The effect of a petroleum lubricant in reducing the static coeficient of friction in a bearing appears to be due primarily to the presence of a small amount of some substance that is .firmly adsorbed on the surface of the bearing to form a film of high lubricating power. This film does not form instantly on a freshly oiled surface; a n appreciable length of time elupses before equilibrium is established. The constituent that is primarily responsible for the lubricating power may be removed from the oil by adsorption on finely divided metalfor example, on powdered Wood‘s metal. I t is retained so tenaciously by the metal that attempts to recover it in a relatively pure form have not been successful. When a petroleum lubricant is heated in the presence of air, a marked loss of lubricating power occurs at about 75’ C. This change appears to be the result of oxidation. It may be inhibited by adding small amounts of certain substances to the oil. Cyclohexanol and /3naphthol are among the addition agents that show this effect. These particular addition agents also improve the lubricating power of the oil at ordinary temperatures.
Factors Affecting Lubricating Properties of a Petroleum Oil F. H. RHODESAND ARTHURW. LEWIS Cornel1 University, Ithaca, N. Y.
W
ILSON and Barnard (3) have shown-that the lubricating power of a petroleum oil, as measured by the effect of the oil on the coefficients of static friction between metallic surfaces to which it has been applied, is due largely to the presence of some component that is very firmly adsorbed by the metal. This as yet unidentified constituent appears to be present in very small amounts only. The investigation described in the present article was undertaken for a twofold purpose: (1) to isolate and to identify the component that is primarily responsible for the “oiliness” of petroleum lubricants; and (2) to discover, if possible, other substances that, when added to an oil, will have a similar effect in improving the lubricating properties. The first of these problems has not yet been solved; the second objective has been attained. MEASUREMENT OF
THE
COEFFICIENT OF STATIC FRICTION bearing the polished lugs were boiled in absolute alcohol, rinsed
The method used for measuring the coefficient of static friction was similar to that described by Rhodes and Allen (2): It consisted essentially of a light rider resting at three points flat metallic surface lubricated by the oil that was being examined, which surface could be tilted slowly until slipping of the rider occurred. The tangent of the angle of inclination at which the rider began t o slide was taken as a measure of the coefficient of static friction. The rider itself consisted of a triangular plate of sheet aluminum, in each corner of which was mounted R polished-steel ball hearing that served as a point of support. The bed plate on which the rider rested was a brass block in which were set three chromium-plated lugs so spaced that each bearing of the rider rested on a separate lug. The flat u p p s faces of the lugs were in the same plane so as to give, in effect, horizontal plane surface. The plate carrying the bearing was mounted on a frame which was hinged at one end and was so arranged that it could be tilted slowly. The bearing block was provided with electrical heating coils for varying and controlling the temperature and was enclosed in a chamber through which either dry air or dry nitrogen could be circulated. on a
With this apparatus it is possible to obtain results that are consistent to within 3 per-cent. For example, in two determinations made with the same oil the following results were obtained: Temp.
RUN 1 Coefficient of friction
c. 24.5 33.5 50 65 81 88.5 90
Temp. c.
RUN 2 Coefficient of friction
O
0.1595 0.156 0.1605 0.157 0.157 0.172 0.177
24.5 33.5 52 io 83 90
0.159 0.159 0.162 0.160 0.164 0.174
with absolute alcohol, and dried in a vacuum desiccator over phosphorus pentoxide before being used in making a determination of the coefficient of friction. The absolute alcohol used had to be carefully purified, since certain impurities, when present even in small amounts, are adsorbed on the metal and decrease the accuracy of the results obtained. In each series of tests the apparatus was allowed to stand for at least a n hour after the fresh oil had been applied to the bearings before the determination of the frictional coefficient was made. The adsorbed film, to the presence of which the lubrication is due, forms rather slowly so that, if the coefficient is determined very soon after the oil is applied, erroneous results may be obtained (1). -4sample of typical petroleum oil gave the following results : TIXEAFTER OIL W A S
APPLIED Mtnutes 5 9 15 18 24
COEFFICIENT
TIME4FTER
OF
OIL W A S
COEFFICIENT OF
0.1535 0.160 0.169 0.211 0.250
Mznutes 30 34 38 50 56
0.242 0.173 0.167 0.151 0.152
FRICTION
APPLIED
FRICTION
It was also observed that, in making a series of determinations with the same sample of oil, less consistent results are obtained if the individual measurements are taken a t intervals of less than 3 minutes. Apparently a short time is required for the film to be restored to its equilibrium condition after the slight disruption caused by the motion of the bearing.
SELECTIVE ADSORPTIONOF LUBRICATING COMPOUNDS ON METALS
It was observed, however, that in order to obtain consistent results, the bearing surfaces had t o be cleaned meticulously before the lubricant was applied. Both the rider and the block
The lubricating oil used in these experiments was a commercial oil of “medium” grade, prepared from pure Pennsylvania crude by the usual procesi of steam-distillation followed by
1011
INDUSTRIAL AND ENGINEERING CHEMISTRY
1012
chilling and filtration to remove wax. ing characteristics: Density at 20' C 0.88 Carbon residue (6onradson), 0.81 % TEMP.
c.
30.6
Sayboll seconds 651 445 205.4 64
87
37.8
100
130 210
98.9
VIBCOBITY
'F.
54.4
It showed the follow-
Very finely divided Wood's metal was prepared by atomizing the molten alloy (melting point, 65" C.) with a jet of steam and collecting the mist in cold water. The resulting powder was dried and passed through a 150-mesh sieve. A
Vol. 26, No. 9
portion of the metal, the clear oil showed a coefficient of friction of 0.1601. The recovered iron was extracted with warm benzene, and from the extract the benzene was removed by distillation in a vacuum. The oil thus recovered from the iron showed a frictional coefficient of 0.1508. These results are qualitatively similar to those obtained with Wood's metal. Any component of especially high lubricating power that was adsorbed by the metal was so firmly retained that it was not removed even by extraction with benzene.
EFFECTS OF TEMPERATURE AND OF
THE
PRESENCE
OXYGEN Two series of experiments were made in which the coefficients of static friction of the lubricating oil at different temperatures were measured. In one set of measurements the gas passed through the apparatus was dry air; in the other, dry nitrogen. The results are shown in Figure 1. In each series the coefficient remained about constant until a temperature of around 75" C. was reached. Above that temperature the sample exposed to air showed an increase in the frictional coefficient, gradual a t first and then more and more rapid. I n nitrogen the coefficient of friction decreased somewhat at 0 the higher temperatures. Temperature 'C. A fresh sample of oil was tested in a current of dry air. FIGURE1. CHANGEIN COEFFICIENT OF STATIC FRICTION The temperature was raised gradually to 99' C. and was ON HEATING then decreased to 34" C. The coefficient of friction inA. In dry air B. In dry nitrogen creased at the higher temperatures but decreased again only portion of the lubricating oil was shaken with an equal slightly when the temperature was again lowered (Figure 2).] weight of the metal and centrifugalized to remove the solid material. The coefficient of static friction of the original oil was 0.149; that of the oil which had been treated with the Wood's metal was 0.1852. After standing for several days exposed to the air, the treated oil showed a coefficient of 0.155. In a second experiment 50 grams of an oil that had a coefficient of friction of 0.141 was treated with two successive portions (50 grams each) of Wood's metal. The treated oil showed a coefficient of 0.189; on standing for 7 days exposed to the air, the coefficient decreased to 0.141. The metal removes a small amount of some substance that is primarily responsible for the lubricating power of the oil; this material is slowly regenerated when the oil stands in the air a t room temperaFIGURE 2. EFFECT OF HEATING AND SU~SEQUENT COOLING, ture. IN AIR,ON COEFFICIENT OF STATIC FRICTION Several portions of powdered Wood's metal that had been A . Results obtained during heating shaken with separate portions of fresh oil were combined and B. Results obtained during cooling dropped into boiling water. The metal melted but did not These experiments indicate that the decrease in lubricating coalesce, even on long-continued heating and stirring. The droplets remained separate, like globules of dirty mercury. power that occurs a t the higher temperatures is caused by Only a part of the adhering oil was liberated to float on the the oxidation of some constituent of the oil. They do not water. The liberated oil showed a coefficient of friction of indicate whether the decrease is due to the destruction of the 0.171, only slightly less than that of the main portion of the material that is primarily responsible for the lubricating power oil that had been separated from the metal by centrifugalizing. or to the formation of some substance that has a marked The experimental evidence indicates that the material which is effect in increasing the frictional coefficient. most effective in increasing the lubricating power of the oil is EFFECTS OF PRELIMINARY OXIDATION A N D OF TREATso firmly adsorbed on the finely divided metal that it is reMENT WITH ALKALIAND ACID tained even when the metal is melted. The presence of If the decrease in the coefficient of static friction that this film of absorbed oil on the surface is apparently responoccurs when the oil is heated in air is due simply to oxidation, sible for the failure of the metal to agglomerate. Attempts were made to recover the adsorbed oil from it should be more pronounced when the oil has been subjected the Wood's metal by heating gently with dilute hydrochloric to a preliminary oxidation-for example, to exposure to oxyacid. Even in the presence of the acid, however, the ag- gen under ultraviolet light. A sample of the original oil glomeration of the finely divided metal was very incomplete. through which oxygen was being passed was exposed for 2 The recovered oil, after washing to remove traces of acid, hours, in a thin flat quartz vessel, to the light from a mercury lamp (Hanovia). The coefficients of friction of the resulting showed a coefficient of friction of 0.175. Somewhat analogous results were obtained when the origi- oil a t various temperatures are shown in Figure 3. At comnal oil was treated with finely divided iron powder. One paratively low temperatures the lubricating power of the hundred grams of the oil were shaken with 20 grams of the treated oil was not much inferior to that of the original iron and allowed to stand for 12 hours in contact with the material; a t higher temperatures the coefficient of static metal. The clear oil was then separated and again treated friction increased very rapidly. These results indicate with 20 grams of the iron. After the removal of the second that the decrease in lubricating power that takes place when
-
OF
September, 1934
INDUSTRIAL AND ENGINEERING CHEMISTRY
the oil is heated in air is not due directly to the primary products of the oxidation but is to be attributed to the presence of substances formed by the polymerization or other secondary reactions of these primary products. A portion of fresh oil was washed with a concentrated solution of sodium hydroxide, then with dilute sulfuric acid, and finally with water, The final clear neutral oil was dried in a vacuum. The coefficients of friction of the washed oil a r e s h o w n i n Figure 3. The washing treatment improves the lubricating p r o p e r t i e s but does not eliminate the rise in the frictional coefficient a t moderately high temperatures. A p o r t i o n of t h e washed oil that had subsequently b e e n exposed to ultraviolet light for 1.5 20 a 60 80 /a, /a h o u r s in the presence of air was someTemb. - 'C what similar in its FIGURE3. EFFECTOF OXIDATION lubricating properUNDER ULTRAVIOLET LIGHTAND OF ties to the original TREATMENT OF ALJIALI A . Original oil treated with 'oxygen under Oil, the inultraviolet light c r e a s e in the fricB. Original oil washed with alkali and subsequently treated with air under ultrat i 0 d CoefficientOCviolet light a t a 'Omem C. Origina! oil washed with alkali and acid D. Original oil untreated what lower temperature (Figure 3). A portion of the original oil through which oxygen was being blown was exposed to the ultraviolet light for 4 hours and was then shaken with a 30 per cent solution of sodium hydroxide, washed, neutralized, and dried. The resulting oil showed a coefficient of friction of 0.177 a t 76" C. and 0.169 a t 94' C. This experiment indicates that the product that is formed when the oil is oxidized in the presence of ultraviolet light and that is assumed to be responsible for the reduction in lubricating power is largely removed when the treated oil is extracted with alkali.
1013
first a decrease and then an increase in the frictional coefficient. The final rise in the coefficient begins only a t temperatures much higher than that at which the rapid loss of lubricating power of untreated oil occurs. Cyclohexanol, @-naphthol, p-cresol, and a-naphthylamine show the most marked effects on the lubricating characteristics. With each of these addition agents the coefficient diminishes rapidly as the temperature rises until i t becomes approximately onehalf as great as a t room temperature. At still higher teniperatures the coefficient again rises rapidly, but this final increase does not occur until relatively high temperatures are reached. Menthol, thymol, and 1,5-dihydroxynaphthalerie have a similar but less pronounced influence on the lubricating power. The effect of the variation of concentration of an active addition agent is depicted in Figure 4 where are plotted the results obtained with oils containing 0.001, 0.01, and 0.05 per cent, respectively, of @-naphthol. Even in extremely low concentrations this substance has an appreciable influence in improving the lubricating power a t moderate temperatures and in retarding the rise in the frictional coefficient a t higher temperatures. With increasing concentrations, these effects become more and more marked. I n general, the addition agents most suitable for commercial use in improving lubricating oil appear to be @-naphthol and cyclohexanol. Both of these substances reduce considerably the frictional coefficient a t ordinary temperatures as well as a t elevated temperatures; both are relatively .I8 P
7ernB.h OC. THE LUBRICATING FIGURE 4. EFFECT OF NAPHTHOL PROPERTIES A . @-Naphtholconcentration = 0.001 per cent The effects of small amounts of various addition agents 0.01 per cent B . @-Naphtholconcentration C. &Naphthol concentration = 0.05 per cent upon the coefficient of static friction of the lubricating oil and upon the variation of the coefficient with temperature stable and nonvolatile; neither should have any corrosiw were measured: action on the metal of the bearings. Both of these materials SUBSTANC~ CONCN. SWBBTANCE CONCN. have been used by the writers as addition agents in auto% % mobile oils with satisfactory results. Oil containing cyclop-Creaol 0.08 p-Hydroxydiphenyl Satd. hexanol has been used in the lubrication of a ruling engine in P-Naph t hol 0.05 m-Dimeth laminophenol Satd. @-Naph th ol 0.05 p-Aminop Keno1 Satd. which the bearings operate a t heavy load and at very low @-Naphthol 0.01 1.5-Dihydroxynaphthalene Satd. @-Naphthol 0.001 Menthol 0.05 speed; the use of this treated oil eliminates the chattering and Thymol 0.05 a-Naphthylamine 0.05 sticking that occurs when a similar oil without addition Cyclohexanol 0.05 agent is employed. Those substances for which no exact value of the concentraThe exact cause of the beneficial effects of these addition tion is stated were not soluble to the extent of 1 part in 200 agents is not obvious. Those substances that decrease the parts of the oil. With these compounds, saturated solutions static coefficient a t low and moderate temperatures appear to were used. owe a t least a part of their effect to the absorption of the The results obtained with oil containing these substances added material on the surface of the bearing, with the formaare shown by Figures 4, 5 , and 6. Of the various addition tion of a film of high lubricating power. This assumption is agents tested, a-naphthol has the least effect. p-Hydroxydi- consistent with the fact that these substances are polar comphenyl, m-dimethylaminophenol, and p-aminophenol have pounds that should be adsorbed rather strongly on metal. relatively little influence on the frictional coefficient. With Minor differences in structure, however, may greatly ineach of these substances an increase in temperature causes fluence either the extent of adsorption or the lubricating
EFFECTS OF ADDITION AGENTSON
-
1014
INDUSTRIAL AND ENGINEERING CHEMISTRY
.20
i l 1
0 ~
.I8r
.I6
I
Vol. 26, No. 9
c .?,I6 .w
.I4
.-0 2.14
.z .I2 LL c 0
3.12
3 .IO
u:
u:
$0
u8.08
Q
.oaL
2 0 40 60 80 100,120 140 160 Temperature - C.
FIGURE 5 . EFFECT OF ADDITIONAGENTSON COEFFICIENT OF STATIC FRICTION A. B.
C.
Menthol m-Dimethylaminophenol p.4minophenol
I
.O 6
20 40 60 80 100 J20 140 160
Temperature - C. FIGURE 6. EFFECTSOF ADDITIONAGENTSON COEFFICIENT OF STATIC FRICTION A. p-Hydroxydiphenyl E . p-Cresol
D.
u-Naphthol
F.
1,5-Dihydroxynaphthalene
E. Thymol
properties of the adsorbed film, or both. For instance, a-naphthol has very little beneficial effect while P-naphthol very greatly increases the lubricating power both at room temperature and a t elevated temperatures. The inhibition of the decrease in “oiliness” on heating may be due to the action of the addition agents as negative catalysts in the oxidation of the oil; this hypothesis is supported by the fact that most of the substances that have beneficial effects are known to act as antioxygens. The use, in lubricating oils, of addition agents that lower the coefficient of static friction a t ordinary temperatures and prevent the rise in the frictional coefficient a t higher temperatures probably has very considerable commercial advantage. The addition of these substances in the small amounts required to improve the lubricating characteristics of the oil would not add appreciably to the cost of manufacture of the lubricant nor should it in any way affect unfavorably the other characteristics of the oil. The use of oil containing
C. a-Naphthylamine D . Cyolohexanol
such addition agents should be of particular advantage in the lubrication of slow-moving machine parts or of bearings on which it is difficult to maintain continuously an excess of oil, as well as on bearings that are operated intermittently and are started and stopped frequently. The advantage would, perhaps, be less evident when the lubricant is used on bearings that are operated at high speed and with flood lubrication, although even under such conditions the addition to the oil of a substance that increases its lubricating power should be of some benefit. LITERATURE CITED (1) Hardy, chapter on “Friction, Surface Energy, and Lubrication” in Alexander’s “Colloid Chemistry,” Vol. I. Chemical Catalog Co., N. Y . , 192G. (2) Rhodes and Allen, IND.EXG.CHEM.,25, 1275 f1933). ( 3 ) Wilson and Barnard, Ibid., 14, 653 (1922). RECEIVED April 9, 1934.
The ketone was purified by distillation under reduced pressure and by two recrystallizations from ethanol. The melting point was 68” to 69” C. best yields of laurone reported in the literature are 10 SHERLOCK SWANN, JR., E. G. APPEL,AND S. S. KISTLER to The 30 per cent from lauric acid over thoria a t 400” C. (S), from lauric acid with phosphorous pentoxide ( 2 ) , and 91 per University of Illinois, Urbana, Ill. cent by heating small quantities of the acid in an iron dish N T H E first part of this communication’ it was shown that for about 4 hours. The preparation of laurone described in ketones could be prepared in excellent yield by distilling this paper gives better yields than the first two methods and the corresponding acid over thoria akrogel a t atmospheric produces laurone a t a much higher rate than the third. Besides laurone, undecylenone, CH, = CH(CH,),COor under reduced pressures. The size of the ketone which (CH2)?,CH=CH2 (melting point, 43” C.) was prepared from can be conveniently prepared is limited by the volatility of the acid and ketone. In order to extend the scope of the ethyl undecylenate in 86 per cent yield. This is a new method to higher ketones, the conversion of ethyl esters was ketone. The microanalysis in per cent is as follows: calcustudied. The esters have a much lower boiling point than lated for C2iH310: C = 82.30, H = 12.42; found: C = their corresponding acids and, therefore, may often be em- 82.39, H = 12.47. This method may be used successfully to prepare in large ployed when the use of the free acid is not feasible. quantities aliphatic ketones of high molecular weight which Ethyl laurate was chosen as a typical example. The results of its conversion to laurone are shown in the following table: were formerly obtained only with difficulty. The microanalysis was carried out by K. Eder of the RATE TEMP. ESTBR Chemistry Department of the University of Illinois. RUN OVER OF RECON-
Thoria Aerogel Catalyst: Aliphatic Esters t o Ketones
I
No. 1
2 3 4 (1
ESTER“ CATALYSTCATALYST Grams 57 57 57 57
Grarna/min. 3-4 3-4 4-6 8
The ethyl laurate was
25 to 30 mm. pressure. 1
COVERED
KETONE
VERSION
Grams % 20.0 27 61.8 92.5 4.0 39 10.0 34.8 82.5 17.0 29.6 70.0 distilled over the catalyst a t 150’ t o 160’ C .at
C. 300 360 360 360
IND. ENQ.CHEM., 26, 388-91 (1934).
LITERATURE CITED
Grams
(1) Gruen, A,, Ulbrich, E., and Krczil, F., 2.angew. Chent., 39, 421 (1926). (2) Kipping, F. S., J . Chem. SOC.,57, 981 (1590). 13) Pickard, R. H., and Kenyon, J., Ibid., 99, 57 (1911). RECEIVED June 9,1934
