Viscosity of Solutions in Branched-Chain
Paraffins E. H. McARDLE AND A. E. ROBERTSON Esso Laboratories, Research Division, Standard Oil Development Co., Elizabeth, N. J.
taring after solution. Transfer was made as rapidly as possiblei. e., in about 5 seconds-to a precision-grade Gardner-Holdt viscosity tube, which was quickly closed with an oversize cork. Repeated comparisons with standard tubes were made a t 25’ * 0.2O c.
Solutions of an alkyd-type resin, and of a natural and two substitute rubbers, at high concentrations in two pentanes, four hexanes, five heptanes, and two octanes show much higher absolute and relative viscosities in the branched paraffins than in the corresponding normal paraffins. An arbitrary method of predicting the viscosity of the resin solutions in branched paraffins has been devised. Large differences between solution viscosities suggest a rough method of hydrocarbon analysis.
TABLE 11. ABSOLUTEVISCOSITIES .4T 25’
c.
(Beckosol No. 19 solids, 8.0 grams in 10.0 ml. of hydrocarbon) Gardner-Holdt n-Pentane Isopentane n-Hexane 2-Methvlpentane 2,3-Dimethylbutane Neohexane n-Heptane 22-Dimethylpentane 2,3-Dimethylpentane 2,4-Dimethylpentane Triptane n-Octane Isooctane n-Nonane n-Decane
A
PREVIOUS paper (3) showed that solutions of certain
resins and linear polymers in isoparaffinic hydrocarbon solvents exhibited much higher absolute and relative viscosities than solutions in corresponding normal paraffins. Comparisons were limited to alkyd-type resins and polybutenes, dissolved in whexane, neohexane, n-heptane, and triptane. Measurements have now been extended to include two pentanes, four hexanes, five heptanes, two octanes, and normal nonane and decane. Physical properties of these hydrocarbons are shown in Table I. Solutes include milled samples of natural and substitute rubbers, and a nonphthalic alkyd resin. Most of the measurements were made with the resin solute.
Centipoises
__
8ii
128 120,120 135,136 165,165 250,250 145, 145 240 172 200 310 193 294,295 240 302
To obtain reproducible results, the resin should not change its viscosity rapidly in storage. The resin selected, Beckosol No. 19 solids, is of the drying type, of 27-gallon oil length. It contains no drier, however, and its behavior a t room temperature in bulk is such that no skin has formed in the can after more than a year. That its viscosity changes extremely slowly upon standing is shown by the fact that check solutions in the four hexanes made 10 days apart, and in n-heptane made 3 weeks apart, gave identical viscosities. Indeed, its solutions made a year ago in identical samples of neohexane, n-heptane, and triptane had viscosities within 2 per cent of the values in Table I1 ( 3 ) . Further, its solution viscosities a t the concentration selected for the present series of hydrocarbons are covered by the standard tubes of the Gardner-Holdt “varnish series”.
Viscosity Measurement The technique used previously was employed in making solutions. In the case of the resin, it involved weighing 8 grams of the slowly flowing material into a 0.03-liter (1-ounce) wide-mouth bottle on a Torsion pharmaceutical balance to an accuracy of 2 mg., carefully pipetting 10 ml. of solvent, closing the bottle with a selected oversize cork, slowly forcing the cork well into the neck, tumbling the bottle overnight a t 15 r. p. m. a t 25’ C., and back-
Table I11 lists absolute and relative viscosities of solutions of high polymers in pairs of hexanes and heptanes. Good agreement between relative viscosities of the , milled crepe in the two normal paraffins is notable together with the fact that in each case the PROPERTIES OF HYDROCARBONS USED‘ TABLE I. PHYSICAL branched-paraffin solution has a higher relative Estiviscosity than the n-paraffin solution. Cottrell mated
Hydrocarbon
O
n-Pentane Is0 entane n-Zexane 2-Methylpentane 2 3-Dimethylbutane ieohexane n-Heptane 2 2-Dimethylpentane 2’3-Dimethylpentane 2’4-Dimethylpentane &iptaned n-Octane Isooctanee n-Nonane n-Decane 3Iethyloyclohexane Ethylcyclopentane a b c
d e
nyb
Boiling Point
c.
3 6 . 0 (36.08) 2 8 . 0 (27.95) 6 8 . 6 (68.74) 6 0 . 2 (60.27) 5 8 . 0 (58.00) 4 9 . 7 (49.73) 9 8 . 6 (98.42) 7 9 . 1 (79.21) 89.8(89.8) 8 0 . 5 (80.70) 80.8 (SO. 88) 1 2 5 . 6 (125.63) 9 9 . 4 (99.23) 150.85 (150.74) 174.2 (174.04) 100.9 103.4
1.3580 (1.3577) 1.3541 (1.3839) 1.3751 (1.3750) 1.3715 (1.3716) 1,3749 (1.3750) 1.3689 (1.3689) 1.3876 (1.3876) 1,3822 (1.3824) 1.3920 (1.3920) 1.3820 (1.3820) 1.3894 (1.3895) 1.3980 (1.3976) 1.3918 (1.3916) 1.4070 (1.4056) 1.4146 (1.4120) 1.4253 1.4197
Spread
Purity
c. None None None None None None None None None None None None None 0.05 0.1 0.2 None
%
+++ +
Estimation of Viscosity
99 99 99 99 99+ 99 99 99 99 99 99 99 99 “SO+” “go+” 97 99.5
+ +++
++ ++
Values in parentheses assembled by Francis ( 1 ) . Determined in Pulfrich refractometer. 2.2-Tlimethvlbutane. 2’5 &T&n&hylbutane. 2:2:4-Trimethylpentsne. 3
484
Within the present series of resin solutions, it is possible to estimate viscosity by an arbitrary method, based on structural formulas written in t.he plane of the paper. A somewhat similar method of correlating physical properties of paraffin hydrocarbons has recently been employed with interesting results ( 2 ) . A plot of log solution viscosity against solvent boiling point is linear in the case of the normal paraffins (Figure 1). Average deviation is b 3 per cent. Thus the viscosity of a n n-butane solution of this resin would doubtless be very close t o 63 centipoises at 25’ C.
August 15, 1943
ANALYTICAL EDITION
485
TABLE 111. VISCOSITIESAT 25’ C. Milled Crepe Milled Butyl 3.0 Grams in 18 Co. 3.0 Grams in 15 Cc. Absolute Relative Absolute Relative Poises
n-Hexane Neobexane n-Heptane Triptane 4
20.1 35.2 26.1 44.6
Poises
6,700 10,350 6,700 8,260
19.3 46.3 28.8 63.4
Milled Buna 5 3.0 Grams in 21 Cc. Absolute Relative Poises
6,430 13,600 7,390 11,750
15.3
5.100
a
2b:l
33.9
5,140 6,280
N o solution when tumbled at 15 r. p. m. at 25O C.
Triptane may be considered the result of adding a branch to neohexane, a t the 3-carbon of the butane chain. I n the case of 2,3-dimethylbutane1 addition of a second branch at the 3carbon caused a decrease of 2 centipoises from the corrected viscosity of isopentane. Thus, since two branches a t the Z carbon of butane add 96 centipoises, or 2.5 times 37 centipoises, it is reasonable to expect that the 3-carbon branch of triptane will cause a decrease of 2.5 X 2 centipoises from the corrected viscosity of neohexane, yielding a value of 154 centipoises for the corrected viscosity of triptane.
FIQURE 1
If it is now assumed that viscosities of branched-chain par-
Increasing Chain Length
affin solutions are related similarly to boiling point, for the Assuming that the contour length of pentane is in the neighborhood of four thirds the contour length of butane, the placesame degree of branching, an increment can be computed for the effect of branching. For example, isopentane can be conment of a methyl branch at the 2-carbon of pentane should produce roughly three fourths the effect it does in the case of sidered as the result of adding a methyl branch to butane. butane. On this basis, prediction of the 2-methylpentane The resulting hydrocarbon boils 28.5’ C. above butane, and solution viscosity is only 4.1 per cent high, as shown in Table shows a resin solution viscosity of 128 centipoises. It follows, V; 2,3-dimethylpentane is 4.6 per cent high; and 2,243then, that if this branched compound boiled a t the boiling methylpentane 3.8 per cent low. It is interesting to note that point of its parent chain-viz., 28.5’ lower-it would show a solution viscosity of 100 centipoises. The estimate is made the absolute solution viscosity of 2,2-dimethylpentane is lower than that of neohexane, although the latter boils 30” C. graphically from Figure 1, as follows: Starting point on the lower. curve is a measured viscosity value (128 centipoises). The curve is followed downward for a difference in boiling points I n Table IV it is seen that the corrected viscosity contribution of each branch in 2,4-dimethylpentane is equal to between the parent chain and its branched homolog (28.5’), and the “corrected” viscosity read. Thus, according to the two thirds of its value when taken alone as the only branch arbitrary assumption, the effect of the degree of branching on the butane chain. The combined branching effect in isowhich makes isopentane out of butane is to increase solution octane may thus be expected to amount to two thirds of 90 viscosity by 37 centipoises (Table IV). (the 2,2-carbon branches) plus 37 (the 4-carbon branch), or Placement of a second branch a t the adjoining chain carbon, 88 centipoises. as in 2,3-dimethylbutane, does not increase corrected viscosity. Over-all precision involved in making the solutions, transApparently, once the additional resistance to flow encountered ferring to viscosity tubes, and measuring viscosities, is seen by adding one branch has been overcome, no further resistfrom Figure 1 to be about 3 per cent, compared with an averance is met on adding a second branch to the adioining chain carbon. This effect is estimated by”follo&g the curve downward from its 165centipoise point for 58.5’, to read the corrected TABLE IV. “CORRECTED” VISCOSITIES AND EFFECT OF BRANCHINQ viscosity of 98 centipoises, 2 less than that of Measured Solvent Net Solution Boiling Temperature Corrected Branching isopentane. Solvent Viscosity Point Correction Viscosity Effect When the two branches are separated by one CP. c. O c. CP. CP. carbon atom, as in 2,4-dimethylpentane1 each C contributes two thirds the effect i t would if 128 28.0 28.5 100 37 alone on the butane chain. Combined effect, of 49 centipoises, is found by following the curve downward from 200 centipoises for 44.5’ (the C C difference between the boiling points of 2,4‘ c c ’ 165 58.0 58.5 98 33 dimethylpentane and n-pentane) to read the cy ’t: corrected viscosity of 134 centipoises. C The corrected viscosity of the neohexane solution is similarly found to be 159 centipoises, c-d-c-c 250 49.7 50.2 159 96 I and the branching effect of two methyls on the C 2-carbon of butane to be 96 centipoises. C C Measured viscosities of the other branched L C d 200 80.5 44.5 134 49 paraffin solutions should now be predictable. 37 + 37 / C \c Table V compares measured and estimated values.
FC-,
INDUSTRIAL AND ENGINEERING CHEMISTRY
486
tubes, and the same standard tubes are used for several comparisons.) Further, it is seen from Figure 2 that when naphthenes are present the log solution viscosity-solvent composition curve is not linear, as in the case of mixed paraffins (Table VI). Moreover] the curve must be smoothed to begin at 152 centipoises, an ideal viscosity for the 50/50 methylcyclohexane-n-heptane blend, rather than a t the actual 140-centipoise point. Thus when less than about 50 per cent of isooctane is present, the expected error is relatively large. A test sample containing 28.55 per cent n-heptane, 28.55 per cent isooctane, and 42.9 per cent methylcyclohexane showed a solution viscosity of 172 centipoises and a refractive index of 1.4055. Taking 1.3900 as the refractive index of the paraffins present] the methylcyclohexane content would be estimated a t 44 per cent. Then, from Figure 2, the isooctane content is read as 33 per cent. (If the isooctane content could be determined accurately, its contribution to refractive index could be subtracted from 1.4055, and the remainder partitioned algebraically between n-heptane and methylcyclohexane. Meanwhile, the round figure 1.3900 is taken as an average.) It is hoped that further work will disclose a stable resin, or other polymer, which exhibits a minimum differential in the viscosity of its solutions between normal paroffins and naphthenes of the commonly encountered
TABLE V. ESTIMATED VISCOSITIES
Solvent
'
Temperature Correction
Solution Viacosityb Estimated Actual
c.
c.
Centipoises
154
80.9
81.4
316
310
Net Branching Effect
Corrected Viscosity4
CP.
CP.
96-5
Solvent Boiling Point
C
/c
C-C-C
i 'c C C
\c-c-c-c
'/a
x 37
113
60.3
24.3
141
135,136
'c-c
'/I
x 35
111
89.8
53.8
180
172
'/I
X 96
157
79.2
43.2
231
240
173
99.2
63.2
300
294,295
C
/
C
'c--c
C/ C
1 I
c-c-c-c-c C
C
I
96 4- 37 1.5
c-c-c-c
I
C
\c
a Corrected to boiling point of parent straight chain, which is butane for triptane, pentane for other heptanes, etc. b Average deviation between actual and estimated, 3 . 3 % .
age deviation of 3.3 per cent between measured and estimated values. Following customary practice in correlating properties of hydrocarbons, the method was fitted to existing data on available hydrocarbons. The present data do not provide for estimates in such structures as &methylpentane, 3-ethylpentane, 3-methylhexane, 3,3-dimethylpentane, or 2,3,4trimethylpentane. These paraffins, however, are less likely to be present as major components of postwar commercial solvent naphthas than some of the products studied.
Trial Analytical Method At the present writing, there exists no wholly satisfactory means of determining the relative proportions of normal paraffins] branched paraffins] and naphthenes in close-cut naphtha fractions. The large spreads in resin solution viscosity between normal and highly branched paraffins offer a basis for such a method. As shown in Table 11, isooctane has a 103 per cent higher solution viscosity than n-heptane, whereas its density is only 1.5 per cent higher, and its refractive index only 1 per cent higher than the corresponding values for nheptane. Indeed, refractive index is customarily used to estimate naphthenic content, since within a narrow boiling range the spread between paraffins averages less than one fifth that between paraffins and the naphthene average (1,d). A number of potential errors preclude basing an accurate analytical procedure upon solution viscosities obtained in the present series, although it is possible to make a rapid semiquantitative estimate of hydrocarbons present. Beckosol No. 19 solids does not behave ideally (Table VI). It contains a terpene derivative, which may explain its high viscosity in methylcyclohexane as compared with its solutions in ethylcyclopentane, n-heptane, or the 50/50 mixture with n-heptane, (Solution viscosity measurements when the solvents boil around 100" C. show a precision of 2 per cent or less, since very little evaporation occurs during transfer to viscosity
Vol. 15, No. 8
CURVE FOR ANALYTICAL METHOD
300.
2 250 v,
a 5
200 '%
W 0
175
TEST SAMPLE
0
25
50 % ISOOCTANE
75
1
I00
FIGURE 2
TABLE
VI.
SOLUTION VISCOSITIES AND Solvent
REFRACTIVE INDICES
Solution Viscosity
ny
CP. n-Hentane lGth>Gylohexane Ethyleyclopentane T _a _n"n_r_t a n P _
50/50 methylcyclohexane/n-heptanea 757' A/25Y0 isooctane 6 7 4 A/33% isooctane 50% A/5E)% isooctane 25% A/7a% isooctane 50150 n-heptane/isooctane 50)50 methylyclohexane/isooctane 33/33/33 n-heptane/ethylcyclopentane/isoootane Test sample
:
a
Mixture A.
145,145 160,155 130,128 294,295 140,140 165 172 188 227 202 191
1.3876 1.4253 1.4197 1.3918 1.4067 1.4031 1.4021 1.3995 1.3958 1.3900 1.4090
165
1.4005 1.4055
172
August 15, 1943
ANALYTICAL EDITION
types, but which when dissolved at a suitable concentration gives a maximum spread between normal and branched paraffns. Tabulated data covering the necessary number of 10' C.-wide fractions can then be taken once, and the viscosity of an unknown sample determined a t a later date concurrently with a control determination on a selected normal paraffin, to correct for resin or polymer aging.
Acknowledgment Opportunity is taken to thank M. Lapeyrouse of the Esso Laboratories, Research Division, for helpful criticism in con-
481
nection with this work. The authors are also indebted t o M. R. Fenske, of the Pennsylvania State College, for supplying the majority of hydrocarbons studied.
Literature Cited (1) Francis, A. W., IND. ENG.CHEX.,33, 554 (1941). (2) Ibid., 35, 442 (1943). (3) McArdle, E. H., and Robertson, A. E., Ibid., 34, 1005 (1942). (4) Thomas, C. L.,Bloch, H. S., and Hoekstra, J., IND. ENG.CHEM., A N A L . ED., 10,153 (1938). P R ~ ~ E N Tbefore E D the Division of Petroleum Chemistry at the 105th Meeting of the AMERICAN CHEMICAL SOCIETY, Detroit, Mich.
N-Methylaniline Point of Viscous Petroleum Oils B. W. GEDDES, L. Z. WILCOX, AND E. H. McARDLE Esso Laboratories, Research Division, Standard Oil Development Co., Elizabeth, N. J.
D
URING the past twenty years, aniline point (1, 9) has become accepted as a reliable quick estimate of the proximate composition of the lighter straight-run petroleum products. It has been widely adopted as a measure of solvency of petroleum distillates for various solutes, and has acquired its popularity largely because among such tests it alone requires no temperature control. For highly aromatic solvent naphthas, it has more recently been modified to "mixed aniline point" (6, 6, 8), wherein the aromatic naphtha is first diluted with an equal volume of a 60" C. aniline point naphtha of 43"/44' API gravity. For testing lubricating oils of relatively high aniline p o i n t 4 e., above 80" C.-a second modification is now suggested, wherein aniline is replaced with N-methylaniline. Preliminary results indicate that the method may also serve in testing other petroleum products. The following simple relation exists between aniline point and N-methylaniline point of petroleum lubricating oils: Range N-Methylaniline points
Aniline Point of Diluted Oil I n an attempt to lower the equivalent aniline point of lubricating oils, a typical 120-viscosity aviation oil was diluted with an equal volume of toluene, and the aniline point of this mixture determined. It was hoped that such dilution would drop the temperature to the half-way point, much as "mixed aniline point" raises the reading for aromatic naphtha3. (Here, the straight aniline point of toluene is assumed to be -40' C., -40' F., based upon its "mixed aniline point"
1
/
LUBRICATING OILS
T
No. of Oils Tested
Aniline points
15
80-130
O
c.
* c. 3-53
Tem erature Di#erence O
c.
77 =+=I
At present, an important use of this type of test is in estimating the relative tendencies of these oils to attack rubberlike materials. Following the development of several synthetic elastomers which are not ewily soluble in petroleum oils, an increasing proportion of flexible transfer lines and storage equipment has been fabricated from these materials, and correlations have been drawn between the aniline point of petroleum oils and the logarithm of the per cent increase in volume of rubberlike materials (2, 4, '7). The proposed test method was developed because laboratory determinations of aniline point become increasingly difficult and hazardous as one passes from the lighter petroleum fractions to such heavy and highly paraffinic stocks as aviation lubricating oils. At 120" C. (248' F.)a hot oil or hot air bath, or both, is required to obtain an accurate reading. Furthermore, experience shows that the vapor pressure of aniline at this temperature is sufficient to volatilize enough from the stirred mixture to cause a progressive change in a series of readings. Moreover, the concentration of aniline vapor in the vicinity of a short test tube may be deemed objectionable by sensitive operators.
=
2oLLL IO
180 200 220 240 ANILINE POINT,OI
260
FIGURE 1
of 10.0" C.) Unfortunately, however, instead of lowering the aniline point of the aviation oil (255" F.) to the half-way point (107.5' F.)-i. e., by 147.5" F.-the drop amounted to only 85' F.,from 255" to 170". When the oil was diluted with two volumes of commercial 10' xylene, to make 10 ml. of mixture, the critical solution temperature with 10 ml. of aniline became 140" F., much more than one third the way from -40' (assumed for xylene, from its mixed aniline point of 10" C.) up to 255". Thus no simple correlation between aniline point and a blended aniline point appeared likely to exist when dealing with such dissimilar materials as lowboiling aromatics and viscous lubricating oils. When N-