V O L U M E 2 1 , N O . 6, J U N E 1 9 4 9 gen cylinder and to the hydrogenation tube and flow of hydrogen is begun a t once a t a rate of about 85 liters per hour. This is continued for 15 minutes, which is usually ample time for the complete removal of the sample from the silica gel; however, it is best to check the completeness of removal by detaching the hydrogen sulfide absorber after 15 minutes and checking the hydrogen stream for hydrogen sulfide by inserting a M.S.A. detector tube. If a positive indication is obtained in l minute the absorber is reattached for 5 minutes, followed bv another check with a M.S.-k. tube. This is continued until a negatlve indication is obtained on a 1-minute trial. If experience shows that more than 1.5 minutes are required for complete desorption, the duplicate sample is run for the full time necessary before any checks fire made. Only samples containing very strongly adsorbed sulfur compounds require a desorption temperature as high as 600” C., but it is best to run all samples at this temperature in order to avoid low results and imperfectly cleaned adsorption tubes. \Then the hydrogenation 1s completed the hydrogen sulfide absorber is removed and its sulfur content is determined. For highest accuracy a blank run should be made by following the procedure using an identical adsorption tube holding no sample. and the result subtractpd from the sample determination. Blanks usually are less than 10 micrograms of sulfur. EXPERIMEbThL
Three adsorbents were tried as sample collectors and several materials as hydrogenation catalysts. It was found that silica gel IS the most satisfactory absorbent because it reversibly adsorbs small amounts of sulfur compounds. iictivated alumina usually contains alkaline substances which react with thiols and other acidic compounds, making it difficult to get quantitative desorption; activated charcoal (coconut) adsorbs small amounts of sulfur compounds so strongly that almost none can be desorbed. .1 catalyst other than the quartz tube is not absolutely iequired. but better contact is obtained with the quartz chips and floir rates may be increased. Catalysts such as cobalt, molvbdenum, copper, and nickel are unsatisfactory because they retain some of the sulfur. Iron reduced on quartz was found fairlv satisfactory but gave some erratic results. ilctivated alumina n a s also fairlv good but tends to retain sulfur and release
237 it slowly. Quartz is fast and nonabsorptive; it gives consistently good results and so was adopted. To test the method, dilutions of various organic sulfur compounds and Calodorant in petroleum ether were vaporized with air or natural gas and passed through the adsorbents and then the procedure given above was followed. Tests were also made with measured volumes of sulfur dioxide, hydrogen sulfide, carbon oxysulfide, and methanethiol. Of all the compounds tested only hydrogen sulfide and carbon oxysulfide could not be quantitatively adsorbed. In fact, they were so slightly adsorbed that following the sampling by a purge of 7 liters of nitrogen or hydrogen entirely removed (xithin experimental error) these two substances from the adsorption tube without affecting others present. A summary of some tests made with silica gel adsorbent and quartz catalyst is given in Table I. A single adsorption tube was found able to hold quantitatively up to 35 mg. of sulfur present in Calodorant (601, sulfur content) but i t is considered poor prac€ice to take large samples unless no low boiling sulfur compounds are present. I n sampling gases containing methanethiol, sulfur dioxide, or carbon disulfide together with higher boiling compounds (not necessarily sulfur compounds) it was found advisable to use two adsorption tubes in series; the presence of little or no sulfur in the second tube then offers proof of complete retention of all but hydrogen sulfide and carbon oxysulfide. LITERATURE CITED
(1) (2) (3) (4)
Brady, A K ~ LC .H E Y . , 20, 512 (1948). Field and Oldach, ISD. ESG.CHEM.,AKAL.ED.,18, 668 (1946). Fogo and Popowsky, Ax.4~. CHEM.,21, 732 (1949).
Huff, Proc. Intern. Conf. Bituminous Coal, 2nd Conf., 1928, Vol. 11, 814. ( 5 ) Lieber and Rosen, I K D .ENG.CHEM.,A s . 4 ~ .ED., 4, 90 (1932). (6) Meulen, H. ter, Rec. trav. chim., 41, 112 (1922). (7) Mnller, I N D .ESG. CHEM.,ANAL.E D . , 13, 673 (1941).
(8) (9) (10) (11)
Rogers and Baldaste, Ibid., 12, 724 (1940). Rutherford, Proc. Pacific Coast Gas Assoc., 31,98(1940). Wilson, I N D .ESG. CHEM.,ANAL.ED.,5, 20 (1933). Zahn, Ibid., 9, 543 (1937).
RECEIVED August 30,1948.
Bearing, Corrosion Test for lubricating Oils Correlation with Engine Performance E. C. HUGHES, J. D. BARTLESON, AND M. L. SUNDAY Standard Oil Company (Ohio),Cleveland, Ohio
T
HE use of “detergent” type additives in motor oils resulted in an increase in the bearing corrosion problem and it became necessary to develop a laboratory test that would predict the corrosive tendencies of these additives. The present study was undertaken with the objective of developing a laboratory corrosion test that could be correlated with the standard Chevrolet laboratory engine test ( 7 ) . An attempt was made to incorporate into the test as many as possible of the factors that are responsible for the corrosion of bearings in engines. Much of the information that had been gained during the development of the Sohio oxidation test ( 6 ) was applied to the proposed test procedure. The Sohio oxidation test yields correlative data concerning oxidation and detergency. I t also differentiates between noncorrosive and extremely corrosive oils, but does not evaluate examples of mild corrosion. A survey of the literature indicated that a number of methods for testing bearing corrosion had been developed. A description
of the tests may be found in a recent review of the subject by Larsen ( 8 ) . Of the existing tests, the Underwood (IS), the MacCoull (9), and the corrosion and stability apparatus ( 1 4 ) have shown correlation with the standard laboratory engine tests. These tests have in common a high rate of shear in the oil adjacent to the bearing. A serious disadvantage of the Underwood test is that a large oil sample is required and the apparatus is difficult to clean. The corrosion and stability apparatus is complex and not applicable to multiple operation. The RIacCoull test possesses no objectionable features, but the original publication was not clear as to the engine test procedure or the nature of the oils used in the correlation study. Xone of these tests was considered to meet the objective of the present study. As a result, a modification of the Sohio oxidation test method was developed in which satisfactory bearing corrosion correlation was obtained by employing the principle used in the above testsproducing a high rate of shear in the oil surrounding the bearing
ANALYTICAL CHEMISTRY
738 A description is given of a thrust hearing apparatus adaptable to the Sohio oxidation test for lubricating oils. The operating conditions and catalyst cornponents have been determined so that the test at 10 hours correlates for oopper-lead corrosion with the 36-hour, L-4 Chevrolet (A.S.T.M.) test for varied groups of inhibitors and oils. At the same time correlative information is obtained on the oxidation
characteristics of the lubricating oils. The test is to he oorrelative for 16 oil-inhibitor com1binations comprising four inhibitor types and two 4mrnmereial oils. The test is useful both in the field
