than that of the approach to equilibrium of the entire mixture.

Data and Discussion The results of the exchange experiments at 100' are shown in Table I and the physical properties of the oxides before and after reaction with water in Table I1 and in Fig. 1. It is evident that there are reactions occurring which are so extensive that, in some instances, there is almost complete equilibration of all oxygen of the oxide with the added water. At the same time, there has been a marked change in the physical structure of the oxides. For clarity, the data relating to silica gel will be considered in detail first. Referring to Table I , exchange is seen to occur in two distinct stages. Initially, there is a rapid exchange involving 1020'333 of all the silica oxygen, while the second stage is a continuing, slow exchange involving a further 30-50y0 of the silica oxygen. After one month some 40-70% of all the oxide oxygen has equilibrated with oxygen in the added water. The rate and extent of exchange are alike for the samples dried a t 450 and 760'. TABLE I EQUILIBRIUM WITH 100-105'

P E R CENT. O F OXYGEN I N

Oxide

SiO2-2 Sample 1 Sample 2 SiG-2 SiO2-1 Sample 1 Sample 2

Max. drying temp,, -Reaction 'C. 0.25 0.50

1.0

100 20 21 22

.. 450 760

. . 1s

8 10

9

7

9 7

10

H2018 AT

time, hours----96

3.0 8.0

168 720

. . 30 . . 61 71 18 23 42 . . . . . . . . . . 25 41 . . 15 . . 31 41

5531

reaching changes in pore structure toward larger average pore diameter as indicated by the change in the nitrogen adsorption isotherms, shown in Fig. 1. TABLE I1 PHYSICAL AND CATALYTIC PROPERTIES OF THE OXIDES

Oxide SiOz-2 SiOn-2 SiOi-1 Ah01 Alto: S i O rAlzOa Kaolin Bentonite Activated bentonite-1 Activated bentonite4

Surface area, sq. m./g. Be- After fore reaction reac- for 1 tion month 148 267 222 276 157 330 217 350 372 366 312 208 29 26 76 70

Cat. activity, wt. % conversionb -7 -7

Maximum drying temp., OC. 100 450 760 100 450 730 100 100

Ignition loss a t 1000Oa 7.0,4.8 2.6 0.93 42.6,43.9 4.5,5.3 0.77 13.2 8.0

565

2.2

5.0

256

247

45

565

1.9

5.3

240

230

48

g H of oxideHn0 slurry a t 280' 5.0 * 0.5 5.0 4.8 4.8 6.3 5.5 7.7 9.1

-7 11 11 53 20

8

Where two values are shown, they are for samples 1 and 2. Alexander, Proc. A m . Petroleum Inst., 27 (111), 51, Nov. (1947). a

From X-ray examination: 1. Silica-no crystal pattern was observed. 2. Alumina-the sample dried a t 100" showed poorly defined lines of Boehmite, y-A1&&0. After one month reaction, the Boehmite pattern became much stronger. The alumina dried a t 450" showed a few lines of y-alumina. After one month reaction, a strong Boehmite pattern was seen; a few weak unidentified lines were also presentthey may be y-alumina. 3. Silica-alumina-no crystal pattern was observed. 4. Kaolin-no change in the kaolinite pattern was seen after one month's reaction. 5. Montmorillonite-for the natural and acid-activated bentonites, the montmorillonite pattern was obtained before and after one month's reaction, with the lines somewhat weak after reaction.

These data are interpreted in the following manner: The initial, rapid exchange involves oxygen 100 A1203 of the oxide present both in hydroxyl form as 26 09 24 . 37 . . 74 87 Sample 1 , , . . 33 37 46 76 . . . . [SOH] and HOH and in an electronically strained Sample 2 configuration which results from the dehydration 450 AlzOa process. The slow, continuing exchange involves 24 24 27 . . 37 . . 57 72 the Sample 1 of the surface oxygen, linked to sili. . . . 49 53 63 60 . . . . con remainder Sample 2 through both valence bonds. Coincident Si02-Al20~ I ow 15 15 12 . . 18 . . 36 44 with the slow exchange, migration of the surface Sample 1 exposing fresh oxygen, previously in the ., . . 11 . . 17 31 . . . . occurs, Sample 2 body of the oxide which becomes available for 100 . . . . 0" . . 0 .. 0 0 Kaolin clay exchange. This involves an hydrolysis : [Si0 Bentonite clay 100 . . . . 0" . . 0 .. 0 0 Si] HzO --+ [3SiOH], and exchange then occurs Bentonite clay, through hydration, as previously established for acid-activated-1 (Fil. . 8 . . 13 21 inorganic oxy anions.8 This explanation is boriic trolTCC) 565 11 . . 12 out by calculations involving the ignition loss and Bentonite clap, acid-actisurface area values, as will be shown. vated-2 565 9 . . 11 . . 13 . . 20 27 The oxide oxygen present as hydroxyl can be = No reaction with experimental error-ca. d0.03 atom calculated from the ignition loss. For the samples % 0'8. dried a t 450 and 760°, the loss of water on ignition The surface areas of the silica samples have de- arises from the reaction [SSiOH] -+ [SiOSi] creased considerably during one month contact H20; for silica dried a t loo', physically bound with water a t 105') as illustrated by a comparison water is lost also. Therefore, the ignition loss of the fifth and sixth columns of Table 11. This may be used to calculate the total hydroxyl oxychange in surface area is accompanied by far- gen content of the oxide, whether present as , ,

, ,

9

13 26

, ,

..

-On

+

+

ide samples, a series of exchange reactions was run using DzO. These reactions were studied a t two temperatures: (1) in sealed ampules a t room temperature, and (2) in sealed ampules a t 563' The data are shown in Table IV and are in general agreement with those obtained by ignition loss measurements. Ii -1) EXCHANX

TABLE IV I)&

BETWEC':

AVD

II-C'OWAISIXG

OXIDES

m n p l e dried for three hours a t 760" I O mm. ib) sample dried for three hours :it 930" 111th evaciidtion at < 10-3 mtn.