The Effect of Sulfiding a Nickel on Silica-Alumina Catalyst - The

Hydrocracking and Hydroisomerization of High-Density Polyethylene and Waste Plastic over Zeolite and Silica−Alumina-Supported Ni and Ni− ...
0 downloads 0 Views 584KB Size
G. E. LANGLOIS, R. F. SULLIVAN, AND CLARK J. EGAN

3666

The Effect of Sulfiding a Nickel on Silica-Alumina Catalyst

by G. E. Langlois, R. F. Sullivan, and Clark J. Egan Chevron Research Company, Richmond, California

(Received January $6, 1966)

When a nickel on silica-alumina catalyst is sulfided, two of the principal changes observed in the hydrocracking of n-decane are: (1) the over-all rate of reaction increases greatly and (2) the predominant reaction changes from isomerization to cracking. This type of cracking is similar to that observed with the silica-alumina support alone. It is concluded from these and other observations that the catalyst before sulfiding contains both nickel metal and nickel salts of the silica-alumina acid sites. On sulfiding, the nickel metal is converted to nickel sulfide, and the HzS reacts with the nickel-silica-alumina salts to regenerate the original strong acid sites of the silica-alumina support.

Introduction Nickel on silica-alumina catalysts catalyze a number of hydrocarbon reactions including isomerization, hydrocracking, olefin hydrogenation, etc.'- lo The catalysts are of a dual functional nature, exhibiting both acidic and hydrogenation activity. If the catalyst is sulfided prior to or during use, substantial changes in catalyst activity occur. For example, Frye, et U E . , ~ report that, in the hydrogenation of n-pentene with nickel on silica-alumina catalyst, sulfiding reducesthe rate of hydrogenation and increases the isopentane content of the product. Ciapetta and Hunter2 studied the reaction of several normal paraffins over nickel on silica-alumina and found that isomerization was the predominant reaction. I n contrast, Flinn, Larson, and Beuther4 found that, with a sulfided nickel on silica-alumina catalyst, cracking of normal paraffins was the principal reaction. They postulate a mechanism involving acid-catalyzed isomerization of the normal paraffin followed by rapid cracking of the branched isomers to form predominantly isoparaffinic products. We have also found that in the hydrocracking of paraffins there is a profound difference in the behavior of normal paraffins with sulfided and nonsulfided nickel on silica-alumina catalyst. I n this paper on the hydrocracking of n-decane, we present data showing more completely the effects of sulfiding and propose a inechanism to explain the changes observed.

Experimental Section The experiments were performed in continuous-flow, fixed-bed microcatalytic units of the type described The Journal of Physical Chemistry

previously.6 Catalysts were prepared by impregnation of commercial alumina and silica alumina (10% alumina). The method of catalyst preparation, pretreatment, prereduction, and sulfiding procedure, together with the method of analysis of the product by gas chromatography and mass spectrometry, were also described previously.6 The n-decane used in these experiments was Phillips Pure grade (99%) which was dried with silica gel. In the tables and figures, LHSV denotes the liquid hourly space velocity, namely, volume of n-decane per volume of catalyst per hour.

Results Reactions of n-Decane. %-Decane was hydrocracked over nickel on silica-alumina and over sulfided nickel on silica-alumina. Results are given in Table I. Shown also for comparison are results obtained with the silica-alumina support and nickel sulfide on alu(1) F. G. Ciapetta and J. B. Hunter, Ind. Eng. Chem., 45, 147

(1953). (2) F. G. Ciapetta and J. B. Hunter, ibid., 45, 155 (1953). (3) F. G. Ciapetta, ibid., 45, 162 (1953). (4) R. A. Flinn, 0. A. Larson, and H . Beuther, ibid., 52, 153 (1960). (5) C. G. Frye, B. D. Barger, H. XI. Brennan, J. R. Coky, and L. C. Gutberlet, Ind. Eng. Chem., Product Res. Dev., 2, 40 (1963). (6) R . F. Sullivan, C. J. Egan, G. E. Langlois, and R. P. Sieg, J . A m . Chem. Soc., 83, 1156 (1961). (7) C. J. Egan, G. E. Langlois, and R. J. White, ibid., 84, 1204 (1962). (8) R . F. Sullivan, C. J. Egan, and G. E. Langlois, J . Catalysis, 3, 183 (1964). (9) R. J. White, C. J. Egan, and G. E. Langlois, J . Phys. Chem., 68, 3085 (1964). (10) W. F. Pansing, ibid., 69, 392 (1965).

3667

EFFECT OF SULFIDING A NICKELON SILICA-ALUMINA CATALYST

Table I : Hydrocracking of n-Decane

Catalyst

0.1

I SILICAALUMINA

I NICKEL ON SILICA-ALUMINA

I NICKEL SULfIDE ON SILICA-ALUMINA

Figure 1. Hydrocracking of n-decane at 288", 8.0 LHSV, and 82 atm.

mina. Some of the results are shown graphically in Figure 1. With silica-alumina catalyst the rate of reaction is low (about 2% conversion). The addition of nickel to the silica-alumina causes the over-all rate of reaction to increase about threefold. When the nickel is converted to the sulfide, the rate of reaction increases even more sharply. The decane conversion is about seven times that with the nonsulfided nickel on silicaalumina catalyst and about 20 times that with the silica-alumina alone. This high rate of reaction is not caused by the nickel sulfide component specifically because nickel sulfide on an alumina of comparable surface area shows essentially no activity. The high rate is not directly caused by a reaction of the hydrogen sulfide with the silica-alumina support either, because treatment of the silica-alumina support with H2S actually causes a slight decrease in the rate of reaction. With the silica-alumina catalyst, the rate of reaction decreases with time. The data given here are for a 6min on-stream period. Activities of the nickel on silicaalumina and nickel sulfide on silica-alumina catalysts, on the other hand, are quite stable, and little change in rate of reaction with time is observed. Despite the very large difference in rate of reaction, the products obtained with the silica-alumina and with the nickel sulfide on silica-alumina catalysts are surprisingly similar. Cracking is the principal reaction in both cases, and only minor quantities of isodecanes are produced. The products from cracking are almost entirely in the C3-CT range, and the distribution by carbon number is quite similar for both catalysts, as shown in Figure 2. This distribution of the

Ni in catalyst, yo Temp, "C Press, atm Feed rate, LHSV HJhydrocarbon mole ratio Total conversion, % Isomerization, 70 Cracking, % Product, moles/100 moles of feed Methane Ethane Propane Isobutane n-Butane Isopentane n-Pentane Isohexanes n-Hexane Isoheptana5 n-Heptane Octanes Xonanes Isodecanes n-Decane 01efiIIs Iso-to-normal ratio, C4-CI Excess butanes, moles/100 moles of decane cracked

Silicaalumina

-

0 -288

NisSn on silicaalumina

Ni on silicaalumina

6.6

6.6 +

82 8

)c-

-

p

5.0 300 82 " L

10

-lo------+

2.5 0.3 2.2

7.8 4.5 3.3

52.8 5.5 47.3

0.01 ... 0.3 1.2 0.2 1.3 0.1 0.9 0.06 0.3 0.01 0.03 0.01 0.3 97.5 0.05 10

0.28 0.05 0.33 0.07 0.90 0.07 1.20 0.12 1.0 0.07 0.81 0.47 0.52 4.5 92.1 0.08

0.02 0.05 7.9 31.1 6.8 21.5 2.6 18.7 2.0 4.6 0.2 1.1 1.0 5.5 47.2 ... 6.6

...

33

20

Ni& on alumina

...

Essentially no reaction

product from cracking is characteristic of acid catalysis proceeding by a carbonium ion mechanism. The formation of methane and ethane does not occur to any significant extent in acid-catalyzed cracking because their formation involves a primary carbonium ion intermediate which is energetically unfavorable. The products from cracking over the silica-alumina catalyst and over the nickel sulfide on silica-alumina catalyst are predominantly isoparaffins. The ratio of branched to straight chain species in the products from cracking (namely, the iso-to-normal ratio) is high, far in excess of thermodynamic equilibrium. Thus, although relatively little isodecane is found in the product, it is probable, as suggested by earlier investig a t o r ~ ,that ~ extensive isomerization to branched species occurs followed by rapid selective cracking of the branched isomers to produce branched products. With silica-alumina catalyst the product from cracking contains only minor quantities of olefins, indicating Volume 70,Number 11 JYovember 1966

3 668

G. E. LANQLOIS, R. F. SULLIVAN, AND CLARK J. EGAN

-SILICA-ALUM1 NA

2

4

6

6

IO

1

CARBON NUMBER

Figure 2. Products from hydrocracking of n-decane at 288O, 8.0 LHSV, and 82 atm; distribution by carbon number.

that extensive saturation by hydrogen transfer or direct hydrogenation occurs. With the nickel sulfide on silica-alumina catalyst, the products from cracking are all saturated, indicating that the nickel sulfide has sufficient hydrogenation activity to saturate the olefinic intermediates completely. Nonsulfided nickel on silica-alumina catalyst gives products quite different from those obtained with silica alumina or with sulfided nickel on silica-alumina. The primary reaction is isomerization to produce isodecanes. Although a significant amount of cracking occurs, the products from cracking are quite different, consisting almost entirely of normal paraffins. The distribution by carbon number is more random, and a substantial quantity of methane is formed. All these features indicate that the cracking here is occurring by hydrogenolysis on the nickel metal surface. The almost complete lack of isoparaffins in the products from cracking, as shown in Figures 1 and 3, indicates that acid-catalyzed cracking does not occur to any appreciable extent. Gradual Suljiding. To obtain a better insight into the effects of sulfiding, a gradual sulfiding experiment was made wherein n-decane was processed over a nickel on silica-alumina catalyst (3.6y0nickel) a t 288", 82 atm, and a hydrogen/decane mole ratio of 10. When the reaction had reached a steady state, sulfur was introduced a t a slow rate. The sulfur was added to the decane as dimethyl disulfide, which is readily converted by the catalyst in the presence of hydrogen to H2S. Sulfur addition was started at a concentration of 1 ppm in the decane. This was later increased to 30 ppm and finally to 3000 ppm of sulfur. The product was sampled periodically during the time of sulfur addition, and changes in the rate of reaction and product composition were determined. Results are The Journal

OJ'

Physical Chemistry

I

I SILICAALUMINA

I

NICKEL ON NICKEL SULFIDE 0 SILICA-ALUMINA SILICA-ALUMINA

Figure 3. Hydrocracking of n-decane at 288O, 8.0 LHSV, and 82 atm.

I

'0

I 10

1 20

I

30

I 40

I 50

I 00

I 70

I 80

I 90

METHYL NONANES DIMETHYL OCTANE ' 0

10

20

30

40

50

60

70

60

90

100

SULFUR ADDED PERCENT AMOUNT FOR CONVERSION TO N13S2

Figure 4. Hydrocracking of n-decane during gradual sulfiding of the catalyst.

shown in Figures 4,5, and 6 as a function of the amount of sulfur added. When sulfur addition is started, the over-all rate of reaction increases very rapidly a t first and then continues to rise more slowly until a stoichiometric amount of sulfur has been added. When only 1% of the stoichiometric amount of sulfur has been added, the rate of reaction is increased about fivefold. Most of the effect on the over-all rate of reaction is realized when about 20% of the stoichiometric amount of sulfur has been added. The increase in rate of reaction is due to an increase in the rate of cracking. The production of isodecanes remains about constant, but the composition changes. With the nonsulfided catalyst

3 669

EFFECTOF SULFIDIKG A NICKEL ON SILICA-ALUMINA CATALYST

+

HzS are given by Rosenqvist.”

SULFUR IN ECANE

I

z

‘.i z2 LL

0



0.01

0.1

I

I

I .o

10

loo

SULFUR ADDED. X OF AMOUNT FOR CONVERSION TO N i S S a

Figure 5 . Hydrocracking of n-decane during gradual sulfiding.

30

1

CONVERSION TO N i 3 . s ~

Figure 6. Hydrocracking of n-decane during gradual sulfiding. Production of “excess butanes.”

only methylnonanes are produced. As sulfur is added, dimethyloctanes begin to appear, and the quantity increases with increasing sulfur addition. Thus, although the total quantity of isodecanes produced increases only moderately, the degree of branching increases substantially, indicating a high level of isomerization activity. The iso-to-normal ratio in the products from cracking also changes dramatically during sulfiding. This is shown in Figure 5 , where the ratio of isopentane to npentane is plotted against quantity of sulfur added. These data show that even traces of sulfur can have a large effect. A detectable change in the iso-to-normal pentane ratio is observed when only 0.01% of the stoichiometric amount of sulfur has been added. It is of interest that the effect of sulfiding is observed at H2S concentrations below that required to convert bulk nickel to its lowest stable sulfide (Ni3Sz). Equilibrium data for the reaction of nickel metal with

At the temperature of the gradual sulfiding experiment (288”), an H2S/Hz mole ratio of 4.5 X 10+ is required to convert nickel to Ni3S2. Definite effects of sulfiding are observed in the catalytic reaction when the H2S/H2 mole ratio is only 4.5 X lo-’ (equivalent to 1 ppm of sulfur in the hydrocarbon reactant). This is about one-tenth the concentration required for the bulk transition, Nickel on a catalytic surface may be more active than bulk nickel and react with HzS a t lower concentrations than a H2S/H2mole ratio of 4.5 X The sensitivity of the catalyst to trace quantities of sulfur makes it difficult to observe the nonsulfided behavior. Reactants must be carefully purified, and in this experiment a new reactor, which had never been in contact with sulfur-containing reactants, was used. With a reactor which had previously been used with sulfur-containing reactants, a nonsulfided catalyst gave products characteristic of a partially sulfided catalyst. Apparently, some iron sulfides were formed on the stainless steel reactor surface, and these slowly decomposed releasing traces of H2S sufficient to affect the catalyst. Excess Butanes. If the cracking of n-decane involves only simple scission of one or more carboncarbon bonds, the distribution of products from cracking should have a relatively simple relationship. For example, with the sulfided nickel on silica-alumina catalyst, where methane and ethane formation is negligibly small and the product from cracking consists only of C3-C7species, the only reactions possible would be

f C+6 cs

The total butane produced should be related to the other species by the equation: moles of butane = moles of hexane moles of propane

+

- moles of heptane 2

(1)

The actual butane content of the product in some cases is substantially greater than that calculated on this simple basis, indicating that other more complicated reactions are involved. The difference between the actual butane produced and that calculated by eq 1 is called “excess butanes.’’ The formation of excess ~

~~~

~~

~

(11) T.Rosenqvist, J. Iron Steel Inst., 176, 37 (1954).

Volume 70, h’umber 1 1

November 1966

3670

G. E. LANGLOIS, R. F. SULLIVAN, AND CLARKJ. EGAN

butanes increases as the catalyst is sulfided as shown in Figure 6. Excess butanes are also obtained with the silica-alumina catalyst, again showing the great similarity in product distribution obtained with silicaalumina and sulfided nickel on silica-alumina. Production of “excess butanes” probably occurs as a result of polymerization or alkylation to form species of higher molecular weight than decane, which subsequently crack to produce additional butane. For example, if a Clo molecule condensed with a cracked fragment, such as hexene, to form a CM,this could then crack to produce 4 moles of butane. We do not suggest that high molecular weight species crack only to form butane, but, rather, that butane is the most common product of cracking and, hence, is the one which appears to be in excess in a material balance calculation of this type.

nickel silica-alumina salts. The reduced nonsulfided catalyst is thus believed to consist of silica-alumina with all or most of the acid sites in the form of the nickel salt and with crystallites of nickel metal dispersed throughout. Such a catalyst would have high hydrogenation-dehydrogenation activity but low acidic activity. A reaction of the type suggested here between nickel and silica-alumina was postulated earlier by C i a ~ e t t a . ’ , ~ On sulfiding the catalyst with HzS,two reactions are believed to occur: (1) The nickel metal is converted t o nickel sulfide. (2) The HzS-combines with the nickel of the nickel silica-alumina sites and, at the same time, regenerates the original strong acid sites of the silica-alumina. In this proposed mechanism, the role of sulfiding, with respect to the acidic function, is simply to regenerate the acid sites originally preseiit on the silicaalumina support. The sulfided catalyst would therefore be expected to have the same acidity as the silicaalumina. (Evidence for this is presented later.) This mechanism, which postulates that the primary effects of sulfiding are caused by reaction of H2S with a nickel-silica-alumina species, can account for the fact that the effects of sulfiding are observed a t H2S concentrations far below that required for reaction with nickel metal. Presumably, reaction of H2S with the nickel-silica-alumina surface species occurs much more readily. It also accounts for the surprisingly small amount of sulfur required to give most of the effects of sulfiding. Only part of the total nickel on the catalyst is complexed with the silica-alumina surface. The balance is present as nickel metal. If the H2S attacks preferentially the surface species, then considerably less than the stoichiometric amount (to convert all the nickel to Ki&) would be required to achieve the effects of sulfiding. The primary function of the nickel component is that of a dehydrogenation-hydrogenation ~ata1yst.l~ A number of investigator^^*'^-'^ have shown that, in the reaction of paraffins over acid catalysts, such as silica-alumina, an olefin is required to form the carbonium ion intermediate. Once a carbonium ion is formed,

Discussion When a nickel on silica-alumina catalyst is sulfided, the principal changes observed in the hydrocracking of n-decane are: (1) The over-all rate of reaction increases greatly. (2) The predominant reaction changes from isomerization to cracking. (3) The products from cracking are primarily C3-C7 isoparaffins instead of normal paraffins. (4)The product from isomerization is more highly branched. ( 5 ) The production of excess butanes increases. This indicates that more complex reactions, such as alkylation and polymerization, are occurring prior to the final cracking. These changes all suggest a greatly increased catalyst aciditiy which cannot be associated with nickel sulfide per se because nickel sulfide on alumina does not have activity. Kickel sulfide has been shown to have mild acidic properties;12 however, this acidic strength appears insufficient to account for the present results. T’he increased acidity cannot be the result of reaction of the H2S with the silica-alumina because sulfiding the silica-alumina alone actually results in a small decrease in reaction rate. The above observations can be accounted for as follows: T’he acidity of the silica-alumina catalyst is believed to be caused by strong Breinsted acid sites, probably acidic hydroxyl groups attached to the aluminum atoms. During impregnation with a soluble nickel salt the acidic protons of these hydroxyl groups exchange with the nickel ions to form nickel salts of the silica-alumina acid sites. Excess nickel above that required to neutralize the acid sites is precipitated and converted to nickel oxide on drying and calcining. Reduction with hydrogen under the conditions employed prior to reaction is sufficient to reduce the nickel oxide to nickel metal but is not sufficient to reduce the The Journal oj Physical Chemistry

(12) H. Pines, M. Shamaiengar, and W. S. Postl, J . Am. Chem. Soc., 77, 5099 (1955). (13) H . Beuther and 0. A. Larson, Ind. Eng. Chem., Proc. Des. De?., 4, 177 (1965). (14) B. S. Greensfelder, H. H. Voge, and G. 31. Good, Ind. Eng. Chem., 41, 2573 (1949). (15) B.S. Greensfelder, H. H. Voge, and G. ;VI. Good, ibid., 37, 514 (1945). (16) V. Haensel, Adcan. Catalysis, 3, 179 (1951). (17) A. G. Oblad, T. H. Milliken, and G. A. Mills, ibid., 3, 199 (1951).

EFFECT OF SULFIDING A NICKELON SILICA-ALUMINA CATALYST

it can react by isomerization and cracking. New paraffin molecules may be adsorbed by hydrogen transfer with carbonium ions already on the catalyst surface. I n the absence of a dehydrogenation catalyst, olefin formation or activated adsorption of paraffin by hydrogen transfer is the rate-limiting step. The nickel, which is a dehydrogenation catalyst, can accelerate the reaction by providing a ready source of olefin intermediates. Other important functions of the nickel are to hydrogenate the olefin fragments formed by the cracking reaction and to prevent catalyst fouling by hydrogenating unsaturated species which are coke precursors. The addition of nickel to silica-alumina increases the over-all rate of reaction of decane by providing a source of olefinic intermediates. However, the reaction occurring changes from predominantly cracking to predominantly isomerization. This is consistent with the hypothesis that the strong-acid sites of the silica-alumina have been converted to weak acids by reaction with the nickel. Isomerization is an easier reaction than cracking and will still occur on the relatively weak-acid sites. Cracking requires the strongacid sites and, hence, is substantially eliminated by the addition of the nickel. On sulfiding the nickel on silica-alumina, the strongacid sites are regenerated, and cracking again becomes the primary reaction. The over-all reaction rate is much faster than that with the silica-alumina support alone because the nickel sulfide, although not so effective a hydrogenation-dehydrogenation catalyst as nickel metal, is still sufficiently active to provide the necessary olefinic intermediates and to saturate the olefinic products. The excess butanes formed as a result of alkylation or condensation reactions prior to cracking also are

367 1

produced only on the sulfided catalyst and the silicaalumina, indicating that the condensation reactions also require the strong-acid sites. In this mechanism, it is postulated that the acid sites liberated by sulfiding the nickel on silica-alumina are essentially identical with those of the original silica-alumina support. If so, an unsaturated species, such as an aromatic hydrocarbon which can be protonated directly by the Br@nsted-acid sites (and, hence, does not need a dehydrogenation step to initiate reaction), would be expected to react to the same extent over both catalysts. Results of contacting o-xylene with silica-alumina and with sulfided nickel on silicaalumina are shown in Table 11. The extent of the acid-catalyzed reactions of isomerization and disproportionation is about the same on both catalysts as predicted. Also, in the case of the sulfided nickel catalyst, some hydrogenation to cycloparaffins occurs consistent with its expected moderate hydrogenation activity.

Table 11: Reaction of o-Xylene over Silica-Alumina and Sulfided Nickel on Silica-Alumina" Silicaalumina

Isomerization to m- and pxylenes, 70 Disproportionation to toluene and trimethylbenzenes, % Hydrogenation to cycloparaffins, % Hydrocracking, % Unreacted o-xylene, %

NisSz on silioaalumina (3.6% Ni)

42

40

21

22

Trace

13