RAL OIL
I
N
JITLLPRED M. HICKS-BRUUN", BRUCE 1,. RITZ, ROBERT E. LEDLEY, JR., ANT) JOHANNEs R. BRIIUY2 Sun Oil Company. Uorwood, Pa type f l o ~ inetw, maintaining n cuii>t A I I ~flo\i ot air r'he tal\ HE importance of mineral oil oxidation studies is illusmag be adjusted to hold the reduced pressure conslmt a t a n \ trated by the volume of work reported during the past ten value between 0 and 5 pounds per square inch. years. This work has been extensively reviewed (9, 10, The flowmeter is a mercui y-filled glass instrument provi with interchangeable capillary tubes for metering a wide I ange nf 14, 16). In many cases types of oils amrerated as to oxidation reflons. The static preasure a t the downstream end of the A o n sist,ance or susceptibility by the measurement of one specific meter is maintained constant by a gas-service back-pressure reg11property or product. Most often used has been oxygen absorplator (American Meter Company) a t a presure somewhat h i g h . tion or sludge formation. The measurement of only one propthan the maximum which may be obtained in the preheater 1111 erty cannot present a complete picture of the behavior of an oil These pressures, measured by oil-filled manometers, are, spectively, 460 and 230 mm. of oil (30 and 15 mm. of meicu upon oxidation. Unfortunately the laboratory tests reported above atmospheric pressure. Since any given differential pi have been limited, either because they were designed for such sure may be maintained a c r o s the flourneter capillary tube small quantities of oils that samples adequate for complete these two valves, the flow through the capillary tube may be h constant. The flowmeter is piovided n i t h a capillary tube n l i examination cannot be withdrawn during the course of the oxidation, or because the conditions of oxidation were not ~ f f i - produces a differential pressure drop of 100 nun. of mercury a t t desired rate of flow (30 liteis per hour) ; and since both the f l i ~ ciently controlled. Work in this laboratory during the past rontrol valve and the back-pressure regulator are sensitive I few years on the oxidation of oils has indicated not only the pressure changes of less than 1 mm. of rriercury, variations in 1 1 1 ~ necessity for measuring by exact methods as many criteria of rate of flow are less than 1%. The preheater through nhich the air patbbes before entriil oxidation as possible, but also the need for developing a suitable the oxidation chamber 1s a section of I/d-inch standard pipe, pr111) laboratory apparatus in which a quantity of oil, comparable in erly insulated and wrapped with sufficient electric heating coil t c size to that contained in the average automobile crankcase, heat the incoming air to the desired reaction temperaturi could be oxidized under carefully controlled conditions of tem(175" '2.). A thermometer ~vellis located in the line close t o I i ( perature, pressure, and oxidizing gas. Furthermore, the ap-, oxidation chamber. The autoclave, complete w t h drive, &\ab constructed b j t i i t paratus should be so designed that samples of the solid, liquid, Blaw-Knox Company. I t has B nominal capacity of 2 gallo~land gaseous oxidation products, sufficiently large for test purand all parts which come in contact with the charge or O X I ~ I L I I poses, could be withdrawn at, definite intervals throughout the gases are made of 18-8 chrome-nickel steel. The agilato time of oxidation. diive, c h g i n g line, gas outlet, and tliermocouple well are mounted on the removable head, leaving ;he interior free of This paper describes ,such an apparatus and includes data projections for easy cleaning. A Johns-Manvalle asbestos w i v u ~ showing the operation precision att'ained and the reaction gasket is used between the head and the aLitocla\e trends observed during the partial oxidation of nine widely differT w o 5-inch intermedmte-type turbine agitator- are mouiltt r i on the drive &haft i n the positions shoan in Fieurc 1 'rlc ent oils. A new oxidation scale i s de. ihed for the comparibottom agitat'or mixe3 t)iP son of the various oxidainc30ming air with t h tion criteria for different I liquid, and the top otic oils. Although certain 1 serves t o recirculate tilt. A high-precision apparatus has beeri developed suitable partiaily spent gases. The related functions will be for oxidizing, under carefully controlled conditions, a shaft speed is 431 r.p.n~ pointed out, it is not quantity of oil comparable in size to that contained in an The drive shaft enteli possible to discuss satisthe autoclave through i~ automobile crankcase. Water and carbon dioxide forfactorily oxidation staspecial self-sealing greawmation can be expressed as functions of the oxygen abbility in relation to oil packed stuffing box, 'ur sorbed. Viscosity changes are due mostly to resins. rounded by a water jacki I constitution until these The amount of precipitable sludge is not related directly to prevent overheating o' effects are better underto the other oxidation data, nor does it form in oxidized the shaft and packing stood. oil in the absence of additional oxygen. Catalytic efThe gas outlet is a 2-fooi fects are found to be important and confirm the work done qectign of I-inch t 8 - h DESCRIPTION OF iteel pipe equipped w.111 elsewhere. An oxidation scale has been deiised for comAPPARATUS roils f o P heating and m" parison of the oxidation resistance of oils bj conversion ing as desired The onI\ The apparatus for the qf selected oxidation criteria. connections to the bo oxidation of oils (Figuie of the autoclave are t I ) may conveniently be
T
INDUSTRIAL AND ENGINEERING CHEMISTRY
June, 1944
563
-
c A
I
BY-PASS (FOR FLOWMETER CALIBRATION )
a
f
WATER
c02
---
WATER
coz
Figure 1. Apparatus for the Oxidation of Oils
,
The exit gases pass through a water-cooled condenser, a crushed ice trap, and two dry ice traps, where normally liquid oxidation products are condensed. The ice trap is provided with a stopcock for draining off the accumulated condensate, and the succeeding dry ice traps are graduated a t the bottom to permit reading of the volume of condensate collected in them. A pair of glass absorption towers containing dehydrite and ascarite are next in line. The dry gases are then conducted through a copper oxide furnace (heated t o a bright red heat or 950-1000" C.)*in which any oxidizable components of the gas are oxidized to water and carbon dioxide. These products of the "secondary oxidation" are likewise absorbed in dehydrite and ascarite towers. A positive-displacement wet-test gasometer (American Meter Company) is used for measuring the volume of the exhaust gases, and a Williams gas analysis apparatus is employed t o measure the oxygen concentration of this spent gas. Auxiliary apparatus is provided for purging the system and the oil sample with pure nitrogen a t the beginning and end of each run. A by-pass line equipped with a pressure gage is provided between the high-pressure air line and the preheater for testing the autoclave and auxiliary connections for leaks. Bypasses around the purification train and the .absorption towers permit calibration of the main flowmeter against the gasometer when the autoclave is empty. METHODOF OPERATION.Charges of 3000 grams of oil were oxidized for 8 to 32 hours at 175" * 1" C. under a pressure about 2 mm. above atmospheric. The air rate for the oxidizing gas was maintained at 0.5 liter per minute and controlled to better than 1%. To standardize the oxidations further, purified nitrogen was passed through the apparatus and charge during the heating period (1.5 hours * 10 minutes) for 30 minutes after the charge reached the reaction temperature and during the cooling period at the end of each run. Both the air preheater and the exit arm were maintained at the reaction temperature of 175" C. Every 2 hours samples of the charge were taken, the amounts of the oil-soluble and water-soluble layers in the traps were recorded, and frnsh absorption towers of known weight were placed in the train. Oxygen analyses were made every hour. Care had to be exercised t o avoid activation effects which developed in the apparatus under certain conditions and were
found t o exert a decided influence on the extent t o which most oils would oxidize. For example, the oxidation of highly refined oils t o high acid numbers seemed to activate the apparatus in such a way t h a t subsequent runs on other oils caused them t o deteriorate much more rapidly than their normal rate. None of the various methods of cleaning which were tried would deactivate the apparatus, but several runs with a less highly refined oil served to bring the rate of oxidation back t o its normal value. EXAMINATION OF OXIDATION PRODUCTS
Oxygen absorption and amounts of carbon dioxide and water formed throughout the oxidation treatment were determined periodically by methods given in the description of the apparatus. The low-boiling hydrocarbons and the trap acids were determined from the condensate obtained in the various traps. Although subject to some error, the trap acids were calculated as acetic acid from the acid numbers of the water layer of the trap material. The total amounts of primary water were subsequently corrected for the weights of acids thus obtained. The undissolved sludge was determined by filtration of the oil through a n asbestos mat in a Gooch crucible. Dissolved sludge was determined by the propane precipitation method of Hall, Levin, and McMillan (IS). The values of the total sludge conpercentages of undissolved and stitute a summation of the weight ~. dissolved sludge. The fixed acids were determined on the propane-soluble oil fractions bv saDonification with sodium hvdroxide, followed by filtration and aqueous extraction of the soaps from a pentane solution of the oil, with subsequent acidification of the soaps. The carbon-hydrogen ratio of the noncondensed gases was calculated from the.weights of carbon dioxide and water formed in the "secondary oxidation" of carbon monoxide, hydrogen, and hydrocarbon vapors which passed through the dry ice traps. Kinematic viscosities were determined by a modified Ostwald viscometer ( I ) . A Klett-Summerson colorimeter with a No. 54 filter was used to obtain optical densities. Determinations of the acid and saponification numbers were made by standard methods and calculated as milligrams of potassium hydroxide per gram of oil.
INDUSTRIAL AND ENGINEERING CHEMISTRY
564
Vol. 36, No. 6
TABLEI. PRECISIOX OR APPARATUS, SHOWNBY TRIPLICATE OXIDATIONRUNSON OIL B Oxidation, hr Run 11-280 Run 11-284 Run 12-14 Average
b
5.65 6.51
12.45 13.43
6.08
12.94
7.1 7.1b
3.8 3.8b
....
Deviation Max. yofrom av. Av. % 0
Mg.H?O Formed/G. of Oil 4 8
.Mg. 0 9 .4bsorbed/G. of Oil 4 8
-
.....
2.10 1.97 1.92 "00
5.76 5.83 5.68 5.76
5.0 3.5
1.4 0.9
Sludge5
13.0 8 7
-8
4
0.73 0.69 0.76 0.73
0.36 0.31 0.43 0.37 16.2 11,
5.5 3.2
1.29 1.23 1.16 1.23
6.13 6.92 7.02 6.69
15.61 16.14 15.12 15.62
6,7
8.4 5.6
3.3 4.2
3..5
3.24 2.88 3.60 3.24 11.1 7.4
7.94 7.94 9.19 8.36 9.9 0.6
Total sludge oil-insoluble -I-propane-insoluhle sliidiae. Triplicate runs on other oils show a similar average deviation of about 6 %
0 9 absorbed, m* /g. oil Primary H ~ O &, / g . oil Primary COP, mg./g. oil Undissolved sludge, w t . % Dissolved sludge, w t . % Fixed acids, w t . % ' %:H ratio of noncondensed and nonacidic gases Trap products, wt. % ' Hydrocarbons Acids (calcd. as acetio) Optical density/mm. Saponification Nu. Acid No. Density, 20' C. Viscosity, centipoise At 210' F. A t 100". F. Viacosity Increase, % A t 210" F. A t l0Do F
..
3.88 0.70 0.070 0,023 0.15 0.43 9.4
7.83 2.00 0.23 0.043 0.33 0.88 14.5
11.94 3. (3(3 0.47 0.13
0.034 0.005 610 1.3 1.2
.
0.120 0,016 3040 2.0 1.6 0,9239
0.224 0,033 0.6330 3.1 2.2
0.336 0,050 8200 4.4 3.1 0,9266
3 . 55 30.9
5.62 51.9
5.73 54.3
5.90
56.6
6.01 58.8
...
1.33 1.99
3.24 0,60
6.31 11.14
8.36 15.62
.
,.
..
, . . , .
..,
si
..
ii.
siii
TABIJ3
Oil designation Jaybolt viscosity, see.
At
looo F.
.,
111.
A
209 99 43
A t 130' F. A t 210: F. Viscosity index
n
0 884 21.1 350 -35 0.9202 1.5098 73.1
21
8
paraffin chains alcd. mol. wt. (16)
0
a
0.22 0.21 0.26 0.23
Viscosity Inciwaee % A t 210' F. .4t l o O D F. -__.. 4 8 4 8
% Total
M g . CO? Formed/G. of Oil 4 8
E 385
..,
15.64 5.76 0.73 0.33 0.90 2.26 19.5
0.55
1.78 17.9
. ..
, . ,
...
... I
.
...
... , .
0
..,
11.62 4.57 0.38 0.032 0.17 1.81 8.9
18.56 7.39 0.71 0,046 0.28 2.54 9.6
24.36 10.25 1.10 0.056 0.44 3.51 10.1
0.058 0.003 61 1.7 1.3
0.166 0,007 286 3.3 1.7 0.9013
0.270 0.012 697 6.4 :3 , 0
...
0.386 0.018 1150 10.3 4.0 0.9048
6.84 65.6
0 97
68.2
7.17 71.5
7.41 76.4
1.89 2.92
3.87 6.97
6.81 12.28
10.38 19.94
..,.
0,89j3 6.7: 63 I
P H Y S I C A L PROPERTIES O F OILS BBFORE
5.80 1.72 0.13 0.023 0.12 0.95 8.3
OXIDXITON
Ra
D
G
F
232 107 44 f 6 0.879 21.6 345 30
293 130 47 -I- 24 0,871 22.3 350 20 0.9169 1.5050 80 7
291 132 47 4-25 0.846 25.5 380 - 20 0.8978 1.4897 96.0
220 107 45 f50 0.833 27.7 370 40 0.8861 1.4828 100.7
2610 275 - 60 0.8952 1.4913 66.6
275 60 0.88.52 1.4866 71.6
205 107 47 -4.105 0.810 31.4 415 -475 0.8644 1.4764 108.8
19 31 50 337
16 34 60 353
3 43 54
0 43 57 380
17 38 46 277
14 36 50 279
2 24 74 413
-
-
875
-
B
I
H
56
56
43 33
43 33
I
.
26:f
-
Oil C was oil B inhibited.
PRECISION OF APPARATUS
T h e precision of the results from oxidation in the autoclave is shown by the data from three runs under the same conditions on a typical lubricating oil (Table I). It is evident that the average deviation for oxygen absorbed, water, carbon dioxide, and sludge formed, and viscosity increase is about 6% or better. This degree of precision is in good accord with t h a t reported by other experimenters (8,11, 17'). The use of a portion of t h e oxygen in the spent gas t o convert any oxidizable components of this gas t o water and carbon dioxide admittedly causes a n error in the subsequent oxygen analyses. This error is small enough t o be well within t h e limit of reproducibility of the oxygen absorption measurements, because the amount of noncondensable gases formed is small and because the high carbon-hydrogen ratio of this oxidizable portion indicates that it is mostly carbon monoxide. Table I1 is a typical set of data obtained by the oxidation for 8-hour periods of a lubricating oil, B, and a more highly refined oil, E. The physical properties of these oils before oxidation are listed in Table III. Comparison of the oxidation results for these two oils shows them t o be similar t o those reported by othei
laboratories; that is, the highly refined oil absorbed considerably more oxygen than the less highly refined oil, formcd more water, carbon dioxide, and saponiiiable material, and exhibited a slightly greater viscosity increase. On the other hand, the less highly refined oil formed much more sludge. Similar data were obtained in duplicate or triplicate on a aeries of and nine oils of varying degrees of refinement. These include an uninhibited paraffin base oil and eight naphthene base oils, of which one is an inhibited lubricating oil (Table
111). RELATION OF On ABSORBED TO H20 AND COP FORMED
D a t a for oxygen absorbed and the corresponding amounts of water and carbon dioxide formed are presented in Figure 2. Although each graph contains data for nine widely different oils, the points all fall in the same definite band and thus indicate a bimilar type of oxidation reaction for each oil. The curves obtained are obviously not linear over the entire range of oxygen absorption, but when the results are plotted over relatively short ranges of oxygen absorption (as in Figure 2), straight lines may be drawn as representing tangents t o the curves; from them may
INDUSTRIAL AND ENGINEERING CHEMISTRY
June, 1944
565
tions 2 and 5 were combined AS FUNCTIONS OF OxYQEN ABSORBED and presented in a similar form TABLEIV. CARBONDIOXIDEAND WATERFORMATION DURING A SERIES OF OXIDATION RUNS (Equation 8). The similarity
between Equations 7 and 8 indicates that, in spite of the dif1 176 NO ferences in oxidation conditions, 2 175 NO the general type of oxidation rea 175 Yes (Cod = 0.075 01) 0 016 4 175 No (HzO) 0.69 - 0.016 action has not changed appreci5 175 No IHz0) = 0.87 0.031 6 176 Yes HnO) 0.81 (On) 0.033 ably. 185 No (Hz0) 7 7 ( C o d 4-0.22 ... 7 Table V presents the experi8 175 No (H20) = 12 5 ( C o d + 0.078 7-60 mental data in a form which shows the distribution of oxygen to water, carbon dioxide, and be derived equations for the formation of water and carbon other products. These results are in general agreement with dioxide as a function of the amount of oxygen absorbed. These those obtained by other investigators (a, 6, 7,8, IO). equations, with others derived from similar data and graphs, are listed in Table IV. Equations 3 and 6 were obtained from 32-hour oxidation runs TABLEV. AVBIRAQE DISTRIBUTION OF OXYGEN A B S O R B ~BY DA SERIESOF OILS made in the presence of 0.2 t o 1.0% iron and copper filings as Length of Range of 0 9 catalyst. Equation 7 was obtained from earlier oxidation runs Oxidation Consumption, %02 %09 'Otoh'ee,"" on the same oils made in a different apparatus a t 185' C. and Run, Hr. Mg./G. of Oil to Hz0 t o Cot Products 100 pounds pressure. Since no oxygen absorption data Were 8 3-20 34.5 3.1 62.4 32 7-60 43.5 6.9 49.8 available for these oxidations, the water formed was derived as a 32 (catalytio) .10-60 40.5 7.5 52.0 function of the carbon dioxide formed. For comparison, EquaEqmtion
No.
TemperaC.
ture,
Time Pressure, of Run, Lb./Sq. In. Hr. Atm. 8 Atm. 32 Atm. 32 Atm. 8 Atm. 32 Atm. 32 100 30 Atm. 32
Catalytio Run
Approx. Equation in Terms of Millimoles (Con) = 0.031 (01) 0.0012 (COz) 3 0.069 (02) 0.0086 5
b$
5
--
Range of Consumption, Mg./G. of Oil 3-20 7-60 10-60 3-20 ' 7-60 10-60
10.0
B
A
/
** 8.0 =!
?m 7.0
-
(H2O) = 0.49( 0 2 ) 0.55
*y.
4 HRS.OXIDATION
0 0
= 8 HRS. OXIDATION
A
= 32 HRS. OXIDATION
-
16 HRS. OXIDATION
*&
P
'5
3.
* .
0 0 A
8
2 HRS. OXIDATION
= 4 HRS. = 6 HRS.
A*
" I
OXlIMllON OXIDATION
L
8 HF& OXIWION
a
o L 1 l ' 1l0 ' lI
'
I
20 ' '
I
I
'
Mg, 0 2 A6SORBED per g.0IL
Figure 2. Water and Carbon Dioxide Formation as a
.u a
dJ
I Function of Oxygen Absorption for Nine Different Oils
0 0
a A
A.
Water formation during %hour runs at 1 7 5 O C.
B.
Water formation during 32 how rune at 1 7 5 O C.
C.
Carbon dioxide formation during 32-hour runs at 175O C.
--
= 4 HRS. OXIDETION 8 HRS. OXIDATION 16 HRS. OXIDATION
= 32 HRS. OXIDATION
L 40 60 Mg. Oe ABSORBED per g. OIL
566
INDUSTRIAL AND ENGINEERING CHEMISTRY
TABLEVI. EFFECT O F UNDISSOLVED c -
Condition of Oil Original Original; resin-free Oxidized. filtered Oxidized f sludge-free Oxidized' sludee- & acid-free Oxidized; sludie-, acid-, & resin-free
Viscosity Centipoises 50.8 49.8 66.2 63.1 62.5 46.9
DISSOLVEDSLUDGE, RESINS,AXD A C I D S O N vISCoSITY O F O X I D I Z E D OIL ___ -Oil I-------. ... Oil Ra t looo F. Oxidation products re___ Viscosity a t looo F. Oxidation products % increase moved, nt. % ' Centipoises ?' & increase moved, wt. 70 AND
_-_
.....
-2.0 30.4 24.2 22.9 -7.7
(I) R--CH,-R'
+ '/z +
0 2
---+R-CHOH-Et' R- C-R' + HpO
0 2 --i,
ti- CHZ-CH2-R'
..I.....,..
Resins, 0.82 Undissolved sludge 2 . 8 Dissolved sludge, d. 79 Acids, 1.33 Resins, 4 . 4
From the standpoint of water and carbon dioxide formation, the theoretically possible equations which repretent those reactions commonly ieferred t o a- oxidation can be divided into three groups: (a) No gaseous reaction products are formed; ( b ) water is the only gaseoua reaction product; (c) water formation is accompanied by the liberation of carbon dioxide, carbon monoxide, or low-membered paraffin, olefin, or acetylene hydroc.arbons. The three different typeq of reaction.; may be illustrated by the following equations: (11) 13- CHZ-R'
Vol. 36, No. 6
+ l l / z BR-CHO + R'CHO + €I& 0 2 --f
Many more equations can be A ritten Ln each ( are sufficient for illustration. Inspection of class I11 equations aho\rs that, in the examples given, for every molecule of carbon dioxide foimed, tJ+o molecules of n a t e r are given off. Hence, if r e absume that this is a general relation, it is possible to calculate the neight of water corresponding t o the weight of carbon dioxide evolved during the primary and secondary oxidations. This water, corresponding to the carbon dioxide formation in class 111 reactions, can be wbtracted from the total mater.. Presumably, the amount of water left reprecents that given off by reactions of the second group referred t o above, or th0.e in which oxygen reacts t o foim watc>r, and in which aldehydes, ketones, acids, hydroxy aldehydes, ketoalcohols, ketoaldehydes, ketoacids, or olefins may be formed. Calculations of this type made from the equations obtained for the 8- and 32-hour oxidations show t h a t during the early stages of oxidation about 96% of the water is formed by class I1 reactions. For longer periods of oxidation this value falls t o about 85y0,indicating a slight change in the type of oxidation ieartion. VISCOYIT1 IRCKEASE
The causes of viscosity increa:? in laboratory-oxidized oils liave been investigated, typical iecults on oils B and I are pre-ented in Table VI. The oxidation products were divided into dissolved sludge, acids, and resins. For this work the dissolved iludge was determined by pentane precipitation. The acid. s e r e removed by neutralization a t room temperature; hence any nonacidic saponifiable material is included, together with other neutral oxygenated compounds, under the grouping of resins. They were determined by adsorption on activated alumina, and include sludge which is propane insoluble but was not precipitated by pentane. For oil B approximately 80% of the viscosity increase (looo F.) nab caused by neutral resins, 16% by pentane-insoluble dissolved sludge, and the remaining 4% by acidic material. For oil I the values run 90% due t o resins, 7% due t o sludge, and 3% due t o acids. However, on a gram for gram basis (Le., assuming that equal quantities of the three classes of products were present in the oil) the sludge would cause the greatest increase in viscosity, with resins next and acids least.
43.7 43.1 48.2 47.7 47.6 41.6
.....
-1.4 10.3 9.2 8.9 -4.8
.I'C-
............
Resins. 0 . 3 3 Undissolved sludge, 0. 93 Dissolved sludge, 0.012 Acids, 0 . 7 3 Resins, 3 . 2
This id the same relative order as the probable molecular weights of these materials (9,14)* These data are in excellent agreement with the reaults obtained by Davis and co-workers (4)in their analysis of a used crankcase oil, although their analysis does not differentiate between acidic and nonacidic resins. VARNISH FORMATION
T o measure varnish formation during oxidation, pi ovision wa. made t o clamp small g1a.s or steel slides t o the drive shaft of the oxidation autoclave. In all cases varnish formation was mea< ured by the gain in weight of the slide after it was nashed f i c r of oil with pentane and carefully wiped nith a -oft cloth to remove loosely adhering sludge. A mass of data was accumulated, but only two generalization. could be made: first, t h a t the type of surface appealed to h a \ ( little effect on the amount of varnish formed, and second, that the n eight of va1ni.h depwited on the qlides roughly parallelcd undissolved sludge formation. Thur, in ordrr to establiih t h e relation between the varnish which plated out on the ilidei and the undissolved sludge, a separate set of experiments was madrt The results are shown in Table T'II. T o obtain the>e dnts, alidcq nere suqpended in highly oxidi7ed oil heated to temperatures u p t o 188" C. under a pressure of 5 to 10 mm. of mercury. Aftrr cooling to room temprrature the slides viere found t o be covered n i t h a haid, adherent, dark brown coating apparently identical n i t h the varnish obtained by Fenykr (8) and other- upon labora-
Condition of Oil Oxidized Oxidized: filtered Oxidized BS filtered, 4- readded dudge Fresh. -1- added sludge
Undissolved Sludge, o/c 2.4 0.0
3.8 3.0
hol. Iarnish, &!-.?)'! ?LCCm. Slide I Slide 2 Av. 5.3 5.8 5.6 0.11 0.0 0.0
5.8 8.0
5.3 4.7
5.6
4.9
tory oxidation of lubricating oils. However, as soon MS all undissolved sludge was removed from the oxidized oil, this type of varnish formation ceased. Furthermore, a f w h oil to which some of this undissolved .dudge had been added shoived similar varnish formation under the ,same conditions. Thus it appea,rs that this type of varnish is undissolved sludge which has, a t least partially, gone back int.0 solution in the oil as the temperature was raised, and has deposited out again as an adherent coating upon cooling. I n support of this conclusion, much evidence has been accumulated which shows that the amount of this varnishlike coating is dependent upon the amount of undissolved sludge present in the oil, the solubility of the oxidized oil for sludge, the temperature t o which the oil has been heated, and the degree of cooling which is permitted before the slides are removed from the oil. If the slides are removed while the oil is still a t maximum temperature, in all cases varnish formation is negligible, and heavily coated slides will even lose a large fraction of the deposit if they are heated t o a high temperature in oil free from undis. solved sludge and removed from the oil a t this high temperature.
0
LOWEST OXID. RESISTANCE
25 50 % CHANGE IN VISCOSITY AT 210’F. I
I
I
I
2.5 5.0 WT. 96 TOTAL SLUDGE FORMED
75 ’
I
7.5
.
I
formation of any one product can fully represent the deterioration of any oil. However, for comparative purposes, within each class of oxidation criteria, the order is so nearly the same that one criterion can be selected as representa-
hydrocarbons ----) resins -+ oil-soluble sludge --+ oilinsoluble sludge. The molecular weights of the resins are usually TABLEVIII. THERMAL STABILITY OF PARTIALLY OXIDIZED about 500, with an oxygen content of 9 to 15%; the various OIL A HEATEDIN NITROGEN sludge fractions have molecular weights over 1000 and contain Sludge Wt. Q Viscosity over 20% oxygen (9, 14). These high molecular weights indicate TreatUndisPropine-insof Resins, a t 100°,F., ment, Hr. solved dissolved Total Wt. Centipoises considerable condensation and polymerization; hence, it seemed Charge 3.21 1.78 4.99 4.85 62.36 desirable to determine whether heat alone was sufficient to bring 8 3.44 1.94 5.38 4.56 62.23 3.15 1.99 5.14 4.82 63.04 about a change of resins t o sludge or, more probably, a con32 3.01 1.94 4.95 5.30 65.89 version of oil-soluble to oil-insoluble sludge. A charge of oxidized oil A was heated in the auto‘lave for 32 hours at 175” ’ and a pressure Of TABLE Ix. RELATIVEORDER O F OXIDATION RESISTANCE O F SIX OILS O N 2 mm. of mercury above atmospheric, with nitroPERCENTOXIDATION SCALE gen bubbling through it a t the rate of 0.5 liter per Experimental Values minute. Samples were taken after 8, 16, and 32 co2 Per Cent Oxidation Scale (Ordinate in Fig. 3) formed, change a t sludge, 4 hours. On these samples the undissolved and disOil wt. % 210 F., % ’ wt. % ’ factor factor factor index solved sludge, resin, and viscosity values were deA 0.359 16.2 5.40 54.5 79.5 31.0 55.0 termined (Table VIII). It is apparent that in C 0.094 12.4 1.50 86.5 88.1 81.0 85.2 D 0.357 21.2 4.16 54.8 73.0 47.0 58.3 an atmosphere of nitrogen the change in composiE 0.644 31.2. 3.63 17.9 60.0 53.8 43.9 55.3 6.26 tion of an oxidized oil is negligible a t 175” C. F 0.778 -2.0 29.5 20.0 15.8 7 -
I
0.325
10.5
1.72
58.5
87.0
78.1
74.5
OXIDATION FACTORS AND INDEX
Oxidation effects vary for different oils. ConLequently, much of the practical value of data which show these effects results from comparisons. It is not only important to measure as many as possible of the criteria indicative of oxidation, but also t o compare them as completely as posbible. The magnitude of such a comparison for even a few oils is obvioug from the preceding tables of oxidation data, with an attempt to keep in mind the comparable numerical
Figure 3 is a graphical representation of the oxidation scale. The conversion units, 0 t o 100, are used as ordinate with the three different scales for the experimental values as abscissas, The plot can be simplified by use of the same numerical abscissa values in multiples of 10 for the three oxidation factors. However, this graph does not imply any relation between per cent change in viscosity, total sludge formation, or per cent carbon
568
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
dioxide formed, since thew are independent variables. A straight line has been drawn between the ordinate value of 100 (the point where no oxidation can be detected) and a n abscissa point arbitrarily selected so t h a t the poorest experimental values found so far for each of the selected oxidation criteria will fall between 15 and 20. Although based on experimental data for oils of low oxidation resistance, this range can be widened or negative values can be used, if oil? of lower oxidation stability are found or more severe conditions of oxidation are used. Also, more extensive work on oxidation may make it desirable t o include other factors in the scale-for instance, a peroxide value. When the ordinate is used t o represent carbon dioxide formation, viscosity increase or sludge formation, the scale reading is called “COZ factor”, “viscosity factor”, or “sludge factor”, respectively. A calculated numerical mean of the three factorb is called “oxidation index”, expressed in per cent oxidation units. Table IX gives the experimental values of the selected oxidation criteria for six of the nine oils, together with the corresponding oxidation scale units.
Yol. 36, No. 6
LITERATURE CITED
Am. Soo. for Testing Materials, Designation I9 445-;39T, Method B. Balsbaugh, J. C., Larsen, R. G., and Oncley, J. L., INU.Ewo. CHEM.,30, 287 (1938).
Burk, R. E., Hughes, E. C., Scovill, W. E., and Bartleaon, J. D., Petroleum Div., B.C.S.,Atlantic City, 1941. Davis, L. L., Lincoln, B. H., Byrkit, G . D., and Jones, W. A , , IND. ENG.CHEM., 33, 339 (1941).
Denison, G. H., Jr., private communication, .June, 1943. Dornte, R. W., and Ferguson, C. V., IXD.ENG.CHEW.,28, 863 (1936).
Dorntc. R. W., Ferguaon, C. V., and Haskins, C. P.,Ibid., 28, 1342 (1936).
Fenske, M .R., Stevenson, C. E., Lawson, N. D., Herbolvheixnet G., and Koch, E. F., Ibid., 33, 516 (1941). Fenske, M. E., Stevenson, C. E., Rusk, R. A., Lawson, N. D., Cannon. &I. R., and Koch. E. F., IND.ENQ.CHEM.,ANAL ED., 13, 5 1 (1941).
b’uchs, G. H.
von,
and Diamond, H., IND.ENO.CHEM.,34, 927
(1942).
Fuchs, G, H. yon, Wilson, N. B., and Edlund, K. R., IND. ESG. CHEM.,ANAL.ED., 13, 306 (1941). Gruse, W. ih., and Livingstone, C . J., J.Inst. Petroleum, 26, 413 (1940).
Hall, F. W., Levin, H., and McMillan, W. A., IND.EXG.CHEW., ANAL.ED., 11, 183 (1939).
ACKN0WLEP)CMENT
Appreciation is expressed t o S. S. Kurtz, Jr., for many helpful ‘suggestions and t o Herbert L. Johnson and other associates in the laboratory for assistance in this work.
Viscosities and
Haus, E., Oel KohEe Erdoel Teer, 14 (E)299, , 321 (1938). Hirsohler, A . E., Petroleum Div., A.C.S., Mcmphis, 1942. Lamb, G. G., Loane, C. M.,and Gaynor, J. W., IND. ENO. CHEM.,ANAL.ED., 13, 317 (1941).
Rogers, ?’ H., and Shoemaker, B. H., Ibid., 6,419 (1934).
ensities of
HYDROGEN
S
HELNIUT WAKEHAM AND FRANK C. MAGNE Southern Regional Rmearch Laboratory, U. S. Department, of Agriculture, New Orleans, LEI. little information has appeared in the literanumber values, free fatty acid contents, melting and solidification ture on the viscosities of cottonseed oils or hydrogenated ranges, and refractive indices of the oils are shown in Table 1. cottonseed oils. Strevens ( 7 ) published the value of Iodine numbers were obtained by the wijs method. The melt. 0.994 poise a t 15.5’ C. for a pale cottonseed oil which was identiing and solidification ranges were determined on samples cooled fied only by B density of 0.925 a t 60” F.;he pointed out t h a t for in sealed capillaries 24 hours before measurement in a melting a wide variety of oils the greater the iodine number, the lower the viscosity at a given temperature. Rawitsch (6) and Boekenoogen ( 2 ) measured the TABLE 1. PROPERTIES OF COTTONSEED OILS USED IN \‘ISCOSlTY AND DENSITY viscosities of unhydrogenated cottonseed oils ovei DETERMINATIONS the temperature range between 20’ and 90’ C., Free Fatty Acid Calcd. as % and Bauer and Markley ( 1 ) reported the viscosiOleic ties a t 98.9” C. of hydiogenated cottonseed oils Itefraotive Before AfterIodine measmeashaving iodine numbers ranging from 6.8 to 58.9. No. Melting Solidification -&!!Z-ureureSample No, (Wijy) Range, 6. Range, C. C. n~ ment ment As yet, however, no complete study has been pub-6.7 to -2.4 -4.4tO - 9 . 6 40 0.06 10 112 1.4658 0.08 lished of the variations of viscosity and density 40 1.4651 20 - 5 . 6 to f 4 . 2 - 4 . 4 to - 9 . 9 0.09 0.10 108 with iodine number and temperature for cotton40 - 2 . 5 t o -8 1.4656 0 . 0 3 - 3 . 0 to f 1 . 9 0.26 101 40 1 9 . 6 t o 14.0 1.4619 0 . 0 5 0.27 16.8to 23.5 78 seed oil. Increased processing and utilization of 50 3 3 . 3 t o 28.1 1.4568 5-A d 3 2 . 4 t o 39.1 0.08 0.11 66 40 6-Bd 3 5 . 5 t o 27.6 1 .4604 0 . 0 8 0.09 32.2to 41.6 65 domestic vegetable oils have emphasized the need 50 3 4 . 8 t o 27.4 1.4566 7-Bd 35.4to 41.0 0.09 0.15 65 for such data which may be used in the design of 43 30.7to 27.8 1.4590 0.05 0.09 8-C l d 66 50 0.09 0.06 3 2 1.9 6to 4 3 9 2 . 7 5 3 3 . 9 t o 2 7 . 6 1.4567 9-C ld 68 processing equipment. 50 1.4555 32.5to 28.4 0.05 0.09 3 5 . 6 t o 42.3 10-C 2d 61 49 1.4558 32.2to 29.1 0.09 0.08 3 4 . 7 t o 42.9 i l - C 2d 61 Samples of hydrogenated cottonseed oils were 40 3 4 . 8 t o 31.6 1.4589 0.08 0.42 3 7 . 0 t o 42.3 12 c 56 prepared in this laboratory from a commercial re0.74 60 50.8to 48.2 52.9to 57.1 1 ,4490 0.17 I3 28 5 3 . 3 t o 5 1 . 2 1.23 6 0 . 6 t o 6 1 . 6 .... 14 6 fined and deodorized cottonseed oil. Properties of 0 Comniercially refined oils from two different refineries of Southern Cotton Oil Co. these.oils were compared with those of commercially 6 U.S.P.grade sold under the label of B . R. Elk & Co., Inc. refined unhydrogenated cottonseed oils from three Hydrogenstid in Southern Regional Research Lab. from a refined oil of 102 iodine number. different sources. These oil samples were suppled Commercially hydrogenated by Southern Cotton Oil Co. A, B , and C refer t o batchee A, B , and C, respectiveiy. Batch C was hydrogenated t o two different iodine numbers. mented with commercially hydrogenated oils of routine batches obtained from a local processor Iodine O
2:
C
C
O