4
Anal. Chem. 1983, 55, 4-7
Although the retention time for thiosulfate is 10 min, the total time for a full cycle, including restabilization of the base line with eluent A, is about 15 min. This is mainly due to the tailing of the thiosulfate peak and to a lesser extent depending on the restabilization of the base line, which requires approximately 1 min. In order to decrease the severe tailing of the thiosulfate peak, experiments were performed with B40$-in both eluents as a modifier. However, the extent of tailing was not reduced and the sulfite/sulfate separation deteriorated. It is possible that the severe tailing of this peak not only is due to adsorption effects in the separator column but also is caused by decomposition of thiosulfate when it is acidified in the suppressor column. This, however, is not confirmed. The use of eluents with concentrations high enough to elute thiosulfate in about 10-12 min shortens the lifetime of the suppressor column. T o cope with this problem, it is possible to use two suppressors in parallel, alternatively using one while regenerating the other. Another possibility is the use of the hollow fiber suppressor, with continuous regeneration, recently described by Stevens et al. (8). Note: Eluent suppression ion chromatography is covered
by patents in several countries.
ACKNOWLEDGMENT The authors wish to thank Lars Lundmark and Svante Jonsson for the electronic and mechanical constructions. LITERATURE CITED (1) Small, H.; Stevens, T. S.; Bauman, W. C. Anal. Chem. 1975, 4 7 , 1801-1809. (2) Holcombe, L. J.; Jones, B. F.; Ellsworth, E. E.; Meserole, F. B. I n "Ion Chromatographic Analysis of Environmental Pollutants"; Ann Arbor Science: Ann Arbor, MI, 1979; Vol. 2, pp 401-412. (3) Trujillo, F. J.; Miller, M. M.; Skogerboe, R. K.; Taylor, H. E.; Grant, C. L. Anal. Chem. 1981, 53, 1944-1946. (4) Gjerde, D. T.; Fritz, J. S.;Schmuckler, G. J . Chromafogr. 1979, 186, 509-519. (5) Gjerde, D. T.: Schmuckler, G.; Fritz, J. S.J . Chromatogr. 1980, 187, 35-45. (6) Llndgren, M.; Cedergren, A.; Lindberg, J. Anal. Chim. Acta 1982, 14 I , 279-286. (7) Nelder, J. A.; Mead, R. Computer J . 1965, 7 , 308-313. (8) Stevens, T. S.; Davls, J. C.; Small, H. Anal. Chem. 1981, 53, 1408-1 492.
RECEIVED for review July 26, 1982. Accepted September 29, 1982.
Determination of Cyanide, Sulfide, Iodide, and Bromide by Ion Chromatography with Electrochemical Detection Roy D. Rockfln" and Edward L. Johnson Dionex Corp., 7228 Titan Way, Sunnyvale, California 94086
Cyanlde, sulfide, iodlde, and bromide are separated and detected by uslng ion chromatography (IC) and electrochemical detection via a sllver working electrode. The detection limlts are 2 ppb, 30 ppb, 10 ppb, and 10 ppb, respectlvely. Cyanide and sulflde can be determined simultaneously, as well as with other anions commonly determtned by IC. Cyanlde contained in Cd and Zn complexes Is quantitatlvely determined as total "free" cyanide, whlle cyanide contained In NI and Cu complexes Is only partlally determlned as "free" cyanlde. The strongly bound cyanlde in Au, Fe, or Co complexes Is not detected.
Although ion exchange techniques can easily separate cyanide or sulfide from a host of common anions, their detection via common methods such as conductivity is very poor. During an IC analysis, the weakly acidic species HCN and H2S are formed in the anion suppressor column. Unlike the halogen acids, they are not detected by the conductivity detector due to their low dissociation and, therefore, low conductivity. This inability to detect cyanide and sulfide has prevented the exploitation of the separating power of ion chromatography (I) for the determination of these ions. In all the analytical methods so far developed for cyanide and sulfide, removing interferences is a necessary first step when analyzing most samples. In addition to interfering with each other, other species interfering with cyanide and sulfide determination include the halogens, thiocyanate, and thiosulfate. The traditional wet chemical analytical method for cyanide, including the removal of interferences, involves precipitating sulfide with cadmium ion, filtering, acidifying,
and distilling the sample (2). The cyanide is trapped in a sodium hydroxide solution, which is usually assayed by argentometric titration, by spectrophotometry, or by an ion selective electrode. The entire process takes approximately 2 h. Samples which can be analyzed directly without distillation are those which are known not to contain significant quantities of interfering species. Sulfide is usually determined by precipitating sulfide with zinc ion, filtering, and then acidifying the precipitate. This solution can be assayed by iodometric titration, spectrophotometry (methylene blue method) or by an ion selective electrode. This procedure takes approximatley h per sample. Electrochemical methods for cyanide determination include amperometry (3-5) and polarography (6). Sulfide can be determined by cathodic stripping voltammetry (7). The polarographic method (6) can determine cyanide or sulfide when in the presence of the other; however, iodide and thiosulfate interfere. Recently Pihlar and Kosta (8, 9) developed an electrochemical method for cyanide analysis by using flow injection analysis (FIA). The method is based on the ability of a silver working electrode in an amperometric electrochemical flowthrough cell to produce a current. The reaction for cyanide is Ag
+ 2CN-
-
Ag(CN)2-
+ e-
The main conclusions from the work of Philar and Kosta are as follows: (1)Current is directly proportional to cyanide ion concentration. (2) The electrode maintains the same sensitivity over long periods of time; i.e., it is not poisoned.
0003-2700/83/0355-0004$01.50/00 1982 American Chemical Society
ANALYTICAL CHEMISTRY, VOL. 55, NO. 1, JANUARY 1983 5 CEL. OUTLET
c
4001
CELL BOD"
COUNTER ELECTRODE
300
I
ctcw FAT C
100
ION EXCHANSE
-.2
0
-.l
0
+.l +.2
+.3
Eapp.
\
WCRKIUG ELECTEODE
for 1 ppm cyanide Flgure 2. Dependence of current response on (dots), 500 ppb sulfide (triangles), and 5 ppm iodide (squares).
CELL INLE
Flgure 1. Diagram of amperometric flow-through cell.
1 us-
(3) Sulfide interference is still a problem. Similar reactions occur for sulfide and the halogens; however, the products are precipitates rather than a soluble complex. 2Ag -k S2 Ag2S 2e-
I I C"
-
Ag -1- X-
-+
II
+ AgX + (3-
In this paper, the results of placing an ion exchange column in front of the electrochemical detector are presented. Cyanide and sulfide are separated and thus are determined simultaneously. Although bromide and iodide can be determined by IC with conductivity detection, the use of electrochemical detection results in greater selectivity as well as increased sensitivity.
EXPERIMENTAL SECTION All chromatography was performed on a Dionex System 2010 (P/N 35201) ion chromatograph. Unless ot,herwise specified, the sample loop size was 50 fiL and the eluent flow rate was 2.5 mL/min. Cyanide and raulfide were separated on an HPIC-AS4 (P/N 35311) anion exchiange column using an eluent consisting of 14.7 mM ethylenediamine, 10 mM NaH2B03(prepared from H,BO:, and NaOH), and 1.0 mM Na2C03. 'The eluent pH is 11.0. Chromatography of iodide was performed by using an HPIC-AS1 (P/N 30827) column with an eluent consisting of 20 mM NaN03. Chromatography of bromide was performed using an HPIC-AS3 (P/N 30985) column with an eluent consisting of 2 mM Na2C03. Electrochemical detection was performed with an IonChrom/Amperometric Detector (P/N 35221). The cell (Figure 1)consists of a silver rod working electrode 1.3 cm long X 0.178 cm in diameter, an Ag/AgCl reference electrode separated from the flowing stream by a Nafion cation exchange membrane, and platinum counterelectrode. (Nafion is a registered trademark of E. I. du Pont de Nemours & Co.) The cell geometry is based on one previously reporteld in the literature (10). The working electrode was occasionally cleaned by mechanical polishing. The applied potential was 0 V for cyanide and sulfide, 0.20 V for iodide, and 0.30 V for bromide. Cyanide standard solutions were prepared from a 1000 ppm NaCN (Mallinkrodt, Paris, KY) stock solution, standardized by argentometric titration. Sulfide standards were prepared by diluting 21% (NH&3 (J. T. Baker, Phillipsburg, NJ). K,Fe(CN),, K,CO(CN)~,and CuCN were purchased from Alfa Products were purchased from (Danvers, MA). K2Zn(C"N).,and KAU(CN)~ Pfaltz and Bauer (Stamford, CT). Ni(CN)4z was prepared from NaCN and Ni(CH3C0z)z(Matheson, Norwood, OH). Cd(CN)42was prepared from lo00 ppm solutions of CdClz (Alfa) and NaCN. Copper cyanide solutions were prepared by adding stoichiometric quantities of NaCN to EL 1.0 X M CUCN solution. RESULTS AND DISCUSSION Applied Potential. Figure 2 shows the dependence of current response on applied potential (Eapp). As the potential is increased from negative to positive, the peak height increases as the diffusion controlled plateau is reached. Beyond this
I
o
4
a 12 Minutes
16
Flgure 3. Simultaneous analysis by using electrochemical and conductivity detection. Concentrations are 300 ppb sulfide, 500 ppb cyanide, 1 ppm fluoride, 4 ppm chloride, 10 ppm nitrite, 10 ppm bromide, 25 ppm nitrate, 30 ppm sulfite, 25 ppm sulfate, and 50 ppm phosphate.
potential, peak height decreases as other reactions compete with the one of interest, probably oxidation of silver to form an oxide or hydroxide, thus poisoning the electrode surface (11). Simultaneous Multianion Analysis. Electrochemical detection can be used in conjunction with conductivity detection to determine, in a simultaneous analysis, ions detectable by either method. This is accomplished by placing the electrochemical detector between the separator and suppressor columns of the ion chromatograph. The electrochemical detector must be placed before the suppressor column, as the suppressor's purpose is to lower the conductivity by decreasing the concentration of the supporting electrolyte. The separation of a 10 anion standard is shown in Figure 3. Sulfide, cyanide, bromide, and sulfite are detected a t the silver electrode, while nitrite, nitrate, phosphate, and sulfate produce no response. Due to the low dissociation of HzS and HCN following protonation by the suppressor column, they are not detected by the conductivity detector. The major advantage of chromatography over other analytical methods is its ability to separate interferences. With one exception, the determination of one of the four ions is not affected by the presence of the others. For example, a solution containing 2500 times as much chloride as bromide has little effect on the determination of bromide (Table I), even though chloride elutes first. Since Eo for the oxidation of Ag to AgCl is positive of E" for the oxidation of Ag to AgBr, at a potential just on the diffusion controlled plateau for bromide, the current response for bromide will be much greater than that
6
ANALYTICAL CHEMISTRY, VOL. 55, NO. 1, JANUARY 1983
Table I. Determination of 1 ppm Cyanide in the Presence of Sulfide amt of added sz-,PPm 0
0.10 1.00 10.0
recovery of CN-, %, t2%
1
100 109 112 104
ihA)
Table 11. Determination of 400 ppb Bromide in the Presence of Chloride amt of added Cl-, ppm
recovery of Br-, %, +2%
0
100
10 100
104
1000
100 115a
a The 1000 ppm chloride solution contains 50 ppb bromide, accounting for 12.5%of the 15% excess.
Flgure 4. Determination of 50 ppb bromide in 1000 ppm chloride from reagent grade NaCl (Baker). A 100-pL sample loop was used.
for chloride. This can be exploited in order to determine small quantities of bromide in a large excess of chloride; a difficult process using other methods of analysis due to the similar chemical properties of the two halides. The determination of 50 ppb bromide present in 1000 ppm of reagent grade chloride is shown in Figure 4. The large negative dip following the chloride peak is caused by the reduction of the AgCl deposited on the electrode. Since there is no longer chloride in the solution next to the electrode (as it has all eluted), and since the applied potential is below the diffusion controlled plateau, AgCl reduction is favored in order to satisfy the Nernst equation. A small dip following the bromide peak can also be seen. The determination of cyanide is affected by the presence of sulfide in the sample, as shown in Table 11. Since sulfide elutes before cyanide (Figure 3), cyanide does not interfere with the analysis of sulfide. The addition of 0.1 ppm sulfide enhances the cyanide peak by 9%. This effect can be minimized or eliminated by using the standard addition method or by matching the cyanide standards as closely as possible to the sample. For example, if the sample is known to contain approximately 100 ppb sulfide, then this amount should be added to the standards. Reproducibility. The reproducibility of peak heights for repeated injections of the four ions is generally 1% , however the first injection of a series usually produces a peak a few
PPB
Flgure 5. log (current)vs. log (concentration)for cyanide (dots),sulfide (triangles),and iodide (squares).
percent lower than the subsequent injections. This effect is most pronounced for sulfide, which produces an initial peak as much as 10% lower than the subsequent peaks. In actual use, errors in concentration measurements can be reduced or eliminated by making two or three injections until the peak height is reproducible. Extremely small quantities of sulfide (e20 ppb) also produce irreproducible peaks, often disappearing entirely. As this problem is independent of the applied potential and is not noticed in FIA experiments, it is thought to be caused by chemical reactions occurring in the column. It is interesting that the deposition of a layer of Ag2S or AgX on the working electrode surface does not poison the electrode, since we have seen no evidence of a decrease in peak height caused by a build up of silver sulfide or halide. Shimizu, Aoki, and Osteryoung (12)have noted that the rate of oxidation of a rotated silver disk electrode in the presence of sulfide is limited by the diffusion of sulfide ions to the electrode, as long as less than 0.08 C cm-2 of charge has passed. This charge, equivalent to about 700 monolayers, is considerably in excess of the integrated current from a 1ppm sulfide peak, about C cm-2, Cyanide forms a soluble product and cannot form a layer (13). Linearity and Sensitivity. Calibration curves for cyanide, sulfide, and iodide are shown in Figure 5. In general, plots of the log of peak height as a function of the log of concentration are linear at low concentration. At high concentrations, an increase in concentration produces a smaller increase in current as shown by the plateau for each ion. This plateau (also observed in FIA studies) is caused by uncompensated resistance in the cell. The upper limit of linearity can be extended by increasing the applied potential or by decreasing the size of the injection loop. With a 1OO-wL injection loop, the detection limit for cyanide is 2 ppb, for sulfide 30 ppb, for iodide 10 ppb, and for bromide, 10 ppb. The detection limits reported here are approximately 2 orders of magnitude lower than those reported with the use of gold or mercury working electrodes (5). Analysis of Metal Cyanide Complexes. The cyanide in inorganic cyanides can be present as both complexed and free cyanide. In order to study the chromatography of metal cyanides, we prepared and assayed solutions of cadmium, zinc, copper, nickel, gold, iron, and cobalt cyanides. Table I11 lists the percentage of total cyanide detected. The results suggest that the complex cyanides can be grouped into three categories depending on the cumulative formation constant and stability of the complex. Category 1 includes the weakly complexed and labile (14) cyanides Cd(CN)?- (log p4 = 18.78) (15) and Zn(CN):- (log p4 = 16.7). These complexes completely dissociate under the chromatographic conditions used; the cyanide being indistinguishable from free cyanide.
ANALYTICAL CHEMISTRY, VOL. 55, NO. 1, JANUARY 1983
Table 111. Percentage of Total Cyanide in Metal Complexes Determined as “Free” Cyanide metal complex Cd(CX),’Zn( CN),’Xi( CS),’C U ( CN),3-
Cu(c N ) 3 a Cu(CNLAu( CNjiFe( CN),3Co(CN), 3 -
log Of0
?&
18.8 16.7
102
102 ai 52
31.3 30.3 28.6 24.0 38.3
42
38 0
42
0
64
0
Formation constants from ref 15.
From ref 16.
7
38.3), Fe(CN)6* (log = 42), and CO(CN)6* (log $(3 = 64) (16). No free cyanide was detected for these complexes. Although these complexes do not elute under the chromatographic conditions used, they can be eluted and determined by using different chromatographic conditions and conductivity detection (17). Samples containing both free cyanide (or weakly complexed cyanide) and strongly complexed cyanide can be analyzed for free cyanide by direct injection. The determination of total cyanide (both free and strongly complexed) requires distillation of the sample with caustic trapping.
ACKNOWLEDGMENT The authors thank Dennis C. Johnson for his help and advice. Registry No. NaC1, 7647-14-5; Cd(CN)42-,16041-14-8; Zn19441-11-3; (CN)?-, 19440-55-2;Ni(CN):-, 48042-08-6; CU(CN)~%, CU(CN)~~-, 16593-63-8; Cu(CN);, 18973-62-1;Ag, 7440-22-4. LITERATURE CITED
Minuter
Flgure 6. Chromatogram of 1.O X
M CU(CN),~-.
Category 2 includes the moderately strong cyanide com(log p4= 30.3). plexes Ni(CN)z- (log p4 = 31.3) and CU(CN)~% Although the complexen are labile, they are retained on the column and slowly dissociate during the chromatography. This slow dissociation produces tailing which lasts for several minutes as the free c!yanide elutes and is detected. The chromatography of C U ( C N ) ~is~ shown in Figure 6 and can be compared to the cyamide peak in Figure 3. As the results presented in Table 111 demonstrate, the tailing and the nonquantitative recovery of cyanide preclude the use of direct injection to determine ,total cyanide in samples containing copper or nickel. These samples may be analyzed after acid distillation and caustic trapping. The cyanide in the caustic solution can then be determined by ion chromatography with electrochemical detection. Category 3 includes those cyanides which are inert and therefore totally undissociated, such as AU(CN)~(log p2 =
(1) Pohl, C. A.; Johnson, E. L. J. Chromatogf.Scl. 1980, 18, 442. (2) “Standard Methods for the Analysis of Water and Wastewater”, 15th ed.; APHA, AWWA, WPCF, 1980. (3) McCloskey, J. A. Anal. Chem. 1961, 3 3 , 1842. (4) Miller, G. W.; Long, L. E.;George, G. M.; Slkes, W. L. Anal. Chem. 1964, 36, 980. (5) Bond, A. M.; Heritage, I. D.; Wallace, G. G.; McCormlck, M. J. Anal. Chem. 1982, 54, 582. (6) Canterford, D. R. Anal. Chem. 1975, 4 7 , 88. (7) Shlmizu, K.; Osteryoung, R. A. Anal. Chem. 1981, 53, 588. (8)Pihiar, 8.; Kosta, L. Anal. Chlm. Acta 1980, 114, 275. (9) Plhlar, 8.; Kosta, L.; Hristovski, B. Talanfa 1979, 26, 805. (IO) k-wn, J. A.; Kolle, R.; Johnson, D. C. Anal. Chlm. Acta 1980, 116, JJ.
(11) Hampson, A. N.; Macdonaid, K. I.; Lee, J. B. J . Necfroanal. Chem. 1973, 4 5 , 149. (12) Shlmizu, K.; Aokl, K.; Osteryoung, R. A. J. Nectroanal. Chem. 1981, 129, 159. (13) Shimlzu, K.; Osteryoung, R. A. Anal. Chem. 1981, 53, 2351. (14) Ford-Smith, M. H. “The Chemistry of Complex Cyanides”; Her Majesty’s Stationary Office: London, 1964. (15) “Lang’s Handbook of Chemistry”, 1l t h ed.; Dean, J. A,, Ed. McGrawHill: New York, 1973. (16) Marteil, A. E.; Calvin, M. “Chemistry of Inorganic Compounds”; Prentice Hall: Englewood Cliffs, NJ, 1952. (17) Fitchett, A. W.; Johnson, E.; Pohl, C. to be presented at the Pittsburgh Conference in Applied Spectroscopy and Analytical Chemistry, Atlantic City, NJ, March 1983.
RECEIVED for review June 28,1982. Accepted September 27, 1982.
