CATALYTIC ACTION OF AX ALUMINUM OXIDE CATALYST* BY WILDER D. BANCROFT AND AVERY B . GEORGE
Introduction There appear in the literature some differences of opinion in regard to the explanation for the action of oxide catalysts when prepared under different conditions. Adkinsl has advanced a hypothesis in which he considers that the catalytic activity of alumina is conditioned by its molecular porosity or the distances between the aluminum atoms. This is determined in part by the size, shape and position of the radicals attached to the aluminum when the aluminum compound goes into the solid state. I n terms of this hypothesis it is held that decarboxylation is favored by large pores in the aluminum] and that ethylene formation is favored by small pores. Large and small in this case refer to units of molecular dimensions. Xdkins says that catalysts which are colloidal give small pores when water is driven off, as the water is adsorbed, so they are favorable to ethylene formation. True hydroxides give less ethylene, and the catalysts prepared from the ethoxides favor the carbon dioxide formation] and these give large pores. The data of Adkins giving the relative amounts of ethylene and carbon dioxide obtained are given in Table I.
TABLEI Catalyst I.
2.
3. 4.
5. 6. 7. 8. 9. IO. 11.
Wt. CnH4gm.
Ethoxide on pumice Isopropoxide on pumice Methoxide on pumice Amalgam interaction with H 2 0 Ethoxide precipitated by H 2 0 Isobutoxide on pumice Propoxide on pumice Butoxide on pumice Ethoxide pills from powder Eitrate pills from powder by precipitation with ",OH Pills from precipitate by action of HC1 on XaAlOs
2.46 1.95
4.57 4.50
0.538
2.30
4.40
0 .522
5.95 6.75
3 ,IO 3.75 4.30
1.92
2 . IO
0.433
1.80
0.488
2 . 50
4.90
0 .j I 0
2.90
0.630
2.80
4.60 4.I O
0.682
5.90
3.20
1.84
5 .oo
3.70
1.35
* This work is done under the rogramme now being carried out a t Cornell University and supported in part by a grant i?om the Heckscher Foundation for the Advancement of Research established by August Heckscher a t Cornell University. J. Am. Chem. SOC.,44, 385, 2175 (1922).
2944
WILDER D. BANCROFT A S D AVERT B. GEORGE
On the other hand Taylor? regards an oxide catalyst as composed, not of a single catalyst, but of two catalysts, metal ions and oxide ions. The nature of the changes induced in the adsorbed reactant is determined by the charge of the ion on which the reactant molecules is adsorbed. The pxtent of the two alternative changes mill be determined by the relative extent of adsorption of reactant on the two ions, on the relative frequency of the t w o ions in the surface, and on their specific individual catalytic activities. These several factors, extent of adsorption, frequency of ions in the surface, and catalytic activity will be determined by the degree of saturation of the lattice ions (i.e. catalytic structure) and by the extent to which the ions are already covered by poisons (salts ammonia, water, etc.) In this explanation alkali poisons favor dehydrogenation or decarboxylation, while acid poisons favor thr dehydration process. K i t h these two different views on the matter, it seemed that this would be a good field for investigation, in order to arrive at the correct explanation. The substance chosen to be decomposed was ethyl acetate, for this was used by Adkins in some of his work. Sabatier3 classifies the decomposition of ethyl acetate a t 4ooOC when using different catalysts, as follows. A1203
z C H ~ C O O C ~ H ~ (CH3)CO
+ 2C2H4 + C O S + H20
TiO?
2CH3COOCZH5+ 2CHSCOOH
+ 2CZH4
From the above reactions it can be seen that a determination of the relative amount of the gases, ethylene and carbon dioxide, would give a means of measuring the course of the reaction. This was exactly the procedure followed, and the result was expressed as the ratio, Keight CzH4/\VeightCO,, which is the relation of dehydration to decarboxylation. Experimental Procedure
Preparation of Cntalysts. The alumina catalysts used in this work were prepared by a number of different methods. Most of the catalysts were prepared by the procedures used by Adkins, in order that the two sets of results might be put on a comparative basis. Catalyst KO.I . zoo grams of A1(S03)3.9H20were dissolved in about two liters of distilled water, and then concentrated ammonium hydroxide was added to the hot solution until the precipitation was complete. The precipitate was washed a few times by decantation, then filtered, washed, and dried in an electric oven a t about 120’C. The dried alumina was broken up into small, fairly uniform pieces, which were used in the runs. Catalyst S o . 2 . -A solution of I O grams of L%1(x08)3.gH20in jo C C . of water was made up, and I O grams of washed pumice were soaked in this hot Colloid Symposium Monograph, 4, 19 (1926). Sabatier: “La Catalyse en Chimie organique,” 341 (1920).
CATALYTIC A C T I O S O F A S A L L 3 1 I h T M OXIDE CATALI-ST
2915
soluticn for about five minutes. Then the solution was drawn off, and 50 cc. cf concentrated amnionicm hydroxide added to the pumice. This treatment caused the hydrnted aluminum oside t o be precipitated on the pumice, after which the ammonium hydroxide was drawn off, and the supported catalyst Lvashed with water a few timrs. T h r catalyst was dried in the electric furnace nnd then over a bunsen burner for a few minutes. at IZO'(', Catalyst S o . 3. 30 grams of A1(S03)3.gH20 were heated in an evaporating dish until the pure AI2O3 remained, thus driving off the water and the oxides of nitrogen. This solid was broken up into small pieces for use, as in the case cf cat:tlyst s o . I . Catalyst S o . 4. I O grams of .41(SOa)a.9H20iyere dissolved in a small volume of distilled water, and t o this concentrated solution 1 2 grams of washed pumice were added. The mixture was heated on n water bath, until the pumice was fairly dry, and then further drying was obtained by placing the catalyst in the electric furnace at about 11j'C. The catalyst was then heated until all of the nitric acid was driven off, after Lvhich it. was ready for use. Catalyst S o . 5 . Some aluminum ethoxide was put on a watch glass, and exposed to the air for a period of time. The ethoxide hydrolyzed with the formation of aluminum oxide, which was broken up into small pieces for use. Catalyst KO. 6. Four grams of aluminum ethoxide were melted and 3 grams of washed pumice was added to the melt. This was allowed t o cool, and then moist air was passed over the catalyst for about five hours, after which it was exposed to the air for a long period of time before use. Catalyst S o . 7. Some of the dry washed pumice, which is a silicate of sodium, potassium and aluminum, was used as a catalyst. d p p a m t z c s a n d Procedure. The apparatus consisted of a flask for vaporizing the liquid, this being connected to a reaction tube heated by an electric furnace, with a system for condensing the liquid products, and an eight-liter bottle, which served as a gasometer for collecting the gases evolved, at the other end. A saturated solution of sodium chloride was used as the confining liquid in the gasometer, as the gases evolved arc only very slightly soluble in this solution. A thin layer of the catalyst was spread along the reaction tube for a length of 2 0 cm. The furnace was heated to a temperature of about qjoo(', at xvhich temperature all of the runs were made. The procedure followed was to put a measured amount of the ethyl acetate into the flask through the funnel at the top. At the end of a run the amount of liquid left was measured, then the amount of ethyl acetate actually used could be determined. The flask containing the ethyl acetate was immersed in an oil bath, and the rate a t Iyhich the ethyl acetate vapors were allowed to pass over the catalyst was controlled by varying the temperature of the oil bath. The bath was kept at a ternperature of 80"-g0"c', in which range about 40 grams of ethyl acetate were vaporized per hour. The ethyl acetate vapors were passed through the catalyst tube, the liquid products being condensed, and the gaseous products collected in the gasometer on emergence from the heated tube. The weight of the liquid product, and the volume of gas were deter-
WILDER
2946
D.
BANCROFT AND AVERY B. GEORGE
mined. Then a sample of the gas was taken from the gasometer, and analyzed for carbon dioxide and ethylene. The carbon dioxide was removed by absorbing the gas in a solution of sodium hydroxide, while the ethylene was determined by passing the gas into a pipette containing fuming sulphuric acid, thus obtaining the necessary data.
Experimental Results The first runs were made using the catalysts prepared from the aluminum ethoxide, numbers 5 and 6 as given above, and the data obtained are given in Table 11. The ratios obtained from Adkins’ data are also given.
TABLE I1 Ratio Wt. CIH4/\Tt. COS
Catalyst
Unsupported S o . Supported KO. 6
j
Adkins’ Ratios
2.18
0.682
1.66
0.538
The percentage increase in the amount of ethylene formed from the unsupported catalyst over the supported one compares very well with that obtained by Adkins. However, the ethylene/carbon dioxide ratios obtained by Adkins are very much lower than the values found in this work. He prepared this catalyst by exposing the aluminum ethoxide to the air for a long time, so this was a case of slow hydrolysis. Since the ethoxide used by him should be the same as that employed in this work, the difference in values might be due to impurities in the air. The catalyst employed here was not exposed directly to the atmosphere of the laboratory. I t seemed that ammonia would be the most likely present, so a catalyst was made from the ethoxide by hydrolyzing it in a solution of ammonia. After drying, this catalyst was used in a run, and the ethylene/carbon ratio obtained was 1 . 2 2 . Xow this value is much less than the 2.18 obtained above, and while it is not as low as Adkins’ value, this shows that the presence of ammonia causes a low ethylene/carbon dioxide ratio. So this shows that the presence of ammonia, and probably other basic impurities in the air, will account for the low values obtained with the ethoxide catalysts. S o w we thought that if Adkins’ explanation was right, it should be possible to take a catalyst which is predominately decarboxylating, thus having large pores, and change it over to a dehydrating catalyst with small pores, by heating to a high temperature thus causing the catalyst to sinter. An experiment was made along this line using some of catalyst KO. I , prepared by precipitation of the hydrated oxide from an aluminum nitrate solution with ammonium hydroxide. The first two runs were made without heating the catalyst above the temperature of the furnace. Then another portion of the catalyst was heated over a 11eker burner for about a half hour, in order t o cause the catalyst to be sintered. The data obtained are given in Table 111.
CATALYTIC ACTION O F AN ALUMSIUJ.1 OXIDE CATALYST
TABLE
Catalyst
111
Ratio Wt. G H J W t . CO?
Not sintered S o .
I
2.14, 2.16
So.
I
2.44
Sintered
294i
These figures show that the sintered catalyst apparently does give a larger ratio of ethylenelcarbon dioxide. The catalyst used was a predominantly dehydrating one, so perhaps this 14% increase in the ethylene/carbon ratio is not fair data on which to base a conclusion. Some of the catalyst prepared by hydrolyzing the ethoxide in an ammonia solution was sintered in the same way as was the KO.I catalyst, and a run made. The ratio obtained was 1.39, the percentage increase of the ratio being practically the same as the 14ci given above. This shows that the increase of the ethylene/carbon dioxide ratio, due to sintering, was practically the same starting with a catalyst of low or high ethylene/carbon dioxide ratio. As shown above, ammonia caused a marked decrease in the value of the ratio, so even at the high temperature of sintering there must be some of the ammonia present. A further experiment was done along this line, for some of the catalyst prepared from the aluminum ethoxide was heated to a high temperature and sintered, then it was used in a run. In this case a value of z . 9 j was obtained for the ratio, as compared to 2.18 for the unsintered catalyst. This increase in the amount of ethylene formed cannot be accounted for entirely by sintering, for part of the catalyst became dark colored on heating. As alumina would not become dark by this treatment, this change was probably due to some organic material present, as carbon, which became charred. This change apparently favors the formation of ethylene. So the values given in Table I11 above, and the data obtained by sintering the ethoxide catalyst hydrolyzed in the presence of ammonia illustrate better the change involved due to sintering, and this amounts to about a 14';c increase of the ethylene/ carbon dioxide ratio. However, the increase in the amount of ethylene obtained by sintering the catalyst was considerably less than the difference between the dehydrating and decarboxylating catalysts. So from this it seemed that the facts could not be explained in this way. h miscellaneous experiment was made using some of the washed pumice as the catalyst. With this substance a ratio of 62. j was obtained, which means that there was only very little carbon dioxide formed. This seemed very interesting, and as pumice is a complex silicate, it may be that substances of this type have possibilities in contact catalysis. The reason for making this experiment was to determine whether or not the support used had any effect on the reaction, and it apparently has. In order to compare the activity of a supported catalyst prepared by precipitation with ammonium hydroxide, the procedure as given for catalyst KO. z above was used. Many people have compared the activity of oxide catalysts prepared by precipitation with ammonium hydroxide, with supported catalysts prepared by heating a nitrate solution containing the support,
2948
WILDER D. BANCROFT A S D AVERT B. GEORGE
as pumice. This is not a just comparison, for in this case the effect of the ammonia is neglected, while it may have a marked effect. The results obtained with catalyst S o . z gave a rather high ratio, much higher than would be expected. I t seems that this was due to the fact that this method of preparation did not give a uniform surface layer of the alumina. I n fact portions of the pumice were not completely covered so the catalyst really consisted of some alumina on the surface and also some of the uncovered pumice. K‘ow that pumice was found t o be a good catalyst for the formation of ethylen? this mill account for the high ethylene ratio. A comparison was now madc of a supported catalyst on pumice and an unsupported one, both being prepared by heating ai1(S03)3.gHz0, thus the catalyst obtained should be free from any adsorbed ions, which is not true in the case of the precipitation methods. (’atalysts number 3 and 1 were used, and the data are given in Table IJ-.
TABLE IT Catalyst
Unsupported S o . 3 Supported KO. 1
Ratlo K t . C2HI’R-t COT
3 84 3 57
I n this case the unsupported catalyst gives a little larger amount of ethylene than does the supported one, but the difference is very small. While the experiments on the sintering of the catalysts did show an increase in the amount of ethylene formed, this could not be the major factor, for the difference was not great enough t o account for the facts. Consequently the next approach was t o study the effect of the adsorbed ions on the oxide surface. According t o Taylor’s theory the basic radicals are primarily decarboxylating, so if ammonium hydroxide mas added to an oxide catalyst the adsorption of the ammonia should increase the amount of carbon dioxide formed. h catalyst was prepared according to the procedure given for catalyst S o . 4 above. This catalyst was assumed to be practically pure, being free from adsorbed ions, and its ethylene ’carbon dioxide ratio was 3.5 j as shown in Table IT- above. Now some of this catalyst was soaked in concentrated ammonium hydroxide for a few minutes, thus causing the ammonia to be adsorbed on the alumina. After u-ashing and drying the catalyst it was used in a run, and an ethy1ene;carbon dioxide ratio of 2.80 was ohtained. This shows that the presence of the ammonia increased the relative amount of carbon dioxide formed. Some of catalyst S o . z was treated in exactlythe same way as x a s catalyst S o . 4 above, and then some runs were made with it. In this case the formation of ethylene was favored, hut this was due to the fact that on soaking the supported catalyst some of the alumina layer was removed from the pumice, thus leaving more exposed pumice, which favors the formation of ethylene. The catalysts prepared by heating A1(?;03)3.9H20,as S o . 3 above, favor the formation of larger amounts of ethylene than do those prepared by the precipitation with ammonium hydroxide. This fits in very well with the general
CATALYTIC .4CTIOX O F A S ALCMISVM O X I D E CATALYST
2949
explanation, that the adsorbed ions have an effect, and that the basic radicals favor decarboxylation. Another series of experiments was carried out in order to determine the effect of adsorbed ions on the oxide surfaces. h catalyst wis prepared using the same procedure as that employed in making catalyst S o . 4 above. This catalyst wis then divided into three parts. One run was made using the pure nluminum oxide catalyst, then one of the other portions was soaked in a zinc nitrate solution, and the other in an aluminum sulphate solution. In the first case zinc was taken up by the surface, while in the second case sulphate was adsorbed, the other tvio ions being common to the initial state of the catalyst. The solutions used contained equivalent concentrations of the snlts, nnd in the case of the one with zinc nitrate, the catalyst Tyas heated until :ill the nitric acid T m s driven off. On inaking runs with these catalysts, the effcct of the acid and the basic radicals could be determined. According to thp theory, the zinc should favor decarboxylation, while the sulphate favors t h c dehydration, or the formation of ethylene. The data obtained are given in Table V. TABLE T’ Ratio Kt. C2HI/IWt.CO?
Catalyst
2.75 I ,
13
12.5
From the data given in the above tahle, it has been shown that the presence of the zinc causes an increase in the amount of carbon dioxide formed, while the sulphate favors an increase in the amount of ethylene formed. These results are in harmony with the theory set forth by Taylor, and as the facts are explained much better with this than by the theory of molecular porosity, it seems that the theory set forth by Taylor is probably the correct one.
Conclusions The increase in the relative amount of ethylene obtained by sintering an aluminum oxide catalyst is considerably less than the difference between the dehydrating and decarboxylating catalysts. 2 . The use of washed pumice as a catalyst gives very large amounts of ethylene, with little carbon dioxide. 3. The catalysts prepared by heatingA1(S03)3.gH20,favor the formation of larger amounts of ethylene than do those prepared by the precipitation with ammonium hydroxide. 4. Khile there m s a slight increase in the amount of ethylene obtained by sintering a catalyst, this was small as compared t o the effect of adsorbed substances. So the results of this paper shox that Taylor’s theory, which states that the extent of the changes induced in the adsorbed reactant will be determined by the specific individual catalytic activities of the adsorbed ions or their reaction products, is probably the correct one. I.
Cornell University.