January 1949
INDUSTRIAL AND ENGINEERING CHEMISTRY
there is little or no effect on the viscosity, the behavior being the same as t h a t of a nonswelling agent on the low conversion sol polymer. Thus, the viscosity is influenced by solvents of varying gwelling action on a given polymer and by variations in the swelling action on a given polymer and by variations in the swelling index of different polymers with respect t o a Liven solvent. High swelling, or solution, by a solvent €or sol polymer distributes the solvent among the small polymer particles. The expansion of the particles reduces the freedom of motion and the viscosity increases. The diffusion of styrene has been found t o be twice as high into a latex containing 33% monomer (67% yield) as into a latex containing only 201, monomer (98% yield). Presumably, the polymer network is less stiff a t the lower yield
‘I) FINAL %ATEX CONCENTRATION AND VISCOSITY
The limiting viscosity and yield point undergo a rapid increase at about 60% polymer concentration as a 50% polychloroprene latex is concentrated by distillation of the water in the presence of 9. silicone antifoam agent. The relation of the viscosity t o the yolids content of the latex is given in Figure 4. I n this case the increase in viscosity results from bringing the particles closer together. If the polymer particles were spheres of uniform diameter, a t polymer coilcentrations above 57.570 by weight or 52.3% by volume, a loose hexagonal close packing arrangement would exist until, at 74% by volume, the particles would touch and have an interparticle distance of zero. Actually, the latex *scomprised of particles of sizes varying from a few hundredths of R micron t o 1 micron, with a n average number size of about 0.2 micron. Thus, the voids relation and interparticle distance are different from those calculated for spheres all of the same size. The same relation between viscosity and yield point and polymer concentration a~ exists for latires concentrated by distill*
197
* DISTILLED 0
POLYMERIZED ONLY
4 . I
t
50
52
Figure 4.
I
I
I
54 56 58 LATEX CONCENTRATION, Y
I
I
80
62
Relation of Viscosity to Solids Content
tion applies t o latices made from emulsions of varying concentration. The results from such a series of latices made in a 4% rosin soap solution are plotted in open circles in Figure 4. LITERATURE CITED
(1) Harkins, W. D., J . Am. Chem. SOC.,69, 1428-44 (1947). (2) Jordan, H.F., Brass, P. D., and Roe, C. P., IND. ENG.CHEM., ANAL.ED., 9,186 (1937). (3) Klevens, H.B.,J . Colloid Sci., 2,365 (1947). (4) Leviton, A.,and Leighton, A., J . Phys. Chem., 40, 71 (1936). (6) Livingston, H. K., Ibid., 51,443 (1947). (6) Mooney, M.,and Ewart, R. H., Physics, 5, 350-4 (1934). (7) Sibree, J. O.,Trans. Faraday Soc., 27, 161-76 (1931). (8) Taylor, G.I., Proc. Roy. Soc., 138,41 (1932). RECEIVED November 7. 1947. Presented before the High Polymer Forum a t the 112th lteeting of the $XERICAN CRBXICAL SOCIETY,New York, N. Y. Contribution 59. Jackson Laboratory.
ilica Gel Fractionation of 300Viscosity Lubricating Oil R. R. HIBBARD’ The Standard Oil Company (Ohio), Cleveland, Ohio
In
studying the factors involved in the oxidation of lubricating oil, a careful chromatographic fractionation was run to resolve a 300-viscosity red oil into its hydrocarbon types. A large number of fractions were taken and physical and chemical properties determined on many of them. Blends containing 1 0 q ~of the various aromatic fractions were run in the Sohio oxidation test and the tendencies of these fractions to form sludge, pentane insolubles, and acids have been related to hydrorarbon type.
c
HROMATOGRAPHIC fractionation over silica gel has been used widely in recent years t o separate mixtures of hydrocarbons into chemical types. This technique has proved especially valuable for isolating and determining aromatic hydrocarbons boiling in the gasoline range, using methods similar t o those suggested by Mair ( 7 ) . Higher boiling fractions have also been analyzed with the aid of silica gel by Gooding and Hopkins ( d ) , Lipkin et al. (6) and Willingham (11) I n the work reported here a 300-viscosity lubricating oil stock was fractionated through silica gel on a scale large enough t o 1 Present address, National Advisory Committee for Aeronautios. Cleveland, Ohio
allow oxidative tests t o be made on relatively narrow chromatographic cuts. In a n attemyt t o obtain cleaner separation, a flowing chromatogram procedure was employed and as much of the oil fractions was washed from the adsorbent with a nonpolar solvent (n-pentane) as was easily possible-a technique similar t o t h a t employed by Mair and Forziati (8). Only the most strongly adsorbed portions were eluted from the silica gel with polar solvents. FRACTIONATION TECHNIQUE
To lessen the amount of solvent required, t h e assembly shown in Figure 1 was constructed. The chromatographic column proper was made from a 20-foot length of 2-inch standard pipe with the lower end capped, drilled, and tapped t o take 0.25-inch fittings. A 60-mesh stainless steel screen on a suitable support rested on the lower cap. The effluent drained via the 0.25-inch tubing t o a n iron pot of about I-liter capacity; the pot rested on a 1000-watt Chromolox ring heater. From the top of the pot a 22-foot length of 1-inch standard pipe carried the solvent vapors t o an overhead condenser and the condensate was recycled to the top of the chromatographic column. The vapor riser was
198
m
INDUSTRIAL AND ENGINEERING CHEMISTRY
\\lapped about four turn. per foot with 0.25-inch _-.CONDENSER copper tubing to permit bteam heating; the iiser aiid pot were suitably insulatcd. All joints were screwed littings set in litharge-glycerol cement. A T hctmeen the column and the pot allowed a slow nitrogen bleed to be introduced, primarily t o keep oxygen out of the bystem but also t o aid i n stripping the solvent from 1 the oil. 20' To provide the 2 to 3 pounds per square inch pressure required t o pel colate the diluted oil through the silica gel at a reaqonable rate, the column was packed with gel to a height ot but 12 feet, leaving ax1 8-foot liquid head to supply the RESISTANCE HEATER driving force. A 4-inch ~ R A C TOI N S layer of 20- 00 40-mesh carFigure 1. Chromatographic borundum was poured into Column the column first and Davison's KO.6595 silica gel (28 to 350 mesh), dried a t 250" F., was packed in; a tamper n7as used from above supplemented by the application of a n air hammer against the outside of the pipe. Eleven pounds of gel were so packed t o make a 12-foot depth of bed. The lubricating oil used in this study was a plant-treated sample of 300 red oil of the specifications shown in Table 1. As developing solvents and eluting agents, Phillip's technical grade n-pentane and I-pentene and C . P . benzene, chloroform, and ethanol were used.
TABLE I. SPECIFICATION OF LUBRICATING OIL STOCK 292 S.U.S.o 100' F. Viscosity Viscosity index Gravity Flash Crude source Acid treatment. Clay contact * Saybolt Universal seconds.
4 9 . 2 S.U.S. 210' F.
69 25.00 4.P.I. 400' F. Mid-continent 11 Ib. HnSOa pe? bbl. 4 . 5 lb. Super Filtrol per bbl,
TABLE 11. TIME-SOLVEKT-YIELD SCHEDULE OF CHROMATOGRAPHIC FRACTIONATION
56-97 97-115
times during the run the rat'e of solvent circulation was dctrr. mined to be between 1.5 and 2.5 liters per hour. Fractions consisting of from 2 to 5 ' 3 of the charged oil R'DW withdra,wii periodically from the pot into tared bottles and when the oil content dropped beloiT 170, the circulating n-pentane wap replaced in pa'rt, first by 1-pentene and then by benzcnc. Finally the hydrocarbon solvent was largely taken from the system b y diverting the condensate from the silica gel column and a.pplying 2 t o 3 pounds per square inch gage nitrogen pressure to t,he top. The residue was elut,ed with a 50:50 chloroform-ethanol mixture. The tinic-solvent-yield schedule of the fractionation is given in Table 11. Sinety-one per cent of the oil was washed througl: t,he gel using n-pentane alone although the rate of aromatic lotv R P shou-11in Figure 2 . Only the most st,r.ongl?;
OIL THROUGH SILICA GEL, WT. %
Figure 2.
Elution of Oil by n-Pentane
adsorbed last 2% required the use of the more polar nonhyirocarbon solvents for their elution. The fractions in their tared receivers were freed of any remaining solvent by evacuation with well trapped Cenco pump while heated with a n infrared lamp. These were then weighed and gave a 98% material halance: the loss was prorated across all h c t i o n s . PHYSICAL PHOPEKTIES ANI) ANALYTICAL P>M2'A O R CHROMATOGRAPHIC FRACTIONS
Refractive index (n22),density (di"), and specific dispersiori (:VP Ab) data were first obtained on all fractions and arc.
-
The pot was brought to a temperature of 225' F., the steam flow started through the coil around the vapor riser, and 1500 grams of oil diluted with 1.5 liters of n-pentane poured into the column. Pentane was added periodically t o keep the level near the top until the oil plus the solvent started t o come through t o the stripper (about 7 hours). Thereafter, the level was maintained largely by solvent recycle with but occasional additions t o make up for oil removal and evaporation losses. At several
tiours 0-7 7-49 49-56
Vol. 41, No. 1
Circulating Solvent n-Pentane n-Pentane 75% n-pentane 25% 1-pentene 37% n-pentane f 13% 1-pentene $5 0 % benzene 50% ethanol 50% chloroform
+
+
Cumulative W t . 70Yield of 011 0 91 94
98
100
plotted in Figure 3. From the specific dispersion data and using the pure hydrocarbon values of Thorpe and Larsen (IO), it is estimated that the break between s a h r a t e s and aromatics occurs at about 62 weight % and that the second break a t 81 weight marks the separation point between single ring and multiring aromatics. Viscosities were run a t 100 e and 210 F.on selected fractions and viscosity indexes calculated. These values are shown in Table I11 along with the pour points of the saturate fractions. Thcsc values show that exceptionally high and extremely low viscosity index fractions were separated by the silica with the high values probably due, in part, t,o a wax enrichmenr in the earlier fract'ions. Using molecular weight .values as determined by the Mills, I h s c h l e r , and Kurtz correlations of gravity, viscosity, arid molecular weight (9) and both densities and refractive indexes. the per cent naphthenic carbon atoms %-ere calculated for se1ect)ea aaturate fractions by the method of Liplrin and Martin (6)and are shown in Table IV. The values given are those obtained by averaging the results from Lipkin and Martins A-density and ( A = density coefficient') equations; the t,wo metliod+
Ianuary 1949
199
INDUSTRIAL AND ENGINEERING CHEMISTRY
TABLE111. VISCOSITIES AT 100' AND 210" F., VISCOSITY INDEXES,AND POURPOINTS OF SELECTED CHROMATOGRAPHIC FRACTIONS Pour Point,
Weight
%
Viscosities, 100' F. 24.0 15.7 32.9 39.6 48.2 55.6 98.9 232 900a 16OOa 2325"
Fraction
c
Centistokes, 210" F. 4.84 3.82 5.48 6.07 6.54 6.71 8.56 11.20 17.05 20.7
Viscosity Index 139 157 113 108 93 74 46 37 - 130" -310"
22.8
-410"
F.
+55 +40 f40 4-40 +2; h b b
-
t h
Nd
Approximate. b Not determined.
a
'rABLE
IV.
WEIGHT P E R C E N T NAPHTHEXE SATURATE FRACTIONS
%
%
CHROMATOGRAPHIC FRACTION, WT.
SELECTED
Weight
Weight
Naphthene Ring 28 32 30 35 38 37
Fraction 0-3 12-16 28--30 42-45 50-53 56-61
0
RISG IS
%
TABLEV. PER CENT SULFURIK COMPOSITES OF
Figure 3. Physical Properties of Chromatographic Fractions
CHROMATOGRAPHIC FR.4CTIOlvS Weight
gave similar results. The table shows t h a t silica.ge1 had little tendency t o separate the saturates into paraffinic and naphthenic concentrates. The first 53% of the oil then was blended together and some of the smaller of the aromatic fractions were combined and the per cent sulfur determined on these composites. These d a t a are shown in Table V. The sharp increase in total sulfur at the 86'7iO mark appears significant in the light of the later oxidation results. For a few of the fractions in the aromat,ic range the carbon t o hydrogen ratio was obtained and the per cent paraffinic, naphdhenic, and aromatic carbon atoms calculated by the method of Deansley and Carleton (2); the results are given in Table VI. T h e 98 t o 100% fraction would not yield rational results t o the Deansley-Carleton analysis, possibly because of the uncertainty of t h e specific dispersion determination in this very dark sample. Colors were not determined on the fractions but were water-white on t h e first SOYO and gradually darkened t o the end. From t h e above analytical data the silica gel fractionation appears t o have separated the lube stock into the following distinct classes: 1.
0-6270 fraction
2. 3 4: 5.
6241%
81-867 86-98% 9%100%
Paraffins and naphthenes containing about 30% naphthenic carbon atoms Single ring aromatics Polynuclear aromatics relatively low i n sulfur Polynuclear aromatic relatively rich i n sulfur Material of high carbon to hydrogen, ratio, moderately rich in sulfur b u t otherwise not well defined
OXIDATION CHARACTERISTICS OF THE CHROMATOGRAPHIC FRACTIONS
The susceptibility of the various aromatic fractions t o oxidation and the character of the oxidation products were determined by running blends of these cuts in a scaled down version of the oil oxidation test of Burk, Hughes, Scovill, and Bartleson (1) The equipment was approximately one fifth the size dedcribed by Burk et al. and the tests were run under the conditions shown in Table VII. The test was run on a series of blends, each consisting of 90 weight % of the saturate 0 t o 53% chromatographic fraction plus 10% of one of the cuts from the 61 to 100% aromatic I
7%
Sulfur,
9&100
0.32 0.53 0.77 0.94 0.84 1.83 1.90 2.08 1.60
Fraction 0-53 63-72 72-76 76-81 81-86 86-91 91-96 96-98
%
CARBONT O HYDROGEN RATIO AND PER CENT CARBON ATOMSIN AROMATIC FRACTIONS % C: H Carbon Atoms, Yo
TABLEVI.
Weight Fraction 76-81 81-86
86-91 91-96 98-100
Ratio 7.36 7.88 7.94 9.08 9.35
Paraffin 25 13 5 7
Naphthene Aromatia 48 27 ' 60 27 64 31 44 49 Sum well over 1 0 0 ~ o
TABLE VII. OXIDATION TESTCOSDITIONS Temperature, F. Oil sample, ml. Air rate, l/hour Time, hours Catalysts Steel, sq. om. Copper, sq. em. Cop er lead, sq. om. L e a l biomide, % Fe, 2-ethyl hexoate as % FeZOs
280 35 70 36 20 0.2
1.0 0.1
0.012
fractions. Also run for comparison purposes was t h e whole 300 red oil and a 300-viscosity (100" F.) solvent extracted oil from a similar crude source. On completion of the test, the degree and type of .oxidation were rated in the following terms: 1. Sludge rating, the appearance of the glass tube holding the oil on a n arbitrary A t o F basis with A+ being a tube perfectly free from sludge. This factor is a measure of t h e amount of oil insoluble products formed during oxidation. 2. Pentane insolubles in the oxidized oil, a measure of the oxidation products which are oil-soluble b u t thrown out of solution on dilution. 3. Viscosity increase, the increase in viscosity of the oxidized
I N D U S T R I A L A N D E N G I N E E R I N G CHEMISTRY
FRACTION UStD IN BLEND VTT" %
FRACTION USED IN BLEND, WT. %
Figure 5 .
Figure 4. Oxidation Test of Blends of C h r o m a t o g r a p h i c Fractions IO',
oi indicated Ciaction
phi-
SUhlMARY
I n summarizing the interpretation of the oxidation test data, it appeared that qaturatrs and single ring aromatic3 yieldpi1
Oxidation Test of Blendr o f
C h r o m a t o g r a p h i c Frdctirrns Ulf70 of indirated irac.tlou v l i i s 90% srttirrnte rrsrtiorr
irOrh w t u r a t p frartior
oil over The c*hargedoil ab iiiea\ured iii Saybolt Uriivrrial at 100 O F and indicating the same type of oxidation as 2. 2. C'orrosion, thti w i g h t loss of the copper-lend te3t piece during the run. 5 . Keutralization number, it meabure of the free acid Iourid in the oxidized oil. The resulting oxidation test data are shown in Figure 4 (sludge ratings and pentane insolubles) and Figuie 5 (viscosity incrpase, corrosion, and neutralization number) The five classes of oil as eepaiated b j chrornatograph~chactionation ieacted as follows in this oxidation test: The saturai es yielded niodersi rly hlgh molecular. 1% rlghl but, oil-soluble polymers on oxidation a? indicated by high valuec: in pentane insolubles and viscosity increase. They yielded acids also and were corrosive but, they had n o tendency l o foFm oil-insoluble sludge. Blends containing 10% of single ring aroriiaiics (the 62 to 81 "/( chromatographic fractions) in 90% of the saturates brhaved substantiallv the Same as the pure iaturates on oxidation. Blends containing the lo^ sulfur, polvnuclear aro tiiatim (the 81 t o 86O4 ftactions) showed a marked increase in qludgr and pentane insolubles n j t h but little change in the other types of oxidation. A catalysis of the reactions yieldjng peiitttrie insolubles was suggested by the fact that ihe increase i n thew insolubles was greater than the a ~ ~ i o i i iofi t aromatics added i o the saturates. The oxidized bleiids containing the high sulfur, polynuclear rnateiial (the 86 t o 98y0 fractions) were low in acidity, corrosion, viscosity increase, and pentane insolubles; thls showed the strong inhibitory action of these fractions toward this type of oxidation. However, they were high in sludge formation. Blends containing the 2 7 , most strongly held by silica gel showed excellent performance in the oxidation-test in all reyxcts. This material served as a useful natural inhibitor without forming 4udgc
Vol. 41, N c 1
riiotlrrttt e ruc!lecular m i g h t rmim a,nd acids or1 oxi.dation and t,k polynuclear aroimtics went t o high molecular weight siiidge w little effect on the above reactions, unlesri they were relatively high in sulfur content. In this latter case, thcy acted as inhibitor^ toward thra acid arid resin-forming reactions but, thcxnsehet oxidized t o sludge. Denison (S) has shown that t h c nat,irraJ inhibitors in lubricating oil are sulfur compounds w1 destroying peroxides and t h i s ivork confirms his ob There was present, also in this lube st material which was strongly adsorbed 011 + useful i inhihitor in all rclsprws. 4CII\OU I,E.I)GWbY I' r ,
I tie author n-isheb t o ~icliriowl~dgc the aisistance grvt'rl Ijini tn h. Koqman arid K. Ward and to thank The Standai d 0 1 1 ('ompany of Ohio for permission t o puhlibh this work. H e i- 111debted t o E. C. Hughes COI hi3 help arid advice thioughout the aork and t o Ii. P. J,ankelina for assiitanw $11 the preparMTio7 (1 t h p nianufcripi. LITEKATUHE CP'I'EL,
it. E., Hughes, E. C.. Scovill, W. I. R. E., A r d . Chem., 20, 180 (1948). (6) Lipkin, > R., 'I Martin, . C . C., a,nd Kurtz. Y. X.,.CKD. EN(:.(:Hv*u. ANAL.ED.,18, 376 (1946). (7) U a i r , E. J., J . Research .Vatl. HUT.Stcindruds, 34, 435 (3948,. (8) >lair, B. J., and Forziati, A. F..I b i d . , 32, 165 (1944). (9) Mill:?,I. W., Hirschlw, .4.E., and Kurtz, 6.S., Ixu. Ex
