1908
Anal. Chem. 1980,
52, 1908-1912
indicator of Cu(I1) complexing and/or chelation than is AGO4. Equations 30 to 32 form the basis of another simple calculation procedure having considerable diagnostic value. Values of dx,/dxsH, and dXSH/dXSH2 calculated by graphical differentiation are listed in Table IV. Three conclusions are drawn from this table about the experimental data and the complexing or chelation reaction: (a) The mole fraction data are internally self-consistent to f 2 . 7 ~ . (b) Cu'+ has reacted with t h e bidentate sites SH and SH2in t h e SH t o SHYmole ratio range 2 Ia 5 3. a gives the reaction stoichiometry in Equations 30. (c) As the degree of protonation increases from 0.08 to 0.33, the increasing a indicates an increasing preference of t h e Cu" for reaction with sites in the SH form. As CU'+ complexing displaces protons, t h e decreases in electrostatic charge on the fulvic acid is 2a
-
(a
+ 1) = ( a
-
1) =
dXSH ~
LITERATURE CITED Buffie, J.; Greter, F.: Haerdi, W. Anal. Chem. 1977, 49, 216. Gamble, D. S.; Schnitzer, M. "Trace Metals and Metal Organic Interactions in Natural Waters", Singer, P. C., Ed.; Ann Arbor Science Publishers: Ann Arbor, Mich.. 1973; Chapter 9. Reuter, J. H.; Perdue, E. M. Geochim. Cosmochim. Acta 1977, 4 1 , 325. Langford, C. H.; Khan, T. R.; Skippen, G. B. Inorg. Nucl. Chem. Lett. 1979, 15, 291. Schnitzer, M.: Khan, S. U. "Soil Organic Matter"; Elsevier Scientific Publishing Company: Amsterdam, 1978. Schnitzer, M.; Khan, S. U. "Humic Substances in the Environment"; Marcel Dekker: New York, 1972. Gamble, D. S. Can. J . Chem. 1970, 48, 2662. Gamble, D. S. Can. J. Chem. 1972, 50, 2680. Bruch, R. D.; Langford, C. H.; Gamble, D. S. Can. J . Chem. 1978, 56, 1196. Gamble, D . S.; Schnitzer, M.; Hoffman, I. Can. J . Chem. 1970, 48, 3197. Stevenson, F. J.; Krastanov, S. A.; Ardakani, M. S. Geoderma 1973, 9 , 129. Manning, P. G.; Ramamoorthy, S. J. Inorg. Nucl. Chem. 1973, 3 5 , 1577. Bresnahan, W. T.; Grant, C. L.: Weber, J. H. Anal. Chem. 1978. 50, 1675. Saar, R. A,; Weber, J. H. Can. J . Chem. 1979, 5 7 , 1263. Gamble, D. S.; Langford, C. H.; Tong, J. R. K. Can. J. Chem. 1976, 54, 1239. Gamble, D. S.; Schnitzer, M.; Skinner, D. S. Can. J . Soil Sci. 1977, 57, 47. Gamble, D. S. Can. J. Chem. 1973, 5 1 , 3217. Langford, C . H.; Khan, T. R. Can. J. Chem. 1975, 53, 2979. Schnitzer, M.; Dejardin, J. S. Soil Sci. SOC.Am. Proc. 1962, 26, 362. Schnitzer. M.; Skinner, S. I. M. Soil Sci. 1963, 96, 86.
(47)
dXSHl
This measures the effectiveness with which the Cu'+ displaces the protons. I t is found in Table IV that the protons are less effectively displaced as xSH, increases. Since this last point is contrary to expectation, this raises questions about changes in conformation and aggregation. Because of this and because there is some suspicion that light scattering may exist in the absorbance curve in Figure 6 beyond 3.4% l o 4m chelate, it is believed that light scattering measurements will be required in the future for this system.
RECEIVED for review February 26, 1980. Accepted June 6, 1980.
Combustion-Ion Chromatographic Determination of Chlorine in Silicate Rocks Keenan L. Evans and Carleton B. Moore* Department of Chemistry, Arizona State University, Tempe, Arizona 8528 1
A method is described for the analysis of chlorine in silicate rock standards. The proposed combustion-ion chromatographic method is compared with previous methods of analysis and optimum working conditions are investigated. Detection limits are presently 0.8 pg absolute and precision is commonly f5% relative standard deviation. The method is rapid and compares favorably with neutron activation analysis and X-ray fluorescence techniques. Chlorine values are listed for 28 silicate rock standards.
technique is rather expensive and the post-irradiation separation procedures can be time consuming and laborious. The X-ray fluorescence method is rapid and yields values which are in good agreement with neutron activation analysis values. However, i t is a secondary method of analysis a n d fresh standard pellets must be prepared and run with each lot of samples, because of C1 adsorption from the atmosphere upon regassing of the pellets after analysis under vacuum conditions. High and inconsistent C1 content of the pellet binding material can also be a problem in t h e XRF method (14). Ion chromatographic techniques ( I 6) offer certain advantages over other methods. T h e proposed combustion-ion chromatographic method (CIC) eliminates many of the above mentioned problems usually encountered in C1 analysis of rock samples. Sample preparation is minimized and so most problems due to lab contamination are avoided. Once sample solutions have been obtained, they may be stored, if necessary, before final IC determination and standards may be saved for subsequent analyses. The technique is rapid, precise, capable of handling large numbers of samples with ease, and does not require a large number of geochemical standards for the preparation of a working curve. In the present study, up t o 60 samples a day including blanks were run with the induction furnace and u p to 40 sample solutions a day were analyzed with t h e IC.
T h e determination of chlorine in rock samples is important because of t h e significant role t h a t chlorides play in the geochemical para-genesis of geologic sequences. Mason ( I ) states t h a t "chlorides are the most important form in which metals are removed from a magma". Numerous workers have commented on t h e role of chlorides in ore deposition (1-71, rock weathering ( 8 , 9 ) ,and the metamorphism ( 1 0 , I I ) . Thus, t h e need for a rapid, accurate method of chlorine determination, which is capable of handling large numbers of rock samples easily a n d relatively inexpensively, is well documented. Previous methods of analysis for chlorine in geochemical samples include spectrophotometry (12), neutron activation (NAA) (13),and X-ray fluorescence (XRF) (14,151. Extensive sample preparation makes chlorine contamination a serious problem in the spectrophotometric method, which also suffers from a lack of precision. While neutron activation techniques have been shown to be highly accurate, precise, a n d avoid many of the contamination problems of other methods, the 0003-2700/80/0352-1908$01 OO/O
EXPERIMENTAL Apparatus. A DIONEX model 10 Ion Chromatograph equipped with a 3 X 150 mm anion precolumn, a 3 X 500 mm anion separator column, and a 9 X 250 mm anion suppressor
6
1980 American Chemical Society
ANALYTICAL CHEMISTRY, VOL. 52, NO. 12, OCTOBER 1980
1909
Table 1. Comparison of Calculated CT Concentrations by Method I and Method I1 weight, sample
split
g
c1, PPm Method Method I I1
NIM-G NIM-G NIM-G
304 304 304
0.1915 0.2143 0.2336
210 209 213
202 209 211
BCR-1 BCR-1 BCR-1
51/19 51/19 51/19
0.2178 0.2073 0.1986
59 63
62
60
58
59
C1 content of the original sample was calculated from the following formula: Figure 1. Induction furnace arrangement for chlorine determination
column was utilized for final determination of C1- in the sample solutions. A 0.5-mL sample loop and full scale sensitivity settings of 1.0, 3.0, and 10.0 pmho/cm were used depending on the C1concentration of the sample solutions. A 0.003 M NaHC034.0024 M Na2C03IC eluent (El) at a flow rate of 138 mL/h was used for all determinations. Chromatograms were recorded on a Heath model EU-20B recorder. A 10-m length of 1.5-mm i.d. Teflon tubing was installed in front of the pressure gauge to act as a pulse dampener and weather-stripping was installed around the column panel door to minimize the effects of air drafts. A LECO induction furnace model 523-700 was used t o burn the samples in preparation for final IC determination. The combustion products delivery tube was modified with glass tubing to accept Nalgene LPE or PP 125 mL bottles containing the El trapping solution as shown in Figure 1. A variety of induction furnace operating conditions were investigated. All glassware was originally leached in 1.0 M HN03, and then thoroughly rinsed with doubly distilled H20. Class A volumetric glassware was used in preparing all IC standards. Prior to IC analysis, all beakers and syringes were doubly rinsed with doubly distilled H 2 0 and then doubly rinsed with sample solution. Nalgene ware was rinsed and leached overnight with doubly distilled H 2 0 prior to use. Sample Preparation. Rock samples were weighed as air-dried powders into ceramic LECO crucibles, ~528-031,which had previously been heated in a well ventilated oven to 1200 "C for a t least 12 h, allowed to stand overnight at 600 "C, and then air cooled to room temperature, whereupon 1 scoop (0.7 g) of LECO iron chip accelerator, 2501-077, was added. One scoop (1.0 g) of LECO copper accelerator, ~501-263,was placed on top of the samples and the crucibles were topped with a porous quartz wafer, 2528-042, to prevent splattering onto the combustion tube. After purging the crucibles as described above, care must be taken to handle them at all times with clean tongs; otherwise the samples are easily contaminated with chlorine. The samples were then combusted in the LECO model 523-700 induction furnace for at least 5 min under an 0 2 flow rate of between 1.5-2.0 L/min, and the resulting combustion gases were trapped in 50 mL of IC eluent El for subsequent IC analysis.
RESULTS A N D D I S C U S S I O N W o r k i n g C u r v e s and Detection Limits. A series of from 5 to 10 Cu-Fe accelerator blanks were run with each series of samples in one of two ways. In the first (Method l),the blanks were caught in 50 mL of El only and the average C1- peak height of the blanks was subtracted from that of the sample solutions. Concentrations of sample solutions were then determined by comparison to a working curve constructed by plotting relative peak heights of a n appropriate series of dilutions of C1- (as NaCl) in El standards vs. C1- concentration. In the second method (Method 2 ) , blanks were caught in a series of NaCl in El standards and the working curve was automatically blank compensated. Table I shows a comparison of results of the two methods and reveals that they yield very similar results on the same subsamples.
p p m in sample = p p m C1- in trapping solution X 50.0 mL of solution sample weight in grams
(1) Detection limits were calculated to be that concentration required to give a value 3 standard deviations above the average blank. For 0.2, 0.4, and 1.0 g samples, the detection limits are 4, 2, and 0.8 ppm C1, respectively, in the rock samples or 0.8 kg absolute. All samples so far analyzed have been well above the detection limits thus calculated. This detection limit corresponds to a concentration of 0.016 ppm C1- in the 50-mL sample solutions. This is well above the IC detection limit for C1- which is about 0.0015 ppm C1- on the 0.3 kmho/cm full scale setting where pump noise and temperature drift are the major factors affecting the detection limit. Thus, if necessary, the overall detection limit of the method may be lowered by weighing of Cu and Fe accelerators and achieving a more precise blank estimate. Choice of Trapping S o l u t i o n and I C O p e r a t i n g Conditions. Possible trapping solutions investigated include El, El + 0.3% H202,El + 3% H202,0.3% H202,0.1 M NaOH, and distilled H20. Ei alone yielded lower and more consistent blank values than any of the H 2 0 2solutions and the large water dips, which occur because of the lower conductivity of water relative t o the eluent, of the 0.1 M NaOH and H 2 0 solutions sometimes interfered with low C1- determinations. T h e choice of El as IC eluent allowed separation of the C1peak from possible interferences and produced reasonably short analysis times per sample of about 15 min for sulfur containing samples and about 10 min for those containing little or no sulfur (sulfate is the last species eluted under these conditions). Samples were originally double trapped in a bubble train arrangement b u t the second trap always contained negligible amounts of C1- and was discontinued on later runs. Fifty milliliters of sample solution was an ample amount of solution to doubly rinse beakers and IC syringes and still have enough solution for multiple determinations. Several sample solutions of BCR-1, Allende, GSP-1, and AGV-1 were re-analyzed after sitting in the Nalgene L P E or PP bottles for 2 months. Except for one sample of GSP-1, which showed a 6.7% increase in calculated C1- concentration, no significant changes in calculated concentration were observed. R o c k Standards. Rock standards from a variety of international institutions have been analyzed by the proposed method. Agencies supplying the standards are listed in Table
11. Accuracy, Precision, and Sources of Variation. Table I11 lists our results for C1 and compares them to available literature values. The neutron activation analyses of Johansen and Steinnes (13) and the XRF analyses of Fabbi and Espos (14,15)are considered to be the most reliable literature values.
1910
ANALYTICAL CHEMISTRY, VOL. 5 2 , NO. 12, OCTOBER 1980
Table 11. List of Agencies Supplying Rock Standards country agency United States Geological Survey Geological Survey of Japan National Bureau of Standards Canadian Certified Reference Materials Project Zen tral es Geologisches Institut National Institute for Metallurgy J.T. Matthey Ltd. Arizona State University NASA
standards supplied
Japan
AGV-1, PCC-1, BCR-1, GSP-1, DTS-1, BHVO-1, RGM-1, QLO-1, G-1 JB-1, JG-1
USA
NBS-91
Canada
MRG-1, S Y - 2
East Germany
ZGI-BM
South Africa
NIM-G, NIM-D, NIM-L, NIM-P, NIM-N
UK USA
“Specpure” chemicals SM-RY, RC 62-9, ASU-3, ASU-4, Allende Knippa basalt
USA
USA
Table 111. Comparison of CIC C1 Values with Literature Values av. C1 this no. of sam- work, sample split rock type ples PPm DTS- 1 2 44/30 dunite 9.3 PCC-1
72/32
peridotite
2
69
AGV-1
11/11
andesite
2
120.5
GSP-1
3019
granodiorite
2
267
G-2
116131
granite
2
47.9
BCR-1 BCR-1 BCR-1 BCR-1 QLO BHVO-1 RGM-1 NIM-G NIM-G NIM-G NIM-D NIM-L NIM-N NIM-P Allende
3/10 5615 8015 51/19 2/22 17/11 411 5 304 (run 1 ) 304 (run 2 ) 304 (run 3) 305 306 307 303 818
basalt basalt basalt basalt quartz latite basalt rhyolite granite granite granite dunite lujavrite norite pyroxenite C3V meteorite
5 5 5 5 2 2
85.2 56.8 67.8 60.8 210.5 85.5 427
2.6 3.3 7.1 2.8
212 211
2.4 1.0 5.5 2.9
2
3 3 3 3 2 2
2 9
213 175 902 34 52 229
method: (literature value ref)= NAA: 9.4 (13), XRF: 11 ( 1 4 ) , 11 (19), SPEC: 33 ( 1 2 ) NAA: 6 6 ( 1 3 ) , XRF: 59 ( 1 4 ) , SPEC: 74 ( 1 2 ) , 60(19), loo? ( 1 6 ) NAA: 1 1 5 ( 1 3 ) , XRF: 108 ( 1 4 ) , SPEC: 319 ( 1 2 ) , 110(19), ZOO? ( 1 6 ) NAA: 311 (13), XRF: 305 (14), SPEC: 342 ( 1 2 ) , 300(19), 400? ( 1 6 ) NAA: 53 ( 1 3 ) , XRF: 54 ( 1 4 ) , SPEC: 1 9 2 ( 1 2 ) , 50(19), loo? ( 1 6 ) NAA: 5 8 ( 1 3 ) , XRF: 64 ( 1 4 ) , SPEC: 6 2 ( 1 2 ) , 50(19), loo? ( 1 6 ) XRF: 1 9 2 ( 1 5 ) XRF: 9 3 ( 1 5 ) XRF: 440 ( 1 5 ) 600 (19), 160? ( 1 6 ) 400 (19), 400? ( 1 6 ) 100 (19),1300? ( 1 6 70? ( 1 6 )
4.8
NAA: 2 3 4 ( 2 2 j 4 151 5.8 167 ( l a ) , 1 9 0 ( Z O ) , 69 (20) basalt 51 2 67 ( 1 9 ) , 57 ( Z O ) , 54 2 0 ) granite 150? ( 1 6 ) gabbro 130 2 140? ( 1 6 ) syenite 99 2 2 61.2 basalt EKG/2 2 100 ( 1 9 ) basalt 68.5 1 4 0 (NBS cert. value) 9 134 M opal glass 7.5 4 opal glass 7.1 140 J ASU in house std. 5 36 rhyolite 5.5 ASU in house std. basalt 13 2 31.5 ASU in house std. basani t e 8062 155 2 ASU in house std. 1 2 238 dacite dacite 2 2 280.5 not available Matthey C 3 21 8.1 “Specpure” not available G Matthey 3 20 Si0 5.5 “Specpure” not available 18 56 3 CaCO Matthey “Specpure” 744 21 n o t available Matthey 3 MgO “Specpure” If method is not listed, literature value is from a compilation of data that does not list the method. JB-1 JG-1 MRG-1 SY-2 KNIPPA ZGI-BM NBS-91 NBS-91 SM-RY ASU-3 ASU-4 RC 62-9 RC 62-9 SiO,
a
% rel. stand. dev.
31236 31237
ANALYTICAL CHEMISTRY. VOL. 52, NO. 1 2 , OCTOBER 1980
Table IV. Analysis of Variance of Four Splits of BCR-1 with Five Samples per Split
Table V. Analysis of Variance of Three Different Furnace Runs of NIM-G
degrees source of variation
sum of squares
between splits within splits
degrees
mean freedom square of
2364
3
S , = 129
16 -
sb =
source of variation
788 8.1
1 9 ratio
S = 2493
=
131.2a
a F a t 99% confidence level = 5.29; therefore variation between splits is highly significant.
~
E
210-
a
L
J
r
a -,
_.
195
-
5
180-
v
r
05
0
15
10
O2 Flow /
25
20
1911
30
min
Figure 2. Variation of calculated CI concentration in NIM-G with Os! flow rate and burn time
between runs within runs
sum of squares sb =
6.1
S , = 330
S = 336
freedom
mean square
2
3.05
of
6 8
55.6
ratio = 18.2n
F a t 99% confidence level = 99.3; therefore variations between runs is not significant. content of three different BCR-1 splits. Two of the BCR-1 splits we have analyzed for chlorine in this paper, 56/5 and 3/10, were found by Flanagan et a1 123)to have significantly different carbon contents, with values of 77 ppm =k4.2% and 89.4 p p m h3.3% respectively. T h e third split which they analyzed, 63/9, had a carbon content of 69.3 ppm *6.0%. The supply of this split was exhausted prior to our present chlorine study. An analysis of variance of the data obtained for NIM-G on three different days by two different induction furnace operators is listed in Table V. T h e variation between the different analyses is not significant a t t h e 99% confidence level. When compared with the overall precision of all nine KIM-G samples, however, the variation within analyses does become significant a t the 94'70 confidence level. This simply tells us t h a t we need more samples to achieve maximum precision.
Sample Size, 0 2 Flow , Rate, and Combusiton Time.
05
10
25
20
15
-
O2
F l o w Rate
30
1 min
Figure 3. Variation of calculated CI concentration in BCR-1, 80/5, with 0, flow rate and burn time Our agreement with these values is excellent as shown in Table 111. T h e spectrophotometrically determined values of Huang and Johns (12),appear anomalously high when compared to X R F , NAA, and the proposed CIC method. The comparison t o NAA values (13)yields a least-squares linear correlation coefficient of 0.9939 with a slope of 0.998, while the comparison to X R F values ( 1 4 , 1 5 ) ,yields a correlation coefficient of 0.9956 with a slope of 1.065 as calculated with a handheld HP-33E calculator. When comparing our results with some of the other literature values (17-22), it should be kept in mind t h a t the C1 values of Abbey (17)are admittedly a questionable compilation a n d those of Flanagan (19) are listed as orders of magnitude. I n order t o check t h e overall precision of the method and possible sources of variation, 5 samples each of 4 different splits of BCR-1 were analyzed. The results in Table I11 show relative standard deviations ranging from 2~2.6%in split 3/10 t o h7.1% for split 80/5. T h e overall standard deviation for t h e combined average of all four splits is h18.5% and the results of a n analysis of variance shown in Table IV reveal t h a t t h e variation between t h e splits is highly significant a t t h e 99% confidence level. This result parallels the findings of Flanagan e t al. (23) with respect to variance of carbon
Eight samples of Allende meteorite from 1 to 0.1 g and 13 samples of NBS-91 from 0.025 to 0.23 g were analyzed by the CIC method and the results in Table I11 indicate little dependence on sample size except for a lack of precision on samples below 0.040 g. Combustion times from 3 to 12 min and O2 flow rates from 0.5 t o 2.5 L / m i n were investigated. Combustion times were handtimed with a stopwatch from .the time the first visible glow of the accelerator indicated coupling taking place. Most samples less t h a n 0.5 g coupled almost immediately upon introduction of t h e crucible to t h e combustion position between the induction coils. Larger samples sometimes took as long as a few minutes to begin coupling effectively. O2flow rates were measured with a LECO rotometer calibrated in units of 0.1 L/min. T h e results of these investigations are illustrated in Figures 2 and 3 and indicate low C1 recovery only a t very low O2 flows and bLirn times below 5 min. Analysis of Reagent Chemicals. Several "Specpure" and reagent grade chemicals were analyzed for C1 by the CIC method. T h e results listed in Table I11 illustrate that while many of these chemicals may be very pure with respect to their trace metal content, their chlorine content can be very high a n d / o r inhomogeneous. This fact points out once again the well known problems of finding appropriate, inexpensive, low, and consistent C1 substrate mat.eria1 for t h e preparation of artificial C1 standards. T h e proposed method should be of great help in characterizing the C1 content of a large number of geochemical and environmental samples including rocks, meteorites, soils, laboratory chemicals, hot springs deposits, volcanic ash, volcanic glasses, incinerator ash, and t h e like.
ACKNOWLEDGMENT T h e authors thank James Tarter (ASU) for helping t o prepare samples, supplying samples, and for helpful discussions; Charles Lewis (ASU) for helping to prepare samples and supplying standards; and Ann Yates (ASU) for supplying standards.
LITERATURE CITED (1) Mason, B. "Principles of Geochemistry". 3rd ed.; John Wiley: New York, 1966; p 145.
1912
Anal. Cbern. 1980, 5 2 , 1912-1922
(2) Krauskopf, K. "Source Rocks for Metal Bearing Fluids", in "Geochemistry of Hydrothermal Ore Deposits", Barnes, H. L., Ed.; Rhinehart and Winston: New York, 1967, p 22. (3) Holland, H. D. €con. Geol. 1972, 6 7 , 281. (4) Tooms, J. S. Trans. Inst. Mln. Metal., Sect. B 1970, 79, 116. (5) Dunham, K. C. Trans. Inst. Min. Metal., Sect. B 1970, 7 9 , 127. (6) Roedder, E. U . S . Geol. Surv., Prof. Pap. 1972, 440-JJ. (7) Goldschmidt, V. M. "Geochemistry"; Clarendon Press: Oxford, 1954; pp 588-589. (8) Behne, W. Geochim. Cosmochim. Acta 1953, 3 , 186. (9) Noble, D. C.;Smith, V. C.; Peck, L. C. Geochim. Cosmochim. Acta 1967, 3 1 , 215. (10) Van de Kamp, P. C. J . Geol. 1970, 78, 281. (11) Boyle, R. W. Geol. Surv. Can. Mem. 1961, 370. (12) Huang, W. H.; Johns, W. D. Geochim. Cosmochim. Acta 1967, 37, 597-602. (13) Johansen, 0.; Steinnes, E. Geochlm. Cosmochim. Acta 1967, 3 1 , 1107-1109. ( 1 4 ) Fabbi, B. P.; Espos, L. F. Appl. Spectrosc. 1972, 2 6 - 2 . 293-295. (15) Fabbi. B. P.; Espos, L. F. U . S . Geol. Serv., Prof. Pap. 1976, 840, 89-93.
Small, H.; Stevens, T. S.; Bauman. W. C. Anal. Chem. 1975, 4 7 , 1801. Abbey, S. X-ray Spectrosc. 1978, 7 - 2 , 99-120. Terashima, S. Bull. Geol. Surv. Jpn. 1974, 2 5 , 175-179. Flanagan, F. J. Geochim. Cosmochim. Acta 1973, 37. 1189-1200. Ando, A . ; Kurasawa, H.; Ohmori, T.; Takeda, E. Geochem. J . 1974 8 , 175- 192. (21) Dreibus. G . ; Spettle, B.; Wanke. H. "Origin and Distribution of the Elements", VoI. 11; Pergamon Press: New York, 1979; pp 33-38. (22) Clark, R . S., Jr.; Jarosewich. E.; Mason, 6.; Nelen, J.; Gomez, M.; Hyde, J. R. Smithson. Contrib. Earth Sci. 1970, 5 , 44-53. (23) Flanagan, F. J.; Chandler, J. C.: Breger, 1. A , ; Moore, C. 8.;Lewis, C. F. U . S . Geol. Surv., Prof. Pap. 1976, 840, 123-126. (16) (17) (18) (19) (20)
RECEIVED for review April 18, 1980. Accepted .June 16, 1980. This research was supported in part by NASA grant NSG03-001-001. This work was presented a t the 22nd Rocky Mountain Conference of the Rocky Mountain Chromatography Discussion Group a t Denver, Colo., August 11-14, 1980.
Determination of Atmospheric Sulfur Dioxide without Tetrachloromercurate(I1) and the Mechanism of the Schiff Reaction Purnendu K. Dasgupta," Kymron DeCesare, and James C. Ullrey California Primate Research Center, University of California, Ua vis, California 956 16
Formaldehyde ( 7 mM) buffered at pH - 4 is used to stabilize atmospheric SO2 as hydroxymethanesulfonic acid. Equilibrium data for the above reaction are presented. Sulfite, liberated from the compound by base, is added to acidic pararosaniline for color development by the Schiff reaction, and absorbance is measured at 580 nm. The procedure has been optimized with regard to acidity and reagent concentrations. The method is comparable to the West-Gaeke method (Anal. Chern. 1956, 28, 1816) in absorption and recovery efficiency, sensitivity, and precision. No unusual interferences are observed due to 03,NO,, and transition-metal ions, except Mn(I1). A novel ion chromatographic procedure to determine hydroxymethanesulfonate and sulfate in the same sample is also described. Enhancement of sensitivity in the colorimetric method by solvent extraction has been studied. Investigations into the mechanism of the Schifi reaction and structure of the products have established the validity of the alkylsulfonic acid theory. Mechanistically an arninocarbinol seems to be the first intermediate, which undergoes subsequent nucleophilic substitution by bisulfite ion. To account for the high absorptivity of the product, we suggest that the sulfonic acid group is significantly ionized.
I n 1866, Schiff ( I ) reported t h e color regeneration in a SO2-bleached fuchsin solution upon t h e addition of an aldehyde. Extensive application of t h e Schiff reagent (fuchsin-sulfurous acid) t o identify carbonyl compounds was pioneered by Schmidt (2) a n d identification of other functional groups, which can be oxidized t o t h e carbonyl moiety, was introduced by Bauer ( 3 ) and McManus ( 4 ) . Kasten (5) and P e a c e (6) have adequately reviewed the histologic applications of this uniquely important reaction in histochemistry. Steigmann (7) turned the reaction around and utilized acid bleached fuchsin and formaldehyde for the qualitative identification of sulfites; a quantitative procedure by Grant ( 8 ) followed. Kozlyaeva (9) reported the determination of airborne SOz and used evacuated flasks for collection. Atkin ( I O ) used an alkaline glycerol solution, an absorber first described by Haller ( I I ) , t o determine relatively high levels of SO?. Urone a n d Boggs (12) reduced the alkali content of Haller's absorber by a factor of 25 and described the first useful method for measuring ambient levels of SO1 (13, 1 3 ) . It !\a