Determination of Sulfur and Chlorine in Petroleum Liquids by X-Ray Fluorescence T. C. YAO' and F. W. PORSCHE Research and Development Department, Standard Oil Co. (Indiana), Whiting, Ind.
b X-ray fluorescence has been adapted to the determination of sulfur and chlorine in gasolines, distillate fuels, residual fuels, lubricating oil additives, insecticides, and herbicides. A commercial x-ray spectrograph, equipped with a rock salt crystal, helium path, flow proportional counter, and pulse height analyzer, is used. The minimum concentration that can b e detected is 0.002 weight %. With single calibrations for each element, the x-ray method is as accurate as the ASTM methods. The x-ray method is much faster, elapsed time per analysis being 5 minutes for each element.
discriminators, the determination of chlorine and sulfur at concentrations below 0.1% should be feasible. A standard spectrograph for x-ray fluorescence, equipped with a pulse height discriminator, a flow proportional counter, and a helium atmosphere, has been used in developing a new rapid method for the quantitative determination of sulfur and chlorine down to 0.02%. The method has been applied t o a wide variety of sample types: gasolines, distillate fuels, residual fuels, lubricating-oil additives, insecticides, and herbicides.
S
The apparatus is a Korelco threeposition x-ray spectrograph with an Atomic pulse height analyzer, or equivalent. Instrumental conditions are listed in Table I. The sample cells are commercially available cylindrical stainless steel cups, each having as the bottom a window of Mylar film 0.25 mil thick. Samples are poured into the cells to a minimum depth of about 0.5 inch; additional sample has no effect on results. The cells are cleaned and dried after use, and a new Mylar window is used for each sample. Sulfur standards are prepared by accurately weighing high purity n-butyl mercaptan (1-butanethiol) into chilled iso-octane. Chlorine standards are similarly prepared from reagent grade carbon tetrachloride and chilled xylene. Starting a t 28 above the peak, the standards are scanned for sulfur or chlorine a t a constant rate of l / p 0 per minute. The base line is established by scanning approximately 1O 28 below the peak. The signals are fed through the ratemeter t o the recording potentiometer a t attenuation settings proper to
in petroleum liquids is normally determined by ASTRI. lamp ( d ) , bomb ( I ) , or high temperature combustion (3) methods, and chlorine is determined by sodium reduction (4). Each method has limitations. The lamp method is sensitive and precise but can be applied only t o samples that can be burned through a wick. The less sensitive bomb method is normally applied to samples boiling above 350' F. Both methods require 2 to 6 hours to complete one analysis, although time per test can be reduced by analyzing several samples simultaneously. High temperature combustion is fast but is not suitable for samplee that are volatile or low in sulfur. Sodium reduction for chlorine is also slow. A single method, with equivalent or better accuracy and with a shorter elapsed time per andysis, v a s sought for the determination of either sulfur or chlorine in the wide rang? of samples previously annlyzcd by the different ASTM methods. The use of x-ray fluorescence with a flow proportional counter has been suggested for the detection of elements of low atomic number, such as sulfur and chlorine ( 6 ) . X-ray fluorescence has been used to determine sulfur in gas oils (b), and chlorine in resins ( 7 ) . T o maximize sensitivity for these elements, the x-ray path can be flushed continuously with helium (8). With more powerful x-ray sources and the conimercial introduction of pulse height ULFUR
1 Present address, Analytical Chemistry Department, Aerojet-General Corp , SRP, Sacramento, Calif.
2010
0
ANALYTICAL CHEMISTRY
PROCEDURE
Table I. X-Ray Instrumental Conditions Primary x-ray FA4-60 tube, tungsource sten target Excitation 50 kv., 40 ma. Counter Flow proportional Crystal Rock salt (flat); 2d = 5.639 A. Arm collimator 0.005 Inch spacing Port collimator spacing
Radiation path
Suflur line, K , Chlorine line, K ,
0.125 Inch
Helium, 1.5 liters per minute 5.373 .4. 4 . 7 2 9 .4.
concentration. A time-constant of 8 seconds on the ratemeter gives the most reproducible scan. A calibration curve is drawn by plotting concentration as ordinate against net counting rate (peak minus background, in counts per second) as abscissa. T o simplify the calculation of results, concentration scales can be made for each attenuation setting and used to read per cent sulfur or chlorine directly from the peak height on the chart of the recording potentiometer. Each scale is made from a n expanded portion of the calibration curve corresponding to the appropriate attenuation setting. The length of the abscissa of this portion of the curve is made equal to the full pen travel of the recording potentiometer. By projecting concentration values from this curve onto a blank rulr, held parallel to the abscissa, the concentration scale is prepared. The same instrumental procedure is used in analyzing samples. Because the pulse height discriminator holds the base line constant for all types of samples, base line is checked only once per &hour shift. Routinely, the scan is stopped immediately after the peak has been established. The x-ray tube output is adjusted slightly from shift to shift to compensate for variations in the calibrations. Complete recalibration is rarely required. All samples arc analyzed without dilution unless interfering elcments are present. PRECISION A N D ACCURACY
The repeatability of the x-ray method was checked by having each of thrce operators determine in replicate the sulfur or chlorine content of three synthetic standards. These determinations were made over a period of several days. A sample containing 0.100% sulfur was analyzed 11 times with a repeatability (95% probability) of 0.006, which compares favorably with the 0.005 quoted for the A S T N lanip method ( 2 ) . For 23 determinations on a sample containing 1.60% sulfur, x-ray repeatability of 0.04 compares well with thc 0.025 and 0.09 reprntabilitics of the ASThl bomb and high-temperature combustion methods ( 1 , 3 ) . For chlorine, ten determinations on a sample
Table
II.
Comparison of Sulfur Determinations
( % sulfur)
2 8
Figure 1.
2 8
Effect of pulse height discriminator
0.00470 sulfur in gasoline
cont:rining 0.1507, chlorine gave a repeatability of 0.003-ten times as good as the 0.03 repeatability reported for the ASThI sodium-reduction method (4). Comparisons \\.ere made between the x-ray and chemical methods on more than 100 samples ranging from lowboiling gasolines to high-boiling residual fuels and semisolid tarry residues. Comparisons were also made on lubricating oil additives containing metals. For each type of sample, the chemical analysis was carricd out by a method suited to its boiling range and viscosity. The x-rays results cornpare excellently with those obtained by the lamp and bomb methods (Table 11). The agreemcnt on residual fuels is good, except for the highly viscous petroleum tar. Sonuniformity of the sample is the probable cause of this discrepancy; the x-ray method is less likely t o be in error because a larger portion of the sample is t,alien than in the hightemperature combustion method. Several lubricating-oil additives, insecticides, antl herbicides were analyzed for chlorine by both thc x-ray and sodium-reduction . methods. Typical analyses, given in Table 111, shon. excellent agreement. DISCUSSION
Several potential sourccs of difficulty must be carefully controlled. For example, background must be reduced, exposure of the sample t o the primary x-ray beam must be minimized, and interferences arising from absorption by other clerncmts must' be overcoinc. i i t low concentrations of sulfur or chlorine, counting r a t w :ire cstrernel!. low antl bsckgrouiitl noise has a ninrlied effcbct on the determination. By using a pulse height discriniinator, the noise level is reduced and kept constant. Figurv 1 is a reproduction of recordcr tracings for 0.00470 sulfur wit'h and n-ithout the discri!ninntor. The arrows point t o the expccted location of the sulfur peak in terms of 38. Without the discriminator, the noise level is so high t h t any signal tlucx to sulfur is not
Sample T>.pe Gasoline Kerosine Diesel fuel Distillate fuel A Distillate fuel B Residual fuel A Residual fuel B Residual fuel C Residual fuel D Petroleum tar Lube oil additive A Lube oil additive B
X-Ray 0,029 0.049 0,122 0.498
0.715 0.83 0.97 1.20 1.45
2.42
0,050
0.119 0.509 0,713 ... . . . ... . . ...
3.08
...
16.6
Table 111.
Lamp 0.026
Bomb . . . . .
HighDiftixreiice: Temperiiture X-Ray Comliustion Clienrieal +0.003 -0.001 +o. 003
...
-0,011
, . .
+0.002 -0.0% -0 6 3 -0 u3 +0.05 -0 2:< -0 03 -to 3
0.55
... ...
1 .ob
1 ,2:1
1.40 ...
...
16.3
2 , (i5 3.13
...
Comparison of Chlorine Determinations
Sample Type Lube oil additive Insecticide Herbicide ( 2 , 4 D ) Herbicide ( 2 , 4 D and 2,4,5-T)
X-Ray 17.4 12.6 14.6
16.6
detectable. With the discriminator, the signal due to sulfur is clearly distinguishable from background. Long exposures to the primary x-ray beam heat the sample, resulting in expansion and loss of volatile components, and cause the decomposition of such unstable sulfur c.ompounds as mercaptans (thiols). The effect of decomposition was observed when a sample containing 0.1% sulfur, as nbutyl mercaptan in iso-octane, was alternately exposed to x-rays for 7 minutes and unexposed for 5 minutes. A precipibate, which settled on the cell window and significantly increased counting rates during successive exposures, indicated a concentration of sulfur compounds at the bottom of the cell. To reduce the errors arising from heat and decomposition, the time of sample exposure is shortened by determining counts per second with a ratemeter rather than a scaler. Absorption phenomena are of two kinds: the mass absorption of all the atoms in the sample, which increases with atomic number, and the proximity of an absorption edge of the interfering clement or clements t,o the analytical r.:idiation. -4lowering of sulfur results is causcd by thc prcwnce of chlorine; this is simply a mass absorption effect. *-Imuc.11 mor(' drastic lowering of chlorine results is caused by the presence of sulfur, because a sulfur absorption edge is located in the ivvnve-lmq$h region of the K , line of chlorine. This effect becomes negligible, however, when the sulfur Concentration is 0.2% or less. Carbon and hydrogen were common to all the authors' samples and interfered despite their low mass absorption. An inerwsc. in carbon-to-hydrogen ratio mnrkcdly loners sulfur and chlorine'
Chemical 17.2 12.9 14.7 16.5
1)i i'f e w i I c e : S-R:iy - C:Iic~iiicnl +0 2
-0.3 -0.1
+o.
1
results, especially a t low concentrations. For example, 0.1% sulfur or chlorine in a n aromatic solvent gives results that are 1501, low when iso-octane standards are used. At 1% sulfur or chlorine, the error drops to 7y0. Correspondingly, a standard sample containing 1.00% chlorine in iso-octane showed 1.07% chlorine when a calibration curve based on xylene standards was used. Lubricating oil additives generally give low results for sulfur and chlorine because of mass absorption by high concentrations of such other elements as barium, zinc, and calcium. If appropriate precautions are not taken, the lowering of results can be so serious that less than half of the element mill be found. Two means are available for reducing absorption effects: dilution of the sample with the solvent used to prepare the standards, and comparison of the, sample with standards having the same matrix. Dilution is much easier and is usually feasible. Tenfold dilution is enough to overcome wide differences in carbon-to-hydrogcn ratio of snml)le and standards. Gasolincs are dilute enough that absorption int,crfcrcne(as :ire negligible. Tho effcctivc~~ic~ss of dilution in o v ( ~ coniing mass absoi,ption is shoivn by tho an:LlJ,sis of a lulxictating oil :ttl(liti\-cb slion n choniicall>. to cont:tin 16.3 ?idfur, 17.2y0ch1orinc, :ind about 3 c w h of phosphorus antl zinc. S-r:iy tinalysis of the undiluted additive g:i~.rs sulfur m d chloriiic~ va1u1.s of 9.1 ani1 3.5yo,respectiwl?.. Tcn-, bnenty-, ant1 hundredfold dilutions F ; L L V ( ~ 14.7, 1G.6, 2nd 16.6y0 sulfur, :mi 14.4, 15.6, : i n c I 17.4% chlorine. Tlic inass atmr1)tion effects are overconic f:iirly c411icbl\~,:is s1imr.n l>y tlw c ~ ) ~ n l ) l c trcwvc'ry c of VOL. 31, NO. 12, DECEMBER 1 9 5 9
201 1
sulfur after a twentyfold dilution. For the complete recovery of chlorine, further dilution was needed to reduce the 5ulfur content below 0.201,. If a sample should contain so much of an interfering clement that dilution to a usable interfercnce level would dilute the sulfur or chlorine below the limit of detectability, it could only be analyzed by comparison with a standard having the samr matrix. CONCLUSION
The x-ray method has been applied successfully for over a year to the routine determination of sulfur and chlorine in samples ranging from gaso-
lines to highly viscous fuel oils. The short elapsed time of five minutes for the determination of each element has markedly reduced analytical costs and irflproved refinery control. The attractiveness of x-ray fluorescence techniques suggests that many new applications for the control of manufacturing processes should be possible.
( 4 ) Ibid., D 1317-57T, 1957. (5) Doughman, W. R., Sullivan, A. P., Hirt, R. C., A ~ A L .CHElI. 30, 1924
LITERATURE CITED
RECEIVEDfor review May 11, 1959. Accepted July 29, 1959. Presented in part a t the Pittsburgh Conference on Analytical Chemistry and Applied Spectroscopy, Pittsburgh, Pa., hfarch 1958. Ninth Annual, Symposium of American Association of Spectrographers, Chicago, Ill., June 1958.
(1) Am.
Soc. for Testing Materials, Philadelphia, Pa., “ASTM Standards on Petroleum Products and Lubricants,” D 129-57, 1957. ( 2 ) Ibid., D 1266-57T, 1957. (3) Ibid., Appendix 11, p. 944, 1956.
119883.
(6) Hendee, C. F., Fine, d Brown, IT.B., Rev. Sci. I m t r . 29, 531 (1956 J ~
(7) Walter, E. R , Pittsburgh Conference on Analytical Chemistry and Applied Spectroscopy, Pittsburgh, Pa., March 1956. (8) Zingaro, P. W.)Sorelco Reptr. 1, 45 (1954).
Colorimetric Determination of Diethyl, Dibutyl, and DioctyI Phthalates in BaII Propellants GEORGE NORWITZ Pitman-Dunn laboratories, Frankford Arsenal, Philadelphia 37, Pa.
b
The hydroxamic acid colorimetric method for esters was applied to the determination of phthalates in ball propellants. Conditions for color development were studied. The behavior of the common constituents of propellants in the method was investigated. To eliminate interference from nitroglycerin and dinitrotoluene, the ether extract from the propellant was treated with titanous chloride in a buffered acetate medium. To eliminqte interference from 2-nitrodiphenylamine, which was not completely rsduced by titanous chloride, the color was measured a t 650 mp rather than a t the point of maximum absorbance, 540 mp. The method was not applicable to samples containing other esters such as triacetin.
R
improved methods for the determination of phthalates in propellants (9,11,12) are still time-consuming and require a fairly large sample, which is not always available. Therefore this laboratory undertook the development of a rapid colorimetric method. Reports of the Phthalate Committee of the Joint Army-SavyAir Force Panel on the Analytical Chemistry on Solid Propellants (7, 10) indicated the desirability of developing a colorimetric method for phthalates in propellants and stimulated the present n ork. The committee has not yet formally evaluated it. .itte~iiptsin this laboratory to apply ECENT
201 2
ANALYTICAL CHEMISTRY
the colorimetric procedure of Sn-ann ( I S ) for the determination of phthalates in coating materials to that of phthalates in propellants were unsuccessful because of interferences. Attention n-as therefore turned to the hydroxamic colorimetric method for esters (3, 6, 6, 14). This method has been applied by Goddu, LeBlanc, and Wright (3) to the determination of many esters, including dimethyl phthalate, in the absence of interfering compounds. I n the proposed procedure the phthalate is extracted from the propellant with ethyl ether, and the nitroglycerin and dinitrotoluene are reduced with titanous chloride in a buffered acetate medium. The phthalate is extracted with petroleum ether, which is then evaporated off. The hydroxamic color is developed and is read at 650 m l to eliminate interference from 2-nitrodiphenylamine, SPECIAL REAGENTS
Hydroxylamine solution in methanol (12.5%) ( 3 ) . Sodium hydroxide solution in methanol 112.5%) ,-, ( 3,) . Hirdroxylamine-sodium hydroxide reagent ( 3 ) . Ferric Perchlorate Stock Solution. Dissolve 1.60 grams of National Bureau of Standards Sample 55c (pure iron) in a 250-ml. beaker with 20 ml. of perchloric acid (70yo) by Farming gently on the hot plate. Transfer to a 200-ml. volumetric flask with 20 ml. of water and cool to room temperature ~
under a tap. Dilute to 200 ml. with 95% ethyl alcohol. Ferric Perchlorate Reagent. Transfer 40 ml. of the ferric perchlorate stock solution to a 1-liter volumetric flask and add 12 ml. of perchloric acid (70’%). Cool to room temperature under a tap and dilute to the mark with 9575 ethyl alcohol. Sodium Acetate Solution (500 grams per liter). Dissolve 500 grams of sodium acetate trihydrate in water and dilute to 1 liter. Titanous Chloride Solution (20%), Fisher Scientific Co. PROCEDURE
Preparation of Calibration Curves. Fill clean, dry 50-ml. dropping bottles with diethyl, dibutyl, and dioctvl - , phthalates. Weigh a 250-ml. Erlenmeyer flask accura&ly to 0.1 mg. and add ihthalate (about 10 or 11 drops) so that about 0.25 gram of the phthalate is present. Add 200 ml. of anhydrous ethyl ether and dilute to volume with ethyl ether in a dry 500-ml. glass-stoppered volumetric flask, Pipet the following aliquots into dry 250-ml. Erlenmeyer flasks having 29/42 ground-glass joints: diethyl and dibutyl phthalate-5, 10 20, and 30 ml.; dioctyl phthalate-10, 20,30, and 40 ml. Use dry pipets and rinse with ethyl ether. Do not work with more than five samples a t a time. Evaporate off the ethyl ether in a flame-free hood by placing the flasks in 2-liter beakers containing 1.5 liters of tap water that has just been heated to 60” C. Disregard the odor of ethyl ether in the flasks.