Determination of Sulfur(lV) and Sulfate in Aerosols by Thermometric Methods Lee D. Hansen, Larry Whiting,' Delbert J. Eatough,'
* Trescott E. Jensen, and Reed M. lzatt
Department of Chemistry and Center for Thermochemical Studies, Brigham Young University, Provo, Utah 84602
A rapid, inexpensive method is presented for the determination of S( IV) and sulfate in airborne particulate samples. The method consists of a thermometric titration of S(IV) with K2Cr207 followed by the direct injection enthalpimetric determination of sulfate with BaCI2 in an HCI, FeCI3 extract of the sample. The precision of the technique is f ( 5 % of the total S(IV) 3 nmol) and f ( 1 0 % of the total sulfate 30 nmol) in 2.25 ml of extractant solution. There are no major interferences with either determination;
+
+
The physiological problems associated with exposure to
SO2 and aerosols containing oxides of sulfur are well-documented and have merited serious concern ( I , 2 ) . Sulfur dioxide is known ( 3 ) to be adsorbed by airborne particulates. Amdur has reported increased pulmanary effects compared to those observed for SOn(g) alone when experimental animals are exposed to sulfate salts or to SOz(g) and particulate matter capable of oxidizing SO2 or catalyzing the oxidation of SO2 by 0 2 ( 4 , 5 ) . Similar effects have been postulated for man when exposed to a combination of SOS(g) and sulfate aerosols (6, 7). (Sulfate is used here to denote any or all species analyzed as sulfate in a hot water digest of the collected sulfate sample.) This irritant response could arise from the reaction either of S02, H 2 0 , 0 2 , and particulates to form H2S04 or sulfate salts ( 4 ) or of SO2 with metal-containing aerosols to form metal-sulfite complexes (8). Epidemiological studies suggest a positive correlation between mean mortality residuals and SO2 and/ or sulfate exposure levels while a decreased resistance to acute respiratory illnesses has been found in individuals who are exposed to high levels of SO2 and/or sulfate for a 3-year period or longer (2, 7). I t would be desirable to have a rapid, simple method for the determination of sulfur speciation in aerosols as it is not known with certainty whether S(IV), S(VI), or other species are responsible for the observed adverse health effects. The only previously reported technique for the determination of both S(1V) and S(V1) in particulate samples is the semiquantitative simultaneous determination of both using photoelectron spectroscopy, ESCA (9).However, the method is not suitable for routine work because the analysis is time donsuming, difficult to perform, and requires costly instrumentation. The commonly used method for determining sulfate in samples of airborne particulates (10, 11) begins with a hot water extraction which apparently leads to the oxidation of virtually all of the sulfur species present and subsequently to an overestimation of the sulfate originally present in the sample (12). In the extraction procedure described here, the compounds of interest are solubilized without oxidation of S(1V). Thermometric methods have been used for the determination of both
Present address, Department of Chemistry, University of Georgia, Athens, Ga. 30601. * Inquiries should be addressed to 158 FB, Brigham Young University, Provo, Utah 84602. 634
ANALYTICAL CHEMISTRY, VOL. 48, NO. 4, APRIL 1976
S(1V) and sulfate but not in the same sample and not in complex mixtures (13-16). The present study is the first to show the feasibility of using these techniques to determine sulfur speciation in airborne particulate samples. Thermometric titration and direct injection enthalpimetry were the methods of choice in the present study for several reasons. First, the reactions used, oxidation of S(1V) with K2Cr20~followed by the precipitation of sulfate with BaC12, are readily detectable by thermometric methods. Second, the method is capable of determining S(1V) and S(V1) in the presence of each other in the nanomole range. Third, the interference of other substances could be detected and appropriate corrections made, if necessary. Fourth, because of the general nature of thermal end-point detection, information could possibly be obtained on other species present in the sample. Last, the procedures developed are inexpensive, rapid, and easy to perform.
EXPERIMENTAL Reagents. K2Cr20: (NBS standard sample 136b), NaHS03 (reagent, MCW), Na2S03 (reagent, MCW), FeClr6H20 (analytical reagent, MCW), FeSO4.7HnO (reagent, B&A), Na2S04 (reagent, anhydrous powder, Baker), BaC12.2HzO (reagent. B&A), and HC1 (analytical reagent, MCW) were used in the preparation of solutions. All solutions were prepared in argon purged 0.1 M HC1, 5 m M FeC13 and stored and used under argon. Apparatus. A Tronac model 450 isoperibol calorimeter (17) equipped with a 3-ml Dewar (18, 191, two extra 0.015-inch i.d. Teflon tubes in the insert, and a 1-ml Gilmont precision buret was used in the study. Details of the instrument design have been published (18, 19). The thermistor bridge on this calorimeter had a sensitivity of -25 mV/OC. T h e bridge output was amplified by a Keithley Model 150B amplifier and recorded on an H P l700B recorder with a 1750A plug-in module. A proton induced x-ray emission spectrometer was used to obtain the elemental analyses of the extraits ( Z O , Z I 1. ComDosition of Extractant Solution. The composition of the extractlron solution was dictated by the following factors: a) it shoul'd have a concentration of H + a t least 0.1 M to prevent oxidation of S(1V) by Fe3+ and Cu2+ (8) which will be present in many samples, but not exceed 1 M H + to prevent reduction of S(1V) which takes place more easily in strong acid; b) the anion of the acid must not interfere with either the redox or precipitation reactions; c) the concentration of Fe3+ must he sufficiently high to prevent loss of SOZ(g) from the acid solution; and d ) sulfite and sulfate salts expected to be present in airborne particulates must be soluble in the solution. The solution must be kept free of 02 to prevent oxidation of the S(1V). A 0.1 M HC1 solution containing 5 mM FeCl 3 and kept under argon met these specifications and was used in the study. This extraction solution would dissolve the sulfates of Ca. Sr. Ba, and P b unless these elements were present in very large amounts. Basic extractant solutions were not considered because of the possible disintegration of some filter materials, insolubility of some sulfite salts in basic solution. and the lack of a convenient oxidizing agent for the sulfite determination. T h e replacement of the FeC13 with HgC12 as used in the WestGaeke method for analysis for SO,(g) (22) was considered but avoided because of possible reaction of the HgCl2 with sulfides in the particulates. Also, current data indicate that the Fe(S03); complex has a larger log K value for formation ( 8 ) than the corresponding Hg(S03); species (23). I t is reported, however, that the Fe(SOd+ complex undergoes an auto-redox process to form Fe2+ and a mixture of SO:- and S202- (24, 2 5 ) . But experiements showed that in 0.1 M HC1. 1 mhl NaHSO?. with an excess of Fe3+
-rl:;.olD~ 0
'\@/
0
=
-___ 0 004 o"y ' F,
I
I
8
Figure 1. Filtration assembly (1) Silicone rubber gasket, (2) depth filter, (3) pore filter, (4) Swinney filter holder, (5) 10-ml syringe (delivery), (6) 2 . 5 4 syringe (receiving), (7) Hamilton connector
present, oxidation of SO:- to precipitable SO:- accounted for less than 1%of the total sulfur during the first 24 h. Furthermore, when these solutions were titrated with CrZ0:- the end point quantitatively corresponded to formation of sulfate by the redox reaction. The sulfate formed was also quantitatively precipitated when excess BaClz was added. C h o i c e of O x i d i z i n g A g e n t for S(1V) D e t e r m i n a t i o n . Several oxidizing agents have been used for the determination of S(1V) (26, 2 7 ) , including solutions of I,, KMn04, HzOz, KI03, Ce(IV), As(V), V(V), Br2, KBr03, and K2C1-207.KZCr207 was selected because it can be weighed as a primary standard and interfering reactions are minimal. Iz and KI03 could not be used because the FeC13 used in the extraction solution (or Cu(I1) and Fe(II1) from the sample) would oxidize the I- product formed. H202, aside from its instability in standard solution, was catalytically decomposedin some of the sample solutions. In addition, the formation of Fe(III)-H202 complexes interfered with the thermometric end point in the S(1V) oxidation. KMn04, Ce(IV), and KBr03 are all unstable in the 0.1 M HC1 solution. Coprecipitation of MnOh, BrO,, IO,, and Cr2O;- with sulfate in the BaC12 precipitation step could also be a problem. This is minimized with Crz0:- since Cr20;- is not isostructural with SO:-. V(V) and As(V) as titrants do not give sharp end points a t low concentrations of sulfite and do not oxidize SO;-in the presence of Fe(II1). Br2 was considered but rejected because of the difficulty of maintaining standard solutions. KzCr207 has been reported to produce dithionate from the oxidation of sulfite, but all of these reports refer to solutions of p H greater than one and usually to the addition of sulfite to dichromate rather than the reverse (24-27). P r o c e d u r e . The extraction was done by placing a filter or fraction of a filter with the particulate material on it in a Vacutainer tube or a test tube with a Twistit stopper. After purging the tube with argon, a volume of extractant solution sufficient to cover the filter was injected through the septum. Monoject 200 stainless steel needles with aluminum hubs were used throughout this study. The tube was then placed in a plate rotating a t 6 rpm, refrigerated a t 5 "C, and extracted for 2 h. The rotating plate was tilted at approximately a 20" angle to prevent contact between the extracting solution and the rubber septum or stopper. After the extraction period, the solution was withdrawn into a 10-ml Hamilton gas-tight syringe and filtered into a Hamilton 2.5-ml gas-tight syringe through Millipore filters (depth, AP 2001000, pore GSTF 01300) held in a polypropylene Swinney holder (Millipore SX0001300) (See Figure 1).An aliquot of the filtrate was injected into the calorimeter reaction vessel through one of the extra Teflon tubes. The Teflon tube and calorimeter reaction vessel had previously been flushed with argon. Any sample left in the injection tube was flushed down with argon and the argon was turned off before the titration was begun. The reaction mixture was then titrated with an argon purged, standard KzCrzOf solution in 0.1 M HC1, 5 mM FeC13. T h e titration was begun when the reaction vessel temperature was still sufficiently below bath temperature that the temperature in the reaction vessel a t the end of the titration would be lower than the bath temperature. The redox thermogram was recorded a t 1 2 in./min and 10 fiV/in. (Keithley 1-mV input, 1-V output, 10 mV/in. on the recorder.) After completion of the redox titration, the reaction mixture was heated to bath temperature, the recorder was switched to 2 in./ min, and 0.25 ml of 0.1 M RaC12 (at the bath temperature) in 0.1 M HC1, 5 m M FeC13 was injected into the reaction vessel. Rapid injection (>0.01 ml/s) through the small diameter (0.015-inch i.d.) Teflon tube resulted in a sizeable heat production which was not reproducible, and therefore a syringe with a threaded plunger was used to deliver the BaClz smoothly over a 1-min period. Also, to promote the rapid precipitation of BaS04, the inside of the Dewar was coated with Bas04 before use and rinsed only with deionized
-0.8L 0 l 2
4
_i
t L L - - - 0 0 3 2 0 2 4
Time ( m i n i
Figure 2. Typical curves for thermometric titration of S(IV) with K2Cr207 and DIE determination of sulfate w i t h BaClp The buret is turned on at point A and off at C. The end point of the redox titration is at B. Between C and D, the temperature in the reaction vessel is adjusted to be the same as the bath temperature. At D, the BaClp is injected. Point E, at which AT is measured, is midway in time between D and F. The slope between A and B is -0.007 "C/fiequiv and the temperature rise at D is -(0.0032 "C/fimol)-0.005 'C
water between runs in order not to entirely remove the coating. Figure 2 details the thermogram obtained and illustrates the titration and precipitation procedures. The amount of sulfate represented by the AT value was found by comparison with a standard curve obtained by the injection of the same BaC12 solution into solutions of known amounts of NazS04 in 0.1 M HC1, 5 mM FeC13. The sulfate initially present in the particulate was taken to be the difference between the sulfate measured in the injection procedure and the S(1V) found in the dichromate titration. The amount of S(1V) in the sample was found from the appropriate end point or end points in the redox portion of the thermogram. Since there may be several different reducing agents (e.g., Fe(II), As(III), and organics) present, the portion of the thermogram representing the titration of S(1V) was identified from the heat of the reaction (proportional to the slope of the thermogram). In those cases, where doubt still existed as to the identity of the reducing agent, standard addition of NaHS03 to the HC1 extract was used to verify the choice of region. In no case were any additional end points observed in the thermogram after the addition of NaHS03 to the sample. Also, the amount of NaHS03 added was always quantitatively ( f 5 % )recovered in the determination. The procedure was tested using two different sample sizes. Most samples were collected from the workroom vicinity of the reverberatory furnaces or convertors of a copper smelting facility near Salt Lake City, Utah, over an 8-h period using either an Anderson Hi Vol sampler (four stages plus a backup) or a small volume personal monitor sampler (28). Other samples (Anderson) were collected a t various locations in Utah, Salt Lake and Weber Counties, Utah (29).One quarter of each Anderson filter (30 cm, glass fiber) containing 3-125 mg of particulates was extracted with 20.0 ml of HC1, FeC13 extractant. A 2.25-1311 aliquot of the extracted solution was titrated with 0.15 ml of 5 mM KZCr207. The personal monitor filters (3.7 cm, PVC) contained 0.05-5 mg of particulate material, and the whole filter was extracted with 6.0 ml of the HC1, FeC13 solution. A 2.25-m1 aliquot of the resulting solution was titrated with 0.15 ml of 1 mM K2Cr207. For both large and small filters, 0.25 ml of 0.10 M BaC12 solution was injected. Blank corrections were made by repeating the above described procedures on new filters from the samelot.
RESULTS Sulfur(1V). The d e t e r m i n a t i o n of S(1V) w i t h K2CrZ07 w a s t e s t e d b y t i t r a t i o n of a l i q u o t s of sulfite solutions which were i n d e p e n d e n t l y checked b y t i t r a t i o n w i t h triiodide. T h e s e c o m p a r a t i v e recovery e x p e r i m e n t s w i t h solutions p r e p a r e d b y weighing NaHS03 and NaZS03 s a l t s i n d i c a t e that all S(1V) i n the H C l , FeC13 e x t r a c t a n t solution is oxidized to s u l f a t e d u r i n g t h e r e d o x t i t r a t i o n and that f o r m a t i o n of o t h e r S - c o n t a i n i n g species (Le., d i t h i o n a t e , etc.) is ANALYTICAL CHEMISTRY, VOL. 48, NO. 4, APRIL 1976
635
Table I. Results of Recovery Experiments for the Thermometric Titration of Freshly Prepared Solutions of Na,SO, and Na,SO, with Cr,O:- or I; Titrate solution pequiv so:taken0
pequiv so:takena
7.59 7.59
...
...
7.59 5.21 2.34
No. of
Calculated % purity of Na,SO,
Solvent
Titrant
replicates
0.1 M HCI 0.1 M HCl, 5 mM FeCl, 0.1 M HC1
Cr,O:Cr,Oi-
5 5
72.7 81.7
i i
11.6 1.6
1 ;
3
82.4
f
4.0
so:-
recovery, @
.. ..
1:. 3 7 8 . 8 i 6.3 ... H,O 2.90 0.1 M HCl, Cr,Oi4 81.9 i 0 . 8 9 9 . 6 i 0.9 5 mM FeC1, 2.34 2.90 0.1 M HCl, Cr,O:3 8 1 . 9 i 0.9 9 8 . 9 i 3.0 5 mM FeCl,, 5 mM CuC1, 2 81.4 i 0.2 ... 2.34 2.90 0.1 M HC1 1; a Calculated from the weight of Na,SO, or Na,SO, taken assuming the material to be pure. b The expected SO:- is the sum of the pequiv SO:- and SO:- in the prepared titrate solution, calculated assuming the purity of the Na,SO, to be 82% and the contaminant to be Na,SO,. Total sulfate after oxidation of SO:- was determined thermometrically as described in the text. CTitrant was in 0.1 M HC1.
...
negligible. Results using a common Na2S03 sample are summarized in Table I. All Cr2O;- titrations with solutions containing 5 m M FeC13 gave results consistent with 13 titrations. The recovery of 99% of the total expected SO:- in the two cases studied indicates only SO:- is formed by t h e redox titration with Cr20;- under t h e conditions employed by this study. Similar results were obtained with other Na2S03 or NaHS03 samples. In the absence of 5 m M FeC13, t h e recovery of SO:- decreased over a 48-h period to less than 30%. Sulfate recovery remained constant in samples stored in stoppered bottles and decreased in samples swept with Ar, indicating oxidation of SO;-to SO,"- and/or loss of SO:- as SOz(g). I n contrast, solutions containing 5 m M FeC13 were stable over t h e same time period. Addition of Na2S03 t o particulate samples gave SO:- and SO:- recoveries comparable t o those in Table I in all cases studied. Based on the agreement among replicate runs for known solutions and for over 250 actual particulate samples, t h e precision of the titration for S(1V) is estimated t o be f (5% of t h e total S(1V) 3 nmol). Sulfate. Calibration data for the precipitation of sulfate by Ba2+ were taken over a range of sulfate amounts from 0.5 t o 12 kmol to cover the complete range of sulfate concentrations likely to be found in airborne particulates. T h e actual calibration curve varies from one instrument to another since the measured temperature change (mV) depends on the thermistor sensitivity and t h e heat capacity of the reaction vessel used as well as on the AH value for BaS04(s) precipitation. Typical calibration data are given in Table 11. The plot of AT vs. pmol of sulfate is linear, with a correlation coefficient of 1.000, a slope of 75.3 i 7.4 pV/pmol, and an intercept of -12.8 f 2.4 pV. T h e nonzero intercept on t h e millivolt axis is due t o the heat of dilution of the BaC12 solution. As indicated by the data in Table I, recovery of sulfate from prepared standard solutions was 99%. Similar results were obtained with solutions of Na2S04 containing no sulfite salts. From the precision of these results and of duplicate trials on over 250 actual particulate samples, the precision of the sulfate determination is estimated to be f (10%of t h e total sulfate + 30 nmol). Comparison with R e s u l t s by Other Methods. ESCA spectra have been run on two of t h e aerosol samples studied to date (28, 30). The spectra verified the presence of both S(1V) and S(V1) in these samples with the relative amounts being comparable to those obtained thermometrically. No quantitative comparison is possible since ESCA is
+
636
ANALYTICAL CHEMISTRY, VOL. 48, NO. 4, APRIL 1976
Table 11. Typical Calibrat.ion Data for the Precipitation of SO:- with Excess Ba2+ in 0.1 M HCl, 5 mM FeC1, Solution SO:-pmol taken 12.473
10.900 9.968 8.680 7.489 6.500 4.984 4.490 4.192 2.515 1.800 1.677 0.850 0.547 0.000
C alculatedc-taken AT,a p V b
9362 795 I 739 i 624 i 561 i 488 i 374 i 334 i 292 i 172i 117 i 107 2 47i 39 i -7 i
3 5
5 18 2 4 6 1 7
13 2 8 2 5 3
So:-, pmol
Error, 7%
0.122 -0.177 0.012 -0.227 0.128 0.148 0.151 0.114 -0.146 -0.062 -0.077 -0.087 -0.056 0.141 0.077
1.0 -1.6 0.1 -2.6 1.7 2.3
3.0 2.5 -3.5 -2.5 -4.3 -5.2 -6.6 25.7
...
See Figure 2. b Values are the average of from 3 t o 5 replicate determinations. Uncertainties are given as the standard deviation of the mean. C Calculated values are (AT + 1 2 . 8 ) / 7 5 . 3 3 . a
a surface analysis tool and t h e techniques described here determine total amounts in t h e sample. The S(1V) was shown t o be present either as a sulfite compound or as adsorbed SO2 depending on the assignment of t h e spectral band a t 168 eV. S(-11) was also shown to be present in t h e samples. Several collected particulate samples were submitted t o a n independent laboratory for independent analysis for total sulfate by a technique (10, 1 1 ) which consists of a n extraction with boiling water, precipitation of B a s 0 4 from t h e supernatant, filtration, dissolution of the filtered B a s 0 4 in EDTA solution, and measurement of the Ba by atomic absorption spectrophotometry. Blank results for t h e two extraction procedures, Table 111, were in good agreement. However, the total sulfate found by the hot water extraction method was consistently higher. In a comparison of 18 samples, the hot water extract results were higher by a factor of 2.4 f 1.1 than the sum of the S(1V) plus sulfate found in the HC1 extract by the thermometric methods. It was verified t h a t the difference was due to the extraction method and not t o differences in the method of determina-
Table 111. Comparison of Results Obtained on Blank Filters and Collected Particulate Samples by Two Different Extraction Procedures mg SO:-/Filter Total collected particulate, mg
Filter
HCl, FeCl, digest thermometric0
Hot H 2 0 Digest Thermometric
Spectrophotometric
...
Not detectable
Polyviny lchloride 0 0.021 [ n o S(IV)] Glass Fiber I 0 2.2 [27% S(IV)] 0 1.0 [41% S(IV)] Glass Fiber I1 0 0.5 [72% S(IV)] Glass Fiber I11 240 37.8 [82% S(IV)] Glass Fiber I1 Glass Fiber I1 88 2.8 [27% S(IV)] Glass Fiber 111 136 22.7 [92% S(IV)] a Total sulfate found in the BaCl, precipitation. The value in [ ] is the mol 5% of Cr,O:- titration.
... 1.1
...
1.4
0.4 0.5 74.2 125.9 11.3 9.1 25.4 29.1 total sulfate detected as S(1V) in the
Table IV. Log K and AH Values at 25 “C for the Principal Reactions Occurring in the Redox Analysis of S(IV) in Solutions Containing S, Fe, and/or As Reaction
A. 3H,S03 + Cr,O:- + 5H+ = 3HSO; + 2Cr3+ + 4H,O B. S
+
Cr,OS-
+
7H+ = HSO;
C. %H,S + Cr,O:-
+
+
2Cr3+ + 3H,O
7%H+ = 3/4HSO; + 2Cr3+ + 4H,O
D. 6Fe2+ + Cr,Of- + 14H+ = 6Fe3+ + 2Cr3+ + 7H,O E. 3FeSO: + Cr,O:- + 11H+ = 3HSO; + 2Cr3+ + 3Fe3+ + ~ H , O F. 3H,AsO, + Cr,O:- + 8H+ = 2Cr3+ + 3H,AsO, + 4H,O G. FeSO: + H,AsO, + H+ = HSO; + H,AsO, + Fe3+ H. 2Fe3+ + H,SO, + H,O = 2FeZ++ HSO; + 3HC I. Fe3+ + H,SO, = FeSO: + 2H+ 0 This study. The Lb7 values with superscript a only were obtained by titration of the indicated reactants. b Calculated from data in A. 3. deBethune and N.A. S. Loud, “Standard Aqueous Electrode Potentials and Temperature Coefficients a t 25 ”C,” published by C. A. Hampel, 8501 Harding Avenue, Skokie, Ill. 1964. C Calculated from data in “Handbook of Biochemistry,” 2nd ed., H. E. Sober, Ed., Chemical Rubber Co., Cleveland, Ohio, 1970. d The AH value is in units of kcali mol Fe3+.
tion by running thermometric determinations for S(1V) a n d sulfate on the boiling water extract after addition of HC1 and FeC13. As expected, no S(1V) was detected in these solutions. T h e thermometric results for sulfate in t h e boiling water extracts are in good agreement with those obtained by t h e independent laboratory, Table 111.
DISCUSSION I n 0.1 M HCl, SO2 is rapidly evolved from a NazS03 solution. However, if Fe3+ is present in a 5:l ratio of Fe3+ t o SOz, no SO2 loss occurs, presumably due t o t h e formation of stable Fe3+-SOi- complexes (8).A 1 : l Fe3+ t o S O z ratio is not sufficient to eliminate SO2 loss from a 1 m M solution of SO2 in 0 , l M HCl. As is illustrated by t h e d a t a in Table IV, both t h e log K and AH values for t h e oxidation of SO,”-by Cr20:- are significantly affected by t h e formation of FeSOi complexes. Thus, in t h e absence of Fe(III), SO;- (Reaction A) is titrated before Fez+ (Reaction D) by Cr2O!-. However, addition of Fe(II1) to t h e solution results in Fez+ being titrated first (compare Reactions D and E). In t h e presence of Fe3+, t h e SO:- produced by t h e reaction of CrzO;-with S(1V) is precipitated quantitatively by BaC12. However, no precipitate occurs in the absence of Fe3+ suggesting that kinetically stable Cr3+-SO:- complexes are produced under these conditions. Formation of kinetically nonlabile Cr3+-SOi- complexes may also explain the less exothermic AH value measured for t h e oxidation of SO:- by Crz05- compared to t h a t
predicted from literature E M F data, Reaction A. Table IV. Dichromate ion has been reported t o react withS(1V) in a nonstoichiometric fashion because of t h e production of varying amounts of S20:- (24-27). Both t h e p H and catalysts used have an effect on t h e amount of SzOi- produced ( 2 5 ) .T h e reaction conditions chosen, Le., p H 1.0 and with Fe3+ present, d o not favor production of dithionate. We found that, within t h e accuracy of t h e calorimetric end point (fO.5%), the titation of a NaZS03 solution in 0.1 M HCl, 5 m M FeC13 with KZCr207 gave t h e same normality as titration with 1, which is reported t o give only sulfate from t h e oxidation of sulfite (25, 2 7 ) . Thus, if dithionate is produced, i t causes a n error of t h e order of 0.5% or less. T h e excess sulfate found by t h e boiling water extraction method could result either from more complete extraction t h a n is obtained with cold HC1 or from formation of sulfate in t h e boiling water from sulfur-containing species other than S(1V) oxides and sulfate. Since extraction times with t h e HC1 solutions several times longer than 2 h did not change the apparent sulfate concentration, and since thermodynamic data ( 2 3 ) indicate B a s 0 4 a n d PbS04 will both be soluble in t h e HC1 extract at the particulate levels studied (>1 mg total particulate/ml extractant solution), the latter explanation seems more reasonable. Reactions which could produce sulfate during t h e boiling water extraction include disproportionation of elemental sulfur and oxidation of sulfides by oxygen from t h e air. T h e production of a new reducing agent during t h e boiling water extraction ANALYTICAL CHEMISTRY, VOL. 48, NO. 4, APRIL 1976
637
does show that chemical changes take place during extraction by this method. Similar results have been reported by Appel and co-workers ( 1 2 ) who showed S(-11) and S(1V) are analyzed as S(V1) in hot water extractant solutions. Interferences. No reducing agents have been found which interfere with the redox titration for S(1V). Ferrous iron, As(III), As(V), HzS, colloidal sulfur (Reactions B-D, F, and G, Table IV), and organic aldehydes were examined. In each case the slope of the thermogram was different than that for S(1V) and, in addition, the potentials for oxidatiori were sufficiently different that no interference with the S(IV) oxidation was seen. In fact, the redox potentials and enthalpy changes are sufficiently different t h a t all of the above can be determined in a given mixture with a single thermometric titration with dichromate ion. The As(II1) in solutions prepared either by dissolving As203 in HC1 or by dissolving As203 in NaOH and acidifying the resulting solution was titrated after S(1V) and the redox reaction was slow so that no sharp end point was seen for the As(II1) species. Similar As(111) species have been identified in aerosol samples from a smelter using an ore pre-roast process but not in samples from a smelter wliich does not pre-roast the ore (28). All Fe(I1) species either present in aerosols or added to extract solutions are titrated before S(1V) as predicted by the data in Table IV (Reactions D and E). As(V) will not oxidize the FeS0; complex (Reaction G) and does not interfere with the determination of S(1V) using the method described here. In the Bas04 precipitation process, experiments showed that excess KzCrz07, and KCl all gave positive deviations of 5%, FeC13 gave a negative deviation of 3%, and CrC136H20, KIO3, NaF, CaC12, and NaN03 did not interfere when added in approximately equimolar quantities with the sulfate (except for FeC13 a t 5 mM). The effects.observed did not depend strongly on concentration above a minimal value or appear to be additive. Thus, systematic errors in the sulfate determination could be as large as 5%. These systematic errors appear to be slightly larger than those previously reported for the thermometric determination of SO,'- with BaC12 in solutions containing more than 1 g of sulfate per liter (14). Interferences which occur during the extraction process are more difficult to evaluate since there are many agents which could coexist with S(1V) in the solid particulate but which would react upon dissolution of the solid in the 0.1 M HCl, 5 mM FeC13 solution. Evidence of such a situation could be obtained in several ways: a) the reduced form of an oxidizing agent might be titrated by the dichromate ion in the redox titration; b) data from elemental analysis of the sample would suggest what possibilities exist; and c) methods such as ESCA could be used to detect specific species in the solid sample as suggested by the first two detection methods.
ACKNOWLEDGMENT Appreciation is expressed to Thomas G. Smith for collection of the aerosol samples, to Robert G. Meisenheimer for
638
ANALYTICAL CHEMISTRY, VOL. 48,
NO. 4,
APRIL 1976
providing the ESCA data, to Nolan F. Mangelson for the
PIXE analysis results, and to Eliot A. Butler for constructive comments during preparation of the manuscript.
LITERATURE CITED R. W. Buechley, W. B. Riggan, V. Hasselblad. and J. B. VanBruggen, Arch. Environ. Health, 27, 134 (1973). J. G. French, G. Lowrimore. W. C. Nelson, J. F. Finklea. T. English, and M. Hertz, Arch. Environ. Health, 27, 129 (1973). B. M. Smith, J. Wagman, and B. R. Fish, Environ. Sci. Technol., 3, 558 (1969). M. 0. Amdur, Arch. Environ. Health, 23, 459 (1971). M. 0. Amdur, J. Air Pollut. Control Assoc., 19, 638 (1969). C. M. Shy and J. F. Finklea, Environ. Sci. Technol., 7, 204 (1973). J. F. Finklea, C. M. Shy, G. J. Love, C. G. Hayes, W. C. Nelson, R. S. Chapman, and D. E. House, "Summary and Conclusions Based Upon CHESS Studies of 1970-71", in: "Health Consequences of Sulfur Oxides: A Report from CHESS, 1970-1971", U.S. Environmental Protection Agency, Research Triangle Park, N.C., Publ. No. EPA-650/1-74004, 1974. L. D. Hansen, D. J. Eatough, L. Whiting, C. H. Bartholomew, C. L. Cluff, R. M. Izatt, and J. J. Christensen, "Trace Substances in Environmental Health, VIII", A Symposium, D. D. Hemphill, Ed., University of Missouri Press, Columbia, Mo., 1974, p 393. N. L. Craig, A. B. Harker, and T. Novakov, Atmos. Environ., 8, 15 (1974). lntersociety Committee, "Methods of Air Sampling and Analysis", American Public Health Association, Washington, D.C., 1972, p 296. A. Wollin, At. Absorp. News/., 9, 43 (1970). B. R. Appel, E. L. Kothny, E. M. Hoffer, J. J. Wesolowski, and R. D. Giauque. Proc. lnt. Conf. Environ. Sensing Assessment, Las Vegas, Nev., September 1975, in press. G. A. Vaughan, "Thermometric and Enthalpimetric Titrimetry", Van Nostrand-Reinhold Co., London, 1973, pp 100ff; H. J. V. Tyrrell and A. E. Beezer, "Thermometric Titrimetry", Chapman and Hall, Ltd., London, 1968, p 119ff. M. Nakanishi and S. Fujieda, Anal. Chem., 46, 119 (1974). M. B. Williams and J. Janata, Talanta, 17, 548 (1970). G. Dube and F. M. Kimmerle, Anal. Chem., 47, 285 (1975). Literature describing this equipment is available from Tronac, Inc.. 1804 South Columbia Lane, Orem, Utah 84057. L. D. Hansen, T. E. Jensen, S.Mayne, D. J. Eatough, R. M. izatt, and J. J. Christensen, J. Chem. Therrnodyn., 1, 919 (1975). L. D. Hansen, R. M. Izatt, D. J. Eatough, T. E. Jensen, and J. J. Christensen, "Recent Advances in Titration Calorimetry", in "Analytical Calorimetry", Vol. 3, Plenum Press, New York, 1974, p 7. N. F. Mangelson, G. M. Allison, D. J. Eatough, M. W. Hill, R. M. Izatt. J. J. Christensen, J. R. Murdock, K. K. Nielson, and S. L. Welsh, "Trace Substances in Environmental Health-VII", A Symposium, D. D. Hemphill, Ed., University of Missouri Press,, Columbia, Mo., 1973, p 379. K. K. Nielson, Ph.D. Thesis, Brigham Young University, 1975. P. W. West and G. C. Gaeke, Anal. Chem., 28, 1816 (1956). L. G. Sillen and A. E. Martell, "Stability Constants of Metal-Ion COmplexes," Special Publication No. 17, The Chemical Society, London, 1964. H. Bassett and W. G. Parker, J. Chem. SOC., 1951, 1540. G. P. Haight, Jr., E. Perchonock, F. Emmenegger, and G. Gordon, J. Am. Chem. SOC., 87, 3835 (1965). L. V. Haff, "Oxygen Containing Inorganic Sulfur Compounds", Anal. Chem. Sulfur its Comp., Part I, J. H. Karchmer, Ed., in Chem. Anal. 29, P. J. Elving and I. M. Kolthoff, ed., Interscience, N.Y., 1970, pp 221ff. W. C. E. Higqinson and J. W. Marshall, J, Chem. soc., 1957, 447. T. J. Smith, D. J. Eatough. L. D. Hansen, and N. F. Mangelson, J. Environ. Cont. Tox., in press. L. D. Hansen, D. J. Eatough, N. F. Mangelson. T. E. Jensen, D. Cannon, T. J. Smith, and D. E. Moore, Roc. lnt. Conf. Environ. Sensing Assessment, Las Vegas, Nev., September 1975, In press. R. G. Meisenheimer, Lawrence Livermore Laboratory, personal communication, 1975.
RECEIVEDfor review September 29, 1975. Accepted December 29, 1975. This work was supported in part by funds from NIOSH.
