V O L U M E 21, N O . 1 2 , D E C E M B E R 1 9 4 9 Table 1.
Representative R e s u l t s
__ Sample A B
C D E F
Vapor Pressure, L h J S q u a r e Inch Rpid method New method 4.3 4.7 4.3 4.4 6.0 6 3 5.9 6.1 7.6 7.9 7.5 7.3 9.6 9.8 9.5 9.4 14.0 14.3 14.1 14.3 18.2 18.5 18.0 18.0
I eading. Inasmuch as air d u b i l i t y decreases with increasing gasoline vapor pressure, this error decreases accordingly. Incwasing the vapor-liquid ratio in the system decreases the error from this source. In the apparatus as shown, in which the vaporliquid ratio is about 20 to 1, the positive error is estimated to be ahout 0.1 pound per square inch for a sample of 7 pounds per square inch vapor pressure saturated with air a t 40’ F. Though for greater precision narrow limits on the vapor-liquid ratio are required (2, s),the use of the 20 to 1 ratio yielded results which uere close enough to those of the ilmerican Society for Testing llaterials, which uses 4 to 1, to be of practical utility. Effect of Vapor-Liquid Ratio. -1negative error in the indicated reading may be caused by a large vapor-liquid ratio. Such an error is due to change in liquid composition resulting from evaporation effects and does not occur for purr compounds 01 constant boiling materials. I t is greatest for samples showing the greatest slope in the front end of the distillation curve and increases with increasing vapor pressure. The Reid method is SUI)ject to this same error, though to a slightly lesser degree
1485 In the present method the influences of dissolved air and of high vapor-liquid ratio counteract each other to some degree. The e\perimental evidence to date indicates that results by this method can be considered equivalent to the Reid vapor pressure. .4lthough both are subject to the errors discussed, they are satisfactory for the usual gasoline vapor pressure control. With the new method it would be possible, if desired, to conduct the test with a zero vapor space and to apply a correction for dissolved air, so that true vapor pressures (9) could he ohtained. ACKNOWLEDGMENT
The author is indebted to Minor C. K. Jones for his guidancr and helpful suggestions throughout this work. LITER4TURE CITE11
(1) Am. Sac. Testing Materials, “Standards,” Part 111, General, p.
637, 1939. (2) Am. Sac. Testing Materials, ”Standards on Petroleum Prod-
ucts,” Committee D 2-1948, p. 222. (3) Barber and Ritchie, ISD. Esa. CHEY., AS.^. ED., 8, 472
(1936). (4) Davis, Ind. Eny. C‘hem., 17, 1136-8 (1925). ( 5 ) Derby and Ilngve. J . A m . Chem. Soc., 35, 1439 (1916). (6) Edgar and Swan, Zbid., 44, 570 (1922). (7) Francis, ISD.ESG.CHEY,,A N - ~ LED., . 1, 38-9 (1929). (8) Menzies, J . Am. Chem. Soc., 42, 2218 (1920). (9) Robertson, A. E., and .Ilbright, R . , S B E Journal, Trans., 52, 466 (1944). (10) Smith and Nenries, J . Am. Chem. ,SIX., 32, 1412 (1910). KECEIYEI)F’ehruarp 11, 1947.
Spectrophotometric Evaluation of Engine lubricants (;.
1,. C:1,4RK AND \I. H. MUELLER, University of Illinois, Crbana, I l l . , AND
T. W. CL LIIER, Ohio Oil Company, Robinson, 111.
I
K WARTIME irivestigatioiii
( 1 ) it was found that the performance of quenching oils for aluminum alloy airplane motor vastings could be determined by means of spectrophotometric curves of the oili, which could be correlated accurately with residual strain in the alloy castings. Increasing deterioration, easily detected in the absorption measurements, resulted in increasing strain, which seriously affected load and safety in motors. The method was adopted a5 a standard test for predicting the pertorniance of new quenching oils, and for working out a procedure tor fortifying such oils by uye of additives and filtration of old dettlriorated oils, resulting in a minimum strain development in castings. It was l o g i d to extend this optical testing method t o wgine lubricants in the hope of simplifying the complex procedure of ten or a dozen tc.st3, not alwavs satisfactorv, adopted aQ standard in the petroleum inciustir.
photometric study. The transmission data, over the complete visible range of the spectrum, were obtained with one of several spectrophotometers such as the Cenco-Sheard, Coleman, or General Electric recording instruments. Oils have a large absorption in the near ultraviolet region and often became dark upon use; therefore it is necessary to dilute the samples with a nwtral solvent in order to obtain maximum sensi-
loo#-
,
I
SAMPLES AhD TECHNIQUE
Through the cooperation of several oil laboratories it was possible to investigate a number of different oils operating under various conditions. Some of the oils were tested under laboratory {aonditions by means of an Vnderwood tester and others were tested under actual operating conditions in automobile and Diesel cngines. After a number of the more familiar and accepted tests of the petroleum labnratorieg, these samples were used for w Spectrn-
Figure 1.
Effect of Dilutions w i t h Chloroforni
ANALYTICAL CHEMISTRY
1486
Wartime investigations showed that the performance of quenching oils for aluminum alloy airplane motor castings could be determined by spectrophotometric curves of the oils wrhich could be correlated accurately with residual strain in the castings. This optical testing method was extended to engine lubricating oils in the hope of simplifying the standard petroleum tests. Some oils were tested under laboratory conditions by an Underwood tester, others under actual operating conditions in automobile and Diesel engines. Performance of new oils may often be predicted from samples prepared by heating in open test tubes in the presence of metal turnings (corresponding to engine walls). The striking similarity between the graphs of optical data and those for such tests as per cent asphaltenes, carbon residues, neutralization number,
tivity of the transmission readings. In the present study, chloroform was found to be a satisfactory solvent. It was possible, in cases where the oil remained comparatively clear, to run spwtrophotometer curves with no diluent. Dilution studies were carried out with the same result as with queriching oils-a shift of the absorption peak toward the shorter n-ave lengths with greater dilution. If t,he same oil sample is diluted in different proportions with chloroform, the spectrophotoniet,ric curves in Figure 1 are obtained. The first requirement, of this method of evaluating lubricating oils is to &elect it desirable diluting ratio.
viscosit!, chloroform-solubles and insolubles, etc., suggests a definite relationship between optical properties and chemical and physical data indicating relative amounts of oxidation, pol? merization, and other deterioration with use under many conditions of temperature and catalg zing materials. Thus the spectrophotometric method offers a rapid means of rating engine lubricating oils with and in advance of their use. Ebidently the change in light absorption is an integratiFe iudication of chemical and physical changes, including development of colloidal sludge. which occur under the conditions of oxidation and heat in an internal combustion engine. The iudeterminate \ariables in engine operation, such as contamination hj incomplete11 Imrned fuel products, carbon deposits, etc., limit application of this or an? other method of used oil anal! sic.
1.5
5 0.5
0.0
0
1000
2000
3000
4000
5000
MILES
Figure 3. Log of Reciprocal of Minimum Transmission on Spectrophotometric Curves Plotted against Miles of Operation in an Automobile Engine for Four Familiar Commercial Lubricating Oils
EXPERIMENTAL
The results obtained with four different samples of oil tested in an automobile engine are shown in the first series of curves. Transmission curves and the usual petroleum tests were made on samples of the oil after a definite number of miles of service. 100,
,
I
tion, and (3) n decrease in the relative transmission. Figwe 3 illustrat~esan effective way to show the change in transmittancy by plotting the logarithm of the reciprocal of the minimum transmission read off from the spectrophotometer curves against miles of service in an automobile engine. Samples of the same four oils were also teated under laboratory conditions in nn Undern-ood tester. In this method the oil is heated at an elevated temperature and sprayed over a metal surface, which produces more rapid deterioration than under service conditions. AX comparison of Figures 2 2nd 4 shon-s a good
.^^ z
u1
320
Figure 2.
520
620
1
420 MILLIMICRONS Transmission Curves for Oil 1 after Use i n Automobile Engine
Typical curves of the relative transmission versus wave length are shown in Figure 2 for oil 1 after various periods of service. These curves were obtained with a Cenco-Sheard instrument using 1 to 200 dilution. An examination of the curves in this graph reveals several trends after the oil has been in use. There are (1) a definite shift toward higher wave lengths a t minimum transmission, (2) a general rounding out of curve at maximum ahsorp-
z $50.. I-
c z w
a 0 320
420
Figure 4.
MILLIMICRONS 520
620
Transmission Curves for Oil 1 after Underwood Test
Compare with Figure 2 for actual engine test
V O L U M E 21, NO. 1 2 , D E C E M B E R 1 9 4 9 2.0
Figure .5.
1487
I
I
residues itnd cwrrosion of cadmium-silver and copper-lead hewings, are shown in Figures 9 to 11. The striking similarity betxeen the graphs of the optical data derived from spectrophotometer curves and those for these familial tests indicates a definite relationship between optical properties and these other chemical and physical data which are generally accepted as measures of relative amounts of oxidation, polymerization, sludge formation, acid formation, viscosity changes. and other deterioration with use under many conditions of temperature and catalyzing materials. The decreasing rating of the oils from 1 to 4 in order indicated by the spectrophotometer (Figures 3 and 5 ) is also generally proved by the standard tests (Figures 6 to 11). Where there is confusion in some standard tests, such a i the insoluhles, the optical data also indicate unre-
Reciprocal of Minimum Transniission Plotted against Hours in Underwood Test
Compare with Figure 3 for same four oils i n a c t u a l rnginr uer
correlation betwevii the transmission curves o l i t i i i i i t d froin oil used in an engine and thiy accelerated method of testing the oil in the laboratory. The data used for the graph i n Figure 5 were obtained from the samples of oil used in the Untie~~n-ood.The .same trend is evident as in Figure 3. On the b photometer tests it niny he predicted that theje four commercial oil$ decreave in stability m d performance in the order 1 to 4. Figurrs 6 to 8 d10w the results of standard petrolpuni tests of the four oils after testing in a service car: per c.riit rhloroformsoluble, per cent c hloroform-insoluble, per writ ri:iphtha-insoluIJIe) per cent asphalteries, neutralization nunilx~r.:tnd viscosity. Similar curves ol)tained from the oils after te4ting i l l the lahorator!- under acwle.r.:i?td conditions. together with t1rit:i on carhon
I
2000
1000 7
i
3000 MILES
5000
4000
'It P
/I
I
I !
/
,-.
1
1000
2000 MILES
.In
l0Oc 0.6
r
2nm
Figure 6.
4000
5000
I
MILES
3000
4000
5002
-
1 2000 3000
0.0 loo0
3000
MILES
4000
5000
Results of Standard Petroleum Tests on Four Oils
I
c
I
,
1000
Figure 7.
I
2000
3000 MILES
Results of Standard I'etroleuni Tests on Four Oils
-
ANALYTICAL CHEMISTRY
1488
Underwood test in indicating performance of lubricating oils in automobiles under controlled conditions. The significance of development of solid particles and then growth with aging, as determined by electron micrographs for quenching oils, is substantiated for lubricating oils. Therefore, this simple rapid spectrophotometric technique is recommended as a valuable supplement or even replacement to other test procedures in petrolrum laboratories. Oil used in Diesel engines is subject to more fuel dilution than in gasoline engines. However, the curves in Figure 12, obtained from samples of oil 5 , which had been in service for a various nuniher of in a Diesel locomotive~show the Same genera' Figure 8. Viscosities Plotted against 3Iiles of Service for Four Oil8 trends which xere evident in Figures 2 and 4. T h r curves on these Diesel oils were obtained on a liability vf these accepted methods, which are generally recugnized (:erieral Electric recording instrument using a 1 to 20 dilution n ith chloroform. as troublesome and unreproducible. There is a strong indication -1w i p s of tests n a s made also on stationary Iliesel eiigirws. that the slope of the optical curves is a remarkably accurate index of the stability of the oil: the greater the slope the less the stahility and the greater the deterioration. These curves also contribute added confidence in the reliahilitv of the accekrated I I I I
v,
w
z
W
-
W
I-
$0.5 v)
a
QO 0
d; A3-1
5
10
15
20
* 0.00
5
10
15
20
10 HOURS HEATED
15
20
-r
'
0'01 0
Figure 9.
5
Results of Laboratory Tests of Three Oils
Figure 10. Results of Laboratory Tests of Three Oils
V O L U M E 21, NO. 1 2 , D E C E M B E R 1 9 4 9
1489 riot too surprising, because of the large oil capacity of the engine. The results of some of the usual petroleum tests made on these samples are given in Table I. Oil 7 was used in two different stationary Diesel engines. The presence of additives in this oil had no appreciable effect on the transmission curve, as shown by making a curve of the base oil without and then with the additive. The curves shown in Figure 13 (lower), plotted from data obtained with a 1 to 20 chloroform dilution and a Coleman spectrophotometer, are as follows: Hours of Operation
Curve 50.
I
Xew oil with additive 37, engine 2 333, engine 2 34, engine 3 331, engine 3
I
i
P
I
l
i
1.2
Figure 12. Tests on Diesel Engine Oils Curve. from G.E.-Hardy recording spectrophotornetcr for oil 5 after test in Dieqel locomotive
Figure 11.
Results of Laboratory Tests on Four Oils
U p p e r . Carbon residues Center. Losses by corrosion of cadmium-silter bearingLower. Losses hy corrosion of copper-lead bearinas
- -
--
--
-
Tahle 1. Petroleum Tests on Oil 6 Conradson carbon, %> 86 A.P.I. insoluhle, R C hlorof urm-solu ble, V0 Neutralization niimber .Ish, % Viscosity increasc .4t 100O F. .It 2 1 2 0 F., %
38 Hour-
734 Hour.
0.161 0.051 o.n5i 0.049 0.012
0.173 0.024 0.0 0.311 0.001
Sone
Xont 5
Sone
Samples were taken periodically for more than a year in order to follow the condition of the oil. Typical curves obtained froni oil sample 6 before use and after a short time of operation ale shown in Figure 13 (upper). The fact that the transmission rurves have not chang;ed R grmt deal even after some 300 hours of operiition i i
Table I1 list5 the results of some of the usual petroleum tests on this oil. In addition to the requirement of a definite diluting ratio in order to make a comparison of the transmission curves of various oil samples, it was also found that there is a considerable change in the curve if the diluted sample has stood for some time. The maximum absorption moves to higher wave lengths upon standing. Curve 6.1 in Figure 14 n a s obtained from unused oil 6 immediately after diluting with chloroform to a ratio of 1 to 20. 6B is thr cuive of the yam? samplr after qtanding for 3 montlis.
'Fable TI. Conradaon carbon, % 86 A.P.I. insoluble, Z Chloroform-soluble, C Z Neutralization nuinher .i..q..,h . CL
I"
Viscosity increase At 100' F., 5% At 210O F., 5;
Petroleum Tests on Oil 7 7A
713
7c
7L)
0.694 0.112 0.065
0.682 0.184 0.075 0.248
0.968
0.893 0:312
0.199 0.199
30.0 10.6
0.248 35.5 12.4
0.407 0,177 0.150 0.218
39.3 11.6
0.108
0.245 0.238 33.7 11.4
ANALYTICAL CHEMISTRY
1490
7E is another new oil diluted t o the same ratio, and 7F is the curve after 3 months. Similar shifts in the curves, though not so pronounced as with chloroform-diluted samples, have been observed with new, undiluted oils after standing. This illustrates the difficulty of obtaining and conserving representative samples and of course evaluates relative stabilities of lubricants entirely apart from the Perere conditions encountered in the engine. CONCLUSIONS
Data from theae spectrometer curves, obtained from oils a t standard dilutions, may be used in a variety of ways. They may be compared from the basis of shift of the minimum transmission, the flattening of the curve at this point, or the logarithm of the reciprocal of the minimum transmission (log 1 T,,,, ) plotted
Figure 14.
Effect of Standing
Oils diluted 1 to 20 with chloroform 6A, 7E. No standing 6B, 7F. 3 m o n t h s ' standinp I
0 ,A 00 IWW
400 I
500
,
600
700
8 I
Figure 13. Coleman Spectrophotometer Tests on Oils Used in Stationary Diesel Engines Curve
Upper. Oil 6
Lower.
Oil 7
No. 7 PA 78 7c 7D
Hoursof Operation New oil with additive 37, Diesel 2 333, Diesel 2 34, Diesel 3 331. Diesel 3
against the time of heating during laboratory 01 Vnderwood tests, or against miles or hours of service in engines. These curves may be compared ~ i t hcorresponding curves made by standard accepted test proceduxes such as per cent aLphaltenes, carbon residues, neutralization number, viscositj , chloroformsolubles and insolubles, naphtha-insolubles, and bearing rorrosion. At the present time transmission curves are being prepared on a
routine basis together with the staiidard tests ill ail oil laboratory with remarkably good correlation. Thus the curves which may be determined in a few minutes may actually serve in place of the laborious, time-consuming conventional tests in evaluating and predicting the performance of Iulxicating oils in us?. Emphasis in this investigation has been on the practical use of the spectrophotometer in evaluating changes and deterioration of engine lubricants. It seems evident that the ( w v e s integrate most, if not all, of the chemical changes in oils under oxidizing, catalyzing, and high temperature conditions which characterize operation of internal combustion engines. S;)ecific interpretation, of course, mould depend on very complete chemical analyses of oils before and after use. This or any other method of used oil analysis is limited by virtue of the complexity of the total problem of engine and lubricant performances. These methods relate primarily to oxidation, stnhility with unknonn relationship to engine cleanliness and detergency, contamination by incompletely burned fuel products, ring and valve sticking. piston crown and carbon deposits, scratching of cylinder walls arid fouling of air-intake ports, protection of bearings by specific additives nithout affecting oil oxidation, acceleration of oxidation by some detergents, sludge left in the engine, etc. Thus, no single method alone can evaluate an oil unequivocally in the face of such complexity. Hut the light-absorption method does seem to have t'he advantages of integrating many changes, including the effects of detergents in producing suspensions of solid products, on oil they are, by a simple, rapid instrumental technique. Because of many indeterminate variables in the use of automotive lubricants, even better results should be expected for turbine and transformer oils and all others used under uniform operating conditions. LITERATURE CITED
I., Seabury, R. L.,and Carl, F., IKD. Exti CHEM.,ANAL.ED.. 16, 740 (1944).
(1) Clark, G . L.,Kaye, W. RECEIVED April 4, 1949.
