Ind. Eng. Chem. Prod. Res. Dev. 1980, 19, 15-19
occurs when the oil is exposed to the combustion products of isooctane. (3) The diffusion of free radicals, formed during fuel combustion, into the oil film on cylinder walls is an important process in the #antioxidantdecay occurring in engines. Acknowledgment The authors acknowledge the assistance of and helpful discussion with hqr. R, E, ~~k~~ and D ~G.~A.. ~ ~s. Gratch, and J. T. Kummer during the course of this work. Literature Cited ASTM Special Technical Publication 315 G , "Multicylinder Test Sequences for Evaluating Automotive Engiiie Oils". ASTM, Philadelphia, Pa., 1977. Butler, J. F.,Henshall, S. H., i6"Piston Ring Scuffing", p p 141-155, Mechanical Engineering Publishers Ltd., London and New York, 1976. Gandhi, H. s., Otto, K., Piken, A. G.,Shelef, M., EnVirOn. SCj. Technol., 11, 170 (1977). Korcek. S., Mahoney, L. R., Johnson, M. D., Hoffman, s., SAE Technical Paper No. 780955 (1978).
15
Lavoie, G. A., Ford Motor Company, Engineering and Research Staff, private communication, 1979. Mahoney, L. R., DaRooge, M. A., J . Am. Chem. SOC.,97, 4722 (1975). Mahoney. L. R., Korcek, S . , Hoffman, S.,Willerrnet, P. A,, Ind. Eng. Chem. Prod. Res. D e v . , 17,250 (1978). Otto, K., Daila Betta, R. A., Yao, H. C., J . Air Pollut. Control Assoc.. 24,596 (1974). Otto, K., Montreuil, C. N., Environ. Sci. Techno/., 10, 154 (1976). Otto, K., Sulak, R. J., Environ. Sci. Technol., 12, 181 (1978). Roux, F.,in "Performance Testing Of Lubricants", Chapter 6, p 75, McCue, C. F.,Cree, J. C. G., and Tourret, R., Ed., Applied Science Publisher Ltd., Essex. England, 1974. ~Steffens, ~R. J., Agnew, i J. T.,~ Oisen, R., A., "Combustion of HydrocarbonsProperty Tables", No. 122, Engineering Extension Series, Purdue University, West Lafayette, Ind., 1966.
Received for review August 2 , 1979 Accepted September 17, 1979 Presented a t the Symposium on Synthetic Lubricants and Additives sponsored by the ~ i ~of petroleum i ~ i chemisty, ~ ~ I ~ ~ , , at the 178th National Meeting of the h e r i C a n Chemical Society, Washington, D.C., Sept 9-14, 1979.
Olefin Oligomer Synthetic Lubricants: Structure and Mechanism of Formation Ronald
L. Shubkin," Marguerite S. Baylerian, and Alfred R. Maler
Ethyl Corporation, Ferndale, Michigan 48220
Low-moleculuar-weight hydrogenated oligomers prepared from middle-range (C,-C,,J I-olefins are finding increasing importaince as synthetic lubricating fluids. For this reason, it is desirable to understand both the mechanism by which they form and their structure. The accepted mechanisms and structures associated with cationic potymerization of olefins are not consistent with the physical and chemical properties of oligomers formed by BF, catalysis. In addition, conventional theories of the mechanism do not explain the unique product distribution resulting from these reactions. We propose and present evidence for a skeletal rearrangement through a protonated cyclopropane intermediate during the course of the oligomerization reaction. This explains many apparent anomalies both in the behavior of these reactions and in the properties of the resulting products.
Introduction Synthetic hydrocarbons are becoming increasingly important as high-performance functional fluids, especially in the area of automoitive crankcase lubrication. A wide variety of such fluids may be made by the oligomerization of linear cy-olefins. A large number of catalysts have been used to oligomerize olefins to useful products. In general, these catalysts fall into three distinct classes-cationic, free radical, and anionic. Cationic catalysts, such as AlC13, were reported for the polymerization of olefins derived from thermal cracking of wax as early as 1931 by Sullivan et al. In 1951, a patent was issued to Montgomery e t al. of Gulf Oil Company describing the use of AlC1, for the oligomerization of 1-octene. Free radical or peroxide initiators for a-olefin oligomerization were patented by Garwood of Socony Mobile in 1960. The anionic or coordination complex catalysts in'clude the ethylaluminum sesquichloride/titanium tetrachloride system, with a patent issued to Southern et al. (1961). One type of cationic catalyst has proven to be particularly well suited for the oligomerization of intermediate chain length (cS-Cl6) cy-olefins. The use of BF3 with a 0196-4321/80/1219-0015$01.00/0
protonic cocatalyst (e.g., RC02H,ROH, or HzO) has been shown to give oligomers that, when hydrogenated, have excellent lubricating properties of high-performance applications (Brennan, 1968; Shubkin, 1973a,b). The fluids have low pour points, high viscosity indices, high flash and fire points, high thermal and oxidative stability, and low volatility. These properties make them attractive candidates for applications requiring performance not economically obtainable with conventional mineral oils. Such uses include some high-performance automotive and diesel crankcase lubricants (both whole and part-synthetic packages), gear and axle grease bases, automatic transmission fluids, hydraulic fluids, transformer fluids, heat transfer media (solar energy application), and others. Early in our work with the BF3-catalyzedoligomerization process, we realized that the accepted reaction mechanism for cationic polymerization (Whitmore, 1934), shown in Figure 1, was inadequate to explain our experimental results. Our initial dissatisfaction with the mechanism outlined in Figure 1 was that we were unable to explain satisfactorily the unique product distribution. At low temperature (30 'C), the reaction product is principally the trimer. At
CZ 1980 American Chemical
Society
16
Ind. Eng. Chem. Prod. Res. Dev., Vol. 19, No. 1, 1980
R-CH,-CH
CH, I
R-CH,-ZH-CH
= CH-Z-R
(+ i s o x e r s )
H+ w
-_
= CH,
R-CH,-CH-CH,-CK-CH,-R
?.-CH,-CH
=
CH, 40
30 I
20
10
I
R-i'?12-CH
0 =
CH2
etc.
Figure 1. Accepted mechanism for cationic polymerization.
elevated temperature (85 "C), peaking is a t the dimer. Arguments based on steric crowding were quite unconvincing. The second concern with the accepted mechanism related to the physical properties of the oligomer products. We found lower viscosities and significantly lower pour points than would have been predicted from the molecular structures. Further investigations (described in the Experimental Section of this paper) revealed additional anomalous behavior. Experimental Section The oligomerization of linear a-olefins using BF, has previously been described (Shubkin, 1973a,bj. Briefly, neat a-olefin is stirred with the catalyst complex (BF,-HAj while maintaining a molar excess of BF,. This may be accomplished by bubbling BF3 gas through the mixture or by maintaining a slight partial pressure of BF, gas in the reactor. To monitor the react,ion progress, samples may be withdrawn periodically, quenched with HzO, and examined by GLC. Hydrogenation of reaction products was under 500 psi of H2 over a 5% Ni/Kieselguhr catalyst. GLC analyses were carried out on a Varian Model 3700 gas chromatograph equipped with a CDS 111 integrator. Results are given as area YO. NMR data were obtained on a Varian A60-A spectrometer, and IR data were obtained on a Perkin-Elmer Model 621 grating spectrophotometer. Product Distribution as a Function of Time. Figure 2 shows the product distribution as a function of time for the BF,-catalyzed oligomerization of 1-decene at 50 "C. Dimer is formed rapidly, peaks, and then decreases as higher oligomers are formed. The products change only slowly after depletion of the monomer. Some increase in tetramer and pentamer with a corresponding decrease in dimer and trimer indicates that oligomers can react with each other, but only very slowly. P r o d u c t Distribution as a Function of Temperature. Figure 3 shows the product distribution for the oligomerization of 1-decene with RF3 at 30, 50, and 85 "C. The reactions were terminated, respectively, at a%, 90%, and 93% conversion of the starting olefin. It is clear that the reaction has a very significant temperature dependence. A t 30 "C, there is very pronounced peaking at the
T i m , Min.
Figure 2. Product distribution vs. time-oligomerization ene at 50 "C.
of I-dec-
-1 80-
50-
t
I M-
ii a
30-
I 10
Products
D.p-r O f 01lpmmr,ulmn
Figure 3. Product distribution vs. temperature-oligomerization of I-decene with BF3.H20.
trimeric product (63.6% of total product). At 50 " C , however, the reaction peaks at the dimer. At 85 "C, only 16% of the total product is trimer, with the dimer accounting for 73.1%. Figure 3 also shows that a small amount of depolymerization takes place a t the higher temperature. P r o d u c t Distribution as a Function of S t a r t i n g Olefin. Three different linear a-olefins were oligomerized at 30 "C. In each case, 70% of the starting olefin was converted. The three olefins used for this study were 1-hexene, 1-decene, and 1-tetradecene. Figure 4 illustrates the dependence of the product distribution on the chain length of the starting olefin. This figure shows that the shorter-chain hexene gives a less peaked distribution of products than the longer-chain tetradecene, although both give trimer as the predominant product. The tetradecene not only has the highest percentage of trimer in the product, but it also has the highest amount of dimer, an indication that the long chain length has an adverse effect on chain growth. Physical Properties of Oligomer Products. Dimerization of a-olefins by trialkylaluminum catalysis is
Ind. Eng. Chem. Prod. Res. Dev., Vol. 19, No. 1, 1980
17
Table 11. Number of Methyl Groups for Different Oligomers methyl groups -
Bo-
G
w-
.s
40
-
30
-
20
-
10
-
I
O L
Dlrn
Tr,M
1 -
apr 01 O l l o r r m u t r n
Figure 4, Product distribution from BF,-catalyzed oligomerization. Three different starting olefins. Table I. Comparisons of Physical Properties of Hydrogenated 1-Tetradecene Dimer Prepared by Two Different Catalysts
catalyst
viscosity at 40 'C, cst
viscosity a t 100 "C, c s t
VI
K, A1 BF,.HA
13.;. 12.3
3.6 3.2
175 129
pour point, "C
+ 30
oligomer
predicted
found
C, pentarner C, tetramer C, trimer C , , dimer
6 5 4 3
6.7 5.6 4.6 3.8
of ketone. A high proportion of methyl ketone was confirmed by NMR. A sample of distilled octene dimer was oxidized with ozone at -78 "C. The product was then reductively cleaved with zinc dust and glacial acetic acid. Examination of the product by GLC showed a minimum of 35 different components. Proton NMR, however, indicated that the ratio of methyl ketone to formaldehyde groups was 63:36. Similar results were obtained when distilled octene trimer was ozonized. The presence of many different products after the cleavage reactions indicates that rapid double-bond isomerization takes place. The high percentage of methyl ketone in the product, however, cannot be explained by the generally accepted structures for the oligomers. Assuming total isomerization of the double bond to the methyl branch, the maximum yield of methyl ketone would be only 50% (eq 2). C
0 (1) 0
-36
RCCC
RCC-CCCR __
0
HCCCR
(2)
( 2 ) Zn,HOAc
known to be very specific to the dimeric product having a vinylidene structure (Ziegler et al., 1960). Hydrogenation of this product gives a linear carbon chain with a single pendant methyl group in the middle. The structure of the hydrogenated product should be identical with that obtained by BF3 catalysis (eq 1).
c11, 1
R,A1
2RCH-CH,
RCCH,CH,R
+
BF,.HA
BRCH=CH,
CH3
3-
RCHCH=CHR
Conclusions The observed nature of the BF3-catalyzedoligomerization reaction, as well as the physical and chemical properties of the products, suggests a mechanism that includes a skeletal rearrangement. We propose that a dimeric carbonium ion rearranges through a protonated cyclopropyl intermediate (eq 3).
RCHCH,CII,R
H2
CH3
(1)
We compared saniples of distilled and hydrogenated 1-tetradecene dimer prepared by the two catalyst systems (Table I). Consideration of the data given in Table I (in light of the well-established relationships between structure and physical propert ies of hydrocarbon chains) indicates that the dimer formed by BF3.HA catalysis has a higher degree of branching than that formed with R3AL NMR Analysis OS Oligomers. Oligomers containing differing numbers of monomer groups were isolated by fractional distillation and examined by proton NMR. These data were used to calculate the average number of methyl groups per molecule. Our findings are shown in Table 11, where they are compared to the number of methyl groups predicted by the accepted (no rearrangement) structure shown earlier in Figure 1. The data in Table I1 indicate that, on the average, two out of every three molecules have one more methyl group (Le., one more branch) than expected. Oxidative Cleavage of Oligomers. A sample of distilled decene dimer was oxidatively cleaved by KMnO?. Only traces of carboxylic acids were removed from the product by extraction. GLC showed a total of 33 products, but only four of these accounted for 70.4% of the total. A very strong IR band a t 1700 cm-' indicates the presence
L
A similar mechanism has been demonstrated by Kramer (1969) in the SbF,-HSO,F catalyzed rearrangement of methylbutyl and methylpentyl cations. Karabatsos (1971) also has invoked the protonated cyclopropyl intermediate in his mechanism for the A1C13-catalyzed rearrangement of methylcyclopentane to cyclohexane and cycloheptane to methylcyclohexane. The first step in the oligomerization reaction is the protonation of the starting olefin by the catalyst (eq 4). BF3.HA
+ RCH=CHZ
+
----*
RCHCHB + BF&
(4)
(where A is a suitable base). Evans, Polanyi, and coworkers demonstrated the need for a protonic cocatalyst in a series of papers published in 1946-7 (Evans et al., 1946a,b; Evans and Polanyi, 1947; Plesch et al., 1947). They found that exhaustively purified dry isobutylene fails to polymerize with BF,. A trace of H 2 0 , however, is all that is necessary to initiate the reaction. The unique advantage of many BF3.HA complexes as catalysts probably lies in the moderate acid strength of the equilibrium shown in eq 5 .
BF3.HA + H+BF& (5) The competing cornplexation (where A- is a suitable base)
18
Ind. Eng. Chem. Prod. Res. Dev., Vol. 19, No. 1, 1980 CX3 1
R-CHI,-C
1
CH, 1
-C-CH,-R
X-CH,-C-
I
FH
1) -Hf
I
2)
I"
'El3 l
CR,
- CH-P.*-E e
P
- +s -I?
I R
+H+
&lid
Migration
I
- 3-C:G-R
I
I
H
,
fH3 3oub1e
CH, R-CH,-C
F"3
C-CH,-R
J
3-1H,-,;-"H;C:3-m,-R
LR3
I
@fH
,am
R-C:3,-7
F
1
-
LT,
I
yk
I
2
R
Figure 6. Rearrangement of trimeric carbonium ions. ,..--T,
