2446
INDUSTRIAL A N D ENGINEERING CHEMISTRY
Polyesters of low molecular weight, such as that supplied by Rohm & Haas as experimental fluid AP-52, represent a n estertype base fluid which has been prepared from dibasic acids and glycols to the desired molecular weight range. These fluids, although they often possess excellent viscosity-temperature properties b y themselves, are susceptible to visccsity increase and viscosity index improvement by means of both acrylic and higher molecular weight polyester additives. The effects of adding these polymers to a polyester lubricant are given in Table XI. The main effect to be desired here is viscosity increase, as the viscosity index for this particular polyester is already excellent. CONCLUSIONS
The general effects on viscosity and viscosity index of blends of polymeric additives and synthetic ester-type lubricants have been presented. I n general, improvement in both properties is obtained where the polymeric additive is soluble in the particular ester involved. These polymers have also been used with lower boiling synthetics which are adaptable to use in hydraulic fluid mechanisms, but these hydraulic fluids, although requiring some lubricity, do not seem to fall within the scope of this paper. It is hoped that this work will serve as a basis for further applications for polymer-thickened synthetics. ACKNOWLEDGMENT
The writer wishes to arknowledge the able assistance of Elsie E. Becker. LITERATURE CITED
(1) Adkins, D. C., Jr., Baker, H. R., Murphy, C. M., and Zisman,
w. A., IND.ENG.CHEM.,39,491 (1947). (2) Ballard, S. A., Morris, R., and Van Winkle, J. L. (to Shell Development Co.), U. S. Patent 2,481,278 (Sept. 6, 1949). (3) Bried, E. M., Kidder, H. F., Murphy, C. M., and Zisman, W. A., IND. ENG.CHEY.,39,484 (1947).
Vol. 42, No. 12
Bruson, H. A., U. S . Patent 2,901,627 (Aug. 31, 1937); 2,100,993 (Nov. 30,1937). Byers, J. H., Natl. Petroleum News, 28, No. 49, 83-9 (1936). Evans. H. C.. and Young. ENQ.CHEM..39, 1676 - D. W.. IND. (1947).
Fenske, M. R., Office of Scientific Research and Development, O.S.R.D. Rept. 1894 (Oct. 6,1943). Fife, H. (to Carbide & Carbon Chemicals Corp.), U. S. Patent 2,457,139 (Dec. 28, 1948); Brit. Patent 606,407 (Jan. 15, 1946).
Gisser, H., Lubrication Eng., 5, No. 4, 176 (1949). Glavis, F. J., and Stringer, H. R., Am. Soo. Testing Materials, “Symposium on Synthetic Lubricants,” 1947. Hain, G. M., Jones, D. T., Merker, R. L., and Zisman, W. A., IND. ENG.CHEM.,39,500 (1947). Hamilton, W. (to Lockheed Aircraft Corp.), U. S. Patent 2,392,530 (Jan. 8, 1946). Hardiman, E. W., and Nissan, A. H., J. Inst. Petroleum, 31, 225 (1945).
Merker, R. (to U. S. Navy), U. S. Patent 2.456.642 (Dec. 21, ENQ.CHEM.,41,2546 (1949). 1948); IND. Morgan, J . D., and Loew, R. E. (to Cities Service Oil Co.), U. 9. Patent 2,396,191 (March 6 , 1946). Morway, A. J., and Zimmer, J. C. (to Standard Oil Development Co.),Ibid., 2,467,147 (April 12, 1949). Otto, M., Blackwood, A. J., and Davis, G. H. B., Refiner Natural Gasoline M f r . , 13, 411-22, 426 (1934); Oil Gas J . , 33, NO.26,98-106 (1934). Otto, M., and Mueller-Cunardi, M., U. S. Patent 2,130,507 (Sept. 20, 1938). Penzig, F. G., Tech. Rept. GS USAF, Wright-Patterson FTR 2206 ND, “Extrapolation of Viscosity Index,” Project DP-151, Supt. Documents, Washington 26, D. C. (May 1948). Ramser, J. H., IND.ENG.CHEM.,41,2053 (1949). Sanderson, R. T., Ibid., 41,368, 375 (1949). Standard Oil Development Co., Brit. Patent 692,214 (July 16, 1943). Thomas, R. N., Zimmer, S. C., Turner, C. B., Rosen, R., and Frolich, P. K., IND.ENG.CHEM.,32,299 (1940). Van Horne, W. L., Ibid., 41, 952 (1949); Petroleum Refiner, 27, NO.3, 90-5 (1948). Zisman, W . A., and Hain, G. M., U.S. Patent 2,448,557 (Sept. 7, 1948). RECEIVED June 23, 1950.
NONCATALYTIC POLYME OLEFINS TO LU F. M. SEGER, H. G. DOHERTY,
ILS AND
A. N. SACHANEN N. J.
Socony-Vacuum Leboratorisr, Paulsboro,
Excellent lubricating oils have been prepared by the noncatalytic polymerization of normal 1-olefink. The variables investigated in this study were: temperature, time, pressure, and olefin chain length and structure. Optimum results were obtained at 650’ F., 10 hours reaction time, and 250 pounds per square inch gage. From 1-decene the yield of synthetic lubrieant was 64.3 weight 70 of the olefin charged. Inspection data were as follows: viscosity, 5.79 cs. at 210’ F.; viscosity index, 142; pour test, -5. Cracked paraffin wax and Fischer-Tropsch product have also been used successfully as charge stocks. The synthetic oils were considered as potentia1 replacements for premium grade motor oils.
ATALYTIC polymerization of olefins or, in general, of unsaturated hydrocarbons has been discussed in numerous artid e s and patents. Production of polymer gasolinefromgaseous olefins and synthesis of plastics or rubber from isobutene, butadiene, and other unsaturates are well known and widely used comniercial processes. Synthetic lubricating oils were produced b y catalytic polymerization of olefins in Germany during World War
C
11. In conbrast, noncatalytic polymetization of unsaturated hydrocarbons has not attracted much interest. The activation energy of noncatalytic polymerization of unsaturated hydrocarbons is usually fairly high, and, a8 a result, relatively high temperatures are necessary to effect the polymerization. Hence various side reactions may occw, simultaneously affecting the yield and properties of polymerized products. The use of proper
INDUSTRIAL AND ENGINEERING CHEMISTRY
ecember 1950 catalysts at lower temperatures is frequently a n answer t o the problem of controlling the process of polymerization and the proporties of the
Table
1.
Equilibrium Constants for Dimerization of Normal 1-Olefins to Normal 1-Olefins Reaction
2C2He (gas) = 1-CIHS (gas) 2CaH6 ( as) = l-nCsHlr ( m) 2 I-ncefis ( 89) = 1-nGfim ( 88) 2 1-nCsHlo &as) = l-nClsHm Tgw) 2 1-nCnHtn (gas) l-nCanHm (gas)
In many cases,
-
the conditions of noncatalytic polyn>5 merization can properly be controlled t o produce a desired polymer in high yields and with practical elimination of side reactions. PRIOR ART
Noncatalytic or thermal polymerization of ethylene has been studied much more thoroughly than that of other olefins. Hague and Wheeler (IO), Pease (go), and Storch ($6)investigated polymerization of ethylene under atmospheric and, lower pressures a t temperatures from 350" to 500" C. At low conversion, polymerization is by far the predominant reaction. High pressure thermal polymerization of ethylene was studied by Sullivan et al. (26). Butenes and higher molecular weight olefins are formed on thermal polymerization of ethylene, along with propene and various other hydrocarbons which are formed undoubtedly rn a result of secondary reactions. Propene was polymerized noncatalytically by Sullivan et al. (BB). The rate of thermal polymerization of ethylene is greater than that of propene and other olefins under the same conditions. Quite reversed results are obtained on catalytic polymerization of these olefins, indicating different mechanisms of the thermal and catalytic polymerizations. Lebedev et al. ( 1 4 ) polymerized isobutene noncatalytically a t 200" C. under pressure. The reaction is slow and produces olefinic polymers. McKinley et al. (16) found, however, that a t higher temperatures, 370" to 400" C., a cyclic dimer, 1,1,3-trimethy1cyclopentaneJwas also formed. Noncatalytic polymerization under pressure using n-2-hexene, %methyl-a-pentene, cyclohexene, and a mixture of normal 1-octene and 2-octene was studied by Nemzov et al. (16-18). At temperatures of 345' to 420' C. and a t moderate conversions the polymerization was found to be a principal reaction. Hugel et al. (11,12) studied the effect of temperature, pressure, and reaction time on polymerization of n-hexadecene, 1-octene, and 2-octene. The reaction products were essentially polymers provided that the temperature of polymerization was below 400"C. Tilicheyev and Feigin (27) investigated pressure polymerization and cracking of caprylene (a mixture of 1-octene and 2-octene) and hexadecene a t 425" C. Polymerization reactions occurred readily in the initial stages of the process, and cracking reactions were predominant in the latter stages. The thermal polymerizatiori of olefins is a typical homogeneous reaction of the second order (18). The reaction is not affected by the presence of oxygen or air or by extended metal surface. The activation .energy of thermal polymerization is about 38,000 to 42,000 calories and is approximately the same for all olefins, at least from ethylene to n-octkne (18). The free radical mechanism of this reaction, however, has not been excluded. Burnham and Pease (1), for example, found that polymerization of ethylene is inhibited by nitrogen oxide. Table
25 1 77 X IO" 4 . 4 6 x 10s 9.91 X 106 1 95 X 101 1 78 X 1W
4
9 1 4
3
127 33 X 46 x 55 X 14 X 90 X
Temperature. 32i le 1 51 X 1 0 2 101 0 . 2 ~ IO* 0 359 102 0 125 10' 0 123
C. 127 8 30 2 '32 x io-* 3 37 X 10-2 1 '29 X 10-Y 1 28 X 10-2
527 0.955
5 . 4 0 ~io-' 5 89 X 10-6
2 37 X 10-2 2 45 X lo-'
Noncatalytic polymerization of olefins has not been operated as a means of producing lubricating oils. It has been mentioned that, in contrast, lubricating oils were produced commercially in Germany by catalytic Polymerization of olefins effected by aluminum Chloride. The same Process was used commercially in this country by the Standard Oil Company of Indiana in the early thirties. THERMODYNAMICS OF POLYMERIZATION
Thermodynamical,ly, polymerization or dimerization of olefins
2CnHan
CnnH,n
is favored by low temperatures. The equilibrium constants of the above reaction for various olefins and a t different temperatures under atmospheric pressure are summarized in Table I (Kilpatrick et al., IS). The equilibrium constants are given for dimerization of normal 1-olefins to normal 1-olefins. Since formation of normal l-dimers is improbable from the structural point of view, the question arises how the equilibrium constants of Table I would be affected assuming the branched structure of dimers. Unfortunately, there are few reliable data on the free energies of formation of branched olefins above hexene. Table I1 includes the equilibrium constants of dimerization of propene and n-1-butene t o certain branched hexenes and octenes. The most probable structures of propylene dimer are assumed t o be 2-methyl-lpentene and, as a second choice, 2,3-dimethyl-%butene. Isooctene or 2,4,4trimethyl-l- or 2-pentene is assumed as a n-butene dimer because of the absence of free energy data for other isomers. The authors recognize that this particular structure of the n-butene dimer is quite improbable. The data for free energies of formation of the two branched hexenes and iso-octene given by Kilpatrick et al. ( 1 3 ) and by Parks and Todd (19), respectively, were used in these calculations. The data of:Table I1 show clearly that dimerization of normal 1-olefins to a branched structure thermodynamically follows the same pattern as that to a normal olefin structure with a terminal double bond. The equilibrium constants for dimers of branched structure and dimers of the normal 1-structure are fairly close above 300" C. There is a greater difference a t lower temperatures. .It must be pointed out, however, that the isomeric hexenes selected as the most probable structures for dimers have the lowest level of free energy. The equilibrium constants of dimerization calculated for other isomers will be much closer t o those calculated for normal 1-structures. On the basis of the data of Tables I and 11, temperatures of the order of 300" t o 350" C. are considered as the upper limit for polymerization of olefins except ethylene under atmospheric pressure. Above these temperatures the reverse reaction of de-
ll. Equilibrium Constapts for Dimerization of Normal l -Olefins to Branched Olefins --Temperature, 25 4 . 4 6 X 1W
ci 2CsHs (gas) = C==C-C-C-C
1.7
(gad
c c
2CaHs (gas) = C--C-C-C (gas) 2 1-nC4H8 (gas) = C=C-C-C-C-C-C--C
c
2447
c
2 I-nCIHs (gas) = C-C-C-C--C C
(gas)
(ass)
x
127 9 . 4 6 X 101
x 10'
100
6.3
6 . 6 X 109 9 . 9 l . X 1v
1.1
1.8
x
10s
x ... ...
10'
327 0.284
C. .427 2 . 9 2 x 10-2
5 . 4 0 X 10-8
527
4.5.
3.4
x
10-1
4 . 6 X 10-2
8.5
2.5
x
10-1
2.5
...
...
... ...
x 10-2 ... . , ..
INDUSTRIAL AND ENGINEERING CHEMISTRY
2448
0
2
6 8 IO 12 14 REACTION TIME. HOURS
4
16
18
20
Figure I. Effect of Temperature and Time on Yield of Synthetic Lubricant Chrree, I-drcme
polymerization is predominant. Ethylene is thermodynamically more polymerizable than other olefins; the upper temperaturcb limit for its polymerization is of the order of 450' C. Higher pressures will increase the values of dimerization constants (Tables I and 11) at a given temperature and thus will shift
Table 400 204 20
F.
Temperature, O
e.
500 260 20
111.
505 263 40
Time, hours Max. pressure, lb./ sq. inch gage < 100 ha similar charge in the bombs. DATA AND DISCUSSION
Temperature-Time Relationship. The data given in Table5 I11 and IV and the curves plotted in Figure 1 illustrate the effect of temperature and time on the yield of synthetic lubricating oil The data were obtained using 1-decene as charge material. I t is t h e yield of oil that is dependent on the selection of optimum temperature-time relationships. Oil quality is not affected appreciably in the temperature range 600" to 650" F. At moderate temperatures, 600" to 650' F , the yield increases with time and appears to reach a maximum in the region of 65 wt. % of thp charge. The optimum conditions are about 650" F. and 10 hours reaction time. At temperatures below 600" F excessively long reaction times are required to effect satisfactory conversions Higher temperatures give appreciable conversions in less than 1 hour with a definite decrease in yield when time is extended to several hours. Under extreme conditions the product quality falls rapidly; as low as zero viscosity index was obtained in some cases. The data reported in Tables I11 and IV and the curves shown in Figure 2 illustrate the effect of temperature and time on product quality as measured by viscosity index. At moderate temperatures, 600' and 650" F., the viscosity index is practically constant, showing only a slight drop with increasing time. However, as the temperature is raised the viscosity index is markedly lowered. Pressure. Relatively low pressures were used throughout this work. In most cases the operating pressure was less than 1000 pounds per square inch a t the working temperature but was sufficient to maintain most of the charge in the liquid phase. Superimposed pressures were used in a few experiments with little effect on the oil yield. An experiment on the polymeriaation of n-1-decene in the large stirrer-type autoclave was performed at 600" F. under pressure of 4000 pounds per square inch gage. The yield and the properties of the synthetic oil were identical with those produced a t the same temperature in several runs made at moderate pressures ranging from 250 to 500 pounds per square inch gage. Effect of Olefinic Charge. The chain length of the olefinic charge affects viscosity index, pour test, and yield. Increase of the chain length from 1-hexene to 1-hexadecene results in higher
+
+
NUMBER OF CARBON ATOMS
Figure 3. Relation OF Pour Test and Viscosity Index to Carbon Number QF Olefin Charge
December 1950
INDUSTRIAL AND ENGINEERING CHEMISTRY
thus showing poorer quality of the olefin. A thermal gasoline produced in current operations on a nonselected charge stock was entirely unsuitable for conversion to lubricating oils-the olefins present were relatively short in chain length, highly branched, :tnd not of the 1-olefin type. The classic Fischer-Tropsch reaction produces straight-chain hydrocarbons with high 1-olefin rontent. The process of Hydrocarbon Research, Incorporated, operating with an iron catalyst a t relatively high temperature, produces somewhat similar material. Hydrocol product as used in experiments was the material taken from the reactor without the usual reforming treatment over alumina. The reforming is necessary if the Hydrocol is to be used as gasoline, but it makes the material unsuitable for use in thermal polymerization t o lubricating oils. Selected Hydrocol fractions weie converted to synthetic lubricants with viscosity indices in the region of 100 or above. Both yield and quality of product were affected by the following variables: boiling range of charge (chain length of olefin), olefin type (1-olefin content), and nonolefinic constituents (dcohols, aldehydes, ketones, acids, esters, aromatic hydrocarbons, and paraffins). The discussion of these variables is too involved to be undertaken in this paper.
245 1
REACTION TIME, HOURS
Figure 4.
Effect of Temperature and l i m e on Specific Gravity of Polymer Oil Chrrgr, 1-decene
unconverted olefins become relatively unavailable for polymerization because of a shift of the double bond to a more central position. Recycling oan be practiced to some extent if care is exercised in limiting the initial polymerization conditions to moderate temperatures. Oil quality is judged chiefly by viscosity index, pour test, color, Table VIII. Thermal Polymerization of Cracked Products and carbon residue. Viscosity index of oils derived from 1-decene Cracked can be held at or above 140 if neither excessive temperature nor paraffin Thermal Cracked foots oil, excessive reaction time is employed. A rapid decline in viscosity wax ( 1 1o Charge material 4400 F.) gasoline Cot 100-400' F . index with long exposure to high temperatures has'already been 610 590 600 625 625 Temperature, F. shown in Figure 2. At more moderate temperatures the viscosity 321 310 316 329 329 C. 9.5 0 10 10 10 Time hours index curves are relatively flat. A sharp rise in specific gravity Max.'pressure, lb./sq. inch of the produrt oil accomlmnies a drop in viscosit,y index. This Rage 400 325 3500 700 2000 is indicated in Figure 4. .41so pyrolytic decomposition of hydrnOil yield, wt. % charge 34.0 1.25 1.6 43 46 carbon oils occurs a t about, the conditions under which the products of poor quality are obtained. Color of the synthetic oil product is of considerable importance Very dark to black oils indicate excessive reaction temperature or the presence of undesirable materials in the charge stock. In extreme cases, the dark color will he associated with appreciable The noncatalytic method of polymerization is applicablenot only A.S.T.M. insolubles, high carbon residues, high gravities, and to pure olefins (24) but to charge stocks containing added styrene, low viscosity indices. Dark color axid the attendant poor qualidiolefins, sulfur, and phosphorus pentasulfide. Reaction of thc ties can be prevented by the usc of low reaction temperatures ole% with styrene or other polymer-forming materials (21, 28) and a substantially pure 1-olefin charge. results in products with properties similar to those of the simplr Synthetic lubricating oils made by thermal polymerization of olefin polymer. Inclusion of sulfur or sulfur-containing com1-olefins (Tables I11 and IV) have moderate viscosities, 5 to 6 cs. pounds in the charge (5-9) gives products with improved resista t 210" F., corresponding roughly to S.A.E.-IO specifications. ance to oxidation. The use of certain nonreactive substances in Vacuum distillation of a synthetic lubricating oil gives the rethe reaction zone allows high yield of oil in a short-time reaction sults summarized in Table IX. The heaviest fractions have a t high temperatures with inhibition of incipient cracking (2-4). viscosities slightly higher than that of S.A.E.-40 specifications. Product Properties. The distillates collected during the topIn terms of molecular weight, noncatalytic polymeriaation of ping operations are highly olefinic and have been considered for normal 1-olefins proceeds t o a comparatively low ceiling correrecycle use. However, under extreme conditions of polymerizad - the sponding t o (ClaH&- I on the average Lnd to ( C l ~ H ~ ~ )for tion temperature or time there is a loss of unsaturation with a heaviest fraction. corresponding increase in specific. gravity. Also, some of the Tt is often implied that lubricating oils containing unsaturation will be less stable toward oxidation than a saturated oil. Although these oils do contain Table IX. Distillation of Hydrocol Polymer Oil measurable amounts of unsatuDistillatibn Temp., F . vel, % Of Color ration, they gave very satisfacFraction Synthetic Actual, Calcd., Viscosity, Cs. Visoosity Gravity Lovi-' No. Lubricant a t 10 mni. at 760 mm. At 100 At 210 Index A.P.I. Soecifi; bond tory results in a laboratory osidation test. Theseveral samples 410-460 686-740 14.83 7.7 3.30 100.3 33.1 10 460-490 740-776 24.30 7.7 4.50 108.5 27 32.3 tested indicated better stability 490-506 77 6-7 93 26.756 7.7 4.75 108.8 18 32.1 506-522 29.65 793-809 108.5 7.7 5.07 32.0 10 toward oxidation in II small 522-524 34.10 809-811 7.7 110.0 5.55 31.6 15 wale bench test than a com524-540 37.32 811-830 5.90 7.7 110.8 31.4 18 540-546 36.91 830-836 5.87 7.7 111.4 31.6 17 mercial solvent-treated paraffin 546-594 55.20 836-890 7.56 108.1 7.7 30.8 17 594-620 890-914 71.65 8.92 7.7 106.5 30.7 24 hase oil of the same viscosity. 620-636 83.63 914-930 O
7.7 23 Carbon residue (Ramsbottom), 0.02. b Fractions joined for pour test, -65O F. Carbon residue (Ramsbottom), 0.1.
...
...
244
9.86 20.02
105.2 107,
30.3 28.9
34
>750
SUMMARY
Purely thermal, noncatalvt,ic polymerization of' l-olefins
2452
INDUSTRIAL A N D ENGINEERING CHEMISTRY
ptwluc~c~sexcellrtit lulJi5a:iting oils. Reactmion conditions are (i00" t o G O " F., rc&knct. t.itnc: :tt)out, 10 hours, and pressules of 2.50 t o 5 0 0 pounds per s q u : t i ~inch g a g . Under such condit>ions I-tloccwc has given as muc*h:ts (5% by weight of an f3.A.E.-10 oil iii :I siti~lrpass, with produc:t irispections as follows: viscosity iiicies, 140; pour test, - I O " h'.: color (Lovibond), 0 . 5 ; carbon l nf 1-olefins is from n-hexene t o rcwiclue, nil. T h e p r i ~ f ~ w wrnnge rr-dodecene., Sudi ole4ns froin utiy source, as cracked p:traffirl NXS or Fisch'er-Tropscfih pi,oduc*t*, ran used for synthesis. If suit.:ible olefinic chnrgc' stlocks lx~conireconomically :~v:~il:hl~+ t h : oils can be :tcc:omconversion t80finished high g i ~ h lutwicating plished in conventional refinery equipnimt consistink of niotlified th(.t~in:tIcrackina and dist,ill:tt,ionunit,s. ACKNOWLEDGMENT
hpprt&ition is rspre-;secl t80G. (:. Slirewshry, 11. S. . \ I o i i t c . niore, Lois B. Buseni:in, :md other inembers of tho lat)orutory st,aff for their valuilhle sssistaricv?in the experimentd work. The Hydrocol product wiis supplied through the courtwy o f thc II!drocarhon Rrsiwch, Inc~orporitted. LITERATURE CITED
Huriiham, H. O., and Pease, I < .
S . . J . .lm. ot the c.h:ti acterktics that are most desirath in a lubricant. Such ploperties as high flash points, low free7ing points, extremely flat viscosity-temperature slspes, arid R high degree of rcktancse to oxidation and thermal decwniposition ale a few of theii m:my excellent physical attributes. Greaaes formulated from these materials exhibit ni:rri> of the pioprities of the basr fluidh m(*h a relative indifference to extreme temperature r h a n g ~ s ,low wlidification points, and :t reni:trknhle heat stability. In ball aiid roller bearings whwr rolling frwtion is encountered, l m t h the J o s a n e fluids and greases have performed remarkably well ( 4 , 8 ) . I n marked contrast to the suwessful use of these materids in ball bearings under actual wrvice cwiditions, the prcr,ent organopolyYiloxanes fail to provide :w :itlcbclu:itr lubricating filii1 between
T
sliding ferrous surfaces under boundary ronditions. Because of this failure, there is a reluc,tance to use the organopolysiloxanes wherever sliding friction is envountered regardless of the metal combinations involved. This reluctance can be traced directly to the dearth of information roncerning the performancue of these materials as luhic-ants for hearing combinations other than steel rubbing against steel. Although data have been published on the methylpolysiloxaries as lubricants between various metal combinations (6) no effort has been made to confirm these data other than the work done by %isman and his associates (1,2 ) . The authors have, therefore, undertaken a study of' ~ e v e r dpolysilosanrs as lul)ric*nnts for various pairs of bearing mrzterials.
