Borislav Bogdanovic / Biserka Henc / Hans-Georg Karmann /
Olefin Transformatians Catalyzed he catalytic synthesis of cyclododccatricnc from buta-
Dimerization, codimerization, oligomerization, isomerization, and cyclization of olefins have been realized in the homogeneous phase using a catalyst system based on organonickel compounds, organic phosphines, and aluminumhalides or organoaluminumhalides
34
INDUSTRIAL A N D E N G I N E E R I N G CHEMISTRY
Tdiene (77) is the basis of the production of n~-lon-12 (7). In 1961 an intcriricdiate in tlic forination of cydododccatrieiic with zero 1-alcnt nickel was isolated (75). This intcrniediate was shown to be the organonickcl coinpound shown in Figure 1. In this compound, thrcc molecules of butadiene form a Cln chain that is attached to the nickel by iiicans of t\vo r-allyl groups a t both ends of the chain. Ring formation takes place under thc influence of electron donors, which causc coupliiis of the termiiial carbon atoms. T h e cyclododccatricnc thus formed is then displaced froin the nictal in thc course of the catalytic reaction by a n excess of butadiene. At ihc time of its discovery, this intcrnicdiatc was the first kno\vii organometallic compound Tvith the his-n-all>-1 iiictal structure. l ' h e parent compound can hc considcrcd to be bis(r-ally1)nickel (Figure 1) ( 1 6 ) . T h e catalytic properties arc changed if onc of the allyl groups in bis(x-all>-l)iiickelis substituted by a halogcn or other anionic groups. n-Allylnickelhalidcs (Figure 1) are dimeric, Ivith halogen bridges, and of lo\\- catalylic activity toward moiioolcfins. Highly active catalysts are obtained by coinplcxing r-all>-lnickelhalidcs with aluniinuinhalides or organoaluminunihalides ( 4 ) . T Allylnickelhalidcs, in conjunc*Lionwith rliese LcLvis acids, catalyze the polymerization of butadiciie, tlic dinicrization of monoolcfins, such as cthylcnc or prop)+lcnc,aiid the dimcrizatioii aiid isoiiicrization of higher olefins. These types of catalytic reactions are by n o mealis unique to nickel a n d catalyLic systems based on palladium (8), cobalt (13), rhodium (5), and otlicr transition metals that are also aeti1.e. The stimulus for the study of the nickel catalyst system was the observation we iiiadc in 1963 that rhe course of the propylene dimerization (Figure 2) c a n bc controlled to give prcdoniiriantly 2-nietliylpe1itc1ics or 2,3-diinethylbutenes ( 4 ) . This is made possible by changing the ligand-usually an organic pliospliiiic-
Hans-Georg Nussel/ Dieter Walter / Gunther Wilke
by Organonickel Compounds introduced the possibility of producing 2,3-dimethylbutenes by a process also involving the dimerization of propylene. These hydrocarbons can be hydrogenated to 2,3-dimethylbutane, which has good antiknock properties, along with a low boiling point, or they can be dehydrogenated to 2,3-dimethylbutadiene, which can be used for polymerizations, Diels-Alder reactions, etc. (4).
Figure 7. I , the intermediate in the cyclododecatriene synthesis; II, bis(n-a1lyl)nickel; III, n-ally hickelhalides
Preparation of Catalyst Instead of r-allylnickelhalides, alkylnickelhalides can be used for the preparation of the catalyst; however, since alkylnickelhalides are usually not particularly stable compounds, the catalyst in these cases must be prePared in situ--bY of salts with Organ@ aluminum compounds ( 7 7). This method is typical for the DreDaration of Zieder catalvsts. T h e advantage ., of using n-allylnickelhalides as catalyst component is that the phosphine can be attached to the metal in a definite manner before adding the olefin. This is of particular importance in “directing” the course of catalytic reaction by means of phosphines. I t will be shown (Figure 11, p 41) that allylnickel conipounds are formed in the catalyst mixture, even if one starts from alkyl nickel compounds. T h e n-allylnickelhalides used can be prepared either by reaction of zero valent nickel compounds with allylhalides, or by the cleaving of bis(n-ally1)nickel compounds with hydrogenhalides (see Figure 3). As nickel (0) component nickeltetracarbonyl was used in the first preparation A
Figure 2.
Dzferent possibilities f o r the dimerization of propylene
attached to the transition metal. We call this type of catalyst a “phosphine-modified catalyst.” Whereas the production of 2-methylpentenes by dimerizing propylene is already performed on an industrial scale using an organoaluminum reaction in the Goodyear Scientific Design Process for the production of isoprene ( 7 ) , the phosphine-modified nickel catalyst
v
A
AUTHORS B. Bogdanovi?, B. Henc, H. G. Karmann, H . G . Nussel, D. Walter, and G . Wilke are with the Max-PlanckInstitut f u r Kohlenforschung, Miilhezm-Ruhr, West Germany. This paper was presented as part of the Symposzum on Novel Processes and Technology of the European and Japanese Chemzcal Industry, 758th ACS National Meeting, New York, N . Y., September 7-72, 7969. VOL. 6 2
NO. 1 2 D E C E M B E R 1 9 7 0
35
Figuie 3. Preparation of n-all~lnickclhalides
Figure 4.
Preparation uf the phosphine-free catalyst
reported by Fischer and Rurgcr (6). Instcad of nickcltetracarbonyl, we use nickel olefin coiiiplcxcs, such as bis(cyc1ooctadiene-1,5) nickcl, since higher J iclds are acliic\-cd, and handling of the highly toxic nickelcarbonyl is avoidcd (2). The second incthod-the cleavage of bis(w-allyl) nickel conipounds (3)-is useful for the preparation of the catalyst on a larger scale (see
Figure 5. Preparation of the phosphine-modibed catallst; reactions of thephosphine-free and of the phosphine-modzjisd catalyst with Lezis bases
36
INDUSTRIAL A N D ENGINEERING CHEMISTRY
below). Both routes can be used to prepare substituted n-allyliiickelhalides. T h e phosphine-iree catalyst is obtained by mixing a w-allylnickelhalide with a n aluminumhalide or an organoalurninumhalide in a halogen hydrocarbon solvent, such as chlorobenzene or nicthylene chloride. \\:hen these conipoiients are mixed, a coniplex is formed (Figure 4) in which the halogen of thc w-allylnickellialidc becomes a part of thc aluminumtetrahalide (or alkyltrihalide) anion. These polar coniplexes are only sparingly soluble in hydrocarbons, but they are soluble in polar, weakly basic solvcnts, such as chlorobenzene. I n the Lewis acid “activated” rr-allylnickclhalidc coiiiplexes, the donating electron pair of the halogcn is shared with the aluminum. This creates free coordination positions around nickel, which become sites for the complexation of olefins. If these free coordination positions are occupied by electron donors (Figure 5) carbon monoxide, phosphines, or cyclooctadiene-l,5, definite but catalytically inactive complexes 111, IV, and V I can be isolated (74). Still catalytically active are thosc coniplexes in which only one coordination position on nickel is occupied by a phosphine-the “phosphine-iriodified catalyst” (11). By changing the phosphine, the course of catalytic reactions can be controlled. T h e phosphinemodified catalyst can be prepared cirher by adding phosphines to a solution of the phosphine-free catalyst, or by combining the crystalline 1 : 1-phosphine adducts of r-allylnickelhalides with the Lewis acid component. This method of prcparatioii of the catalyst would bc too cxpensil-e for pilot plant scale or even technical production. Therefore, we dcvclopcd a “onc-pot process” for preparation of thc catalyst solution, starting with tcchnicalll- available raw materials (Figure 6 ) .
Figure 6 .
"One-pot process" for the preparation of the catalyst
A suspension of anhydrous nickelchloride in chlorobenzene is treated in the presence of excess butadiene with triethylaluminum, whereby the nickel intermediate of the cyclododecatriene synthesis (Figure 1) and diethylaluminumchlor ide are formed alniost quantitatively. Excess butadiene is pumped off and gaseous hydrogenchloride-5 mol per mol of nickel-is introduced. One allyl group of the intermediate is cleaved, resulting in a Clz-7r-allylnickelchloride. At the same time, aluminumchloride is formed, which is complexed with (21%-x-allylnickelchloride. In the last step, the desired phosphine is added to this solution. T h e catalyst thus prepared has been used for the dimerization of propylene in a pilot plant with a capacity of about one ton of propylene per 24 hr (Figure 7) ( 4 ) . Liquid propylene is pumped under a pressure of 15 atin into the reactor through a drying tower containing silica. T h e reactor is a 15-m long, double-walled tube of 20-1. v01, which is water cooled. T h e catalyst solution is injected together with propylene a t the bottom of the reactor. The temperature in the reactor rises to 30" to 40". T h e product is taken off the top of the reactor, and the pressure is reduced to 5 atm. After passing a washing tower, in which the catalyst is destroyed with aqueous alkali and air, and a settling tower, the mixture is taken to a stabilizing
column, where unreacted propylene and product are separated. Reaction rates of dimerization of about 10 to 15 kg prod/g Ni/hr have been realized under these conditions, with the catalyst consumption of 50 mg Ni/kg prod. This means that about 30,000 propylene molecules are chemically transformed on one molecule of the catalyst by passing through the reactor. T h e product consists, with a conversion of about 95'%, of 887Z0 propylene dimers, 10% trimers, and 1-2% tetramers. By using tricyclohexylphosphine, the formation of trimers is somewhat higher. Influence of Phosphines
T h e study of the influence of phosphines on propylene dimerization has been performed on the laboratory scale as well as in the pilot plant, and shows no significant differences. The variation in the sum of n-hexenes, 2methylpentenes, and 2,3-dimethylbutenes on changing the phosphine is shown in Table I. In the series (CsH5)zP to P(i-CzH7)z the yield of n-hexenes decreases gradually from 21.6 to l.8yo,and the yield of 2,3-dirriethylbutenes increases u p to 67.97,. Still higher yields of 2,3-dimethylbutenes are achieved by using phosphines with tert-butyl groups. However, by introducing two tert-butyl groups in combination with one isopropyl group in the phosphine, the yield of 2,3dimethylbutenes suddenly drops to 29.1%. At the same time, the yield of n-hexenes decreases also so that a high yield of 2-methylpentene results. There are obviously two effects operating. T h e first one-the increase in the yield of 2,3-dimethylbutenes-parallels the inductive effect of the groups R attached to the phosphorus, that is, the increasing basicity of the phosphines PRS, so that this might be considered to be an electronic effect of the ligand on the course of the catalytic reaction. [h'ote Added in Proof: According to recent investigations, the phosphine effects described are considered to be steric
Figure 7. Pilot plant f o r the dimerii:ation of propylene
VOL. 6 2
NO. 1 2
DECEMBER 1 9 7 0
37
TABLE I. INFLUENCE O F PHOSPHINES ON PROPYLENE DIMERIZATION AT -20°C AND ONE ATM
EnHexenes, Phosphine
OaP chzPCH24 chzP(CH2j3P02 +d'CHzP+z P(CH3)3 chzP(i-C3H7) P (C zH 5 1 3 P(n-CdHu)s P(CHaO)3 P[N(CzH,)zla
P(i-CsH7)s CH3P(i-C4Hg)* CZH~P(~-C~HD)~ (i-CaH,)p(t-C~Ho):!
CZ-
Methl-
pentenes,
CT3-
Dimethylbutenes,
7%
"ic
"/c
21.6 19.2 20,l
73.9 75.4 73.3
4.5 5.1 6.6
12.2 9.9 14.4 9.2
83.0 80.3 73.0 69.7
4.7 9.8 12.6 21.1
7.1 6.7
69.6 63.6
23,3 29.2
5.5
51.4
43.0
46.5
49.2
3.3
37.9
58.8
1.8 1.2 0.6 0.6
30.3 24.5 22.3 70.1
67.9 74.0 77.0 29.1
rather than electronic effects of the ligand.] This inductive effect should he the highest with phosphines having tert-butyl groups. Hohvever, by introducing tert-butyl groups a new effect-the decrease o i 2,3-diiiiethylbutenes-appears, and we ascribc this effcct to steric hindrance of the bulky tert-but)-1 groups. Better insight into the nature of the effect of the phosphine on the catalytic reaction can be obtained if one takes into account the assumed reaction mechanism and the composition of
Figure 8.
38
Dimerization of propylene
INDUSTRIAL AND ENGINEERING CHEMISTRY
the dimers, in respect to the position of the double bonds. From recent investigations we assuiiie that the nicchanism is of an organometallic growth reaction, the active intermediates being hydride and alkyl nickel compounds. We will later present some of the experimental support for this mechanism, but for the moment we ~ 7 i l coiifine l ourselves to the effect of the phosphine on this mechanism. T h e scheme for the propylene dimerization is shown in Figure 8. In the first step of the reaction the nickel hydride catalyst adds across the double bond of propylene to give two intermediates-a propyl nickel and an isopropyl nickel complex. Both of these intermediates can react further with propylene by insertion of the double bond into the nickel carbon bond, resulting in the four intermediates shown in Figure 8. Beta-elimination of the nickelhydride from these interniediates produces the possible primary products of propylene dinierization : 4-merhylpentene-l , cis- and trans-4-methylpentene-2, 2,3-dimethylbutene-1, hexenes-I and 2, and 2-methylpentene-l . All these olefins are actually found among the reaction products. I t should be noted that the primary nickel alkyl intermediates can eliminate nickelhydride in only one direction, leading to 2,3-dimethylbutene-1 and 2-methylpenteiie-l , while the secondary nickel alkyl interniediates can eliminate nickelhydridc in three different ways. Terminal unbranched olefins are very rapidly isomerized under the influence of the catalyst by a process of repeated nickel hydride addition and eliiriiiiation to the internal olefins. Thercforc, under ordinary reaction conditions thc yield of 4-nicthylpentene-l is low. 2-Methylpentene-I and cis- and trans4-methy-lpentene-2 isomerize to 2-niethylpentcnc-2, and 2,3-dimethylbutene-I isomerizes to 2,3-diniethylbutene-2. I n connection with the influence of the phosphine
Figure 9. Codimerization of ethylene and propylene
molecule the question arises as to which step of the catalytic reaction is controlled by the phosphine. Is it the first step, the addition of nickelhydride to propylcnc, or the second step, the addition of nickelalkyl to propylcnc, or both steps, and how is the direction of addition influenced by the phosphine? Under the assumption that the isopropyl nickel complex and thc propyl nickel complex have the same reactivity towards propylene, it can be seen from Figure 8 that the direction of the addition in the first reaction step (in yo)is given by the ratio
tion stcp, together with unchanged direction of addition in thc first step, gives rise to the synthesis of 2,3-dimethylbutenes from propylene. Now lct us consider thc influcncc of tert-butyl groups. Since thc inductive cffect of a tert-butyl group is greater than that of an isopropyl group, wc should expect that the influence in the second step should be further increased; this is in fact found (from 14/86 to 2/98).
+
4-methylpcntene-1 and -2y0 2,3-dirnethvlbutene-1 and -297, B hexenes % 2-methylpentene-l%
+
TABLE I I . INFLUENCE OF PHOSPHINES ON H-Ni
similarly the average direction of the addition in the second reaction step is given by the ratio
AND ON -C-
+ 2 hexenes yo + 2-methylpentene-lTo
I
Ni ADDITION ON PROPYLENE
4-methylpentene-1 and -2?4 2,3-dimethylbutene-l and -2y0
P [N(C2Halz13 C
To obtain these ratios for different phosphines, it is necessary during the dimerization to avoid the isomerization of 2-methylpentene- 1 and 4-methyl pen tene-2. With strongly basic phosphines the isomerization rate is low (see Table 111) ; in other cases the isomerization can be suppressed by using low conversions. T h e influence of some representative phosphines on H-Ni and on
\
-C-Xi
/
C& .
.
.
75
81
25
19
19
18
69
70
81
41
14
3
2
yo
19
59
86
97
98
.
H--P\'i
C
d. = C. %
.
.
H-Ni C
addition on propylene is shown in Table 11.
In the series of phosphines bis(dipheny1phosphino)methane, tris(diethylamino)phosphine, and triisopropylphosphine, the direction of addition of nickelhydride on propylene does not change appreciably (from 75/25 to 81/19). I n the second reaction step the direction of addition of nickelalkyl on propylene is almost reversed (from 81/19 to 14/86) on changing the phosphine. This change in the direction of addition in the second reac-
C=C \I
I
i
-C-Ni
/ C
I
C=C
\:
i
-C-Ni
/
VOL. 6 2
NO. 1 2
DECEMBER 1970
39
TABLE 1 1 1 . T H E ORDER O F PHOSPHINES OF DECREASING ISOMERIZATION RATE AND OF DECREASING DIMERIZATION VS. 0 L I GOM ER I ZAT I ON RATE
However, by using a phosphine with two tert-butyl groups a n d a n isopropyl group: we see that the direction of addition in the first step has suddenly been changed. We now have only 3156 addition forming isopropyl nickel and 69y0 addition forming propyl nickel. This influence,is quite obviously a steric one; it can be easily recognized that the addition forming propyl nickel is sterically less hindered. T h e phosphine thus exerts two distinct influences; by changing the basicity of the phosphine we can vary the dircctioii of the insertion of the olefin into the nickel-carbon bond over a wide range. T h e direction of addition of nickel hydride is not influenced by the baiicity of the phosphine. Control of the direction of the nickelhydride addition is possible by using more or less sterically hindered phosphines. We are now in a position to better understand the course of propylene dimerization. T h e value of this understanding also lies in the possibilities it creates in predicting the course of similar catalytic reactions on changing the phosphine. O n e such example is that of codimerization of propylene with ethylene. T h e selectivity of codimerizatioii is not high, because along with C j codiiriers of propylene and ethylene larger amounts of ethylene and propylene diiiicrs are formed. T h e composition of the C ; fraction is now of interest. T h e scheme of ethylene-propylenc codinierization is shown iii Figure 9. T h e formation of C;, olefins can take place in two different ways. In the first case, nickelhydride first rcacts with ethylene, leading to nickelethyl, which then reacts with propylene in two different ways. I n the second case, the iiickclhydride first reacts with propylcne, gi\,iiig propyl- and isopropylnickel, which then react with ethylene. By using the knowledge on the influence of phosphines from the propylene dimerization, we can now forecast that only in the first case will the codiriierization be iiifluenccd by the basicity of phosphines, enabling u s to produce a t mill pentenes or methylbutenes. By use of the catalyst containing tricyclohexylphosphinc, wc obtain a C S fraction with more than 90% methylbutenes, the main component (85y0)being 2-nieth>-lbuteiie-l. This result clearly shows that ethylene reacted first, and the insertion of propylcnc in the second step took place in the predicted manner. By using triphenylphosphine, a roughly 1 : 1 mixture of n-pentenes and methylbutenes is obtained. T h e amount of higher oligomers formed during the dimerization of propylene is also influenced by the nature of the phosphine; their yield increases with the basicity of phosphines. Also, the propylene dimers obtained with basic phosphines are 0111). isomerized to a slight extent. I n other Lvords, predominantly primary reaction products are obtained, while those obtained using less basic phosphines, e.g., triphenylphosphine, are 40
INDUSTRIAL A N D ENGINEERING CHEMISTRY
to =
-20°C;
Dimeriration/Oligomerization o j Ethylene 7 atm; catabst: .rr-C3HeAVzBr.PR3 (C2Hb)3Al2Cl3
+
Pi?3 P(CH3)3 p43
CZH~P(~-C~HD)~ (i-C3H7)P(t-CdWg)2
90
-1 10
...
70
2.5
5
65 25
25
10
25
50
>98
...
strongly isonxerized. These observations led to investigations of the influence of phosphine on thc rate of isomerization, and on the ratios of isomerization, dinicrization, and oligomerization rates. T h e order of the phosphines shown in Table I11 does not parallel exactly the increasing basicity of phosphines, sincc triinethylphosphine is morc basic than triphenylphosphine. T h e application of this phosphine effcct can be illustrated with tu70 examples. By using the catalyst n-allylnickelbroniide-trimeth7-lphosphine plus aluminumsesquichloride a t -20°C and 1 atm, more than 98yGof diniers of ethylene, about 1% of trimers, and no higher oligomers are formed (see Tablc 111). T h e amount of oligomers increases with basicity and/or steric hindrance of phosphines in the series given above so that finally with di-tert-butyl-isopropJ7lphosphine abaut 50% of products are tetramers and higher oligomers. M'c have here a phosphine-controlled organometallic "growth reaction." T h e oligomers however are not only linear: but also branched olefins. T h e second example (Table IV) represents the isomerization of propylene dimcrs. T h e composition of propylene dimers as obtained with tricyclohexylphosphine catalyst under pilot plant conditions is shown in the left column. High yield of primary reaction products such as 2,3-diniethylbutene-l) and 2-methylpentene-1 in this case should be noted. If the mixture is additionally treated with the catalyst containing trimethyl phosphine, the migration of double bonds to the thermodynamically more stable positions occurs and the composition of dimers, shown in the right-hand column, is obtained. T h e main components arc now 2-methylpentene-2 and 2,3-dimethylbutene-2. T h e purpose of this isomerization is to enable a distillative separation of 2-methylpentenes from 2,3-dimethylbutenes. T h e formation of n-allylnickel compounds during catalytic dimerizations can be demonstrated in a clear man-
TABLE IV.
ISOMERIZATION O F PROPYLENE DIMERS
a
b
0.1 6.9 4-Methylpentene-1 4-Methylpentene-2-cis 5.5 0.3 63.9 2.2 2,3-Dimethylbutene-1 2.3 4-Methy lpentene-2-trans 5.0 11.4 2-Methylpentene-I 0.6 0.1 Hexene-1 0.2 Hexene-3-tran~and cis 0.6 Hexene-2-trans 1.0 ... 2-Methylpentene-2 2.7 28.6 Hexene-24s 0.5 1.3 1.1 2,3-Dimethylbutene-2 63.5 a Comfiosition of propylene dimers (in wt yo) obtained with the catahst containing tricycloke.xy;ylpho@hine at 40°C and 7 5 atm. b Isomerized profiylene dimers ( a ) after treatment at 0°C with the catahst containing trimethylpliosphine.
ner with the example of the catalytic dimerization of cyclooctene (Figure 10). If cyclooctene is treated with the prepared from nickelacetylacetonate, and ethylaluminumsesquichloride in a smooth dimerization reaction takes place, the products being 1-cyclooctylcyclooctene and dicyclooctylidene. If the reaction mixture is now treated in absence of air with aqueous ammonia, the aluminum component precipitates, and from the organic layer a brown sublimable nickelorganic compound can be isolated in high yields. This compound is a a-allylnickelacetylacetonate, with the structure as shown in Figure 10. I n the course of the catalytic dimerization, a a-allylnickel compound is formed in which the dimer of cyclooctene is combined with the metal by means of a 9-allyl
grouping. By adding a n aluminumhalide to the solution of this compound, we obtain a n active catalyst for the dimerization and isomerization of olefins. Analogous results are obtained if, instead of cyclooctene, other olefins, for example propylene, are used. I n this case, however, we do not obtain one .rr-allylnickel compound, but a mixture of them. Reaction Mechanism
These results lead to the question: what happens to the allyl group of allylnickel compounds if they are introduced as the catalyst component a t the beginning of the catalytic reaction? This question led us directly into the problem of the reaction mechanism. T h e problem was studied again with the example of the dimcrizarion of c)-clooctene. and as rhe allylnickcl com\vas used (Figure 1 1 ) . ponent 9-all).lnickclacer).lac.eron~~~ a-Allylnickelacetylacetonate activated with aluminumsesquichloride was treated with a n excess of cyclooctene, and a t the end of the catalytic reaction aqueous ainmania was added, T h e substituted 6~a-allylnickelacety1acetonate,, was again isolated in high yield. T h e question remains : in which stage of the catalytic reaction does the exchange of the original allyl group occur?
In subsequent experiments the cOurSe of catalytic cyclooctene dimerization was followed by removing
Figure 7 7 . Exchange of allyl groups bonded to nickel zn the course of cyclooctene dimerization
Figure 10. Dimerization of cyclooctene
Figure 72. Relation between the dimerization of cyclooctene and the concentration of ally lnickel compounds VOL. 6 2
NO.
12
DECEMBER 1970
41
samples during the reaction, destroying the catalyst with ammonia, determining the quantity of the 7r-allylnickel compound in the solution, and determining the conversion to dimeric cyclooctene, as Me11 as the dimerization rate. T h e results are shown in Figure 12. Here the concentration of allyliiickcl conipounds, conversion, and the dimerization rate are plotted against time. If a catalyst sample is treated with ammonia before the addition of cyclooctene, 80 to 90% of the original allyl group is found in forin of bis(7r-ally1)iiickel. This is proof that ~ - a l l y l n i c k e l a c e t ) - l a c ~ t ~ iis i astill t c presciit in the solution, since 7r-allyliiickelacctylacetonate disproportionates under the influence of ammonia according to the reaction shown in Figure 12. If, however, thc catalyst is destroyed imiiicdiatcly after addition of cyclooctene, 0111)- traces of allylnickel compounds are found iii thc solution, and with the addition of ammonia irictallic iiickcl is precipitated. At that momcnt the dinierization rate is \ w y high. Subsequently, the dimerization rate decreases, whereas the concentration of allylnickel compounds is still practically zero. Only after most of the cyclooctene is dimerized does the forniatioii of allylnickcl compound of the dimeric cyclooctene occur. After complction of the catalytic reaction, the concentration of this allylnickcl conipound remains constant ai the value of about 70 to 80Yc of the theory. Froiri this expcriiiiciit it can bc concludcd that (1) the acti1.c species of thc catalytic reaction docs not include a allylnickcl conipound, since a t the point of highest reaction rate practically no allylnickel coinpound is present in the solution, (2) the active species is fornied from the allylnickcl coinpouiid by addition of the olefin, and (3) after the ciid of thc catalytic reaction, the active species is transforiiicd into thc allylnickcl conipound of the cyclooctene dimer, and in this way a stabilization of the catalyst systeni is achieved. T h e role of the allylnickel compounds is to stabilize the catalytic system in the absence of reactive olefins. This result, howel-er, has not answered the original question : what happens to the original allyl group of the cataly-st on addition of the olefin?
F i p r e 73. Formation of the (‘actiue’) cataljst b j rcncticn of the .rr-ul[iminickel compound and cjclooctene 42
INDUSTRIAL A N D ENGINEERING CHEMISTRY
Figure 14. Reaction path ,for the formation of bicyclo [6.3.0]undecenes
At a temperature of -40°C (Figure 13): the dimerization rate of cyclooctene is very low, and by adding amnioiiia immediately after cyclooctene addition the reaction can be stopped at the moment in which the allyl group has already reacted, but before any appreciable amounts of dimers have been formed. Under these conditions, 80-85%;, of the allyl group can be detected in thc form of olefins of the formula CllH18; this means that the allyl group reacted in the first reaction step with cyclooctenc. T h e fact that the formation of CllHI8 olefins from allylnickel compound and cyclooctene takes place without hydrogcn evolution can be taken as indirect evidence for the formation of a nickelhydride species. T h e C11 olefins obtained from the reaction of allylnickel catalyst with cyclooctene were expected to be allylcyclooctenes, formed through insertion of cyclooctene in the nickelallyl bond and subsequent P-elimination of nickelhydride. O n hydrogenation, however, a saturated hydrocarbon of the formula C llH20, which prolred to be bicyclo [6.3.0]undecane,was obtained. T h e main component of C11 olefins (>80$,) is bicyclo[6.3.0]undeeene-3 (Figure 14). T h e formation of this hydrocarbon from the allylnickel compound and cyclooctene is assumed to take place by cyclization and subsequent isomerization of the initially formed 3-allylcyelooctene, as shown in Figure 14. Similar cyclizations of 1,5-dienes by means of organoaluniinum compounds have been reported by Ziegler and coworkers (18). Formation of a five-membered ring compound by means of a nickel catalyst has been reported by Muller e t al. (10). When 3-allylcyclooctene is treated with the catalyst prepared from s-allylnickelchloride, tricyclohexylphosphine, and ethylaluminumsesquichloride (Figure 15): a mixture of bicyclo[6.3.O]undecene-Zand -3 (35-40yc,) and 3-propenylcyclooctenes is obtained, since the 1,5-diene cyclization reaction is competing with double bond isomerization. T h e isomerization/cyclization rate of 1,5-dienes can be changed when using different phosphines; this is shown in Table V for cyclooctadiene-1,5. This hydro-
TABLE W.
ISOMERIZATION OF CYCLOOCTADIENE-l,5
0
H Ni
4
YTNi(8a' 2%
Bicyclo [3.3.0]-octene2, % Cyclooctadiene-1,3 and 1,4, %
cyclooctene, cis- a n d trans-3-propenylcyclooctene, and 3-allylcyclooctene. 3-Allylcyclooctene reacts again with the catalyst (probably it never separates from the catalyst, and the elimination and readdition of nickelhydride takes place within the chelated r-complex), giving bicycle [6.3.0]undecenes. Since the direction of insertion of propylene in the nickel carbon bond is strongly influenced by phosphines (see Table 11), it is possible in the present case (Table VI) to change the yield of 3-propenylcyclooctenes us. the yield of 3-isopropenylcyclooctenes from 68.5/17.7 to 8.9/73.8 on changing triphenylphosphine for tricyclohexylphosphine. T h c influ-' ence of phosphines is in agreement with results obtained from the propylene dimerization. T h e fact that the yield of bicyclo[6.3.0]undecenes does not change appreciably on changing the phosphines (14.17%) can be explained if we consider that here two effects are operating in conflict with each other. A phosphine of low basicity (e.g., triphenylphosphine) favors the formation of the secondary nickelalkyl intermediate, which is the precursor of bicycloundecenes, but thc isomerization tendency of this catalyst is high, so that a high yield of 3-propenylcyclooctene results.
