1673
V O L U M E 2 5 , NO. 11, N O V E M B E R 1 9 5 3 Steels containing up to 0.1 % phosphorus were successfully analyzed in the presence of as much as 1.4% manganese, 0.3% silicon, 1.0% chromium, or 0.2% vanadium. ACKNOWLEDGMENT
The authors are indebted t o the Atomic Energy Commission, the Purdue Research Foundation, and E. I. du Pont de Nemours & Co., Inc., for financial support of this work. LITERATURE CITED
Akkseev, R. I., Zavodskaya Lab., 11, 122 (1945). Allen, R. J. L., Biochem. J . , 34, 858 (1940). Berenblum, I., and Chain, E., Ibid., 32, 287 (1938). Beraelius. J. J.. Ann. P h u s i k u, Chem... .131 . 6.. 369 (1826), Bolta, D.’F., and hlellon,-M. G., ANAL.CHEM.,19,‘873 (1947). Ibid., 20, 749 (1948). Brabson, J. A,, et al., IND.ENG.CHEM.,ANAL.ED., 16, 705 (1944). Copaux, H., Compt. Tend., 173, 656 (1921). Farrer, K. T. H., and Muir, S. J., Australian Chem. I n s t . J . and Proc., 11, 222 (1944). Filippova, N. A., and Kuznetsova, L. I., Zavodskaya Lab., 16, 536 (1950). Grant,E.L.; “Statistical Quality Control,” New York, JZcGrawHill Book Co., 1946. Hure. J.. and Ortis, T., B u l l . soc. chim. France, 1949, 834. Jacobs, W. A., J . Biol. Chem., 12,429 (1912). Keggin, J. F., Proc. Roy. SOC.( L o n d o n ) , 144, 75 (1934).
Kinnunen, J., and Wennerstrand, B., Chemist A n a l y s t , 40, 35 (1951). Kitson, R. E., and Mellon, hf. G., IND. ESG. CHEM.,-4x.i~. ED., 16, 128 (1944). Ibid., p. 379. Koto, T., et al., Technol. Repts. Tohoku Univ., 15, No. 1, 70 (1950). Lundell, G. E. F., Hoffman, J. I., and Bright, H. A., “Chemical Analysis of Iron and Steel,” New York, John Wiley Rr Sons, 1931. Maksimova, N. V., and Kozlooskii, >I. T., Z h u r . A n a l . Kh,im., 2, 353 (1947). Mervel, R. V.,Zacodskaya Lab., 11, 135 (1945). Poussigues, M., Ann. chim. anal. chim. a p p l . , 5, 265 (1923). Rainbow, C., N a t u r e , 157, 268 (1946). Rosenheim, A., in Abegg and Auerbach’s “Handbuch der anorganischen Chemie,” Bd. IV, 1 Abt., 2 Halfte, pp. 977106.5, Leipzig, S.Hirael, 1921. Schaffer, F. L., Fong, J., and Kirk, P. L., A N A L . CHEM.,25, 343 (1953). Scroggie, A. G., J . Ant. Chem. Soc., 51, 1057 (1929). Sideris, c.,I S D . ENG.CHEN., ANAL. ED.,14, 762 (1942). Stoll, K., Z. anal. Chem., 112, 81 (1938). Wadelin, C., and Alellon, 11. G., A n a l y s t , 77, 708 (19.52). Woods, J. T., and hlellon, 11.G., IND. ENG.CHEM.,. ~ N A I . . E:D.. 13, 760 (1941). Wu, H., J . B i d . Chem., 43, 189 (1920). RECEIVED for review May 28, 1953. Accepted August 18, 1953. Taken f r o m a thesis submitted to the faculty of Purdue University by Coe Wadelin in partial fulfillment of t h e requirements for the degree of doctor of philosophy, May 1953.
Determination of Trace Elements in Fuel Oils Direct Spectrographic Method C. W. KEY AND G. D. HOGGAN Richjield Oil Corp., Wilmington, Calif. This spectrographic method was developed to provide a rapid and accurate means of determining the concentration of vanadium, sodium, nickel, and calcium in residual fuel oils without prior ashing. Because of the adverse effects which certain combinations of these elements may have during combustion in high temperature boilers or gas turbines, i t is essential that the manufacturing processes be controlled by precise analyses. Conventional wetchemical or spectrographic methods performed on the ash of such fuels are very time-consuming and are subject to loss or contamination errors during the ashing phase. A rotating disk electrode composed of graphite and lithium carbonateintroduces a solution of fuel oil, internal standard, and buffer
into a high voltage spark discharge. During the discharge, a n inert atmosphere is maintained in the arc-spark stand. By using Type 103-F film, and positioning the spectrographic camera, a single exposure covers a spectral range of 3265 to 5915 A. on a 25-inch strip of film. Standards are prepared by adding metal salts of 2-ethyl hexoic or oleic acid to a fuel oil of known composition. Line pairs of the elements and internal standard were chosen to provide good reproducibility and sensitivity. A single sample may be analyzed in 40 minutes, as compared to approximately 16 hours by chemical methods. More frequent analyses by this method provide better control over processing, transportation, storage, and utilization of residual fuel oils.
KXOWLEDGE of the type and amounts of metallic elements
or as inorganic salts and they generally teiid t o concentrate in t h e
oils is often essential to the successful utilization of this source of energy. I n some concentrations the metals in fuel oils may collect as ashes on furnace tubes or in gas turbines and cause serious deposits or corrosion. Vanadium, sodium, and sulfur are the major constituents commonly found in fuel oil ashes which have a pronounced effect on collecting of deposits or corrosion. Nickel, iron, chromium, magnesium] and barium are of secondary importance. Naturally occurring or added calcium in optimum amounts may reduce or entirely eliminate the deleterious effects of the other metals. These elements may be present in crudes as metal-organic compounds
heavier portions during fractionation and processing. The concentration of each will vary with the source of oil, but in all c n v s their total concentration is low. Even the heaviest grade of normal Bunker C fuel contains a maximum of 0.1% ash (1). Generally, it is considerably below this amount; however, when ronibustion improvers or additives have been included, this amount may be substantially increased. Sulfur may be present in the elemental form or it may be combined in an organic or inorganic matrix. Vanadium in the ash of fuel oils is generally present as vanadium pentoxide and sodium as sodium sulfate, with lesser amounts of sodium-vanadate complexes (6). Because of the critical dam-
A that may be deposited during the combustion of residual fuel
ANALYTICAL CHEMISTRY
1674
age that may result from such ashes in certain combinations, it is essential that those present in major concentrations be known. In the authors' 1aborat.ory a knowledge of the concentrat.ion of vanadium, sodium, nickel, and calcium in Bunker C fuel oil was required. Conventional dry-ashing techniques are subject to losses or to contamination during combustion. Webashing techniques are prone to give erroneous results because of contamination from the reagents or containers used. . b y ashing procedure followed by chemical or spectrographic analysis is very timeconsuming. It was evident that the greatest saving in time would result from eliminating ashing and introducing the sample directly into the electrical discharge with little or no prior sample preparation. Furthermore. this would eliminate losses or errors that might result during this part of the analysis. For these reasons, investigative xork toward developing a direct spectrographic method of anal!-sis for trace elmnients in fuel oils was UIIdert.aken. INVESTIG.4TION
A rotating disk electrode has been used by the authors for nonmetallic matrices, such as cracking catalysts ( 3 ) , and for hydrocarbon matrices of gas oil, lubricating oil, and greases. Ot'her authors have reported this same usage (4). Consequent,ly, it was believed to offer the best solution for this problem. Studies indicated that with a rot.ating disk electrode introducing the fuel oil with its metallic conetituents into a high voltage discharge in an inert atmosphere, vanadium, nickel, and calcium lines of adequate sensitivity were available in the ultraviolet spectral range. However, no sodium lines with sufficient sensitivity were found in this region. This difficulty was overcome by using Eastman Type 103-F film in the ARL 2-meter spectrograph and positioning t,he camera to cover t8he3265 to 5915 .i,range in a single expo~urf'.
Table 11. Reproducibility Improvenient by Barium Buffer Fuel Oil plus Ba %Ethyl Hexoate (25 Determinations)
Element
Fuel Oil Only (50 Determinations)
Ca -4v. deviation, % Range, % '
54
4.4 21
Xa Av. deviation, % Range, '%
18 120
6.4 28
Ni Av. deviation, To Range, 7%
70
V
r l v . dei,iation, Tc Range, '%
11
3 0
11
14
5.7
3.4
,50
15
The arc-spark stand on the spectrograph used for this work wae modified to permit the introduction of nitrogen into the electrode housing. An inert atmosphere was found necessary to prevent ignition of the fuel oil during the discharge. It also increased the sensitivity. The nitrogen was admitted into the arc-spark stand through a l/a-inch copper tube in the base and another from the side nearest the spectrograph just above the optical axis. The exit end of the base tube was compressed to prevent the flow of nitrogen from this point. Ita walls were perforated by approsiniately 12 l/le-inch holes distributed over the surface. The tube above the optical axis had an open end and was adjusted to direct the flow of nitrogen into the discharge zone. A suction tube was inserted through a hole in the observation glass opposite the spectrograph and in line with the nitrogen inlet tube, thereby providing a means of removing any vapors as soon as they were formed. The nitrogen flow into the arc-spark stand was controlled by a pressure regulator in conjunction with flowmetera, 0.7' cuhic foot per minute entering along the axis and 2.6 from the base. EQUIP?vIENT
Table I.
Sensitivity Enhancement by Buffers Graphite Only Fuel oil Fuel oil only with Baa Intensity R a t i o s , I I
Element
Graphite plus LisCOa Fuel oil Fuel oil only with Baa
?&?::r:'2,
A v , of 4 Values
____
The Applied Research Laboratories 2-meter spectrograph, with the camera positioned to cover the 3265' to 5915' F. range on a 25-inch strip of 103-F spectral film, is used for this work. The prespark and exposure times are controlled automatically. An *\RIAPrecision source unit with a high voltage spark case is utilized. A three-tray, ARL film-developing machine, thermostatically maintained a t 65" F., is used to process the film, which is dried in a film-drying machine.
Table 111. Effect of Disk Weight on Reproducibility Disk Weight, Grams .. o I I 1 J I Element Line Intensity Ratios, _ _ _ _ _ I Cobalt Line ' Av. of 5 Values
Barium 2-ethyl hexoate. h Barium compound contained sodium as a n impurity; therefore, sodium ratio3 were not determined in these instances.
This encompasses reasonably sensitive lines for the elements sought. Further improvements in sensitivity for some of the elements were obtained by incorporating lithium carbonate into the disk and by using barium 2-ethyl hesoate as a buffer. These improvements are shown in Table I, where an increase in ratio of line to background intensity denotes a comparable increase in sensitivity. The improvement in reproducibility derived from the barium buffer is shown in Table 11. These data are average results obtained from 50 analyses of the fuel alone and 25 analyses of the fuel plus the barium 2-ethyl hexoate. More desirable working curves were obtained TThen the barium buffer was used. Disk composition was established a t 1 part of lithium carbonate to 4 parts of S.P. 1 graphite and the briquetting pressure a t 100,000pounds per square inch. The effect of disk weight on reproducibility is shown in Table 111, which indicates that a 1.3gram disk offers the best compromise. This quantity of graphitr and lithium carbonate in a mold 0.5 inch in diameter produces a disk approximately inch thick; however. the weight of the diak is the critical measurement.
1
.
1 1
Eleiiient Ca A T deviation, % SIax. deviation. %; Sa
.AT. deviation, %
1 I a x . deviation % '
S I
h v . deviation, %
SIax. deviation, 79
7-
Av. deviation, % SIax. deviation, 70
.
4 3 10
5 0 7 8
1 8 4 1
fiO 8 0
7 2 10.
2 4 4 0
;,;
1 7
3 2
2.0 2.2
4 2 9 7
1 3
3.1
3 1
6 2
The spectrograms are read on an ARL projection comparator densitometer. A specimen briquetting press, with a mold l/z-inch in internal diameter equipped with a pin to produce a '/g-inch hole in the center, is used to compress the disk electrodes. A Dunn-Lowry calculator is used for the calculations. No. 6A porcelain boats with a notch cut in the right side, approximately 1.7'5inches from the handle end, are used to hold the sample during discharge.
V O L U M E 25, N O . 11, N O V E M B E R 1 9 5 3 Auxiliary equipment, such as mechanical stirrers, a hot plate, an analytical balance, general analytical laboratory facilities, bottled nitrogen, pressure regulator, and flowmeters are required. STANDARDS, INTERNAL STANDARD, AND B U F F E R
The standards were prepared by adding to a fuel oil of known composition vanadium oleate and the nickel sodium and calcium salts of 2-ethyl hexoic acid dissolved in a refined naphtha. The vanadium was a commercial preparation diluted to the required concentration. The hexoates, which were prepared in the authors' laboratory, contained a 100% excess of 2-ethyl hexoic acid to ensure their solubility in the heavy naphtha. This naphtha, which was predominantly paraffinic-type hydrocarbons with a boiling range of approximately 355' to 385' F., was used to reduce the viscosity of the hexoates sufficiently to permit satisfactory handling and to assure uniform distribution throughout the fuel oil. Aliquots of these solutions were blended with a residual fuel oil to give known concentrations of the four metals. The base fuel had been previously analyzed by spectrographic addition procedures, whereas the nxphtha solutions of the metallic salts xere analyzed chemically.
16?5
sodium was removed from the hexoic acid by vacuum distillation. The naphtha diluent was washed with dilute hydrochloric acid, followed by water washing. Polyethylene bottles were used to store prepared solutions wherever practical. These efforts reduced the sodium contribution of most preparations to an undetectable amount a t the concentrations used in an analysis. The buffer-internal standard solution contributed approximately 0.5 p.p.m. of the total sodium in an analysis; however, a correction is made for this amount. PROCEDURE
The fuel oil sample to be analyzed is thoroughly mixed in its original container, by heating the sample in an oven until its temperature is between 150" and 175' F., then thoroughly agitating it with a mechanical stirrer having a glass or stainless steel shaft and propeller. Unless the oil is known to be dry, a portion 800 700 600
t
500 Table I\'.
Discharge and Exposure Conditions
High precision case Capacity, microfarad Inductance, microhenrys Current amperes Intensity Gontrol stand position yo Transmittance Grating shutter setting Prespark, seconds Exposure, seconds Disk, r.p.ni.
0.007
360 8.5
0.3 48 12 14
The barium salt, used as the buffer, was prepared from 2-ethyl hexoic acid and barium hydroxide. Cobalt, the internal standard, w m also introduced as 2-ethyl hexoate, which was prepared by a double decomposition reaction between lithium 2-ethyl hexoate and cobalt nitrate. *kgain,as in the preparation of the standards, a 100% excess of the 2-ethyl hexoic acid was necessary to ensure complete solubility in the hvdrocarbons. Both the barium and cobalt hexoates were simultaneously dissolved in the paraffinictype naphtha used to add metal salts to the fuel oil standards. The finished solution contained 120 mg. of barium and 1.5 mg. of cobalt per gram of solution.
I n order to obtain standards and an internal standard-buffer solution with an unobjectionable sodium concentration, the following precautions were taken. Prior to its use, all glass\\-are was meticulously cleaned with an acid cleaning solution, followed by multiple rinses with demineralized water of apn r o x i m a t el v 1.bO0,OOO ohms per rc. resistance. W a t e r of t h i s purity was used wherever needed. R e a c t i o n s involving the prod u c t i o n of t h e hexoates were c a r r i e d o u t in Vycor vessels. SVhen borosilicate g l a s s containers were used for this purpose, t h e sodium concentration was excessively high. All reagents were analyzed for sodium and only those which contributed less than 0.5 p.p.m. to the total amount of this element in an analysis were acFigure 1. Arc-Spark Stand ceptable. T h e Assembly
w
Go 4302.5
V 4379.2
400 300
250 200 I50
100 90 00 70
60 50 3
.4 .5 .6 .7 .8 .9 1.0
1.5
2. 2.5 3.
4.
INTENSITY RATIO
Figure 2. Working Curves for Direct Fuel Oil Method Barium buffer
of about 25 grams is dehydrated by heating with absolute alcohol. Sixteen grams of this sample is then xeighed into a tared, 100-ml. Berzelius beaker. To this is added 4.92 grams of the naphtha solution containing barium and cobalt 2-ethyl hexoates. For ease in weighing, this solution is usually added by means of a pipet. Complete homogeneity is assured by using a mechanical stirrer to blend the sample thoroughly while it is being warmed on a hot plate.
Table V.
Approximate Lower Limits of Detectability (Direct fuel oil method) Element of Line Lower Limit, P.P.M. Ca 4226.7 0.4 Na 6890.0 0.2 Xi 3619.4 10 2 V 3184.0
il No. 6.4 porcelain boat is filled to the bottom of the notch with the blended sample-internal standard-buffer mixture. The boat is positioned in the arc-spark stand, so that the stainless steel shaft which holds the rotating disk fits into the groove without binding. Figure 1 shows the arc-spark stand assembly; however, for the sake of clarity in the sketch, the boat has been lowered and the door, with its suction tube, has been removed. The counterelectrode is a '/,-inch spectrographic carbon rod with a 120" cone and is gapped a t 3 mm. Prior to the discharge the door is closed, the vacuum system started, and the flow of nitrogen regulated. Discharge and exposure conditions shown in Table I V are used.
ANALYTICAL CHEMISTRY
1676 Table VI.
Analyses of Sample of Bunker C Fuel Oil Chemical
Ca No. of analyses Element, p.p.m., av. Range p.p.m. Std. ddv., p.p.m. of sample Std. dev., % of element Range, Yoof element
9 317 290-340 16.9 5.3 16
Spectrographic Na Ni 9 9 12.0 96 10.7-13.5 91-102 0.88 3.16 7.3 3.3 23 11
5’ 9 51 48-54 1.94 3.8 12
Laboratory A Na Ni
Ca 302 lo 287-316 8.05 2.7 9.6
FILM DEVELOPMENT
The Eastman Type 103-F film is developed for 2 minutes in Eastman D-8 developer with constant agitation. The temperature of the developing trays is maintained a t 65’ F. The film remains in the shortrstop solution for 0.5 minute and in the Kodak Liquid X-Ray fixer for 2 minutes. I t is washed for 5 minutes in a washing tray. Drying is accomplished in 4 minutes in the ARL dryer. WORKING CURVES AND CALCULATIONS
Exposure conditions for residual fuel oils with an ash content ranging between 100 and 1500 p.p.m. are adjusted to give a negligible background; consequently, this is ignored during subsequent densitometry. Working curves were prepared from the ratios of line pair intensity us. parts per million concentration, using logarithmic coordinates. Typical curves for the fuel oils mentioned above are shown in Figure 2. Fuel oils which have an ash content below 100 p.p.m. are rarely analyzed in the authors’ laboratory. For such oils different line pairs with greater sensitivity are used and temporary working curves are drawn. Discharge and exposure conditions shown in Table IV are used, except that the background transmittance is adjusted to give approximately 80% of clear film transmittance, by changing the grating shutter setting to 1.0. The lower limits of detectability for this method, shown in Table V, apply to the most sensitive, usable lines for the elements concerned from 2400” to 5930 A. In the densitometry the background transmittance is read and a correction is made for it in the calculations. The calculations for fuel oils with a normal ash content are made on the Dunn-Lowry calculator. Those for fuels of very low ash content are made from the temporary working curves. Use of the Dunn-Lowry calculator considerably reduces the time required for routine calculations. DISCUSSION
The chemical determination of the metallic constituents of a residual fuel oil must be made on the ash after all the carbonaceous portion has been completely oxidized. Allowing approximately 8 hours for ashing and a minimum of 8 hours for analyses, the time required for the determination of the vanadium, sodium, nickel, and calcium in a single sample is approximately 16 hours. A comparable analysis may be made by the spectrograph in a p proximately 10 hours, including ashing. When a combined ashing microtechnique and spectrographic method is utilized, the total time may be reduced to approximately 3 hours. However, the time required to ash the sample and the potential errors inherent in ashing techniques make these methods undesirable for routine applications. The direct spectrographic method described requires 40 minutes for the analysis of a single sample. This time may be further reduced by grouping and analyzing several samples simultaneously. The rapidity with which such analyses may be made greatly facilitates the processing or blending of fuel oils to meet specifications covering their metallic constituents. Shown in Table VI is a comparison of results obtained on a sample of Bunker C fuel oiI by this spectrographic procedure and
9 15.2 13.2-17.6 1.38 9.1 29
10 89 85-92 2.26 2.5 7.9
V 13 50.8 48-53 1.39 2.7 9.8
Laboratory B Na Xi
Ca 8 288 285-292 2.67 1.6 2.4
8 14.0 12.7-15.0 1.13 8.1 16
8 74 72-78 2.24 3.0 8.1
V 10 57 53-59 1.7 3.0 10.5
chemical methods by tn-o laboratories. All results listed under “Chemical” are based upon ashing, with subsequent analyses of the ash. Included among the procedures are gravimetric, titrametric, colorimetric, polarographic, and flame photometric methods. The chemical methods, or modifications thereof, which were used by Laboratory A are as follows: 1. Sodium was determined gravimetrically by weighing a precipitate of sodium uranyl acetate ( 7 ) . 2. Vanadium was determined colorimetrically by measuring the spectrophotometric absorption of a phosphotungstovanadate solution ( 8 ) . 3. Nickel was determined colorimetrically, following the procedure reported by Wrightson (8). 4. Calcium was determined by precipitation as the oxalate and titrating with permanganate (6). The spectrographic method has also been used to determine sodium, vanadium, nickel. and calcium in crude oils, in the same manner as with fuel oils after the gasoline fraction had been removed by a laboratory distillation. The metals were concentrated in the bottom fraction, which was then blended with the naphtha-buffer-internal standard solution and handled like a residuum fuel oil. The accuracy is comparable to that obtained on the Bunker C type fuel oils. CONCLUSION s
-4direct spectrographic method for the determination of van& dium, sodium, nickel, and calcium in fuel oils has been developed, which requires 40 minutes for a single determination, compared t o several hours or days by spectrographic, chemical, or physical methods based upon the ash. The precision and sensitivity are comparable to those obtained by the longer methods. This method is well suited for routine analyses where processing or blending must be closely controlled. ACKNOWLEDGMENT
The authors wish to express their gratitude to R. J. Wilson, E. D. Alpert, R. G. Jewel], and the Sinclair Research Laboratories, Inc., for the chemical analyses. They are especially indebted to C. E. Marquart and Eskil Gross for the spectrographic analyses and t o the Richfield Oil Corp. for permission to publish this report, LITERATURE CITED
(1) Bass, E. L., Lubbock, I., and Williams, C. G., Third World Petroleum Congress, The Hague, Proceedings, Section VII, pp. 99-118 (1951). (2) Burdett, R. A., and Jones, L. C., ANAL.CHEM., 19,238-41 (1947). (3) Key, C. W., and Hoggan, G. D., Ibid., 24, 1921-5 (1952). (4) Pagliassotti, J. P., and Porsche, F. W., Ibid.,23, 198-200 (1951). (5) Scott, M. O., Stanfield, R., and Tait, T., J. Inst. Petroleum, 37, 504-9 (September 1951). (6) “Scott’s Standard Methods of Chemical Analysis,” 5th ed., T’ol. 1, pp. 211-12, New York, D Van Xostrand Co., 1939. (7) Ibid., PP. 879-81. (8) Wrightson, Frances, ANAL. CHEY., 21, 1543-5 (1949).
RECEIVED for review M a y 6, 1953. Accepted August 20, 1953. Presented before the American Petroleum Institute, X e w York, N. Y.,1953.