Table 111. Consecutive Acidity Constants of Citric Acid at 25 "C Medium 0.1 A4 KCl 0.1M KNOj 0.1A4 Me4N+C1-
PK~ 4.375 i 0.013 4.30 4.36
PKI 2.892 i 0.013 2.79 2.88
PK~ 5.685 5 0.017 5.65 5.84
Ref. This work (5) (6)
Table IV. Determination of the First Basicity Constant of Carbonate Ion in 0.1M KCI at 25 "CQ bo', mol X lo5
hi',
a',
mol X lo4
mol X 106 3.680 3.315 3.505
2.000 3.000 4.000 For notation, see the end of
1.200 1.300 1.500 the theoretical part.
[OH-] M x 104
2.459 3.069 3.753
Z
PK
0.308 0.256
3.962 3,975 3.938
0.23
(5) K. S. Rajan and A. E. Martell, 1/?org.Cliem., 4,462 (1965). (6) S. S. Tate, A. K. Grzybowski, and S . P. Datta. J . Cliem. Sac.,
When studying the results in Tables I-IV, it must be noted that they have been obtained using a normal laboratory titrator intended for routine purposes. The smallest division of the dial scale of the titrator is 0.1 p H unit, and thus it may be safely concluded that no conventional method could provide as reproducible results with this instrument. N o extreme precautions were taken in the titrations because the main purpose was only to demonstrate the usefulness of the method. The main reasons for the somewhat varying results were the slow drift of the potential and end-point setting and uncertainty in setting the end point because of the rather coarse one-turn potentiometer and insensitive Schmitt trigger circuit. However, if more accurate results are required, a more stable instrument could be easily constructed using commercially available highly stable operational amplifiers. This work is in progress in our laboratory.
1965, 3905. (7) R. Nasanen, Sirom. Kemistiltlzti B, 19, 90 (1946).
RECEIVED for review July 7, 1972. Accepted August 28, 1972.
and a digital computer. The computed equilibrium constants and their standard deviations are given in Table 111. The values of these constants determined at exactly the same conditions could not be found in the literature. The values reported in Table I11 are those determined at the same temperature and ionic strength (5,6). The first basicity constant of carbonate ion has been previously determined by Nasanen (7) using Kilpi's differential potentiometric method. His interpolation formula gives pK = 3.900 in 0.1M solution of potassium chloride at 25.0 "C. The potentiostatic method gave somewhat variable values as shown in Table IV. The reason may be the slightly erratic and slow response of the glass electrode in the alkaline region.
Direct Determination of Sulfur in Oils by Atomic Absorption Spectrometry Using an Inert Gas Shielded Nitrous Oxide-Acetylene Flame G. F. Kirkbright, Maurice Marshall, and T. S. West Chemistry Department, Imperial College, London S. W.7, U.K.
MOSTCRUDE MINERAL OILS contain appreciable amounts of sulfur; many refined petroleum products also contain significant amounts of this element. As a result of the toxic and corrosive effects of the sulfur dioxide produced when these materials are used as fuels and lubricants, the determination, and subsequently control, of the sulfur content of crude oils and refined products is of great importance. A number of sensitive, precise, and accurate methods for the determination of total sulfur in oils are applied routinely. These usually depend on prior combustion of the oil (by the Wickbold method) followed by volumetric or gravimetric determination of the sulfur dioxide evolved as sulfate ( I ) . A variety of ( I ) "IP Standards for Petroleum and its Products," Pts. I and 11, Institute of Petroleum, London, 1970.
instrumented techniques, such as molecular absorption spectrophotometry ( 2 ) , conductimetric and high frequency titration, radiometry (3), or microcoulometry ( 4 ) have been proposed t o replace the simple volumetric or gravimetric determination of the sulfur dioxide recovered after combustion. Although these methods are sensitive, the combustion process may be time-consuming and their reliable routine application requires considerable operator skill. The direct determination of total sulfur in oils without recourse t o sample combustion has been successfully undertaken by X-ray fluorescence (2) Wataru Machida, Tudao Okutani, and Satori Utsumi, Burisrhi Kagakir, 17, 1128 (1968); Clwm. Abstr., 69, 98169 (1968). (3) E. V. Good, Aiinlyst (Loridon),93, 663 (1968). (4) F. C . A. Killer and K . E. Underhill, ibid.,95, 505 (1970).
ANALYTICAL CHEMISTRY, VOL. 44, NO. 14, DECEMBER 1972
2379
Table I. Variation of AAS Sensitivity with Vapor Pressure of Sulfur Compounds AAS Approx. vapor sensitivity for sulfur, Compound pressure. mma ppmi I absorption Dibenzyl disulfide Negligible 27 rrrt-Butyl disulfide 0 5 27 fert-Butyl sulfide 5 2 0 Thiophene 80 0 6 Calculated at 25 ”C
spectrometry (5, 6 ) and by neutron activation analysis (7). These techniques involve high capital cost and again require considerable expertise if accurate results are t o be obtained. There is a need for a simple and rapid direct method for the determination of total sulfur in oils without preliminary sample combustion. For the purposes of determination of sulfur in most crude oils, the sensitivity of the method does not need t o be extremely great. We have recently demonstrated the remarkable transparency below 200 nm of the inert gas separated premixed nitrous oxide-acetylene flame (8). The use of this flame permits the direct determination by atomic absorption spectrometry of elements such as arsenic, selenium (8), sulfur (9), and iodine (10) whose most useful resonance lines are strongly absorbed by the flame gases in other hot flames. The high temperature and reducing properties in the interconal zone of the slightly fuel-rich flame provide high atomization efficiency and freedom from solute vaporization interferences for these elements. The attainable sensitivity reported in our early studies of the direct determination of sulfur a t its 180.7 nm resonance line, when the sulfur sample was introduced into this flame in a n organic solvent medium, was sufficient to allow its determination in crude oils; additionally it seemed probable that flame temperature would ensure efficient atomization of the sulfur irrespective of its bound form in the oil and provide a reliable estimate of the total sulfur content of the samples. This paper describes the results of our investigation of the use of an experimental atomic absorption spectrometer for the direct determination of total sulfur in crude and fuel oils. A sulfur microwave-excited electrodeless discharge lamp source and inert gas separated nitrous oxideacetylene flame have been used in conjunction with a vacuum monochromator equipped with a photomultiplier and digital frequency meter to permit detection by photon counting. EXPERIMENTAL
Apparatus. The apparatus employed was similar t o that described previously ( 9 ) , but a number of modifications were made for this work. The spectral source was a sulfur electrodeless discharge lamp (EDL), powered by a microwave generator operated a t 2450 MHz (Microtron Mk. 11, Electromedical Supplies Ltd., Wantage, U.K.) at 50 watts using a -6 dB attenuator (5) “1971 Annual Book of ASTM Standards.” 02622-67 Part 17,
American Society for Testing and Materials, Philadelphia, Pa., 1971.
(6) R . J. Bird and R. W. Toft, J . Insf. Perrolewn, 56, 170 (1970). (7) C. Elejalde and F. Albisu, Eiisnyos I m z s r . , 3, 48 (1968). (8) G. F. Kirkbright and L. Ranson, ANAL.CHEM.,43, 1238 (1971). (9) G. F. Kirkbright and M. Marshall, ihid., 44, 1288 (1972). ( I O ) G. F. Kirkbright. T. S. West, and P. J. Wilson. A t . Absorption Nrwslett., 11, 53 (1972). 2380
so that the effective power to the EDL was cu. 10 watts. The burner employed was a 7-cm water-cooled nitrous oxideacetylene slot burner (Beckman-R.I.I.C., Glenrothes. Scotland) with facility for separation of the flame by inert gas shielding (ZI). Sample solutions were nebulized on nitrous oxide (flow rate 8.2 I.jmin), and the sample uptake rate was 6.7 ml!min. An acetylene flow rate of 3.0 1.’min was employed, the remainder of the fuel being supplied by the organic solvent, so that a slightly rich flame was obtained. Improved flame stability was obtained by fitting a second reducing valve in the nitrous oxide supply line, immediately before the flow meter; this minimized fluctuations in the gas flow. The flame was separated by nitrogen. In order to provide a light path appreciably more transparent than the atmosphere, glass tubing (25-mm o.d.), purged with nitrogen, was placed between the source and the fluorite lens (20-tnm diameter. 65-nim focal length) used to focus the radiation into the flame. Simiiar tubing, fitted with optically flat fused silica windows, was placed between the lens and the edge of the nitrogen-shielding gas stream for the flame, and also between the flame and the collimating lens of the spectrometer. The fluorite lens provided slightly better transmission a t 180.7 nm than the silica lens used in our earlier work. At shorter wavelengths. the advantage of the fluorite lens would. of course, be even greater. The monochromator used was a Hilger and Watts E796 vacuum polychromator (Hilger and Watts Ltd., London, U.K.) evacuated to CCI. 0.15 Torr and preset to the sulfur 180.7-nm resonance line, with a fixed spectral bandpass of 0.03 nm. The spectrometer was fitted with an EM1 6256B photomultiplier tube (E.M.I. Electronics Ltd., Ruislip, U.K.) supplied with high voltage by a Brandenburg Model 472R power supply (Brandenburg Ltd.. Thornton Heath, U.K.). The output from the photomultiplier was led through a very short (ca. 8 inches) length of co-axial cable, via a 470ohm load resistor to the input of a digital frequency meter (Model TSSA 6636i3M, Venner Electronics Ltd., New Malden, LJ.K.). Ac mains power to the digital frequency meter and EHT supply unit was filtered via an K F mains filter (Model L1829, Belling and Lee Ltd., Enfield. U.K.) to exclude the possibility of mains-borne interference. Preparation of Standard Solutions. A 10000 Gg:ml stock solution of sulfur was prepared by dissolving dibenzyl disulfide (9.616 grams) (Fluka AG, Fluorchem Ltd., Glossop, U.K.) (purum grade) in isobutyl methyl ketone (250 ml). Working standards were then prepared by dilution, as necessary. Sample Preparation. DILUTION PROCEDURE.Samples of oil, 0.2 to 1.0 gram, were weighed using a short length of clean glass tubing as a disposable pipet. Each oil sample was then dissolved in isobutyl methyl ketone, and diluted to volume in a 100-ml flask. The weight of oil taken was chosen on the basis of two criteria: The anticipated sulfur content of the final solution should fall in the convenient working range 40-400 Mglml, as the calibration curve was known to be linear up to 400 pg;ml; and second, the nebulization characteristics of the oil sample solution should not differ significantly from those of the standard solutions because of different viscosity, density, or surface tension. Difficulties of this type should not be experienced with solutions containing less than 0.5% w/Iv of oil, but this point can readily be checked by comparing the sample uptake rate and nebulizer efficiency for both samples and standards. Each oil solution was nebulized into the separated nitrous oxide-acetylene flame, and its absorbance measured a t 180.7 nm. The concentration was obtained by comparison with _ _ _ _ _ _ _ _ _ ~ . ~ ~~
~~~~~~
~~
~
~
(1 1 ) G. F. Kirkbright, M. Sargent. and T. S. West. T d f i m . 16, 1167 (1969).
ANALYTICAL CHEMISTRY, VOL. 44, NO. 14, DECEMBER 1972
Table 11. Results Obtained for Determination of Sulfur in Oils Sulfur content, yo wjw Combustion methoda Leaded gear oil Fuel oil Fuel oil Light crude
AAS determination,
X-ray methodb
0 49
1.97 2.65
Crude
> 1.45 1.47 1.75 > 1.74 1.74 2.70 > 2.69 2'68
Fuel oil A
2.64
Fuel oil B
2.69
this work 0.50,0.46,0.52 1.96, 1.92, 2.01 2.65, 2.67, 2.71 1.25, 1.31, 1.26
Mean for AAS and standard deviation 0.50 i O . 0 2 1.96 =t0.06 2.68 =k 0.03 1.27 j= 0.12
Error relative to standard method, :d 1-2.0
-0.5 +1.1 -12.4
1.50, 1.59, 1.60
1.56
?C 0.05
-10.3
2.55, 2.41, 2.51
2.49 i 0.06
-7.4
> 2.66
2.46, 2.59, 2.44
2.50 i 0.07
-6.0
> 2.73 2.77 2.87 > 2.92 2.97 1.63 > 1.65 1.67
2.43, 2.37, 2.25
2.35 i 0.07
-13.9
2.52, 2.44, 2.85
2.60
0.18
-11.0
1.56, 1.48, 1.44
1.49 i 0.10
-9.7
Crude A
0.106 =t0 . 1lC
-1.9
Crude B
0.159 L 0.009c
10.6
Heavy crude
2.69
Fuel oil C' Fuel oil D
=k
'' As in reference I .
As in reference 6. Standard Addition Method.
the absorbance of a known standard solution: 100 pg/ml was a convenient concentration standard for purposes of calculation. The mean value of six absorbance readings was taken for each solution. A weighed quantity of STANDARD ADDITIONS PROCEDURE. oil, chosen to give a final concentration, after dilution, in the working range 40-400 pgirnl, was made up to volume in a 50-mI flask with isobutyl methyl ketone. Two 20-ml aliquots of the solution were then transferred to 100-ml flasks. To the second of these, a standard addition equivalent to 50 pg/ml of sulfur was made, and the solutions were diluted to volume with isobutyl methyl ketone. The solutions were then nebulized into the separated nitrous oxide-acetylene flame and the concentration of sulfur in the sample solution was calculated from the ratio of their absorbances. To calculate the absorbance due to a particular solution, it was necessary to measure the incident intensity, i.e., number of counts, at 100% transmission ( N r ) ,zero transmission ( N O ) and , while nebulizing sample (Ns). Values for N T , NB, and N S were obtained by taking the total of ten 10-second counts for each. Absorbance was then calculated. RESULTS AND DISCUSSION
Instrumentation. Although the instrumental arrangement employed was similar to that described in our earlier report of the atomic absorption characteristics of sulfur in the nitrous oxide-acetylene flame ( 9 ) , some modifications were made for its use in this study. Thus a 6-dB attenuator rather than a 3-dB attenuator was employed between the microwave power supply and the 3,'4-wavecavity t o permit operation of the microwave power generator at higher power outputs a t which better stability is obtained. The replacement of the microammeter and potentiometric chart recorder detection system used previously by the digital frequency meter employed here, permits detection of the radiation from the E D L by the photon counting technique. The E D L source was shown to exhibit a signal: background intensity ratio a t 180.7 nm of greater than 100 :1 as described previously. The fuel
flow rate, burner height, shielding gas flow rate, EDL operating power, and P M T voltage were optimized t o produce the lowest detection limits for sulfur introduced into the flame as a solution of dibenzyl disulfide in methyl isobutyl ketone (MIBK). The high flame transparency (64% of that of nitrogen) demonstrated previously ( 9 ) at 180.7 n m line a t the optimum fuel flow rate for the determination of sulfur in aqueous or ethanolic medium was also shown t o be achieved when MIBK is nebulized. A lower acetylene flow rate is required t o produce the slightly fuel-rich flame, however, when this solvent is employed. Selection of Sulfur Standard Compound for Oil Analysis. The slightly fuel-rich inert gas shielded nitrous oxide-acetylene flame provides for efficient atomization when sulfur is introduced into the flame as sulfate (9), thiosulfate, thiocyanate, or thiourea in aqueous solution. Equal absorbance response, corresponding to a similar degree of atomization, is obtained for each of the above species. A similar experiment performed with a series of different organic sulfur compounds of low volatility introduced into the flame as their solutions in MIBK revealed that similar 1 % absorption sensitivities were obtained for each compound. The sensitivities obtained using MIBK are somewhat greater than the sensitivity reported earlier for sulfur as thiourea in ethanol (4.4 p p m / l z absorption); this is attributed to the greater nebulization efficiency obtained with MIBK. For sulfur compounds of high volatility a t room temperature, the extensive vaporization in the indirect nebulizer results in appreciably higher sensitivity. As shown in Table I, whereas similar sensitivities are obtained for tert-butyl disulfide and dibenzyl disulfide, that observed for thiophene is substantially greater. The liquid di-n-butyl sulfide has been employed as the standard compound for the determination of sulfur in oils by X-ray fluorescence spectrometry (5); we prefer, however, to employ dibenzyl disulfide as it is available in pure crystalline form. Difficulties were not expected due to the enhanced 1 absorption sensitivities obtained at 180.7 nm with volatile
ANALYTICAL CHEMISTRY, VOL. 44, NO. 14, DECEMBER 1972
2381
sulfur compounds, as in refined products the sulfur compounds of high volatility are largely removed in processing. For crude oils, the presence of HsS and high volatility sulfur compounds might be expected t o lead to high results when a nonvolatile sulfur compound is employed as standard; as described below, however, this effect is not observed. Presumably the volatile sulfur components in crude oil are lost before analysis. Dibenzyl disulfide was chosen for use as the standard compound for use alone in MIBK to prepare the sulfur calibration graph and for addition of known amounts of sulfur to those oils analyzed by the standard additions technique. Results. The results obtained for the determination of sulfur in oils by direct comparison with a calibration curve produced from dibenzyl disulfide in MIBK and also by use of the standard additions technique are shown in Table 11. Each of the values in column 4 represents a n oil sample taken through the sample dilution and recommended procedure. The absorbance for each of these diluted samples was measured six times and the mean value used to calculate the sulfur content. The three values shown for each oil were obtained in different experiments conducted on different days over a period of several weeks; the standard deviations shown in column 5 therefore also take into account the long term repeatability. The values obtained by AAS for the oil samples whose sulfur contents were determined by reference to the calibration curve for dibenzyl disulfide in MIBK show a small systematic negative error compared to the mean values obtained by X-ray fluorescence. This most probably results from a slight mismatch in the viscosity characteristics between the diluted samples and standard solutions. This could possibly be eliminated if a sulfur-free base oil of similar viscosity was available t o add t o the standard solutions to match the nebulization characteristics. Even without matching the standard and sample solution viscosities closely, however, the accuracy attained compares quite favorably with the spread of results obtained between laboratories for sulfur determination in similar oils using the X-ray fluorescence spectrometry method recommended by the American Society for Testing and Materials (5). No significant systematic error was observed in the values for sulfur content obtained for the oils treated by the standard additions technique. This suggests that the
assumption that a mismatch between sample and standard viscosity characteristics is responsible for the error observed in the conventional calibration technique is correct; when suitable sulfur-free base oils are not available, therefore, use of the standard additions technique is preferable, particularly for examination of samples of low sulfur content (
