1326 INDUSTRIAL A N D ENGINEERING CHEMISTRY Vol. 22, No.

12

Chemical Structure of Lubricating Oils'

G. H. B. Davis and E. N. McAllister

DEVELOPMENT
A N D RESEARCH
DEPARTMENT,
STANDARD OIL COMPANYOF LOUISIANA.
BATON
ROUGE,LA.

This paper presents some preliminary data and con-
clusions concerning the chemical structure of lubricat-
ing oils. The work here presented has been confined to
Pennsylvania oil, because this type of oil has been the
most studied and also because of its relatively less com-
plex composition.
A correlation of data on paraffin and cycloparaffin hy-
drocarbons has indicated t h a t the relation of molecular
volume to molecular weight offers a means of estimating
the relative proportions of these structures in petroleum
products. This relationship has been expressed in the
form of the equation:
. . . . . . ....
UCH time has been spent by various investigators
N = 0.358 mol. wt. +
7.7 - 0.3 mol. vol.
where N is the number of carbon atoms per molecule in
naphthene rings.
This method has been applied to a series of specially
narrow cut fractions of Pennsylvania oils and a very close
correlation has been obtained for the empirical formulas
calculated from this molecular volume-molecular weight
relationship and those from the carbon-hydrogen analysis.
I t has been found t h a t the viscosity index of a Pennsyl-
vania lubricating oil is a good indication of the relative
proportions of the hydrocarbon molecules that are in ring
structures or in paraffinic side chains.
and that for Pennsylvania oils this lower value is in the

M on the chemical structure of lubricating oils in an

attempt to further the knowledge of the interrelation
of the various properties of these oils. I n particular, Mar-
cusson (6), Mabery ( 5 ) , Zal'kind (?), and Kyropoulos (4)
have made outstanding contributions. However, the con-
cept of the molecular structure of a lubricating oil and its
neighborhood of C,H2,-s.
The interpretations placed on these analyses are varied.
For the Pennsylvania oils it is generally agreed that they are
practically free from unsaturation. Mabery concludes that
Pennsylvania oils are composed of naphthene nuclei with
paraffin side chains, whereas Xyropoulos lists them as
isoparaffins. Other types of oils, with hydrogen deficiencies
of from -8 to -20, may contain olefinic or aromatic struc-
tures.
Theoretical

It is thus seen that Pennsylvania lubricating oils are

believed to be composed of hydrocarbons containing naph-
thenic (cyclic) structures with probably paraffinic or iso-
paraffinic side chains.
T a b l e I-Characteristics of Hydrocarbons
(Data from International Critical Tables, Vol. I)
CARBON
ATows
HYDROCARBON WT. d i z
MOL. 6" MOL.VOL.

5 Pentane 72
6 Hexane 86
7 Heptane 100
8 Octane 114
9 Nonane 128
10 Decane 142
11 Undecane 156
12 Dodecane 170
13 Tridecane 184
14 Tetradecane 198
16 Pentadecane 212
16 Hexadecane 226
20 Eicosane 282
24 Tetracosane 338
27 Heptacosane 380
32 Dotriacontane 451
36 Hexatriacontane 507
6 3-Methylpentane 86
7 3&Dimethylpentsne 100
8 Iso6ctane 114
9 4-Ethylheptane 128
10 2-Methylnonane 142
6 Cyclohexane 84
7 Methylcyclohexane 98
8 1,3-DimethyIcyclohexane 112
8 Ethylcyclohexane 112
9 1 2 3-Trimethylcyclohexane 126
100 Dekahydronaphthalene 138
Flgure 1-Relation of Molecular Volume and Molecular 12 a ,u-Dicyclohexylethane 196
Welght t o Molecular C o n s t i t u t i o n 144 ~.
Perhvdroohenanthrene 192
a Conjugated ring structure.
relation to its physical properties is still indefinite. Most
investigators agree that the empirical formula for lubricating A survey of the physical properties of pure paraffin hydro-
oils may be represented by the series C,HZo-2 to C,,HP~--, carbons as given in the International Critical Tables has
1 Received September 4, 1930. brought out an interesting correlation-viz., that when
December, 1930 INDUSTRIAL A N D ENGINEERING CHEMISTRY 1327

molecular volume is plotted against molecular weight a
straight line is obtained for the various paraffin hydrocarbons.
This includes both normal and isoparaffins.
When the data on naphthene hydrocarbons are plotted in
like manner it is found that cyclohexane and its homologs
fall in another line parallel to the paraffin line. Likewise,
dicyclohexyl compounds fall in another parallel line offset
by the same amount (Figure I). This suggests that the
molecular volume-molecular weight relationship is a straight
line for a given homologous series including the isomers
in that series. Data in Table I and Figure 1 illustrate this
point.
Table 11-Constants of Naphthene Hydrocarbons
CARBON ATOMS
HYDROCARBON
MOL. MOL. NAPHTHENE
WT. VOL.
determine what relationship existed between the physical
properties of a lubricating oil and its chemical structure.
This study has been limited to Pennsylvania oils, since
they were known to consist of saturated hydrocarbons,
but there was some question as to whether they contained
ring structures or were isoparaffins (4).
A high-grade, well-refined, heavy motor oil was fractionated
very carefully under high vacuum and representative cuts
with average boiling points of 293" C, (560" F.), 322" C.
(612' F,),344" C. (651" F.) and 359" C. (678" F.) a t 10 mm.
absolute pressure were taken. Each cut was then blown
with steam a t 232" C. (450" F.) until 5 per cent of the cut
had distilled, in order to insure complete absence of cracked
products. Each cut was then further fractionated by sol-
IN vent extraction into four equal cuts. The yields and analyses
RINGS of these products are presented in Table 111.

Cyclopentane 70 92.6 5 5.0

Meth ylcyclopentane 84 111.0 5 4.5
Cyclohexane 84 107.2 6 5.6
1,l-Dimethylcyclopentane 98 129.2 5 4.0
N-Propylcyclobutane 98 131.0 4 3.5
Cycloheptane 98 119.8 7 6.8
1,l-Dimethylcyclohexane 112 142.7 6 5.0
1-Methyl-2-ethylcyclopentane 112 145.8 5 4.1
Ethylcyclohexane 112 143.3 6 4.8
Propylcyclopentane 112 144.3 5 4.5
1,1,3-Trimethylcyclohexane 126 158.5 6 5.3
1,2,3-Trimethylcyclohexane 126 158.1 6 5.4
Cyclo6ctane 112 133.6 8 7.7
Isopropylcyclobexane 126 158.8 6 5.2
N-Propylcyclohexane 126 159.5 6 5.0
Isobutylcyclohexane 140 173.5 6 5.8
1,2,3,6-Tetramethylcyclohexane 140 171.1 6 6.5
Ethylcycloheptane 126 154.8 7 6.4
m-Methylisopropylcyclohexane 140 174.1 6 5.6
~-Methylisoprop ylcyclohexane 140 173.9 6 5.7
o-Methylisoprop ylcyclohexane 140 171.9 6 6.3
2-Cyclohexyl-2-methylhutane 154 187.0 6 6.7
Dimethyl-1,3,5-isobutylcyclohexane 168 203.0 6 6.9
1-Methyl-2-isoamylcyclohexane 168 206.0 6 6.0
Decahydronaphthalene 138 156.0 10 10.3
Perhvdrouhenanthrene 192 206.4 14 14.5

The fact that the parallel lines are spaced equidistant

suggested the possibility of continuing the series for com-
pounds containing more than two cyclohexyl rings. This
extrapolation is represented by the dotted lines in Figure 1.
The relationship shown in this figure may be readily
expressed in the form of an equation-viz.:
N = 0.358 mol. wt. 4- 7.7 - 0.3 mol. vol.
0

where N is the number of carbon atoms per molecule in naph-
thene rings. This equation is applicable to hydrocarbons
containing naphthene rings other than cyclohexane, as
illustrated in Table 11, in which it is shown that the calcu-
lated number of carbon atoms in the rings check within one
of the actual number.
Experimental
With the relationship developed above a study has been
made of various cuts of a Pennsylvania motor oil in order to
Figure 2-Relation of Molecular Weight, Molecular Volume, and
Viscosity Index
The analyses for gravity, viscosity, flash, and pour were the
usual inspections as prescribed by the A.S.T.M. The
viscosity index was determined according to the method of
Dean and Davis (1). The molecular weights were deter-
mined by the cryoscopic method ( d ) , using cyclohexane as a
solvent. The carbon and hydrogen determinations were
made by the usual organic combustion method ( 3 ) . The

Table 111-Characteristics of Solvent Extracts of Pennsylvania Motor Oil

c u t No. 1 I
Orig. 3 I 4 I 5 I 6
Av. boiling point (10mm.H g ) ( .. 1 293' C. I 322' C. I 3440 c. I 3590 c.
Extract No. ..
Orig. 1 2 3 4Orig. 1 2 3 43rig. 1 2 3 4Orig. 1 2 3 4
22.5 22.5 22.522.5 22.5 22.5 22.5 22.5 22.522.5 22.5 22.5 22.5 22.5 22.5 22.5
$ $?gins1A. P. I.
Gravity
io0
5:40 1.22 1.221.22 1.22 4:oo 0.90 0.90
27.8 30.222.0 30.0 33.4 34.7 29.822.828.6
0.90 0.90 3:io 0.70 0.700.700.70 i:50 0.340.34 0.34 0.34
31.3 34.3 28.6 22.826.5 31.0 33.3 27.5 20.327.6 31.0 32.3
Viscosit;at 100" F. (Saybolt) 870 213 440 207 175 224 269 558 309 267 215 586 1235 726 440 393 984 3300 855 634 650
Viscosityat210°F.(Saybolt) 85.3 47.454.6 47.446.451.0 51.060.052.8 51.7 49.6 58.0 86.073.5 64.062.0 89.0 140.6 85 4 76.880.6
Viscosity index 101 93 58 103 120 125 101 62 96 107 120 97 70 91 111 118 98 62 101 114 124
Flash, F. 460 420 435 425 425 450 445 450 460 460 470 515 510 510 505 510 550 535 535 550 570
Pour, F. 45 40 30 40 50 45 45 35 45 45 60 50 45 50 55 60 55 40 50 60 65
Mol. weight
Mol. vol.
.
. 398 365 384 395 425 425 395 4045
. . 449 396 438 461 499 484 431 456a
413 445 480 453 455 500 505 551 452 536 612 597
477 522 543 494 508 572 589 619 484 603 703 690
Ultimate analysis:
Hydrogen, % .. 13.6 12.3 13.7 13.7 .. 13.4 12.4 13.2 14.0 .. 13.3 12.9 13.4 13.8 13.3 12.8 12.213.2 .. ..
Carbon, % .. 86.4 87.7 86.3 86.3 .. 86.6 87.6 86.8 86.0 .. 86.7 8 7 . 1 86.686.2 86.7I 87.2 87.886.8 .. ..
Doubtful average molecular weight.
1328 1.VDUSTRIAL A S D ENGILVEERING CHEMISTRY 1~01.22, KO. 12

carbon-hydrogen re- Correlation of Viscosity Index and Chemical Structure

sults are presented
on the basis of 100 It has previously been suggested that viscosity index,
parts of hydrocar- which represents the change in viscosity of the oil with
bon, since the actual temperature, may be used as a guide to the relative compo-
analyses varied from sition of the oil, and that a definite relation must therefore
t,
99.5 to 100.3. exist between the viscosity index and the physical properties
m
By usiflg the data of the oil.
m
presented in Table From the data presented in Table I11 a plot has been
I11 and the formula drawn showing the molecular volume-molecular weight
derived previously, relationship for lubricating oils of various viscosity indices
it is possible to cal- (Figure 2). This plot shows that this relationship is a straight
culate the average line for the oils of different viscosity indices, However, the
number of carbon lines tend to converge a t low molecular weights. This
-
atoms in naphthene suggested the idea that the viscosity index was a function
ring structures for of the percentage of the carbon atoms in the molecules that
each of the oils given were in naphthene rings. The relationship found is shown
in the table. These in Figure 3. The correlation is good.
results are presented
in Table IV. Here
the empirical for-
mula for each sample
is also c a l c u l a t e d
from the number of
c a r b o n atoms in
naphthene rings by
assuming an aver-
age of 6 c a r b o n
atoms per ring and
a deficiency of 2 hy-
drogen atoms per
ring. The empirical
formula calculated
N I
in this manner is
found to correspond
very closely to that
calculated from the
c a r b o n-h y d r o gen
analyses. The wid-
est discrepancies are
in the first extracts
of each f r a c t i o n .
These show a greater
deficiency in hydro-
gen determined by
c a r b on-h y d r o gen
analysis than those
determined by the
molecular volume-
molecular weight re-
l a t i o n s h i p . This
difference is prob- Figure 3-Relation of Viscosity Index t o Per Cent of Carbon Atoms
ably due t o the con- i n Naphthene Rings i n Aromatic Free Lubricating Oils
centration of such
unsaturated or aro-
matic compounds as Since this entire study was made on Pennsylvania oils,
the results are only applicable in the higher viscosity-index
might be present in
ranges (75 to 125). The lower viscosity index oils obviously
the first e x t r a c t . must contain some olefin or aromatic compounds, whereas
The close correla- this study is based primarily on saturated oils. The extrapo-
tion of the empirical lation, such as is made in Figure 3, is of benefit as an
f o r m u l a s in the indication, but will require substantiation before it can be
other cases indicates accepted.
that the assumption
that 6 carbon atoms Conclusions
was t h e a v e r a g e
number in a naph- (1) Pennsylvania-type oils (saturated hydrocarbons)
thene ring was prob- are composed essentially of a grouping of paraffin chains
ably correct. with naphthene rings.
December, 1930 I N D U S T R I A L A N D ENGINEERING CIiEhfISTRY 1329

(2) For this type of oil the number of carbon atoms in Literature Cited
naphthene rings may he determined from the molecular (I! D~~~ and ~ ~aem. ~&rei.Ei - ~ . ,~~6 , 8 1 ,8(1929)
weiEht and densitv.
Y I
( 2 ) Findlay, "Practical Physical Chemistry," Longmans. 1925
(8) Fisher, "Laboratory Msnual of Organic Chemirlry." Wiley, 1924.
(3) viscosity index may he used as an indication of the (4) Kyropoulos, 2. phyiik. Chem., 144, 22 (1929).
chemical structure of the oil and for that reason should (j) Mabcry. IND. c,,~~., 1233 (19~3).
serve as a basis for the correlation of the various physical ( 6 ) ~ a r c u n r o n11fiii.
, .M~rrri=ipruiunaromr,40,308 (19221.
properties of lubricating oils. (7) Znl'kind, Pcirolevrn Shale . I(Ruisio), Ill, 154 (1922): J . Insf. Pibo-
[rum Tach., 9. l57A (19221.

N e w l y Discovered Microscopic Structural Units

of Wood Fibers'
Geo. J. Ritter and R. M. Seborg
U. S.Fonasr PXODUCTS
LABORATOXY,
MAUISON,Wrs

Optical Properties
R ESULTS published by Ritter (1) regarding the dis-
section of wood fibrils by chemical means showed that
the smallest structural 1mit.s of the vood fiber which
had been isolated at that time were fusiform bodies. They
l'reviously described structural cellulose units ( 2 ) sliowcd
between crossed Xicol prisms an inmasing degree of sharp-
ness in the angles of minimurn and maximum luminosity
are short, spindle-shaped units which, whcn arranged parallel with decreasing size of unit. I n other words, the parallelism
to one anot,her with an overlapping of the pointed ends, of the crystalline structure which exhibits the effect in
form the fibrils. The fibrils are long, slender, filament-like polarized light was greater in the smaller units. Such a
structures, arranged to form the fibers. condition would be expected from the fact that t,here is less
Careful microscopical examination of the fusiform bodies opportunity for variation in the parallelism of crystalline
after various treatments suggested that they are composed arrangement in the smaller than in tlie larger units which
of still smaller structural units. A t the bime of the meeting are composites oi the small ones. It was found, however,
of the American Association for the Advancement of Science, that between crossed Nicol prisms the isolated spherical
Des Moines, Iowa, December, 1929, sufficient progress had units were uniformly luminous in all positions, indicating
been made in the study to warrant tlie following prediction (3) :
I t is quite probable that still smaller microscopic building
units in the fusiform bodies will be discovered in the near future,
and that the source of thc optical properties described in t h e
preceding paragraphs will then be found in the newly discovered
units.
Recently, separation of the fusiform bodies into smaller
units was accomplished a t the Forest Products Laboratory,
but suitable photomicrographs of the results were obtained
only after much experimcntation aiddifficulty.
Materials

The materials used in this study consisted of two batches

of mixtures of fibrils and fusiform bodies vhich were pre-
pared from delignified white spruce fibers. One batch was
prepared by the sulfuric acid method; the other, by chlori-
nation in the sunlight.
Procedure a n d Results
Small samples of the mixtures containing fibrils and
fusiform bodies were treated on glass slides wit11 phosphoric
acid (78 per cent) a t approximately 75" C. When slight
pressure was applied to the cover glass over the specimen,
the fusiformbodies separated into smaller units.
The newly discovered units (Figure 1) are spherical in
form when detached from the mother unit. They are
approximately 0.45 micron, or 4500 A. in diameter, which
is nine times the length of the cellulose micelle rcrcaled by
the x-ray diffraction pattern. Since those units h a ~ cnot
been observed in the fusiform bodies, their original fonn is
unknown, but from observed optical properties it would
seem that tlie shape is other than spherical.
1 Received September 20, 1930. Presented belore the Division oi
Cellulose Chemistry a t the 80th Meefine of the American Cheoilcal Society,
Cincinnati, Ohio, September 8 t o 12, 1930
Fleure I-Spherical Units. 700 X
a random arrangement of tlie crystalline structurc. Such a
disarrangement in tlic internal stmcture of the new units
would result during the deformation of an angular body to a
spherical one. This conclusion is confirmcd by the fact
that it would be impossible to build up a solid structure
with spherical units.
The resu1t.s lierein described suggest a future study in
which the spherical units may be isolated under minimum
swelling condit,ioiis, so as t,o reduce the possibility of de-
forming them. If the s8mc units can be obtained by such n
method, which the writers believe is possible, their optical
properties sliould be siioilar to those of tlic fusiform bodies.
Literature Cited
(I! ~ i t t e i IN".
, RNO. C H E W . . a i , 289 ( 1 ~ 2 9 )
?a! R i t t r i . 3 . P'o"?stry, 28. 533 :l9:3'>!