Determination of aqueous fluoride with a helium microwave-induced

Analytical Chemistry 1992 64 (1), 50-55. Abstract | PDF ... Miyazaki. Analytical Chemistry 1990 62 (22), 2457-2460 ... Journal of the Chinese Chemical...
1 downloads 0 Views 500KB Size
Anal. Chem. 198% 61 674-677

674

I

of the predicted shift should fall within roughly twice the standard error of estimate of the corresponding model. Assignments of carbons whose chemical shifts differ by less than this expected accuracy cannot be viewed with confidence.

CONCLUSIONS On the basis of the evaluation procedures outlined above, the chemical shift models of the disaccharides are judged highly accurate and useful for prediction. Potential application areas for the models lie in both structure elucidation studies and in the determination of chemical shift assignments. The principal limitation of the models is the availability of only 40 disaccharide spectra for use in computing the models. As the prediction of the spectra of compounds 41 and 42 illustrates, the inclusion of more structural diversity in the model generation step would be expected to improve the versatility of the models for use in predictions. Additionally, the computed models are based on the use of the MM2 force field to compute three-dimensional structural geometries. We have not investigated the potential error involved in chemical shift predictions based on the models described here and structures modeled with a force field other than that used by MM2. This is a current topic of investigation in our laboratory. On the basis of the disaccharide work, we are confident that chemical shift modeling can be extended to more complex saccharide systems. In such systems, it is expected that long-range structural effects will become increasingly important. Our distance-based parameters easily allow the incorporation of long-range effects into the derived models. LITERATURE CITED (1) Rosenthal, S. N.; Fendler, J. H. A&. phvs. Org. Chem. 1976, 13, 279-424.

Llpklnd, G. M.; Shashkov. A. S.; Knlrel, Y. A.; Vinogradov, E. V.; Kochatkov, N. K. Carbohydr. Res. 1988. 175. 59-75. Veregln, R. P.; Fyfe, C. A.; Marchassault, R. H.; Taylor, M. G. Carbohydr. Res. 1987, 160, 41-56. Bock, K.; Brlgnole, A.; Sigurskjold, B. W. J . Chem. Soc. Perkin Trans. 2 l986# 1711-1713. Dorman, D. E.; Roberts, J. D. J . Am. Chem. SOC. 1971, 93, 4463-4472, Smith, D. H.; Jurs, P. C. J . Am. Chem. Soc. 1976, 100,3316-3321. Small, G. W.; Jurs, P. C. Anal. Chem. 1983, 5 5 , 1128-1134. Small, G. W.; Jurs, P. C. Anal. Chem. 1984, 56, 2307-2314. McIntyre, M. K.; Small, G. W. And. Chem. 1987, 59, 1805-1811. Usul, T.; Yamaoka, N.; Matsuda, K.; Tuzlmura. K.; Sugiyama, H.; Seto, S. J . Chem. SOC. Perkin Trans. 11973, 2425-2432. Colson, P.; King, R. R. Carbohyd. Res. 1976, 47, 1-13. Pfeffer, P. E.; Valentlne, K. M.; Parrlsh, F. W. J . Am. Chem. SOC. 1970, 101, 1265-1274. Voeiter, W.; Bilk, V.; Breitmaier, E. Cdlecf . Czech. Chem . Commun . 1973, 3 8 , 2054-2071. Briigger, W. E.; Jurs, P. C. Anal. Chem. 1975, 47, 781-784. Small, G. W.; Jurs, P. C. Anal. Chem. 1963, 55. 1121-1127. Allinger, N. L. J . Am. Chem. Soc.1977, 99, 6127-8134. Stuper. A. J.; Brugger, W. E.; Jus, P. C. CompufefAssMed Studies of Chemlcal Sfructve and 8iobgicaI Function; Wlley-Intersclence: New York, 1979; pp 83-90. Reeves, R. E. J . Am. Chem. Soc. 1950. 72, 1499-1506. Angyal, S. J. Angew. Chem., Znf. Ed. Engl. 1989, 8 , 157-226. Angyal, S. J.; Plckles. V. A u t . J . Chem. 1972, 2 5 , 1695-1710. Leyden, D. E.; Cox. R. H. Ana!v#calApplications of NMR; Wlley-lnterscience: New York, 1977; p 28. DelRe. G. J . Chem. Soc. 1958, 4031-4040. Randic, M. J . Am. Chem. Soc. 1975, 97, 6609-6615. Kier, L. B.; Hall, L. H. J . pherm. Sci. 1978, 65, 1806-1809. Draper, N. R.; Smith, H. Applled Regression Analysis, 2nd ed.;WlleyIntersclence: New York, 1981; Chapter 6. Belsley, D. A.; Kuh, E.; Welsch, R. E. Regression Ciagmstlcs: Identiwing Znfluenf&l Date and Sources of Collinear&; Wiley-Interscience: New York, 1980 Chapter 3. Lizotte, P. A.; Poulton, J. E. 2.Nafurforsch.. C: Biosci. 1988. 41C, 5-6.

RECEIVED for review August 18, 1988. Accepted December 19, 1988. This research was supported by the National Institutes of Health under the Biomedical Research Support Grant program.

Determination of Aqueous Fluoride with a Helium Microwave-Induced Plasma and Flow Injection Analysis J. M. Gehlhausen and J. W. Camahan* Department of Chemistry, Northern Illinois University, DeKalb, Illinois 60115

The determlnatlon of aqueous fluoride by flow injection analysls (FIA) wlth a heHum microwave-induced plasma (He-MIP) is described. Thls system operates at 500 W and utilizes a malwled resawitor cavlty wtth a demoratable plasma torch. Beth dkect nebulbatkn and FIA in conjunction wlth ultrasonic nebulization (USN) were investigated. FIA was found to be the most reliable method because extended nebulization of aqueous fluoride was found to cause memory effects. Detection Ikn#s for aqueous ttuorlde of 35 and 4 ppm were observed for FIA and direct USN, respectively. The interference effects of pH and selected elements were also studied.

In general, the determination of nonmetals by atomic emission spectrometry (AES)has been quite limited. A major reason for these difficulties is that the resonance lines for most 0003-2700/89/0361-0674$01.50/0

nonmetals are in the vacuum-UV making AES difficult. Secondly, the high excitation energies of nonmetals cause excited-state populations to be low. However, analytical plasmas do show promise in the area of nonmetal determinations. Nonmetals have been determined in gaseous samples or as gas chromatographic eluates with the microwave-induced plasma (MIP) (1-9) and the inductively coupled plasma (ICP) (10-12). The determination of fluorine by AES has proven particularly difficult. The resonance line for fluorine appears at 95.5 nm (13). Consequently, work with fluorine has utilized nonresonance lines, particularly the 685.6-nm F I line. Use of the helium-MIP has been the most successful for the determination of nonmetals such as fluorine because of the improved nonmetal excitation ability compared to the ICP. However, past studies were generally limited to low power (50-100 W) He-MIP precluding direct aqueous nebulization for nonmetal determinations. Reasons for this restriction are 0 1989 Amerlcan Chemlcal Society

ANALYTICAL CHEMISTRY, VOL. 61, NO. 7, APRIL 1, 1989

875

1°.Or

Flgure 1. A schematic of the He-MIP with flow injection analysis.

41

'TI

Time

Fluoride 1 667.82nm He(l) 2 685.60nm F (I) 3 690.25nm F ( I ) 4 700.22nm 0 ( I )

2

I

Figwe 2. A spectrum obtained by nebulizing a 1 ppt fluoride solution into the He-MIP.

that nonmetal emission is quenched or the plasma is extinguished by the aqueous mist. Moderate power (500 W) HeMIPS have shown promise in the determination of aqueous nonmetals. These systems allow aqueous samples to be nebulized directly into the plasma. Successes have been seen in the determination of aqueous chloride, bromide, and iodide (14, 15).

This paper describes the use of a moderate power He-MIP in the determination of aqueous fluoride. The system discussed here has been modified relative to previous work with aqueous halides in this laboratory (14, 15). Flow injection analysis has been utilized to accommodate the reactive nature of fluorine (Figure 1). The detection limits for fluoride by direct nebulization and FIA are presented, and the matrix effects for three easily ionized elements, three metals, and variable pH are reported.

EXPERIMENTAL SECTION The helium plasma system consisted of a modified TWlo cavity with a demountable quartz torch and was maintained with a 500-W microwave generator at 2.45 GHz (16).In earlier publications involving this system a quartz cooling jacket was utilized (14).The cooling jacket was removed without reduction of the torch lifetime or detrimental effects to plasma performance. For calibration plots and matrix effects studies a 0.75 m focal length Spectrometrics (Andover, MA) Echelle Spectrospan I11 spectrometer was used with entrance and exit slits set at 300 pm height X 200 pm width. A 0.35 m GCA/McPherson (Acton, MA) scanning spectrometer equipped with a holographic grating was used to study the spectral region near the 685.6-nm F I emission line. The photomultiplier tube in each case was a Hammamatsu (Middlesex, NJ) R758 operated at 900-1000 V. To introduce analyte to the plasma, a CETAC (Ames, IA) ultrasonic nebulizer was utilized. The sample was introduced directly or in discrete 0.5-mL amounts with an Altex Scientific TEFZEL coated sample injection valve and Teflon sampling loop (Beckman Instrument Co., Arlington Heights, IL, Model No. 243269). A Gilson Minipuls 2 peristaltic pump delivered analyte solution to the ultrasonic nebulizer and sample injection valve. RESULTS AND DISCUSSION Line Selection. The 685.6-nm F I emission was used exclusively in these studies of aqueous fluoride emission. The spectrum shown in Figure 2 was observed by nebulizing a 1 mg/mL solution of fluoride (as KF) into the plasma. The

-.

Figure 3. Effect of the peristaltic pump rate on the fluoride emission

signal. 690.25-nm F I line was also seen but was considerably less intense than the 685.6-nm line. A number of other fluorine emission lines have been detected in analytical plasmas by introducing gaseous samples. However, the work presented here involved aqueous samples. The introduction of water vapor into the plasma reduces the ability of the plasma to excite the analyte, and consequently, fewer fluorine emission lines can be seen with this system. Optimization. The plasma system was operated at a power of 480-500 W, and the plasma gas flow was 17.5 L/min. The flow of helium through the USN was 1.2-1.4 L/min and sample solution was pumped to the nebulizer at a rate of 2.6 mL/min. The fluoride signal did not change significantly with a varying solution pump rate; however, reproducibility was generally better at flow rates of 2 mL/min or higher with a maximum signal a t 2.6 mL/min (Figure 3). Use of Flow Injection Analysis. Comparative studies of the ultrasonic nebulizer (USN) and a variety of pneumatic nebulizers have shown that the USN can improve detection limib significantly, as much as 10-fold. The plasma system discussed in this paper has demonstrated these improvements. Consequently, all work presented in this discussion utilized ultrasonic nebulization. However, with direct ultrasonic nebulization of aqueous fluoride, a serious problem arose; after nebulization of aqueous fluoride was discontinued, a residual fluoride signal (or memory effect) was observed. This residual signal remained for several minutes and the problem worsened after continuous operation of the system. This effect has been attributed to fluorine coating the tip of the sample introduction tube of the plasma torch. The reactive nature of fluorine in the high temperature region of the plasma torch has been noted elsewhere (11). Flow injection analysis (FIA) was implemented to minimize the memory effects. Flow injection analysis allows for the introduction of discrete amounts of sample, thus significantly reducing the amount of fluoride reaching the plasma. For this work, 0.5-mL samples of fluoride were nebulized into the plasma. The use of FIA eliminated the memory effects and reduced analysis time. A sample could be introduced into the plasma every 1-2 min with FIA, compared with 3-5 min for direct nebulization. Linearity and Detection Limits. Defined as a signal intensity equal to twice the standard deviation of the background noise, the mean detection limits for aqueous fluoride by flow injection analysis and direct nebulization were 36 and 10 ppm, respectively. The best detection limit obtained in this study was 4 ppm with direct solution nebulization. The calibration plot obtained by using FIA is shown in Figure 4. The solutions contained 1% "Os. The analytical working range for the calibration plot was slightly more than an order of magnitude. The upper limit of this range is seen as a "roll off" at concentrations slightly less than 1000 pg/mL fluoride. This "roll off" was also observed when the fluoride sample was introduced by direct nebulization and, consequently, is not related to the use of

676

ANALYTICAL CHEMISTRY, VOL. 61, NO. 7, APRIL 1, 1989

K

I 0

1

I

1

1

500

1,000

1,500

2,000

i

0.00.00

0.05

0.10

0.15

i

Concentration bgirnL)

Flgure 4. Calibration plot for fluoride by FIA.

Concentrationof Added Intert.rent(hrd/L)

Flgure 6. Matrix effects of several I A elements on the fluoride emission signal. The fluoride concentration is 500 ppm.

1.5r

I

B

ii

I

00

I

1

I

20

40

60

Metal Concentration (rnglrnL) Interference effects of various metals on the fluoride emission signal. The concentration of fluoride is 500 ppm.

Flgure 5.

FIA but is probably caused by reduced ability of the plasma to excite fluorine with high salt concentrations. There is a loss of signal for FIA relative to direct nebulization as shown by the higher detection limit for FIA. In one experiment, a 0.5 mg/mL fluoride sample was introduced into the plasma under nearly identical conditions by FIA and direct nebulization, and a 22% lower emission signal was observed for FIA. However, the signal intensities were quite similar for FIA and direct nebulization at concentration levels where the signal "roll off" was observed. Matrix Effects. The effect of pH on the fluoride signal was studied. This study demonstrated that from pH values of 1.8 to 11.2 the effects on a 500 ppm fluoride sample were small and within experimental error. Results yielded recoveries of 100.0 f 4.9%. Consequently, most biological or environmental fluoride samples would not require careful consideration of the pH. However, the addition of 1% HN03 (correspondingto a pH of approximately 0.7) produced a signal enhancement of about 27 % . This effect may be caused by either a change in the nebulization characteristics of the solution or other analyte vaporization effects. The effects of zinc, aluminum, and copper were also studied. The ions were provided by zinc nitrate, aluminum chloride, and copper nitrate, respectively. Copper and zinc enhanced the signal of a 500 ppm fluoride sample while aluminum suppressed the signal (Figure 5). The influence of these salts on the background was studied and found to have no discernible effect. At present, it is not possible to present a detailed explanation for the interference effects caused by these metals. However, it is likely the high bond strength of AI-F is responsible for fluoride signal depression in the presence of Al. As shown in Figure 6 the addition of potassium caused a small increase in fluoride signal while lithium and sodium both showed more substantial apparent signal enhancements. The apparent enhancement of the fluoride signal was compared

to the increase in the background signal at a wavelength 2 nm from the 685.6-nm F I line. These experiments showed clearly that the signal enhancement is caused by an increase in the background signal. In the case for each IA element, the signal enhancement and the base-line shift were of a similar magnitude. A fluoride solution containing 3.13 mg/mL potassium exhibited a signal enhancement approximately equal to 10% of the background signal while the increase in the base line was also 10% of the background signal. For a 0.34 mg/mL sodium solution, the signal enhancement was 58% and the base-line shift was 50%, and a 0.1 mg/mL lithium solution gave a 76% signal enhancement and a base-line shift of 68%. It is important to note that the resonance lines of lithium, sodium, and potassium which are at 670, 589, and 766 nm, respectively. The base-line shift was greater for lithium, which has the resonance line closest to the fluoride emission line. The resonance lines of sodium and potassium are further from the 685.6-nm F I line and exhibited smaller increases in the background signal. These observations support the assertion that the increase in background is probably caused by stray light. Consequently, the use of a background correction method will probably eliminate this problem. CONCLUSIONS The He-MIP shows promise for the determination of aqueous fluoride. This study has demonstrated that fluoride can be determined without detrimental effecta to the plasma torch with flow injection. The presence of zine or copper in a fluoride sample enhances fluorine emission signal while aluminum depresses the signal. The addition of sodium, lithium, or potassium to a fluoride sample results in an increase in background signal that is probably due to stray light from the resonance lines of these IA elements. pH changes in the range of 2-11 do not affect the emission signal of fluoride. Consequently, this technique might be useful in instances where detection limits are less critical than problems associated with pH since the ion-selective electrode, which is the most common technique for determining aqueous fluoride, is affected by pH. Thus, the He-MIP shows promise as an alternate technique for determining aqueous fluoride. Registry No. Fluoride, 16984-48-8. LITERATURE CITED (1) Mclean, W. R.; Stanton, D. L.; Penketh, G. E. Analyst 1973, 98, 432-442. (2) Quimby, B. D.; Uden, P. C.; Barnes, R. M. Anal. Chem. 1978, 5 0 , 21 12-2 118. (3) Quimby, B. D.; Delaney, M. F.; Wen, P. C.; Barnes, R. M. Anal. Chem. 1980, 52, 259-283. (4) Mulllgan, K. J.; Caruso, J. A.: Fricke, F. L. Analyst 1980, 105, 1080- 1067. (5) Brenner, K. S. J. Chromatogr. 1978, 167, 365-380. ( 6 ) Estes, S. A.; Uden, P. C.; Barnes, R. M. Anal. Chem. 1981, 53, 1829- 1837.

Anal. Chem. 1989, 6 1 , 677-683 ( 7 ) Chiba. K.; Yoshkla. K.; Tanabe, K.; Oraki, M.; Haraguchi, H.; Winefordner, J. P.; Fuwa, K. A m l . Chem. 1982, 5 4 . 761-764. ( 6 ) Slatkavits, K. J.; Wen, P. C.; Hoey. L. D.; Barnes, R. M. J . Chromatw.1984, 302, 277-207. ( 9 ) Hagen, D. F.; Bellsle, J.; Johnson, J. D.; Venkateswarlu, P. Anal. Blochem. 1981, 118, 336-343. (10) Windsor. D. L.; Denton, M. 8. J . Chromatogr. Sci. 1979, 17, 492-496. ( 1 1 ) Fry, R. C.; Northway, S. J.; Brown, R. M.; Hughes, S.K. Anal. Chem. 1980. 52, 1716-1722. (12) Keane, J. M.; Fry, R. C. Anal. Chem. 1988, 5 8 , 790-797. (13) Mavrodlneanu, R.; Bolteux, H. Flame Specboscopy; Wiley and Sons: New York, 1965.

877

(14) Michlewlcz. K. G.; Carnahan, J. W. Anal. Chem. 1985. 5 7 , 1092- 1095. (15) Michlewicz, K. 0.;Carnahan, J. W. Anal. Chem. 1988, 5 8 , 31 22-3 125. (16) Michlewicz, K. G.; Uhr, J. J.; Carnahan, J. W. Specbochim. Acta, Part B 1985, 408,493-499.

RECEIVEDfor review September 27,1988. Accepted January 5,1989. Presented in part at the 21st Great Lakes Regional Meeting of the American Chemical Society, Chicago, Paper 4,1987.

Fourier Transform Infrared Least-Squares Methods for the Quantitative Analysis of Multicomponent Mixtures of Airborne Vapors of Industrial Hygiene Concern Ying Li-shi’ and Steven P. Levine* The University of Michigan, School of Public Health, Ann Arbor, Michigan 48109-2029

Alr monltoring methods suitable for use in the workplace, though accurate for monltorlng Individual compounds or classes of compounds, cannot be used to monitor several compounds or classes of compounds slrnultaneously. I n the past few years, Fourler transform Infrared (FT-IR) spectroscopy has been lnvestlgated for use as a method for multicomponent quantltatlve analysis. This work focuses on quantltatlve analysis of SIXmlxtures In ambient air. The concentratlon ranges of the two- to sixtomponent mlxtures are from 50 ppm to 100 ppb. The optimal least-squares fit (LSF) method selected, background reference flle chosen, and quantltative peak windows plcked were evaluated In this effort. The quantttatlve results of six mlxtures were accurate at the 50, I O , and 1 ppm levels. There were some components for whkh the analysis was also accurate at the 0.1 ppm level. The data Indicate that the LSF program could be used to quanttfy strongty overlapping mutllcomponenlmlxtwes. The results support the concluslon that the FT-IR spectrometer Is approprlate for the dlrect quantlflcatlon of multicomponent mixtures of many alrborne gases and vapors of industrlal hygiene concern.

INTRODUCTION The applications of classical multivariate least-squares analysis have resulted in improved applicability, precision, and accuracy in multicomponent spectral analysis (1, 2). Assuming a linear relationship between concentration and absorbance (i.e. assuming Beer’s law is obeyed), least-squares fit (LSF) has been successful in the quantitative analysis of multicomponent mixtures even in those cases where there is strong overlap of the infrared spectral features (3-9). The inclusion of all the data in the spectral region of interest also signiticantly improves the precision and accuracy of the results. The LSF methods of Haaland et al. have been applied with success in the quantitative determination of CO, COz, and Present address: Shanghai Medical University, School of Public Health, Shanghai 200032, People’s Republic of China.

NOx in dry air, condensed phase xylene isomers, and boron oxide in borosilicate glasses (3, 10, 13). The growing awareness of the use of toxic organic gas and vapor mixtures in science and industry has increased the need for more sensitive and versatile air monitors for environmental and industrial hygiene use. Previously, Herget and Levine have evaluated a Fourier transform infrared (FT-IR) spectrometer equipped with a long path gas cell for monitoring semiconductor processing areas for sub-threshold limit value (TLV) levels of process gases (11). Recently, the hydride gases and organic vapors were accurately and precisely determined by FT-IR in simulated semiconductor manufacturing work place situations (4). The LSF method provided an accurate quantitation of phosphine at a resolution where the peak shape was partially lost and single point quantitation methods were no longer effective. Earlier work performed by our group demonstrated that the LSF-FT-IR method could quantify single compounds at their TLV or below. The compounds used in this study are frequently found in hazardous waste sites and are of general concern to the field of industrial hygiene (5). Not surprisingly, those data also indicate that, for organic vapors in ambient air, the limit of detection (LOD) of single compounds in mixtures is higher than the LOD for individual compounds (12). However, all of the above-cited references were focused on the condensed phase, on low molecular weight compounds, or on workplace situations simulated by using electronic mixtures of the spectra of organic vapors. This work focuses on quantitative analysis of six mixtures in ambient air. The concentration range of the two- to sixcomponent mixtures are from 50 ppm to 100 ppb. All the calculations of vapor concentrations were accomplished with the LSF computer program. The optimal least-squares method selected, background reference file chosen, and quantitative peak windows picked were evaluated in this effort. The quantitative results of six mixtures were accurate at the 50,10, and 1ppm levels. There were many components for which the analysis was also accurate at the 100 ppb level. The data indicate that LSF program could be used to quantify strongly overlapping multicomponent mixtures. The results support the conclusion that the FT-IR spectrometer is ap-

0003-2700/89/0361-0677$01.50/00 1989 American Chemical Society