ANALYTICAL CHEMISTRY
198 Tebt for ammonia with Nessler reagent. Sessler reagent was prepared as described by Jacobs ( 7 ) . T h e reaction of a droplet of reagent with a crystal of a n ammonium salt produced an orangered solution in which was usually a fine orange-red precipitate. Test for nitrate with a solution of nitron acetate ( 3 ) . T h e reagent consisted of nitron dissolved in 30% aqueous acetic acid. Tvpical needles of nitron nitrate were obtained. Test for carbonate with concentrated nitric acid (evolution of bubbles). 1Io.t of these tests arp not specific, but usually mean that t h r ion indicated is present. The following tests were found to be unsatisfactory for particles below 100 microns in size and indicate the limitation of the method: The diphenylbenzidine-sulfuric acid test for nitrates ( 5 ) .which depends on the development of an intense blue color. The color changes were difficult to observe. The barium rhodizonate ( 6 ) test for sulfates, which depends on the disappearance of the red color of barium rhodizonate with the formation of barium sulfate. T h e color changes were difficult to observe and the precipitate of barium rhodizonate was difficult to handle. The zinc uranyl acetate test for sodium ( 5 ) ,which depends on the formation of sodium zinc uranyl acetate precipitate. Precipitates were obtained when the droplets of reagent were placed under oil on apparently clean microscope slides. Possibly this difficulty would have been reduced if the slide had been of polystyrene, vinvlite, or quartz. Much of the obviously crystalline material collected on tht. slides in 1,os Angeles was 5 microns or larger in diameter. Most of this was ammonium sulfate. Apparently, at least part of this material had originally been dissolved in droplets collected on the slides and had dried. Testing such particles with various reagents is iiot particularly difficult. However, testing particles much smaller than 5 microns requires considerable experience and patience. Some halide, probably sodium chloride, was also col-
lected, but there was little of this even among particles of submicron size. ACKNOWLEDGMENT
Acknoi\-ledgment is due the Smoke and Fumes Committee of the Western Oil and Gas Association, whose financial support and encouragement for the study of smog in Los Angeles made this work possible. Acknowledgment is also due Carsten Steffens of the University of New Mexico, who suggested this approach to the problem. LITERATURE CITED
(1) Renedetti-Pichler, A. -4., and Rachele. .J. R.. Ixo. ENO.CHEM. ANAL.ED.,12, 233, (1940). (2) Chambers, R., J . Rou. Microscop. SOC.,60, 113 (1940). (3) Chamot, E. M., and Mason, C. W., “Handbook of Chemical Microscopy,” 2nd ed., Vol. 2. p. 326, Sew- York, John Wiley & Sons, 1946. (4) Crozier, W-.D., and Seely, B. K.. “Proceedings of First National .\ir Pollution .~~ SvmDosium.” Stkiford Research Institute. Stanford, Calif., 45. 1949. (5) Feigl, F., “Qualitative Analysis by Spot Tests,’’ 3rd English ed., New Yolk, Elsevier Publishing Co., 1946. (6) Green, H. L., and Watson, H. H., Great Brit., Medical Research Council, S p e c i a l R e p l . Serirs, 199 (1935). (7) .Jacobs, M. E., “.l\nalytical Chemistry of Industrial Poisons, Hazards, and Solvents.” pp. 289-90, New York, Interscience Publishers, 1941. ( 8 ) Knott, C. G.. “Collected Scientific Papers of .John hitken,” London. Cambvidge University Press, 1923. (9) Magill, 1’. I,.. d i n . Ind. Hug. .I.ssoc. Quart., 11, 55-64 (March 1950). (10) Magill. P. L., “Proceedings of First National Air Pollution Symposium,” Stanford Research Institute, Stanford, Calif., p. 61, 1949. (11) Sonkin. L. S.,J . Ind. Hug. T O Z & J ~28, . , 269 (1946). (12) Stanford Research Institute, “Smog Problem in Los .l\iigeles County,” Second Interim Report, 1949. (13) Taylor, C. V., 1Jniv. Calif. Pub. Zool., 26, 443 (1925). (14) Titus, R. N., and Gray, H. L., IND.EKG.C H E Y . , h 4 L . ED.,2, 368 (1930). ~~~~
~
~~
i.
RECEIVED ?Jay 22. 1950.
Spectrographic Determination of Phosphorus in lubricating Oil by Solution Excitation J. P. PAGLIASSOTTI AND F. W. PORSCHE Research Department, Standard Oil Co. (Indiana), Whiting, Ind.
HI‘IIICAL methods for the determination of phosphorus in hbricating oils are long and tedious. Because a fastcr and simpler method is desirable for control purposes, ipectrographic techniques m err investigated. In applying spectrographic techiiiques t o the analysis of oils, one of the problems to be solved is how best to introduce the sample into the analytical gap betueen the discharging electrodes. Calkins and 11-hite ( 4 ) impregnated their electrodes by heating and quenching them in the oil to be analyzed. More recently, Gassmann and O’Scill ( 6 ) introduced the sample into the spark through the bottom of a porous graphite cup used as the upper electrode. I n the method presented here, the oil is introduced into the spark by means of a solutions-eucitation apparatus, in which the lower electrode is a rotating graphite disk partially immersed in the oil sample. Such a technique was suggested by Pierucci and and Barbanti-Silva (9) for the analysis of solutions. Later Blank and Sventitskii (3) used a similar technique, employing a copper disk, for the examination of aqueous salt solutions. EQUIPMENT
The spectrographic equipment (commercially available through Applied Research Laboratories, Glendale, Calif. ) includes a 2-
meter grating spectrograph providing a dispersion of 5.2 A. per mm. in the first order, a high precision source unit, ( 7 ) , a Universal arc-spark stand, a solut,ions-c,scitation apparatus, and a comparator-densitometer. The solutions-elicitation apparatus, hecause of its relative newness, warrants description. This apparatus, illustrated in Figuw 1. is niouiitd on the Universal arc-spark stand and is used to rotatv :t disk-shaped graphite electrode through the liquid sample. -1disk 0.125 i- 0.005 inc.h thick cut from a high-purity graphitts rod 0.5 inrh in diameter is mounted on the tapered rotating sh:tft. .I fresh disk is usrd for each esposure. h st’eppedpulley arrangc~mentalloivs a (ahoicr of four disk speeds from 2.5 to 15 r.p.m. The sample is hrld in a combustion boat, mounted on t,he Ion-rr electrode clip of thr arc-spark stand. The periphery of the disk carries s port,ion of the sample up iiito the. tfiwharge, where i t is consumed. Untreated commercial nitrogen is piped to thv arc-spark stand through 0.25-tubing to a fineporosity fritted-glass disk 0.75 inch in diameter. The fritted disk is mounted 0.75 inch to the side of the discharge, as shown in Figure 1, so that the nitrogen blows across the analytical gap. The rate of nitrogrn flow is cmtrolled by regulating the line prrssurt. PREPARATION OF PHOSPHORUS ST4YD4RDS
T h e standard solutions used for calibrat,ing the spectrograph were prepared by mixing the same type of additive concentrate and solvent-extracted base oils as used in the blends of finished
V O L U M E 2 3 , NO. 1, J A N U A R Y 1 9 5 1 plant-production oils submitted for analysis. The particular phosphorus additive used had been analyzed dozens of times by a total of four operators in two different laboratories. Both a gravimetric method ( 1 ) aiitl the following modification of A.S.T.;11. method D 809-44T ( 2 )were used: T h e temperature of the potassium nitrate wash solution is controlled a t 32" t o 40" F. A definite solubility of the ammonium phosphomolybdate precipitate in the potassium nitrate wash solution has been demonstrated in the authors' laboratories. It has also been found that this solubility can be held constant and a t a minimum if the temperature of the wash solution is controlled at 32' to 40' F. Sodium peroxide is not employed during the initial burning step, and the amount of zinc oxide added in this step is increased from 0.5 t o 1.0 gram. The elimination of the sodium peroxide makes it possible to burn the oil in a porcelain crucible instead of in the nickel crucible specified in the A.S.T.RI. method. Tests have shown that neither precision nor accuracy is sacrificed by this change. With the elimination of the sodium peroxide, i t is no longer necessary to go through the chemical steps to destroy the ewess.
Table I.
Exposure Conditions
Prespark period, sec. Exposure, sec. Filter rinalytical gap, mni. Spectrograph slit width,
Table 11.
15 55
Xone 3 50
p
Phosphorus in Synthetic Oils
Synthesis
Modified D 809-44T
%
%
%
0,0502
0.0500 0.03i6 0.0243
0,0499 0.0376 0.0267
0.0374 0.0262
Spectrographic
cal gap, and the resulting disrharge is focused on the grating by means of a 5-inch cylindrical lens. The optical axis of this lens is vertical. Exposure conditions are as shown in Table I. The resulting spectrum is photographically recorded on Eastman spectrum analysis No. 1 film. It' is developed for 3 minutes in Eastman D-19 developer a t 69" F., placed in a 5 % acetic acid stop bath for 10 seconds, and fixed for 30 seconds in Kodak rapid liquid fixer. On the developed film, measurements are then made of the phosphorus line a t 2535.65 ii., and of the internal-standard manganese line a t 2610.20 A. In the calculation of intensity ratios, both lines are corrected for background in the manner described by Pierce and Xachtrieb (8). Conversion to percentage of the sought component is t'hen carried out by means of analytical working curves in a conventional manner ( 5 ) . PRECISION AYD ACCURACY
The average deviation based on repeatability of srventeen runs on a single sample containing 0.0380% phosphorus was 0.0010% phosphorus. The maximum deviation was 0.0021 %. The precision thus demonstrated is sufficiently high t o enable usc of single analyses reliably for routine control purposes.
Table 111. Phosphorus in Typical Lubricating Oils Modified D 80944T
W Figure 1.
Solutions-Excitation Apparatus
%
0 050 0.050 0.047 0.042 0.032 0.030 0.025
0.054 0.048 0.048 0.046 0 039 0.033 0 02i 0 027 0 022 0 019 0 021 0 022 0 022 0 019 0 019 0.021 0.019 0.018
0.024
These changes in procedure necessitate the use of 0.120 instead of 0.135 as the empirical constant in the calculationq. l'ROCEI>URE
Fifteen grams of the oil samplt: arts wighed into a glass vial and 5 ml. of the interri:il-staridard solution are added. The internal-standard solution contains O.lC/b mangarirse and 0.9% lithium, both its nnphtlienates, in a petrolvum fraction boiling between 500" and 700" F. The (.ontents of the vial are t,hen thoroughly mixed with tlie aid of a mechanicd s1i:tker. -1portion of t,he mixed sanipl(5 is iiiseiteti into a small ceramic combustion boat and p1:icc~I o i i thrb Ion-c,r electrode clip of the arc-spark stand. The e~lectt~odcclip is r:iisrd until the lower 0.125 inch of the rotating disk dips into thcl oil. This disk, which is of positive polarity, i> rot:tttd at a spcrtl of 7 . 5 r.p.m. A hemispht~ricali~~ tippcd c:ouritt.ir,lec~tro~~e cut from 0.25-inch graphite rod i+ usc.tl. Th(, c*Iip holding t,hv c.oiinterr.l~,c.trodris water-cooled. The, d r & w l nitrogeii ~ t n i o s p h c ~isr te1st:hlishd ~ :it thr malytical gap by opening thc, nitrogen trnripfcr line and regulating the pressure to 6 inches of mercury. The source unit is adjusted to provide a primary voltage of 125 volts and a secondary inductance of 360 microhenries. A Sorensen regulator is used t o control the input voltage. A 240cycle high-voltage spark discharge is applied across the analyti-
Spectrographic
%
0.023 0.022 0 021 0 021 0 021 0.020 0.019 0,019 0.019 0.017 0.017
0.01Y
Both chemical and spectrographic analyses of three synthetic samples are shown in Table 11. These sainples n-rre prepared i n exactly the same manner as the standards from which the analytical curves were drawn. The spectrographic results are averages of duplicate determinations carried out without previous knowledge of the phosphorus content. T h e chemical data are averages of duplicate determinations performed by each of two operators using different equipment (total of four analyses per sample). Accuracy of the spectrographic method is indicated by these data to be a t least equivalent to that of the chemical method and is within the limits of the photographic and instrumental errors inherent in spectrographic analysis.
ANALYTICAL CHEMISTRY
200 .4s a further demonstration of reliability, analyses of a group of typical lubricating oils are given in Table 111. All results shown are averages of duplicate determinations. The average deviation between chemical and spectrographic analyses of the 19 samples is 0.002% phosphorus; only two results show deviations greater than 0.003% phosphorus, and the largest single deviation is 0.007yo phosphorus. These data indicate that, throughout the range of phosphorus concentration for which the new spectrographic method is applicable, the results are in generally good agreement with those obtained by the chemical method. DISCUSSION
Early experimental work \vas centered on the development of a method in which the background adjacent to the element line would serve as the internal standard ( 4 ) . 4 higher degree of precision and accuracy was possible, however, with the internalstandard procedure described here. -4search for a good internal standard for phosphorus led to the consideration of elements that would be satisfactory not only for phosphorus but also for other metals which may be present in motor-oil additives. It was found that manganese was satisfactory for phosphorus and could also be used for barium. Manganese, however, was not satisfactory for zinc, while lithium was. For this reason, both elements have been included in the internalstandard solution.
Table IV. Known %
Phosphorufi 0.0112 0.0280 0.0700
Deviations from Analytical Curves --
Average % Deviation from Anal. curves for
AI.. anal. curve for fire S h E grades 0.0013 0 0024 0.0020
individual SAE grades 0.0003 0.0012 0.0018
Preliminary investigations demonstrated that a one third gain in precision requlted through the use of a nitrogen atmosphere (IO) a t the analytical gap. It is thought that prevention of oxidation during the discharge may account for this improvement. As evidence of the resulting lack of oxidation, the rotating disk after a run in nitrogen is black and soot,y; whereas, after a run in air, the disk is glossy and clean. T h e use of synthetic standards as the basis for calculation in the present method introduces no significant source of error, as these standards closely resemble the total composition of the unknowns to be analyzed. Five analytical curves for phosphorus were drawn using for each a set of Ytandards prepared from each of the SAE-grade base oils from 10 through 50. All five of these curves were similar, but they could not be superimposed. The variations belween curves appeared to be random, so that i t was not possible t o establish corrections for the effect of oil viscosity. An average analytical curve was also constructed on the basis of data from oils of all viscosity grades (ShE 10-50). Results obtained with this average curve have been considered acceptable for some routine applications. T h e procedure making use of individual viscosity curves was used to obtain the data given in Tables I1 and 111. T h e data presented in Table IV compare the deviations observed by the two methods of translating intensity ratio to phosphorus concentration. It is clear from these data that a higher degree of accuracy was possible when analyses were calculated by means of working curves made up from base oils of the same viscosity grade. This fact becomes increasingly evident as the phosphorus content being determined approaches 0.01%. The porous-electrode technique of Gassmann and O’Neill (6) as published after the present work was well under way. This technique was not included in the present studies because the a-
curacy reported for oils containing less than 0.05% phosphorus was not considered adequate for the authors’ application. It will be n o k d further from H study of Table IV that analyses coinparable to those claimed for the porous-cup technique were obtained on samples containing less than 0.05% phosphorus hy use of the average analytical curve. Greater accuracy can be realized by the solutions method in t,his range when use is made of standards of the same Y.4E viscosity grade, whereas the porous-cup technique is reported to lie insensitive to changes in the viscosity of the oil (6). The solutions-escitation technique introduces the sample into t,he discharge more simply, with fewer variables, and with less sample handling than the quenched-electrode method of Calkins and White ( 4 ) . Moreover, the accuracy of the new method is better than that of any of several variations of the latter that were tried. Solutions escitation has, therefore, superseded the quenched-electrode method formerly used in this laboratory. Spectrochemical analysis has great speed as its outstanding advantage. The spectrographic technique presented here makes i t possible for one man to determine phosphorus in oil in 7 minutes per sample if the .samples can be run in groups of fourteen. When a single determination is desired, results of an analysis can be available within 30 minutes. An esperienced analyst using A.S.T.M. D 809-44T will require 8 hours to analyze six samples for phosphorus. h single chemical determination requires 4 hours. Investigations on the influence of the type of base oil, the nature of the phosphorus compounds, and the presence of other metals are planned. It is hoped that lithium, in addition to being of value as an internal standard, will function as a spectrochemical buffer and will offset the interinfluence of the several elements that may be encountered. T h e method is now being extended t.o include determinations for barium, calcium, and zinc in the concentration ranges encountered in motor oils. Although apparently satisfactory analytical curves have been developed for these metals, data obtained thus far are insufficient for precise evaluation of this extension of the method. 4CKYOW‘LEDGMENT
The authors arc 1ndel)tcd to P . C. White for his encouragement and helpful suggrstions. Acknowledgment is made to W. I. Smith for assistance in the spectrographic analyses; to Paul ISvitn, R. J. Fluschc,. m d hndre\r Puhek for the chemical analyses; and to 1., 12. Grecri for the dcvrlopinent of the modifications to A.S.T.,\I. I)809-44T. LITERATURE CITED
(1) Am. Soc. Testing Materials, Philadelphia, “B.S.T.M. Stand-
ards on Petroleum Products and Lubricants,” Appendix 11, December 1946. (2) [bid., D 8 0 9 4 4 T . p. 3 5 i , November 1948. (3) Blank, 0. V., and Bventitskii, iY,S.,Compt. rend. acad. an‘. U.R.S.S., 44,58-9 (1944). (4) Calkins, L. E., and JVhite, XI. XI., .Vatl. Petroleum S e w s , 38,
KO,27, R519-30 (1946). (5) Churchill, J. R . , IND. E N G . CHEM.,ANAL. E D . . 16, 653-70 (1944). (6) Gassmann, A. G., and O’Seill, W. R., ANAL. C H E M . ,21, 417 (1949). (7) Haeler, M. F., Kemp, J. JY., and Miller, W. H., J . Opticel SOC. A m . , 3 7 , 9 9 0 (194i). (8) Pierce, JV. C., and Nachtrieb, N. H., IND. ENG.(;HEM., .LNIL. E D . ,13, 774-81 (1941). (9) Pierucci, I f . , and Rarbanti-Silva, L., Nuouo cznwito, 17, 275-9 (1940). (10) Scribner, B. F., Corliss, C. H., and Cavanagh, M.B., “Influence of Gases on Spark Spectral Excitation of Aluminum Alloys,” Eighth Pittsburgh Conference on Applied Spectroscopy, Pittsburgh, Pa., 1947. RECEIVED March 11, 195 ) ,