tial decrease in the accuracy of this method of analysis. However, the incorporation of standard isotopic dilution analysis would greatly increase the accuracy of this method.
helpful guidance in its use. Thanks goes to Tony DeRoo at Dow Chemical Company for donating the polyethyleneimine monomers and to Richard Hagstrom at Olin Corporation for the toluenediisocyanate.
Much appreciation goes to Tom Zajicek and Chuck Meade for the use of their atomic absorption spectrophotometer and
RECEVIED for review January 3, 1972. Accepted March 17, 1972.
Determination of Atmospheric Sulfur Dioxide by Differential Pulse Polarography Robert W. Garberl and Claude E. Wilson* Department of Chemistry, Unicersity of Pittsburgh, Pittsburgh, Pa. 15213 The procedure outlined describes a simple, fast, and sensitive method for the determination of atmospheric sulfur dioxide down to 0.1 ppm sulfur dioxide in air. This procedure has been applied to samples of air and nitrogen containing sulfur dioxide at low concentrations. The sulfur dioxide is absorbed into solution by bubbling the air sample through dimethyl sulfoxide containing a supporting electrolyte, e.g., lithium chloride, at a concentration of 0.1M. After deaeration, a differential pulse polarogram is run cathodically on the solution. A peak, between -0.7 and -0.8 V VI. the silver-silver chloride reference electrode, results from the electroreduction of sulfur dioxide. The height of the peak is a linear function of the concentration of sulfur dioxide in the solution. This method is free from interference from sulfides and sulfates; however, oxidants such as nitrogen dioxide have been found to interfere, probably through oxidation of the sulfur dioxide.
RAPIDANALYSIS of atmospheric samples for sulfur dioxide (SOs) is a subject of considerable interest in analytical chemistry and related fields (1-3). Although electrochemical methods have been suggested for such analysis (2, 4, existing methods are not well suited to the purpose. Polarography is not sufficiently sensitive and conductivity is not sufficiently selective for the analysis of typical air samples. The method described below for the analysis of atmospheric sulfur dioxide is both selective and very sensitive. The improved sensitivity of the methods results primarily from the use of pulse polarographic methods which are inherently more sensitive than other voltammetric methods (5, 6). In addition, the use of dimethyl sulfoxide (DMSO) as collecting agent and solvent permits almost full exploitation of the sensitivity of the electrochemical Present address, Tennessee Valley Authority, Sheffield, Ala. 35660. 2 To whom correspondence should be directed. Present address, Department of Chemistry, Indiana-Purdue University at Indianapolis, Indianapolis. Ind. 46205. (1) A. P. Altshuller, ANAL.CHEM., 41, 3R (1969). (2) Arthur C. Stern, Ed., “Air Pollution,” 2nd ed., Vol. 11,
Academic Press: New York, N.Y., 1968, pp 55-75. (3) Morris B. Jacobs, “The Chemical Analysis of Air Pollutants,” Interscience, New York, N.Y., 1960, pp 17G9. (4) H. Dehn, H. Kirch, V. Gutmann, and G. Schober, Monafsh. Chem., 93, 1348 (1962). (5) E. P. Parry and R. A. Osteryoung, ANAL.CHEM.,36, 1366 (1964). (6) Ibid., 37, 1635 (1965).
technique. Measurements of the solubility of sulfur dioxide in dimethyl sulfoxide carried out by Smedslund (7) indicate that quantitative collection of sulfur dioxide from other gases by scrubbing with the solvent is possible. Essentially quantitative removal of sulfur dioxide from the carrier gas was found even at low sulfur dioxide concentrations and relatively short carrier gas to dimethyl sulfoxide contact times. The pulse polarographic methods have been described in recent literature (8,9). EXPERIMENTAL
Purification of Dimethyl Sulfoxide. Commercial reagent grade dimethyl sulfoxide (e.g., Fisher Certified, Catalog No. D-128) occasionally contains sulfur dioxide as a trace impurity. Therefore, it is recommended that the solvent be further purified by the user. Vacuum distillation at a pressure of 1.0 Torr or below and at a temperature of 30 OC has produced a solvent with no detectable sulfur dioxide. The water in the condenser should be kept at 20 “C and not be permitted to go below 18.5 “C as dimethyl sulfoxide freezes at about 18 “C. Other Chemicals. Lithium chloride, Fisher Certified Reagent, was recrystallized from ethanol and dried at 100 “C in vacuo. Sulfur dioxide anhydrous, Matheson Company, was used directly from the lecture bottle. Nitrogen, prepurified,Airco,99.97z pure containing 0.001 % oxygen and 0.0012 water. Apparatus. A PAR (Princeton Applied Research) Model 170 Electrochemistry System equipped with PAR Model 172 mercury drop timer was utilized. The electrolysis vessel was a Brinkmann titration vessel, water-jacketed Model EA875-5 having a capacity of ca. 10 ml equipped with penton upper portion Brinkmann Model EA874. The counter electrode was a platinum square Sargent Model S-30515D. The reference electrode (AgR) was a silver-silver chloride electrode in a dimethyl sulfoxide solution of 0.1M lithium chloride. The junction between the titration vessel and the reference electrode was an asbestos fiber sealed in a glass tube. The electrical resistance of the fiber tip was about 0.5 Mohm. Calibration. A standard solution of about 1 M sulfur dioxide in dimethyl sulfoxide is prepared by passing anhydrous (7) T. H. Smedslund, Finska Kemistamfundets Medd., 59, 40 (1950). (8) G. C. Barker and A. W. Gardner, 2. A m / . Chem., 177, 79
(1960). (9) E. P. Pairy and R. A. Osteryoung, ANAL.CHEM.,37, 1634 (1965). ANALYTICAL CHEMISTRY, VOL. 44, NO. 8, JULY 1972
Sample Blank 1 2 3
6 7 8
Table I. Calibration Concn sulfur Peak dioxide, pM current, nA 0 3.00 16.2 1 .oo 34.3 2.50 10.0 133 267 20.0 50 649 1630 125 ie4oc 800 1100 14300
Calibration figure, nA/pM 13.2 12.5 13.0 13.2 12.9 13.0 13.0 13.0
sulfur dioxide through 10 ml of purified solvent in a midget impinger. A flow rate of approximately 200 cm3/min for about one minute yields a solution of the desired concentration. CAUTION: Sulfur dioxide is extremely soluble in dimethyl sulfoxide. A trap must be placed between the gas cylinder and the solution to prevent back up of the solution into the tank. This trap or equivalent is absolutely essential for safe operation. A second potential problem is the large enthalpy of solution of SOz in DMSO. If the impinger is placed in a large water bath at about 20 "C and if the flow rate of SO2 is below 300 cm3/min,the temperature rise of the DMSO is not excessive. The concentration of the solution is calculated from the weight of SO2taken up by the DMSO. To calculate molar concentration, a small correction of 5 must be made for the volume increase of the solution. The correction is known to i about 2 percentage units. Because the correction is small, the uncertainty in it does not appreciably influence the accuracy of the result. From the standard solution described above a series of solutions are prepared, by serial dilution, in the concentration range 1 X lO+M to 1 X 10-3M sulfur dioxide using a solution of 0.1M lithium chloride in purified dimethyl sulfoxide as solvent. It is recommended that enough of each solution be prepared to allow duplicate determinations on two different samples of the same solution. Each solution and a blank are treated individually as described in the procedure outlined below. A cathodic differential pulse polarogram is obtained for each. The peak current is measured for each solution and the blank subtracted. These current-concentration data are used to calibrate the method. Typical results are given in Table I and are discussed below. The calibration should be repeated at regular intervals in order to correct for instrumental changes. With the instrumentation used for this work, weekly calibration was found to be sufficient. Procedure. Samples of air are taken by bubbling the air through 10 ml of a solution of 0.1Mlithium chloride in purified dimethyl sulfoxide. The samples are taken for exactly 10 minutes at an accurately known flow rate of ca. 250 cm3/minute. This may be accomplished by sucking the air through the impinger with an aspirator or pump controlling the flow rate with a rotameter. The rotameter is best placed between the impinger and the suction device. The air need not be filtered unless it is extremely laden with particulates. The sample solution is then transferred to the electrolysis vessel, the electrodes are positioned for the electrolysis, and the sample is deaerated by bubbling (Airco) prepurified nitrogen through a 1-mm i.d. capillary for two minutes at a flow rate of ca. 250 cm3/minute. It is unnecessary to presaturate the nitrogen with dimethyl sulfoxide as this solvent has a vapor pressure of less than 1 Torr at 25 "C. To prevent redissolution of oxygen from the atmosphere, a stream of nitrogen at 250 cm3/minuteis passed over the surface of the solution after deaeration until electrochemical measurements are complete. A cathodic differential pulse polarogram is made from - 0.6 to -0.9 volt DS. AgR at a scan rate of about 1 mV/sec using a 1358
ANALYTICAL CHEMISTRY, VOL. 44, NO. 8, JULY 1972
pulse height of 50 mV, a drop time of 2 sec (mechanically controlled), and a mercury flow rate of >0.15 mg/sec. The pulse width is 55 msec and the current is sampled during the final 15 msec of the pulse. Because of the resistance of these solutions, a three-electrode system with positive feedback IR compensation is used. Also, to minimize noise caused by pickup from the ac line with solutions of very low concentrations of the electroactive species, the drop timer and the sampling circuitry are synchronized to the ac line modulation. The sulfur dioxide in the sample gives a peak between -0.70 and -0.80 V GS. AgR. The peak current is measured correcting for the blank and the concentration in solution is found from the calibration curve. The concentration of sulfur dioxide in air is then calculated on the basis of the volume of air, corrected to STP by the ideal gas law, which has passed through the impinger assuming 100% collection efficiency. RESULTS AND DISCUSSION
A log [(id - i)/i]os. E plot of the polarographic data yields
Eiiz= -0.73 V 6s. AgR and a slope of 90 mV. The diffusion current constant is 0.48 p A mmole-1 1. mg--213sec-l/*. The diffusion current is taken at the end of drop. The peak of the differentialpulse wave occurs at -0.737 V cs. AgR. In our preliminary studies, we found that of the pulse modes available to us the differential pulse method was best suited for this type of determination. The peak form of output is convenient because it is relatively free of interference from other electrochemical reactions. In addition, the signal to noise ratio is most favorable in the differential mode. Typical results for the calibration procedure are given in Table I. The peak current is linear in concentration within experimental error over the entire concentration range. The average value of the calibration figure is 13.0 nA pM-I. Table I1 represents the data obtained in the analysis of gas samples containing SO,. The samples were produced by passing the carrier gas (air or nitrogen) over a permeation tube containing liquid sulfur dioxide. The permeation tube was designed and constructed by R . Cheng and J. 0. Frohliger (10). They found that the tube had a constant SO2 permeation rate of 1.79 pg/min at 30 "C. Cheng and Frohliger determined the SOzcoulometrically. In preparing gas samples for this work, the temperature of the permeation tube was held at 30 i 1 "C. The flow rate was held essentially constant. Approximate values of this parameter are given in Column 2 of Table 11. The gas sample was passed through a measured volume of DMSO solution for a time given in Column 3 of Table 11. The total amount of SOn to reach the DMSO can be calculated from the contact time and the permeation rate. The concentration given in Column 5 is calculated on the assumption that all of the SO2 is trapped by DMSO. For each solution, a differential pulse polarogram was obtained (Column 6). The SOz concentration was calculated from the peak current and the calibration data which were obtained under similar conditions (Column 7 ) . In general, the agreement between the predicted concentration (Column 5) and the experimentally determined concentration (Column 7) is excellent except at very high flow rates. The magnitude of the SOz concentration in the gas is given in Column 4. This figure is calculated from the permeation rate and the flow rate. Because the flow rate is not known accurately, Column 4 should be regarded as approximate. (10) R. Cheng and J. Frohliger, Department of OccupationalHealth, Graduate School of Public Health,University of Pittsburgh, per-
sonal communications. 1969.
Flow rate, cm3/mina 560 1240 105 3200 470
Table 11. Analysis of Air and Nitrogen Samples Gas concn Soln concn Flow time, min SOZ,ppm SOz p M b 5.60 1.12 2.00 2.80 1 .oo 0.50 28.0 10.00 6.2 1.40 0.20 0.50 1.34 6.30 2.25 0.62 2.80 1 .OO 35.0 14.00 51.5 0.31 1.40 0.50
Sample Nitrogen 1 Nitrogen 2 Nitrogen 3 Nitrogen 4 Air 1 Air 2 1050 75 Air 3 Air 4 2025 0.25 6100 Air 5 a Estimated from Rotameter type flow meter. * Determined from the permeation of permeation tube (8). c Determined by method presented here.
Interferences. Traces of hydrogen sulfide and/or sulfuric acid yielded no detectable interference with the SO2determination. This is as expected since the sulfide is reduced at a much more negative potential than sulfur dioxide and the sulfate is not electroactive at the dropping mercury electrode in this system. In addition, neither species shows evidence of direct chemical reaction with sulfur dioxide in this system. Nitric oxide and/or nitrogen dioxide interfere with this analysis, probably by direct chemical oxidation of the sulfur dioxide. This effect should be expected whenever the analysis of SO2is carried out in a condensed phase. The SO2 peak is reduced by an amount roughly proportional to the amount of nitrogen oxides present. For this reason, the oxides of nitrogen cannot be tolerated at concentrations greater than 1 / 1 ~the concentration of sulfur dioxide if the analysis is to be accurate. Oxygen, if removed by nitrogen deaeration, does not interfere. If deaeration is not carried out, the peak from the sulfur dioxide will be obscured by the oxygen peak which occurs less than 100 mV more positive than the sulfur dioxide. In addition, it appears that the sulfur dioxide at the electrode surface reacts with the products of the oxygen electrolysis. Upon deaeration, the sulfur dioxide peak is fully restored. Water does not appear to interfere and results are reproducible when the concentration of water is greater than that of the sulfur dioxide but less than about 5 wjw. At very low water concentrations, the path of this electrochemical reduction of SO2 may be different. At high water concentrations, the solvent properties of the DMSO are altered. Solution Stability. The S02-DMS0 solutions are quite stable if kept sealed. Good reproducibility is obtained with the same solution over at least two weeks. Thus samples may be collected over a period of time and analyzed at a convenient time and place. Furthermore, samples may be kept for a period of time to permit subsequent verification of results. The solutions should be stored in all glass and Teflon (Du Pont) containers at a temperature of 25 "Cor less. Accuracy and Precision of the Method. The precision of the calibration based upon measurement of four samples of each solution gives a relative standard deviation of 2 on all solutions IO-jM sulfur dioxide or greater. For solutions between lop6and lO-5M the relative standard deviation was 5 %. The relative standard deviation as applied to gas samples obtained with the permeation tube (four samples of each solution) was 5 for solutions with a sulfur dioxide concentration of 10-5M or greater and 10% for sulfur dioxide concentration of 10-6-10-5~. The ultimate uncertainty of this procedure is probably limited by the precision in the standardization step, if all of the
Soln concn Peak current nA 69.8 36.8 380 15.5 82.0 38.0 463 21.4 10.0
so2 pMC 5.36 2.48 28.5 1 .o
6.31 2.78 35.0 1.36 0.53
SOz is removed from the gas by the DMSO. That 100% scrubbing efficiency is not actually obtained is indicated by the data in Table 11. Consider the case of nitrogen as a carrier gas. The concentration calculated from the pulse polarographic determination (Column 7) is systematically lower than the concentration calculated from the permeation tube data (Column 5). Significantly, the difference is essentially linear in the gas flow rate. These data indicate that the method described here for the analysis of SO2 in gases is limited by the precision of the measurements if the flow rate of the gas is less than 500 cm3/min. If higher flow rates are required because of low SO2concentration, a more efficient scrubbing technique is required. The data for air as the carrier gas show the same general trends as nitrogen data. The difference between the predicted SO2 concentration (Column 5) and the experimental value (Column 7) is, however, significantly smaller than that found for nitrogen. In fact, for the lowest flow rate of air, the sign of the difference is reversed. This is almost certainly the result of a small residual of oxygen in the air samples. There is a fortuitous cancellation of errors. The effect of residual oxygen can be reduced by increasing the deaeration time. Because of the low vapor pressure of the SO?over the solution, a much longer deaeration time could be used without seriously affecting the accuracy. Sensitivity. The noise level of the instrument used in the differential pulse mode is typically 2 nA (rms). The sensitivity of the determination predicted from a faradaic current of twice the noise is 0.3 p M in SO,. However, several factors limit the practical sensitivity limit to a higher figure. There are the usual losses associated with reactions with trace impurities (including oxygen and nitrogen oxides) and adsorption of SO, on the vessel walls. Although the vapor pressure of SO, over DMSO is very small, deaeration with nitrogen does lead to significant losses at very low concentration. The accumulation of these errors with the method described above leads to a sensitivity limit of about 5 X lO-'Mat the 25 error level. For gas samples, the sensitivity depends on the minimum detectable concentration indicated above and the maximum usable flow rate and contact time. The flow rate of the gas is limited by errors generated by inefficient trapping of the SO?. The losses of SO, in the DMSO solution described above are time dependent. This fact restricts the maximum contact time that can be used with gas samples of low SO2 concentration. Taken together, the sensitivity limit of this method (flow rate = 250 cm3/min,contact time = 20 minutes) is about 0.04 ppm SO, in the gas sample at the 25 error level. ANALYTICAL CHEMISTRY, VOL. 44, NO. 8, JULY 1972
The authors thank R. Cheng, J. 0. Frohliger, and Morton Corn of the Department of Occupation Health, Graduate School of Public Health, University of Pittsburgh, for aid in making measurement of the flow system; R. J. Bertozzi of the Department of Chemistry, University of Pittsburgh, for aid in preparation of the solvent and for aid in construction of the reference electrodes; and G. W. O’Dom and J. B. Flato of
Princeton Applied Research Corporation for aid in setting up System used. and modifying the RECEIVED for review December 23, 1971. Accepted March 14, 1972. R.W.G. was supported by training grant No. Tl AP-41, National Air Pollution Control Administration, Environmental Protection Agency. This work was supported, in part, by National Science Foundation grant GP-7531.
Bipolar Digipotentiogrator for Electroanalytical Uses Direct Conversion of Charge to a Digital Number William W. Goldsworthyl and Ray G. CIem Lawrence Berkeley Laboratory, Unioersity of California, Berkeley, Calif: 94720
A bipolar digipotentiogrator has been built for the first time. I t functions as a potentiostat through pulsed injection or extraction of charge to maintain a control potential, and simultaneously serves as a current to digital converter. Counting and summing these pulses in time allows the instrument to serve as an integrator. It is capable of a current measuring precision of 0.01%. This device is the heart of a new system which includes a digital wait-gate, a pulse height analyzer, an analyzer interface, an incremental differentiator, a voltage-step ramp generator, and a program timer. Uses of this system in polarography and anodic stripping analysis are illustrated and possible uses in controlling or digitizing charge in other systems are discussed. GREAT TECHNOLOGICAL ADVANCES in the field of solid-state physics have resulted in the development of small, inexpensive electronic components which are extremely fast in operation and are capable of controlling either hole or electron current with equal ease while consuming very little power. Although these properties portend digital type applications, most designers prefer to employ the devices using the analog circuit designs and signal measuring techniques developed in the vacuum-tube era. It seems redundant t o convert charge, liberated in a detection process, into an analog of current or voltage, perform various analog operations on these quantities, then finally convert them to numerical form. If at all possible, it would be far better t o deal directly and immediately with this liberated charge on a digital basis a t the output of the detector system itself, since the number of steps required in the conversion of detector output signals into useful digital numbers would be reduced. Several advantages are gained by employing the direct conversion of charge to a digital number technique described herein. Because of the reduction in the number of steps required between detection and numerical output, considerable savings can be realized in the weight, size, cost, and power requirements of the instrument. Phase shifts, which result on passing a signal through many operational amplifiers and can lead t o deleterious oscillations, are generally obviated. And, since the detector output is immediately converted to numeiical form, data can be easily and promptly transmitted to reRetired on Dec. 31, 1971. All correspondence must be addressed to the second named author. 1360
ANALYTICAL CHEMISTRY, VOL. 44, NO. 8, JULY 1972
mote locations at signal levels much greater than background for more sophisticated analysis. The presented bipolar digipotentiogrator is a direct outgrowth of the previously developed digital integrator which served as a voltage t o frequency converter in our first electroanalytical effort ( I ) . The same principle was later applied to the digitizing of charge liberated in a nuclear-event detector and resulted in the development of a digital nuclear spectrometer (2). Both of these initial instruments served passively to neutralize incoming charge to a preset null through digital feedback of charge of the opposite polarity. Neither was designed t o perform a control function in the external system, although in some systems it seemed reasonable that they could be made to d o so. The validation of this reasoning took the form of the monopolar digipotentiogrator described in a recent correspondence (3). The present bipolar device is a more sophisticated instrument and is more generally applicable t o electroanalytical problems. It is presented as the heart of a new system which includes a digital wait-gate, a pulse height analyzer for data storage, an analyzer interface, a n incremental digital differentiator, a voltage-step ramp generator, and a program timer. Possible uses of this instrument in controlling or digitizing charge in systems other than electrochemical are also briefly discussed below. It is quite likely that a small computer could be programmed t o perform the same control and timing functions as the presented device through the use of software. INSTRUMENTATION
The system shown in Figure 1 is a complete divergence from previous electroanalytical systems. The familiar train of instruments, a potentiostat, a current to voltage converter, and a n analog t o digital converter are absent. In their place, one finds a single device, the bipolar digipotentiogrator, which functions as a potentiostat through pulsed injection or extraction of charge to maintain a desired cell control potential, and (1) R. G. Clem and W. W. Goldsworthy, ANAL.CHEM..43, 918 (1971). (2) W. W. Goldsworthy, Niicl. Instrum. Methods, 94,221 (1971). (3) W. W. Goldsworthy and R. G. Clem, ANAL.CHEM.,43, 1718 (1 971).