1154
Anal. Chem. 1991, 63,1154-1158
(14) RuthVOIl, D. Me;(bdderd, M. ZsayraS 1966, 6, 275-282. (15) Lln, B. C.; Ma, 2.: GoQharrSMrazl. S.; Gulochon, 0. J . 1969, 475, 1-11. (16) Myers, A. L.; Rausnk, J. M. A I C M J . 1966, 1 1 , 121-127. (17) Henson. T. L.; Kabel, R. L. Chem. €ng. Frog., Svmp. Ser. 1967, 74 (83), 36-41. (18) Radke. C. J.; Rausnitz, J. M. A I C M J . 1972, 18, 781-766. (19) LeVan, M. D.: Vermeulen, T. J. phys. Chem. 1981, 85,3247-3250. (20) Cox. Q. B.: Snyder, L. R. J . chrometcg. 1069. 483, 95-110. (21) GoQhan-Shkazl. S.; Gubchon, G. Anal. Chem. 1990, 62, 217-220. (22) Hueng, J. X.: Gubchon, G. J . CdlOM Interface Sci. 1969, 128. 577-591. (23) Newburper, J.; Gulochon. G. J . Chrometog. 1969, 484, 153-166. (24) Golshan-Shlrari, S.; Quiochon, G. J . Chrometogr. 1990, 506, 495-545. (25) GukChon, G.: &IShan-Shlrazi, S.; Jaulmes, A. Anal. W”. 1986, 60, 1656-1866. (26) Czdc, M.; Gulochon, G. Anal. Chem. 1990, 62, 189-200. (27) Schwartzenbach, R. J . ChrometogV. 1960, 202, 397-404. (28) James, D. H.; Phllllps, C. S . G. J . Chem. Soc. 1954, 1086-1070.
Ch”m.
’
(29) Jacobson. J. M.; Frenz, J. H.; Horvath, Cs. Ind. Eng. Chem. Res. 1987. 26, 43-50. (30) (knzelez. M. J.; Jaulmes, A.; Valentin, VldeCMedjar, C. J . chsometom. 1966. 386. 333-344. (31) Kgttl, A. M. PhD.’Dissertatkn, The Unlverslty of Tennessee, Knoxville, 1990. (32) Kiselev, A. V.; Yashln, Ya. I . Gas-SOW C h ” e m @ y ; Masson: Paris, France, 1968.
RECEIVED for review November 30,1990. Accepted February 21, 1991. This work has been supported in part by Grant CHE-8901382 of the National Science Foundation and by the cooperative agreement between the University of Tennessee and the Oak Ridge National Laboratory. We acknowledge the support of our computational effort by the University of Tennessee Computing Center.
Internal Standardization Technique for Capillary Zone Electrophoresis Eric V. Dose and Gorges A. Guiochon* Department of Chemistry, University of Tennessee, Knoxville, Tennessee 37996-1600, and Analytical Chemistry Division, Oak Ridge National Laboratory, Oak Ridge, Tennessee 37831-6120
A new method of Internal standardization for caplllary zone electrophoresis (CZE) k founded on the Ilnear relation between each bn’s effective vdume Injected and lts inherent “Iy.Tho use oftwointmd standards dkwrthe analyst to WaMkh thk relation quantltatlvely for each sample and to correcl the analyte concentratlons for variations In each Inl.ctkn. Tho computatbnal method given k rlmple, and we d ” t r a t e that lt can give reproduclMlltIe8 for manual CZE hydrodynamlc and electrostatic InJectlons of under 1% relative standard devlatlon.
INTRODUCTION Capillary zone electrophoresis (CZE) will not find the same kind of acceptance that chromatography has found in the broad analytical realms of quality control, purity assessment, and trace analysis until much more attention is paid to its quantitative aspects. Detector, instrument and column reliability, and quality and speed of separations have benefited greatly from the last decade’s numerous studies. We feel it is time to address CZE quantitation as well. Electrokinetic (EK) injection is popular among CZE users largely because it requires little or no instrumentation other than that required to effect the separation itself. In contrast, hydrodynamic (HD) injection requires that one apply a pressure differential between the column ends during injection. This requires in turn either the application of vacuum, pressure, or vertical sample displacement. This last approach requires that the column be moved, increasing the size of the instrumental and complicatingtemperature equilibration, for example. Since EK injection merely applies a potential difference between the column ends, injection automation requires only that sample be moved to the buffer cell and that potential pulses be timed precisely. Thus, EK injection equipment should be simpler and more reliable than HD 0003-2700/81/0363-1154$02.50/0
injection equipment. However, since EK injection draws each analyte into the column at a rate proportional to its electromigration velocity, faster ions are overrepresented in the electrophoregram. This velocity discrimination leads directly to apparent bias in detector peak areas. The expected effects of sample solution conductivty and ion mobility differences on the quantitative results obtained using EK injection were described recently (1,2). In the short time since those works were published, bias has been mentioned in several reviews (3-5) and other articles (6-13). Though HD injection seems to be preferred by many workers (4-6,9,13-15), there is little evidence that it offers greater precision than does EK injection. Two problems arise in using EK injection for quantitative analysis. First is the mobility bias problem (Figure l),which can be corrected for if the ion mobilities are known or can be calculated. Second, run-to-run variations in the injection voltage, injection time, and sample conductivity also generally exist. These variations could be corrected for by using internal standards in the same manner as done in chromatography if the relative contributions of electroosmotic flow velocity and electromotive migration velocities were constant. However, variations in the electroosmotic and electromotive injection contributions are not perfectly correlated. To correct for both sources of error, we have developed a new and easily performed internal standardization technique, using two internal standards added to each sample. In this article, we describe this technique and give initial data supporting the technique’s ability to correct for common sources of run-to-runvariations, especially where electrokinetic injection is used.
EXPERIMENTAL SECTION The electrophoretic system consisted of a Hipotronics 25-kV high-voltagepower supply, untreated silica capillary tubing (length 65 cm, working length 42 cm, and inside diameter 75 Mm, Polymicro Technologies, Phoenix, AZ), and an on-column UV detector (JASCOModel UV-100-111) operated at 230-nm wavelength. A Data Translation Model 2801 board digitized the analog detector 0 1991 Amerlcan Chemlcal Soclety
ANALYTICAL CHEMISTRY, VOL. 63,NO. 11, JUNE 1, 1991
1155
Volume Injected Electrokinetic
Flgwo 2. Dependence of effecthre volume Injected to Ion mobility. F , is an Ion mobility scale placed 80 that F A = 0 and F , = 1 for Internal standards A and 9, respectively.
I
,
,
,
,
I
/
,
I
I
I
O
10 Time (minutea)
I
I
I
/
I
t
,
20
Flgure 1. Relative responses of two lnjectlon modes.
output. The Symphony spread sheet (Lotus Development)was used to graph and measure the peak areas by integrating the digitized, recorded voltages above a baseline placed manually across each peak’s base. The buffer solution used throughout was 40 mM sodium phosphate with 0.007% sodium azide in water, pH 7.23. All stock solutions and the sample solutions were prepared by using this buffer solution as diluent. Internal standards benzoic acid and 2-naphthol and analytes N-benzoylphenylalanine, N-benzoylalanine, p-hydroxycinnamicacid, and p-aminobenzoic acid were reagent grade and used as received. The sample solution contained 138.7 pM 2-naphthol (2NOH), 1065 pM benzoate (B-),478 rM N-benzoylphenylalanine anion (BzPA-), 993 pM N-benzoylalanine anion (BzA-), 2967 pM p-aminobenzoate (AB-), and loo0 pM p-hydroxycinnamate (OHC-). At the start of each analysis day, the column was flushed with buffer solution at approximately 1 atm of pressure and then operated at 15kV potential drop. No effort was made to increase the separation efficiency. Hydrostatic injections were performed by manually elevating the reservoir and electrode assembly; a stopwatch was used for timing. Electrokinetic injections were performed by applying electric potential to the capillary;timing was performed by counting seconds verbally.
RESULTS AND DISCUSSION Method Description. The foundation of the present method is the linear relationship between each ion’s effective volume injected Vi and its mobility mi. For nonprogrammed injection, we write Vi = tinjui(r3)where tinjis the injection time, vi is the velocity with which each ion enters the capillary inlet during injection, and r is the capillary inlet radius. It is helpful to partition vi into three independent components: the hydrostatic, siphoning component uHD due to the difference in pressure a t the cspillary’s ends; the electroosmotic component uEO = k{Einj where k is a constant for constant buffer composition and column geometry, { is the column wall’s potential, and E , is the electric field applied during injection; and the electromotive velocity UEM,~= Einjmiwhere E , is the electric field at the injection end of the capillary. Only um,i differs among the ions; UHD and um do not depend on mi and therefore do not differ. Regardless of the magnitude of UHD, UEO, E,,,, or mi, one can write ui = (uHD + uE0) + Ewmi (1) which is linear in mi. Similar expressions have been given previously (1,16). Implicit in eq 1is the assumption that E , is spatially uniform and temporally constant. For hydrostatic injection, E, = 0, and a plot of each ion’s effective injection volume vs ion mobility (Figure 2) is horizontal. However, for electrokinetic injections, Einj# 0, and the plot of Vi vs mi has slope rr2thFinj (and intercept (uEo +
k{Ehj)thj). Values of Vi may be interpolated with known VA and VB for internal standards A and B as Vi = FiVB (1- FJVA (2) where Fi is the derived interpolation constant
+
The most important result is that Fi is independent of all injection conditions that do not alter the mobilities mi. Now the problem for a given analysis is to determine the concentration Ci of each ion i in a sample. For on-column detection and nonprogrammed injection conditions as generally practiced in CZE, the peak area Ai is Ai = RiViCiti (4) where the detector response is proportional to concentration, Ri is some response factor (e.g., molar absorptivity) of ion i, and ti is the time between the start of separation migration and the time that ion i is detected. We assume that the length of the injected sample zone is short compared to the column’s working length. For interpolation factor Fi as defined in eq 3,
Ai
Analytical electrophoregrams can provide values for all the variables in the right-hand side of eq 5 except values for the response fador ratios and Ri/RB;these must be obtained from a calibration electrophoregram. To obtain the response factor ratios from the calibration electrophoregram, either of two conditions must be satisfied: (1)the injection must have zero applied potential, or (2) for electrokinetic injection, the relative ion velocities u i / u A and for all i, must be the same in the injection as it is in the separation phase of the calibration run. Condition 1 is easy to satisfy for hydrostatic injection. Note that the actual injection volume is not important and need not even be known even on a relative basis. Alternative condition 2 simply requires that the injection buffer give the same relative mobilities as the running buffer. This condition is very easy to satisfy for routine electrokinetic injection and, in fact, is violated in only two practical circumstances: (1) where the two buffers differ in the solvent used to make them up, for example, when a sample dissolved in methanol-based buffer is injected into an aqueous buffer; or (2) where the buffers’ pH values differ and at least one analyte or standard ion has a relevant protonation equilibrium.
1156
ANALYTICAL CHEMISTRY, VOL. 63,NO. 11, JUNE 1, 1991
Table I. Manual Hydrodynamic Injection Reproducibility" RSD, W BzPA- BzA- pOHC- pABno int stds
5.6
int std A int std B
0.5 1.0
5.6
1.0 1.3
6.0
1.3 1.2
5.9
1.4 1.2
lit.
ref
llBb
18 16 13 8 20 12 13 18 19 17 8 15
l.0-7.6b 4.0-7.8b 6.1-8.3' 3.7' ca. 2d 0.6-2.3d 2.9d 4.0-5.3d 0.9-2.8* 2.3-4.6' 0.8-1.8"'
present method 0.7 0.6 1.2 1.3 a Retention times in minutes: internal standard A, 5.3;internal standard B, 14.9;BzPA-, 9.1;BzA-, 10.1;pOHC-, 11.0;pAB-, 13.2. Injection time about 60 e; injection vertical displacement about 3 cm. *Manual injection. Unknown whether manual or automated iniection. Automated iniection. e Micellar electroDhoresis.
If the calibration run begins with hydrostatic injection, then all Vi are equal, and from eq 4,
For an electrokinetic calibration injection where portional to u-,~,
is pro-
so that
Ri AiCB -Ri= - AiCA and =RA
AACi
RB
(8)
ABCi
All the response ratios are obtained from a single calibration run. These combined with data from an analysis electrophoregram provide all the data required for eq 5 (CA and CB are known since they are internal standards), and each Ci is easily solved for. Tests of the Present Method. We present two kinds of confirming experimental results, reproducibility, and sensitivity. Reproducibility results demonstrate the present method's ability to compensate for the kinds of random variations and errors in operator technique and instrument conditions expected in actual CZE analyses. Sensitivity study results demonstrate the method's ability to compensate for very large changes in injection time, electrokinetic injection voltage, and relative contribution of electromotive velocity and bulk flow. We used peak areas, rather than peak heights, as commonly practiced in chromatography and as recommended by comparative CZE studies (14,17-19). The relative standard deviation (RSD) and accuracy results discussed below include all integration errors. Peak signal to RMS noise ratios were on the order of 200-to-1 for the electrophoregrams given, except for a few very small injections made in the injection time sensitivity studies discussed below. For hydrodynamic injection, the present internal standardization method offers about the same precision advantage over analysis with no internal standards as does the use of a single internal standard (Table I). Our raw peak area precision, about 5 4 % RSD, falls within the range of manual-injection RSD values of other workers (14, 18, 20) and is about twice the RSD magnitude as those from automatedinjection results reported previously (13,14,20,21). When the analyte peak areas in each electrophoregram are divided by internal standard peck areas and the ratios are employed as responses, RSD values drop to about 1%, in agreement with
Table 11. Manual Electrokinetic Injection Reproducibility"
RSD,W BzPA- BzA- pOHC- pABno int stds
6.8
7.8
8.5
lit.
10.8 13.4b 4.lC 1.1c 2.2-2.8' 7.4 2.6 0.6
ref
18 18 5 19
int std A 3.2 4.1 4.8 int std B 6.8 5.9 5.1 present method 0.8 0.9 0.7 OInternal standard and analyte retention times as in Table I. Calibration injection conditions as in Table I; electrokinetic injection conditions 15 kV, about 5 s. bManualinjection. cAutomated injection. literature RSD values for analyses with internal standards (9, 17, 19). The fact that RSD values from corrected data presented here are substantially less than RSD values from uncorrected data from automated injections suggests that the present method is likely to improve substantially data taken from automated instruments. Confirming experiments are in progress. We fiid for hydrodynamic injection that there is very little further improvement when using two internal standard peak areas and eq 5 compared with simply using one internal standard in the customary manner. This is expected because all ions are treated equally in hydrodynamic injection, and a single internal standard should remove all the systematic run-to-run injection variations provided the conditions of eq 6 hold. In this case, adding a second internal standard may be introducing as much added random error as using eq 5 removes. There may still be an advantage to applying eq 5 to hydrodynamic injections if detection and integration errors can be greatly decreased (so that standardization errors dominate) or if some drift in instrument conditions occurs during runs rather than strictly between them. In electrokinetic injection, no single internal standard can account for more than a fraction of the run-to-run variations commonly encountered. In Table 11, our raw peak area RSD values fall between literature values for manual and automated injections. When we include a single internal standard as commonly done in chromatography, the RSD values do decrease, but they are still larger than raw RSD values from unstandardized automated electrokinetic injections. Our single-standard RSD values are too large for many applications including quality control, an application for which CZE otherwise has great potential. We note that each analyte's RSD value increases with increasing difference between the analyte's retention time and that of the internal standard used. When we apply the present internal standardization method to our manually obtained electrokinetic-injection data, reproducibility improves radically. Application of eq 5 lowers each RSD value to below 1%. Even with the very primitive operating conditions under which these experiments were conducted, reproducibilities compare favorably with the best CZE reproducibility values in the literature. As any internal standard method should do, the present method eliminates sensitivity to large deviations in the amount of sample injected. To confirm the sensitivity of computed concentrations to the duration of hydrodynamic injection, we first obtained the required response ratios using a 60-s hdyrodynamic injection and then determined sample peak areas from hdyrodynamic injections of durations 30,50, 70, and 90 s. The data in Table I11 show that the 3-fold range of injection times is reduced in computed concentrations to errors of a few percent. The largest errors are seen in the smallest injections, as reported by others (9,14). It is unclear whether this is a systematic error arising from effects like flowless,
ANALYTICAL CHEMISTRY, VOL. 63, NO. 11, JUNE 1, 1991
Table V. Dependence of Injection Accuracy on Mode of Injection"
Table 111. Dependence of Hydrodynamic Injection Accuracy on Injection Time" inj time,s
BzPA-
30 50 70
2.7 0.1 -4.1 -0.1
90
relative mncn error, % BzApOHC3.9 0.5 -3.2 -0.9
3.9 1.9 -2.2 -0.5
pAB3.3 0.4 -1.8 1.0
aInternal standard and analyte retention times as in Table I. Results are from single electrophoregram. Injection vertical disDlacement about 3 cm. Table IV. Dewndence of Electrokinetic Injection Accuracy on Injection Time' inj time,s
BzPA-
2.5 5 10 15
-18 11 1.6 2.8
relative concn error, % BzApOHC-20 6.0 3.8 3.9
-18 8.8 1.8 3.2
1157
relative concn error, % BzPA- BzA- pOHC- p A B -
inj mixture
+ 0-sHD + 15-s HD 4-sEK + 30-s HD 2-s EK + 4 5 4 HD 8-s EK 6-s EK
-0.5 2.0 0.6 0.4
-1.2 1.8 -0.4 -0.1
-0.4 1.2 -1.0 -1.2
-1.7 1.5 0.4 0
"Internal standard and analyte retention times as in Table I.
Results are from single electrophoregram. Electrokinetic injection voltage 5 kV. Hydrodynamic injection vertical displacement about 3 cm. EK = electrokinetic, HD = hydrodynamic.
small errors in Table V confirm that the unusually large relative errors in the top rows of Tables I11 and IV are in fact due simply to insufficient injection volumes.
pAB-8.6 3.1 7.3 2.4
"Internal standard and analyte retention times as in Table I. Results are from sinale electroDhorearam. Injection voltme 5 kV. ubiquitous injection (15) or perhaps simply a random error as generally seen for the areas of smaller peaks. We also note that there may be a delay in the start of electroosmotic flow immediately after the application of potential and that such delays may allow ions to diffuse into or out of the column end (15). We plan extensive accuracy-and-precision tests to determine whether the errors seen in Table I11 are systematic or random. The present method also corrects for large deviations in the durations of electrokinetic injections (Table IV). Response factors were obtained from hydrodynamic injection calibration. We find that accuracy suffers when the injected volumes are very small (as in the first two rows of Table IV). However, for injection volumes of roughly the same magnitude as that of the calibration electrophoregram (bottom two rows of Table IV), accuracies are much better (average error of 3.4%). As in the case of hydrodynamic injection time sensitivity, we plan to perform extensive reproducibility studies on electrokinetic injection time sensitivity. The data in Table IV do confirm that large differences in electrokinetic injection times are accounted for by the present standardization method, even when hydrodynamic injection is used in the calibration run. In typical CZE analyses, the relative importance of the bulk flow and electromotive migration components of effective injected volumes Vi may change from run to run because of changes in electroosmotic flow rate due in turn to changes in the surface of the silica capillary wall in contact with the buffer solution. It is further possible that the relative physical heights of the two solutions could differ from run to run, causing variable hydrostatic contributions to the ions' effective injection volumes. Of course, automation or careful manual operation would lessen this risk. We tested the present method's ability to correct for fluctuations in the relative importance of bulk and electromotive velocities by artificially introducing just such fluctuations into the injection method used. The calibration injection was hydrodynamic. For each test electrophoregram, we performed an electrokinetic injection followed immediately by a hydrostatic injection. We chose the two injection times to keep the average peak areas roughly constant while varying greatly the relative contributions of the two injection modes. The results in Table V indicate that the present method was entirely successful in correcting for changing modes of injection. Further, the very
CONCLUSIONS The method described herein uses h o internal standards to correct electrophoretic responses for common run-to-run variations. It is used to its best advantage in electrokinetic injections but also works well for hydrodynamic injections. The calculations are simple enough to be performed with a hand calculator. Use of the present method allows one to simplify instrument design by eliminating the mechanical apparatus required for automated hydrostatic injection. Electrokinetic injection, which allows for the minimum possible mechanical complexity, now provides high accuracy and sub-170RSD precision. Conveniently, the method appears to maintain this degree of accuracy and precision even when significant errors in injection time, injection voltage, hydrodynamic flow, and injection mode occur. The method should yield accurate results even if those errors are never discovered. Limitations on the method's use are not very restrictive. The calibration run's injection conditions must be the same as its separation conditions, or the injection may simply be hydrostatic. Analysis run buffers must be made up in the same solvent as that of the sample (usually water), and the pH must be the same if any anal@ or standard ion is partiauy dissociated near the buffer or sample pH. If no analyte or buffer ion mobility is pH-sensitive, sample pH changes may be allowed even though the electroosmotic velocity changes and even if the extent of that change is unknown. ACKNOWLEDGMENT We thank Roswitha Ramsey of the Oak Ridge National Laboratory for loaning us the instrument used in this work. LITERATURE CITED (1) Huang, X.; Gordon, M. J.; &re, R. N. Anal. Chem. 1988, 6 0 , 375-377. (2) Olivares, J. A.; Nguyen, N. T.; Yonker, C. R.; Smith, R. D. Anal. Chem. 1987, 59, 1230-1232. (3) Gordon, M. J.; Huang, X.; Pentoney, S. L., Jr.; &re, R. N. Scknco 1088, 242, 224-228. (4) Knox, J. H.; McCormack, K. A. J . Liq. Chrometog*. 1088, 12, 2435-2470. (5) Drossman, H.; Luckey, J. A.; Kostichka. A. J.; D'Cunha. J.; Smith, L. M. Anal. Chem. 1990, 62. 900-903. (8) Smith, R. D.; Udseth, H. R.; Loo, J. A.; Wright, 6. W.; Roos, 0. A. Talenta 1088, 36. 161-169. (7) Yu, M.; DoviChl, N. J . Appl. SpeCtros~.1989. 43, 198-201. ( 8 ) Cheng. Y.4.; Dovlchl. N. J. Scknce 1989, 242, 582-564. (9) . . Huana. X.: Luckev. J. A.: Gordon. M. J.: a r e . R. N. Anal. Chem. 1989: 6 1 , 788-7fO. (10) Zhu, M.; Hansen, D. L.; Burd, S.; Gannon, F. J . Chrometogr. 1989, 480, 311-319. (11) Huang. X.; Gordon. M. J.; &re, R. N. J . Chrometog*. 1989, 480, 285-288. (12) Qoss, L.; Yeung, E. S. J . C l " a t o g * . 1989, 480, 189-178. (13) Rahn, P. C. Am. Blotechnd. Lab. 1900, 22-29. (14) Honda. S.; Iwase, S.; Fujlwera, S. J . Chmmatugr. 1007, 404, 313-320. (15) Grushka, E.; M c h m i c k . R. M. J . Chrometog. 1989, 471, 431-428. (18) Jorgenson, J.; Luckacs, K. D. Anal. Chem. 1981, 53, 1298-1302. (17) Fujlwara, S.; Honda, S. Anal. Chem. 1887, 59, 2773-2778.
AMI. them. 1991, 63,1158-1164
1158 (18) (10) (20) (21)
Otaka, K.; Tambe, S.; Ando, T . J . -tow. 1987,396,350-354. Fujwara. 5.; Honda, S. Ann/. Chem. 1888, 58, 1811-1814. Robe, D. J.; JWgWISOn, J. W. AMI. chsm.1888, 80, 642-648. schwa&, n. E.; h i e r a , M.; B o w n h , R. G. J . "a*.1868, 480, 129-139.
RECEIVED for review December 26,1990. Accepted March 1,
1991. This work is supported in part by Grant DE-FGOB86ER13487 from the U.S. Department of Energy, Office of Energy Research, by Grant CHE-8901382 from the U.S. National Science Foundation, and by the cooperative agreement between the University of Tennessee and Oak Ridge National Laboratory.
Effects of Surfactants on Cathodic Stripping Determination of Iodide A. R.Harman and A. S.Baranski* Department of Chemjstry, University of Saskatchewan, Saskatoon, Saskatchewan, Canada SN7 0 WO
The d f ~ &of Trtton X-100 On kdkle m k n etltckncy, the shape of the strlpping peak, and the multlng callkatbn pbb were studied under cycllc vokanwnetrlc coditlam. Thdynamk parameters of mercurous iodide adwrptlon were determined in the presence and in the absence of the wrfactant by a new method based on the convolutbn transform. The rebulk rhow that the M#ltkn of lrtton X-100 affects only thermodynamic propertles of the system In the interfacial regbn. The pmence of the surfactant Improves signlfkantly the detection ilmit of the stripping determlnation of Iodide because the efficiency of lodlde deposition Ir increased due to a stronger adsorption of mercurous iodide on mercury electrode8 (a synergk effect), the stripping peak Is narrower (thereforehlghw and easier to measure) because the energy of admrptbn I s h affectedby potonilal, and thebackground current Is reduced due to the lower capacitance of the doubk layer.
INTRODUCTION The analysis of halides, especially at trace and ultratrace levels, is becoming increasingly important in various areas of human activities. For example such analysis is often undertaken in the food industry (1-3), where levels of iodide are of particular concern, and in the analysis of environmental samples such as natural waters and particulate air pollution (4-8). The methods most often employed for the determination of halides, and in particular iodide, include gas chromatography (9),liquid chromatography (IO),ion-selectiveelectrodes ( 2 , 6 , 9 ) ,flow injection spectrophotometry (11,121, and differential pulse polarography (13). When it is necessary to analyze for halides at the ultratrace level, the method usually employed is neutron activation analysis (NAA). Although this technique offers both the sensitivity and precision needed (14), it suffers from the disadvantage of the very high capital expense associated with the required facilities. This has led a number of researchers to evaluate the use of cathodic stripping voltammetry (CSV) as an alternative method, incorporating either linear sweep (151, differential pulse (I, 15,17), or square wave (16) techniques, with working electrodes made of either silver (17-19), mercury (1, 15, 16,
* To whom correspondence should be addressed. 0003-2700/9 110383-1 158$02.50/0
20),or copper amalgam (21). Such methods of analysis have
proven to have the advantage of experimental simplicity, sensitivity, and low unit cost of equipment. In a recent publication, Luther et al. (16) reported that cathodic stripping analysis carried out in the presence of low concentrations of the surfactant Triton X-100(TX) enhanced the stripping peak of mercurous iodide under square wave voltammetric conditions. The use of this method enabled determinations of iodide in environmental samples a t concentrations more than 1 order of magnitude below previously reported detection limits for either electrochemical techniques (22) or ion chromatography (23). The discovery made by Luther et al. (16) is extremely interesting because in most cases the adsorption of organic molecules on mercury electrodes causes a decrease in sensitivity due to inhibition of charge-transfer processes. In order to better understand the action of these surfactants in cathodic stripping analysis, a detailed study was undertaken. In this publication, we report on the possible mechanism by which cathodic stripping peak enhancement is possible by the presence of adsorbed surfactant on the surface of a mercury working electrode.
EXPERIMENTAL SECTION All experiments were performed by using a conventional three-electrode arrangement. The reference electrode used was either a commercial saturated calomel electrode (SCE) or a laboratory fabricated Pb/PbSO,(,, electrode; however all potentials reported in this work are vs SCE. The auxiliary electrode consisted of a platinum wire with surface area ca. 0.5 cm2. With the exception of the hanging mercury drop electrode (HMDE) (Metrohm Model 6.03351,all other electrodes used in this work were laboratory fabricated, as described in a previous publication (24.
The equipment used for cathodic stripping has been described elsewhere (24). Altemating current impedance experiments were performed by using a EG&G PAR Model 273 potentiostat coupled to a Model 5301 lock-in amplifier. In measurements requiring high sensitivity, a built-in current follower of the Model 273 potentiostat was bypassed and replaced with a EG&G Model 181 current-sensitivepreamplifier. Data acquisition and analysis was performed by using an AT&T Model 6300 microcomputer. All computer programs for data acquisition, processing, and numerical simulations were devised in this laboratory (copies of numerical simulation programs can be obtained from authors). All chemicals used were of analytical grade. The stock solutions (0.1 M) were prepared by using water distilled in a Corning Mega-Pure system, and all dilute solutions were made just prior to their use. The surfactantsemployed in this study were obtained 0 1991 American Chemical Soclety
