V O L U M E 28, N O . 4, A P R I L 1 9 5 6 (Nuclear Instrument and Chemical Corp. Model D-35) was fastened t o the Lucite block, so that the tube window fitted against the glass cell. An 0 ring seal prevented water from entering the space between the polystyrene film and tube window. This assembly is illustrated in Figure 1. The high voltage lead to the counter tube was encased in flexible rubber tubing; the joints were covered with Scotch electrical tape and finally sprayed with Ilrylon plastic spray. RESULTS AND DISCUSSION
Results obtained by using the first modification have been described (2-4), and data collected with the second modification nil1 be available soon (1). This communication compares the usefulness of both modifications for hosphorus-32 and demonstrates the sensitivity of cell B for ca&um-45. The data shown in Figure 2 were obtained by adding a solution containing phosphorus-32 to the unmodified column of the cells and measuring the activity as the solution rose in the adjacent column and passed the counter tube. T o reduce the effect of scattering, a small amount of hexane u-as floated on the surface of the solution. A , Figure 2, is an illustration of how the counting I ate changed as the phosphorus-32 passed the semicylinder (chord length a ) ground in the side of cell A, whereas I3 of Figure 2 shows the rise in activity as the same solution passed the slit in cell B (slit width b).
It is apparent that the counter tube assembly (Rnytheon CK 1020) with cell A measured 8 times as much phosphorus-32 activity as the end-window tube assembly with cell B. This was done, however, at the expense of resolution, for in the first case, 90% of the total increase in counting rate occurred in a 7-mm. rise in level, while in the second case the same percentage was obtained in only a 4-mm. band. This means that although higher activities may be needed, better resolution may be obtained by using cell B. I t s greater resolving power and lowered counting rate were due to the fact that in the second case the radiations were collimated and the tube was a t a greater distance from the radiation source. I n addition, the gradually curved surface of the semicylinder of cell A allowed some radiations to pass through before the thinnest portion of the cell wall was reached. As was shown in a few cases, a greater resolution could be attained in the case of cell A when a thin lead shield containing a slit was placed between the counter tube and the glass cell mall.
545 I n the case of calcium-45, however, a different set of conditions existed. The combined thickness of the counter tube wall and glass mall of cell -4was such as to preclude its use with low energy beta emitters such as calcium-45, sulfur-35, and carbon-14. A graph of counting rate us. liquid height for a solution of calcium-45 with cell B is given in Figure 3. I n this case over 90% of the increase in counting rate occurred in a 2-mm. band. With the low energy beta-particle emitters, therefore, it mas possible to resolve and measure very narrow bands of activity when cell B mas used. Counting efficiency factors of 0.007 and 0.002 yo for calcium-45 and sulfur-35, respectively, have been determined for cell assembly B. Although the amount of glass between the sample and the counter prevented the detection of low energy beta emitters with cell A, it proved satisfactory for phosphorus-32, particularly when the electrophoretic peaks were well resolved. The cell B modification was better suited for the low energy beta emitters or less n-ell-resolved peaks. Intermediate steps in the development of cell B included the fabrication of a Lucite cell similar to cell B. Its inferior optical properties, however, caused it to be replaced with cell B. ACKNOWLEDGMENT
The work was supported in part by research grants from Eli Lilly and Co., Indianapolis, Ind., and the United States Atomic Energy Commission. LITERATURE CITED
(1) Clem R. E., Ericson, A. T., Hein, R. E., RIcFarlsnd, It. H., Leonard, G. FV., J. Biol. Chem., in press. (2) Clegg, R. E., Hein, R. E., Science 117, 714-15 (1953). (3) Clem R. E., Hein, R. E., Suelter, C. H., RIoFnrland, R. II., Poultru Sci. 34, 210-14 (1955). (4) Hein, R. E., Clegg, R. E., Nucleonics 10, 70-1 (1952).
RECEIVED for review June 20, 1956. Accepted January 11, 1950. 16th Midwest Regional Meeting, ACS, Omaha, iYeb., 1966. Contribution No. 510, Department of Chemistry, No. 57, Deprirtrnent of Physics, Kansas Agricultural Experiment Station.
Total Sulfur in Hydrocarbons by Monochromatic X-Ray Absorption BARTON H. ECCLESTON and MARVIN L. WHISMAN Petroleum Experiment Station, Bureau o f Mines, Barhesville,
A Bureau of RIines stiidy involving the separation and identification of sulfur coinpounds in petroleum created the need for a rapid and nondestructive method of determining to tal sulfur on small samples. X-ray absorption methods met these requirenients very well. By modifying existing equipment a method has been developed that makes calibration simple and rapid. Monochromatic radiation is obtained by using the fluorescence attachment to an x-ray spectrograph. Samples can be analyzed in about 20 minutes with precision comparable with other x-ray methods as well as some chemical methods.
x-
RAY science is about 60 years old. Roentgen performed the first crude analysis by x-ray absorption methods in 1895. The more refined use of x-radiation in qualitative and quantitative chemical analysis has been known since 1923, when von Hevesy ( 2 ) published a book on the subject. In more
Okla.
recent years, however, further developments in this field have been stimulated by the availability of better commercial cquipmerit and by demands from indnstry for speedier analytical methods. Total sulfur determination by x-ray absorption has advantages over gravimetric and titrimetric methods because it is more rapid, the precision is good, and the sample is not destroyed. This last factor can be very important t o a research chemist who may spend much effort and time isolating small sulfur concentrates from crude petroleum. DERIVATION O F RIETIIOD
This procedure is based on the well-known law of absorption of monochromatic electromagnetic energy passing through a homogeneous absorber. Prom the absorption law it is known that lnI/Io = -
”a W
(1)
vliere Io is the intensity of the beam entering the absorber, I the intensity after passing through the absorber, .u the m a ~ s
ANALYTICAL CHEMISTRY
546
absorption coefficient a t a given wave length, a the cross-sectional area of the absorber, and w the weight of the absorbing material. When the sample is a mixture of materials, I.( is calculated from pi of each component i and its mass fraction, Fi, as follows: p
= ZiPiFi
(2)
When the absorber is a pure hydrocarbon, Equation 2 becomes: PCH
= pcFc
+ PHFH
The sulfur-calibration curve is based on Equation 9. A t a wave length of 0.587 A,, bs and ~ C are H constants. The crosssectional area of hydrocarbon and hydrocarbon plus sulfur are constants by virtue of cell construction. The mass of the absorbing material w can be made a constant by weighing identical amounts of samples into the cells. Under these conditions Equation 9 is linear, with its slope essentially
(3 1
Carbon and hydrogen are reported to have equal mass-absorption coefficients (6) a t a wave length of 0.53 A. For the conditions used in this laboratory the wave length a t which equal absorption occurred was found to be closer to 0.59 A. Hughes (3)in 1950 found fairly definite evidence that liquid hydrocarbons in long cells caused scattering of the x-ray beam that resulted in mass absorption coefficients differing as much as 15% from pub lished values.
(ps
W
- ~ C H-.)
APPARATUS
The apparatus used is a North American Philips x-ray spectrograph unit consisting of a fluorescent analysis unit, a power supply capable of 60 kv. a t 50 ma., and a scaler unit. Figure 1 is a diagram indicating the principle of the fluorescent analysis unit. The primary beam from the tungsten x-ray tube is directed upon a stationary specimen having a fixed angular relationship with respect to both the primary beam and the goniometer axis on which a crystal is mounted. By substituting a sample cell for the first collimator in Figure 1 and moving the crystal and Geiger tube to the proper angle, the absorbed radiation of the desired wave length can be obtained and measured. Figure 2 shows the modifications made to the goniometer t o accomplish this. In addition to adding a cell holder which allowed two cells to be placed in the beam alternately, the whole fluorescent attachment was fastened to a tilting table, so that the cells could be held vertically. By taking constant weights of sample and hydrocarbon this cell position eliminated the necessity for density measurements and errors due t o temperature variations. A constant mass of material is in the beam regardless of temperature fluctuations during analysis, using a vertical cell partly filled with sample. This vertical cell arrangement is similar to that used with the double-beam x-ray photometer (4).
\SPECIMEN
Figure 1. Basic geometry of fluorescent analysis unit
As the absorption of a pure hydrocarbon is independent of the carbon-hydrogen ratio a t a wave length of 0.587 A. Equation 3 becomes: PCH
= PC = PH
(4)
If sulfur is present in the hydrocarbon Equation 2 combined with 4 gives: PCES
= PCHFCII $. P S F S
(5)
where Fs, the mass fraction of sulfur in the hydrocarbon, is the quantity to be measured. Combining Equations 1, 2, and 5 and 1, 2, and 4 gives: ldCHS/ZO
- (PCHFCH f
W
PSFS)
2
(6)
and ldCH/lO =
If
W
-PCH
Figure 2.
iModifications to fluorescent analysis unit for x-ray absorption analysis
(7)
is held constant for both hydrocarbon and hydrocarbon
plus sulfur, Equation 7 can be combined with Equation 6:
Teflon Gasket B e r y l l i u m Windows
and, as FCH
+ F s = 1, Equation 8 becomes: Figure 3.
-4 Adsorption cell for x-ray sulfur determination
I
V O L U M E 28, N O . 4, A P R I L 1956 The cells used in this work are shown in Figure 3. They were constructed from aluminum tubing closed on both ends with brass screw caps fitted with high-transmittance beryllium windows. Teflon gaskets prevented leakage or evaporation of the sample. The cells were calibrated by weighing identical amounts of a hydrocarbon into each and comparing the transmittance of xrays through them.
547 c0unt.s through each cell four times, alternating the cetane and the standard cell to minimize intensity fluctuations. A calibration curve was plotted. Unknown samples were analyzed in the same manner. Eight grams of sample were weighed into a cell and compared in the x-ray beam v i t h 8 grams of cetane or other pure hydrocarbon. The natural log of the ratio of transmittances was applied to the calibration curve to obtain the weight per cent sulfur. A complete analysis requires about 20 minutes.
OPERATING CONDITIOKS
A counting rate of 350 counts per second through a cell filled with 8 grams of a hydrocarbon was obtained by adjusting the accelerating voltage a t 45 kv. and the current a t 35 ma.
Table I.
Summary of Experimental Conditions for the Determination of Sulfur
X-ray tube target material Material in specimen holder X-ray tube accelerating voltage X-ray tube current Crystal monochromator Goniometer angle Scaler operation Scale factor Wt. of sample Aluminum absorption cell Cell length Inside diameter Volume (approximate) Beryllium window Beryllium window
Tungsten Palladium 45 k v . 35 ma. LiF 28 = 16.60' Fixed count 256 (25,600counts) 8 grams
35/4 inches
n / l s inch
14 ml. 0.750 inch O.D. 0.015inch thick
Table 11. Effect of 1% Impurity upon Sulfur Analysis by ' X-Ray Absorption Calculated as Wt. % Sulfur 0.02 0.07 0.3
Impurity Nitrogen Oxygen Sodium Chlorine Calcium Iron Lead
1
2 4 19
As indicated previously, it is necessary to find the wave length a t which the mass-absorption coefficients for carbon and hydrogen are identical. Victoreen (6) data indicated that this occurs at 0.53 A. Cadmium metal placed in the specimen holder would give a Kor radiation near this wave length (0.536 A,). T o check the identity of mass-absorption coefficients at this wave length 8 grams of two hydrocarbons with widely varying carbonhydrogen ratios were weighed into two cells, respectively. Using the scaler the transmittances of each cell was measured. Theoretically, transmittances should have been identical after corrections for cell difference were applied. Identical transmittances were not obtained, however, so the K a radiation of silver of 0.561 A. was used. Transmittances obtained a t this wave length were more nearly identical, and by plotting and extrapolating these data it WEIS determined that a wave length of about 0.59 A. was that at which mass-absorption coefficients for carbon and hydrogen were apparently equal. As palladium has a K a radiation of 0.587 A., this metal was used, and the calculations of equal absorbance a t th-is wave length vere confirmed experimentally. A lithium fluoride analyzer crystal was used to diffract the K a radiation of palladium into the Geiger tube. This condition was obtained a t a stationary angle of 16.60" (28). Table I summarizes experimental operating conditions for determining total sulfur in hydrocarbon-base stocks. PROCEDURE
A calibration curve was obtained by comparing 8 grams of cetane in the x-ray beam with 8 grams each of six standard sulfur solutions. The time was recorded for the passage of 25,600
INTERFERENCES
Any element, other than carbon, hydrogen, and sulfur, present in the sample being analyzed will cause an error in sulfur determination. Most impurities will give high results. Nitrogen and oxygen,, the most probable impurities in petroleum fractions, interfere less than most other impurities. Table I1 lists a few of the elements that might be found as impurities in distillate fuels. The interferences shown were calculated from the massabsorption coefficients of each element. The interferences of nitrogen and oxygen mere checked experimentally, however, and were found to be in very close agreement with the calculated results. As noted in the table, the interference increases with the atomic number of the element, lead giving a large error in the sulfur determination; however, these interferences are not so serious as they might appear. Table I11 lists some of the impurities for a few distillate fuels that were analyzed by this Iaboratory. These results represent the extremes in impurity concentrations found in a series of 40 fuels analyzed. The greatest interference found in all fuels analyzed was caused by oxygen. The highest oxygen content, 5100 p.p.m. of oxygen, would be interpreted as 0.03 weight yosulfur. PRECISION AND RESULTS
An estimate of the maximum error that would be expected from this method considering the statistical counting error a t a 95% confidence level as the only variable factor, would be 0.04 weight % sulfur. This value is based on obtaining 102,400 counts and assuming that the terms ICHS and ICH (Equation 9) have equal maximum error. That is,
ICH
The assumption is also made that the coincidence counting errors are compensated by calibration of the equipment over the range of counting rates t o be expected.
Table 111. Nitro en, P.P.~. 3660 430 430 80 400 400 300
Oxygen P.P.M.' 4470 1930 1480
200 100
5100 2500
1400
1ioo
Some Distillate-Fuel Impurities Iron, P.P.M. 1.32 22.1 0.08 0.29 0.16 0.93
Calcium,
...
3.97 0.44
Sodium
P.P.AI. P.P.M:
.
0.04 0.03 0.95 0.03 0.03 0.06 0.02 0.29 0.03
0.18 0.36 0.25 0.48 0.33 0.26 0.16
..
..
Accumulative Interference as Wt. % Sulfur 0.039 0.023 0.012
o:bii 0.001 0.011 0.038 0.018
The precision of the x-ray and bomb results was calculated from the data presented in Table IV. The estimated standard deviation for the x-ray results was 0.036 and for the bomb results 0.074. The F test ( 1 ) applied t o these data indicates this difference to be significant at a 95y0confidence level. The sulfur-concentration values listed in the column "known" in Table V were obtained by blending sulfur compounds of established purity with pure hydrocarbons. Using these known
548
ANALYTICAL CHEMISTRY
values and the results as determined by the x-ray and bomb methods, straight lines mere fitted t o these data. The known weight per cent of sulfur was the independent variable, and the s-ray and bomb results mere the dependent variables. An estimate of the standard deviation of differences between the dependent variables and independent variables was calculated (1, 6). These values are 0.038 for the x-ray results and 0.089 for the bomb results. SUMMARY AND COXCLUSION
The data presented in this paper show that this x-ray method for total sulfur is comparable with other x-ray methods and certain chemical methods in precision. It has the advantage of rapidity and conservation of samples over chemical methods Table IT’.
Data Used to Calculate Prccisioii X-Ray, Wt. % Sulfur
Sample A
1.928 1.955 1.910 1.928 1.867 1.918 0.867 0.807 3.359 3.342
B C
D
1.556 1.465
E
v
ASThl D 129-52 Bomb, Wt. % Sulfur 1.856 1.819
0.892 0.873 3.099 3.275 1,400 1 . 660 2.398 3.330 0.594
0.494
G Estimated standard deviation 0.036 Calculated F = 4.23 Critical F7, 8, 0.06 = 3 . 5 0
0.418 0.465 0.074
Table 1‘. Comparison of Sulfur Analysis by X-Ray Absorption and Bomb Methods 4ST f - X-.
Known D 129-52 X-Ray, Deviation of Wt. 7% Bomb, Wt. c/o Bomb from Sulfur Wt. 70Sulfur Sulfur Iinown 0.272 0.269 0.297 0.352 0.300 0.277 0.343 0.366 0.442 +0.097 0.387 0.642 0.360 + O . 155 0.513 0.544 0.488 4-0.031 0,520 0.539 0,566 +0.013 0.765 0.762 -0.003 0.734 0.816 0.770 -0.046 0.873 0.824 0.867 0.883 fO.059 0,892 0.925 +0.033 0.834 1.014 1.04 +0.020 1.064 1.052 0.965 -0.087 1.027 0.990 1.054 -0.064 1.113 1,202 1.13 1.169 -0.072 1.513 1.578 1.465 +0.065 1.688 1.54 1. 656 -0.148 1.809 1.778 1.770 -0.031 1.907 1,838 -0.069 1.955 1.989 +0.041 2.03 1.907 3.295 2.365 2,279 +0.070 3.114 3.02 -0,094 3.098 3.17G 2.98 3.146 -0.190 3.342 3.333 3.187 -0.146 Estimated standard deviation of differences 0.089 Calculated t I r a , 0.06, critical
Deviation of X-Ray from Known
Deviation of X-Ray froni Bomb
8
0.038
0.097 0.68 2.09
adaptable to this problem for laboratories that may have the same or comparable equipment available for other work. LITERATURE CITED
(1)
Bennett, C. A., Franklin, N. L., “Statistical Analysis in Chemistry and the Chemical Industry,” pp. 181, 194, Wiley, New York, 1954.
Hevesy, G. yon. “Chemical Analysis by X-Rays and Its Applications,” p. 86, RfcGram-Hill, New York, 1932. (3) Hughes, H. IC., Wilczemski, J. W.,Proc. Mid-Year Meeting, Am. (2)
(4)
Petroleum Inst. JOM(III), 11-15 (1950). c. E., Jr., . h a L .
RIottlau, A. Y., Driesens,
CHChK.
24, 1852-4
(1952).
It has an advantage over other x-ray methods in the ease and repeatability of obtaining monochromatic radiation. The equipment used, although perhaps too expensive t.0 marrant its purchase for this purpose alone, is easily and economically
(5) (G)
Tictoreen, J. A., J. A p p l . Phus. 20, 1141 (1949). Youden, W. J., “Statistical RIethods for Chemists,” pp. 40-2, R7iley, Kew York, 1951.
RECEIVED for review Septeiiiber 6, 1955. .Accepted January 4, 1956.
Oxidation of Platinum Electrodeie in Potentiometric Redox Titrations JAMES W. ROSS and IRVING SHAIN Department of Chemistry, University of Wisconsin, Mad,son, Wis.
Ilrif~ingpoLcntials observed in potciitioinctric redox titrations using platinum electrodes can be explaiiied I>yoxidatioii of the surface of the electrode. The forniatioii of the oxide coating and its dissolutioii are slow processes, dcpeiidcnt 011 the concciitratio:i and iiaturc of tlic oxidizing and reducing agents prcsciit in the solution. Automatically recorded titration curves sliow an end-point error coiisistcnt with the drifting of potentials studied by means of potential-time curves.
Nu“E
4 ROUS investigators have observed drifting potentials with platinum electrodes in oxidizing solutions ( 1 , 2, 8). The magnitude of the drifting depends on the nature of the F O ~ U tion and the past history of the electrodes. Forbes and Bartlett (1) found that a platinum electrode immersed in a dichromate solution showed a slom drift to more positive (oxidizing) potentials
over a period of 43 minutes. Winter and Moyer (8),titrating iron( 11) with dichromate, found a positive drift in the vicinity of the end point and beyond, n~hichthey attributed t o the low rate a t which the chromium(II1)-dichromate system reaches equilibrium. Furman (2) mentions a positive drift near the end point in titrating iron(I1) with cerium(1V) and permanganate, and a negative drift during the back-titration. He also attributes this drift to the low rate of the chemical reaction in the solution. Because drifting potentials are observed in titrations which give sharp and permanent end points when indicators are used, it %-odd seem that the drift is caused by a slow reaction of the electrode v i t h the solution and not by a slow attainment of equilibrium in the solution. That platinum electrodes cannot always be considered inert is pointed out by Lingane (7). He notes the possibility of attack of platinum electrodes in concentrated chloride solutions by cerium(1V) and permanganate with resulting uncertain potential measurements. Kolthoff and Tanaka ( 4 )