582 Anal. Chem. 1982, 54 , 582-585

Simultaneous Determination of Free Sulfide and Cyanide by Ion Chromatography with Electrochemical Detection

A. M. Bond,* I . D. Heritage, and G. G. Wallace

Division of Chemical and Physical Sciences, Deakin University, Waurn Ponds 32 17, Victoria, Australia

M. J. McCormlck

Laboratory Services Branch, Environment Protection Authority, 240 Victoria Parade, East Melbourne 3002, Victoria, Australia

A method for the slmultaneous determlnatlon of sulfide and cyanide has been developed by uslng Ion chromatography wlth electrochernlcal detection. Most existing methods are subject to considerable interference. However, separatlon of cyanlde and sulfide is conveniently untertaken uslng ion chromatography. Electrochemical detectlon, rather than conductometric detectlon, commonly used in Ion chromatog- raphy, enables basic medla to be used at ail stages of the experlrnent, thereby elimlnatlng problems assoclated with the formatlon of volatile and highly toxic hydrogen cyanide or hydrogen sulflde. A detailed study empioylng a wide range of electrode and electrochemical cells has been lnvestlgated for detectlon of sulfide and cyanlde. Dropplng mercury electrodes or mercury-plated solld electrodes provlde the optlmum response, but the electrochemical cell design and nature of liquid chromatographic apparatus are also Important.

Because of their high toxicity (1, 2) a knowledge of sulfide and cyanide levels is important environmentally and many methods have been developed for their determination (3). Ion-selective electrodes (4-8) have been widely used for sep- arate determination of sulfide or cyanide but unfortunately respond to both species. Similar interference problems arise with many amperometric, spectrophotometric, and other methods (9-11).

Polarographic techniques offer some scope for determining mixtures (12) of sulfide and cyanide. Both direct (12-15) and indirect (16-18) polarographic methods have been employed. A polarographic method for the simultaneous determination of cyanide and sulfide has been developed by Canterford (13) but is restricted to limited concentration ratios and is not totally free from other interference effects.

Generally, all methods are subject to interferences effects and a distillation step is usually involved in analytical pro- cedures for determining sulfide and cyanide (4-6,19). How- ever, although distillation removes many interfering species it does not separate cyanide and sulfide from each other and their simultaneous determination can prove to be difficult. Thus, from the data available, separation of sulfide and cyanide appears to be a prerequisite to the implementation of a successful method for simultaneously determining these anions in aqueous solutions.

Ion chromatography (IC) is a recently developed technique for the separation and determination of both anions and cations (20). It differs from conventional ion exchange chromatography in that low capacity separator columns are used and an additional column (suppressor column) is fre- quently included to enable the system to be used with con- ductometric detection.

Applications and limitations of IC with conductometric detection are described in detail elsewhere (21, 22). Two important properties of sulfide and cyanide make their de-

0003-2700/82/0354-0582$0 l .25/0

termination difficult using the conductometric detector system employed with IC. Acidic conditions are required and pK, values are greater than seven so that conductivity would be poor. Furthermore, the reaction they would undergo on the suppressor column (acid conditions) would produce volatile, toxic substances, namely, hydrogen cyanide and hydrogen sulfide. Changing the detector system to an amperometric device removes the need for the suppressor column (acid conditions) and therefore the above limitations no longer apply.

It is the aim of this work to explore the possibility of sep- arating sulfide and cyanide using IC and simultaneously de- tecting the anions using electrochemical detection (ampero- metrically). Other workers (23) have briefly mentioned the possibility of detecting cyanide and sulfide using IC with postcolumn derivitization and coulometric detection. This appears to be a complex procedure using suppressor columns requiring relatively low pH which is generally not regarded as appropriate for cyanide or sulfide. An acidic ion exchange column coupled with an electrochemical detector has also been briefly examined for determining sulfide (24). This work will show that a simpler, safer electrochemical method can be developed in very basic media using ion chromatography with amperometric detection.

Dropping mercury, mercury-coated platinum and solid gold working electrodes are considered and various commercially available electrochemical detector cells are investigated to provide optimum results.

EXPERIMENTAL SECTION Reagents and Standard Solutions. Analytical grade reagents

were used throughout this work and distilled, deionized water was used to prepare all standards and eluents. A 0.005 M hydroxide solution was prepared by dissolving appropriate amounts of NaOH in water. Sulfide and cyanide standards were prepared freshly every day by dissolving Na2S.9H20 and KCN, respectively, in 0.005 M hydroxide solution.

Prior to use, eluents were filtered through a 0.45-gm fiiter. They were then degassed before and during chromatographic runs with high-purity nitrogen.

Instrumentation. Initial chromatographic work was carried out by using a Dionex (Sunnyvale, CA) Model 10 ion chroma- tograph except that the supplied detector was replaced by an electrochemical (amperometric) detector.

In subsequent work a Waters (Milford, MA) Model 6000A pump in conjunction with a Model U6K injector was examined as an alternative to the Dionex system.

The separator column used for data reported in this work was a Dionex anion exchange column (internal diameter = 3 mm, length = 250 mm). Amperometric detectors employed were as follows: (i) Bio Analytical Systems (B.A.S) (West Lafayette, IN) TL4 detector cells. Gold working, platinum auxiliary, and Ag/ AgCl(3 M NaCl) reference electrodes were employed with this detector. (ii) Metrohm (Herisau Switzerland) Model EA 1096 detector cell. The electrodes were as in (i). (iii) A Tacussell (Villeurbanne, France) DELC-1 detector cell. The use of both a dropping mercury eletrode and a mercury-coated platinum

0 1982 American Chemical Soclety

ANALYTICAL CHEMISTRY, VOL. 54, NO. 3, MARCH 1982 583

(A)

4 working electrode were considered with this cell. Hg was plated onto the platinum at -1.00 V vs. Ag/AgCl for 5 min and then the electrode was dipped into a mercury pool. A glassy carbon auxiliary and Ag/AgCl (3 M NaCl) reference electrodes were employed. A cathode ray oscilloscope was used to trace cur- rent-time curves emd hence determine the drop time in the Ta- cussell cell. (iv) E.G.&G. Princeton Applied Research Corp. (P.A.R.) (Princeton, NJ) Model 310 polarographic detector cell with Model 303 static mercury drop electrode. Auxiliary and reference electrodes were the same as (i).

Polarographic data were recorded by use of a P.A.R. Model 174A polarographic analyzer, with a P.A.R. Model 303 static mercury drop electrode.

Voltammograms in a conventional electrochemical cell were recorded by using ”mini-electrodes” obtained from Metrohm (Herisau, Switzerland). A gold working electrode, glassy carbon auxiliary, and Ag/AgCl (3 M KCl) reference electrode were em- ployed in the voltammetric studies.

Microprocessor-controlled equipment, partially described elsewhere (25) was used to apply the short pulse widths required in the normal pulse work. The complete details of this instru- mentation will be described in a future publication (26).

All experiments were undertaken in a controlled-temperature laboratory at 22 A: 1 OC.

RESULTS AND DISCUSSION Separation. Calculation based on known acidity constanixi

for HCN and H2S (27) indicated that pH values between 11 and 12 should ensure that the predominant species present are in the anionic forms of CN-, HS-, or S2-. In subsequent discussion the term sulfide will refer to either HS- or S2- as they are in dynamic equilibrium. A 0.005 M OH- solution (pH 11.7), as the eluent in IC using the column described in the Experimental Sed;ion and the Dionex ion chromatograph, gave retention volumes of 1.40 mL for sulfide and 2.80 mL for cyanide. This pH was adequate for separation and ensures the absence of HzS and HCN.

The fact that sulfide elutes before cyanide suggests the pH of the eluent is such that sulfide is present predominantly in the HS- form. This is consistent with calculations based on acidity constants. Increasing the pH (inc [OH-]) causes problems on the low capacity column. A 0.008 M OH- solutiort elutes both sulfide and cyanide with the solvent front. This is due to saturation of the active sites on the column by OH-. ions which are retained on the column. This problem is not as severe on a higher capacity column or on a larger Dionex column (internal diameter = 3 mm, length = 500 mm).

Samples were injected, either dissolved in the running solvent or a t a similar pH so as to minimize peaks due to hydroxide. Hydroixide gives an electrochemical response close to the solvent front (retention volume 0.96 mL).

Amperometric Detection. (i) Detection at Gold Elec- trodes. Voltammograms Ion gold electrodes (Figure 1) in so- dium hydroxide showed a well-defined oxidation process for sulfide but a more complicated process for cyanide. Sulfide presumably involves oxidation of gold to form a gold sulfide complex in an analogous fashion to the electrode process a t mercury (see later).

For cyanide, an irreversible wave is observed at 0.70 V vs. Ag/AgCl but only in the presence of the gold oxide layer (28). Neither of the cyanide or cwlfide electrode processes are fully understood, but as has been demonstrated with silver elec- trodes (9), gold electrodes may be used as an amperometric detector for sulfide and cyanide. By use of ion chromatog- raphy with electrochemical (amperometric) detection (ICEC), the B.A.S. thin-layer cell, the Dionex 10 ion chromatograph, and a gold working electrode a reproducible current response could be obtained for sulfide but not cyanide. A detection limit of 100 ng of sulfide was found and the calibration curve was linear over the concentration range tested (0-5 pg). In- jection volumes of 100 pL were employed. Presumably the

u I I 1 I I -0.4 -0.2 0.0 0.2 0.4 0.6 0 8 10

E I V o I t s )

Flgure 1. Cyclic voltammogram In a conventional electrochemical cell1 at a gold electrode for (A) 3 X M cyanlde in 0.005 M NaOH. Scan rate = 200 mV/s.

complicated nature of the cyanide electrode process involvin,g a gold oxide film mitigates against successful development of a reproducible method.

With the Dionex 10 ion chromatograph in its commercially available form, it was found to be essential to use a thin-layer electrochemical detector (e.g., B.A.S.). Flow noise associated with the single piston pump provided with this instrument is too severe for electrochemical detection used in the wall jet mode, e.g., Metrohm EA 1096 detector cell. Damping could be used to minimize this problem, although we did not attempt to extensively modify the Dionex system and pursue this possibility in depth. The different cell designs and operating principles of different electrochemical detectors are discussed in ref 29. The noise does not effect the conductometric de- tector for which the Dionex system was designed nor a cou- lometric detector (23).

(ii) Detection at Mercury Electrodes. Cyclic voltammo- grams of sulfide and cyanide (Figure 2) at stationary mercury electrodes show well-defined responses in sodium hydroxide as expected (13). The electorde processes at mercury have been well studied (12,13,30-32) and can be represented a!3

M sulfide and (6) 3 X

Hg + nCN- + Hg(CN), + ne- n = 2-4 (1)

Hg + S2- + HgS + 2e- (2 ) The electrode process for sulfide is very complicated at high sulfide concentration (13, 15).

The first dropping mercury electrode (DME) system to be examined for use with the Dionex ion chromatograph was the P.A.R. Model 310 detector cell. By use of this detector, which is based on the wall jet principle, the flow noise from the single piston pump (supplied by Dionex) was unacceptable. This is the analogous problem observed a t gold electrodes where the “wall jet” electrode also was found to be unsatisfactory. In order to use the DME detector (wall jet), we substituted a Waters Model 6,000A pump for the Dionex pump. Thit3 reduced the flow noise considerably but designs based on the “thin layer” method (Tascussell) DME detector were still superior. For the remainder of the results presented in thiri work a Waters pump and “thin layer” electrode (Tacussell) were used.

For optimum performance (29, 33) the DME was set up (short, horizontal capillary) so that the natural drop time was; 0.2 s. Dc currents were monitored at 0.00 V vs. Ag/AgCl iri

584 ANALYTICAL CHEMISTRY, VOL. 54, NO. 3, MARCH 1982

(A)

-05 -04 -03 -02 -31 90

E V o l t s

01pA

i

+$.I

- - 0 9 - 3 8 -07 - 5 6 -05 -C4

E V o l t s

Flgure 2. Cyclic voltammograms at statlonary mercury electrodes in a conventional electrochemical cell of (A) M Na,S in 0.005 M NaOH. Scan rate = 100 mV s-'.

M KCN and (B)

T "r i

0.1 uA

Lu 0 2 4 T i m e Iminl

Flgure 3. Chromatogram for injection of standard containing 300 ng of sulfide and 750 ng of cyanlde using a dropping mercury electrode. Dc response was monitored at 0.0 V vs. AglAgCl (3 M NaCI) in a Tacussell detector cell. Drop time = 0.2 s. Eluent (0.005 M NaOH), flow rate = 0.9 mL/mln. Injection volume = 50 pL.

order that both anions could be determined simultaneously. Results are shown in (Figure 3). Detection limits of 100 ng of sulfide and 250 ng of cyanide were determined when using an injection volume of 100 pL. The detection limit was cal- culated as the concentration where the signal to noise ratio (SIN) is 2:l. Calibration curves are shown in Figure 4. Linearity is observed in the lower concentration range exam- ined. The nonzero intercept for cyanide will be discussed subsequently.

Since the DME response is well defined'and reproducible but solid electrodes are more convenient to use and generally yield better SIN ratios, a mercury-coated solid (platinum) electrode was considered as an alternative to the DME using the Tacussell cell. If the dc detection mode was employed with a mercury-coated electrode (0 V vs. Ag/AgCl), the peaks tailed badly because of formation of insoluble mercury sulfide on the mercury film.

/ 4.01,

A m o u n t in) ( f i g )

Flgure 4. Calibration curves for (0) sulfide and (0) cyanide using the dropping mercury electrode and Tacusseli detector cell. Dc current monitored at 0 V vs. Ag/AgCI (3 M NaCi). Other parameters are as in Figure 3.

2.0 r

I I I I I I 0 2 0 4 0 60 80 100

Pulse Width ims l

Flgure 5. Peak current from chromatogram of cyanide vs. pulse width at a mercury-coated platinum electrode in a Tacusseil detector cell. Pulse height = 400 mV (-0.40 V to 0 V) vs. Ag/AgCi (3 M NaCI). Delay between pulses = 1 s. Current was sampled at end of a 20-ms pulse using sample and hold circultry with a time constant of 4 ms. Injected, 50 pL sample containing 10 pg/mL cyanide. Other param- eters are as in Figure 3.

Use of the normal pulse mode minimized the electode contamination problem (34) and increased the magnitude of the sulfide and cyanide responses. A pulse was applied from -0.80 V to 0.00 V vs. Ag/AgCl. Pulse widths of 100, 80,60, 40, and 20 ms were considered. The optimum response was obtained for both cyanide (Figure 5) and sulfide determination (not shown in Figure 5) at the shorter pulse widths (20 ms). At these short pulse widths the charging current does not have time to decay fully (decays slowly due to high resistance in thin layer cell) and noise is therefore introduced. However, the SIN ratio is still more favorable than with longer pulses. Detection limits are considerably improved using the normal pulse method with mercury-coated platinum electrodes as compared with DME. Thus 25 ng of sulfide and 150 ng of cyanide using a 100 pL injection could be detected. Repro- ducibility of standards was better than 5%.

Despite a higher sensitivity, the mercury film method has some disadvantages compared to the DME. Day to day re- producibility of the mercury film and hence the analyte signal is harder to maintain than at the DME. If the film layer is too thin, bad tailing occurs (electrode saturates); also the electrode can eventually become poisoned. The length of time it lasts will depend on the nature and content of samples injected. Consequently more frequent calibration is required

ANALYTICAL CHEMISTRY, VLL. 54, NO. 3, MARCH 1982 585

inate mercaptan interference by chemical oxidation and cyanide complex formation problems by oxidation and dis- tillation. The oxidation-distillation method enables total cyanide to be determined in the presence of sulfide when using the technique described in this paper. H2S codistills with HCN so that ion-selective electrode methods suffer from in- terference. The need for determining free or total cyanide or sulfide depends very much on the application ( 4 , 14).

CONCLUSIONS A method has been developed for simultaneous determi-

nation of cyanide and sulfide in aqueous solutions, with de- tection limits in the l ppm M) range. With the use of concentrator columns, considerably lower detection limits should be obtained. Use of a higher capacity column or a longer column of the same capacity results in better separation of the anions than reported in this work, if this is required.

2ooL n

4oL 4 16 0.0 0.4 0.8 1.2 1.6

A m o u n t inj ipgl

Figure 6. Calibration curves for (0) sulfide and (0) cyanide obtained in the Tacussell detector cell when using a normal pulse waveform at a mercury-coated platinum electrode. Pulse height = 800 rnV (-0.80 V to 0.00 V) vs. Ag/AgCI. Other parameters are as in Figure 5.

L L L i -2 ! 1 1 ' I 0 1 2 3 4 0 1 2 3 4 0 1 2 3 4

T i m e (mi n)

Figure 7. Chromatograms for sulfide and cyanide in deionized water and drinking water. Experimental condltions are as in Figure 4. Delonized water (A) and drinking water sample (B) contained 250 ng of sulfide and 850 ng of cyanide. I n (C) the drinking water sample as in (B) was spiked with copper ions (2 pglmL) and the decrease in the cyanide response can be seen.

than at a DME. A calibration curve for this method is shown in Figure 6.

Determination of Sulfide and Cyanide in Water. Drinking water was spiked with sulfide and cyanide after blank injections showed no peaks. Figure 7 shows that the response in this water is not the same as in deionized water suggesting that only free sulfide and cyanide are being determined. This is not surprising and is characteristic of most methods for determining sulfide or cyanide. Since the drinking water was known to contain a significant amount of copper when ex- amined by a previously developed method using liquid chromatography with electrochemical detection (35) some of the cyanide may be bound to the metal. Deliberate addition of copper to the water reduced the cyanide peak, supporting this hypothesis. The nonzero intercept for cyanide calibration curves is believed to be a result of complexation by impurities.

Essentially identical results were obtained by using short drop times at the DME, dc mode (0 V) or using the mercu- ry-coated platinum electrode, normal pulse mode (pulse -0.80 to 0 V, pulse delay = 1 s, pulse width = 20 ms). Examination of industrial effluents known to contain sulfide and cyanide demonstrated that the proposed method works extremely well and that free cyanide of sulfide rather than total is being determined.

Essentially, all methods for determining total cyanide re- quired the destruction of metal cyano complexes (19). For example, Bernal et al. ( 4 ) in determining cyanide in the presence of mercaptans with an ion-selective electrode elim-

LITERATURE CITED Dreisbach, R. H. "Handbook of Poisoning"; Lange Medical: Los Altos,

Casarett, L. J.; Doull, J. "Toxicology. The Basic Science of Poisons"; Macmiilan: New Ywk, 1975; pp 157, 202. Williams, W. J. "Handbook of Anion Determination"; Butterworths. London, 1979. Bernal, J. E.; Pardo, R.; Rodriguez, M. J. Anal. Chlm. Acta 1980, 120,

Cusbert, P. J. Anal. Chim. Acta 1978, 87 , 429-435. Peniand, J. L.; Fische, G. Metall. Angew. Chem. 1972, 10, 391-395. Naumann, R.; Weber, C. Fresenlus' 2. Anal. Chem. 1971, 253,

Vesely, J.; Jensen, 0. J.; Nicoiaisen, B. Anal. Chim. Acta 1972, 62. 1-13. McCloskey, J. A. Anal. Chem. 1961, 33, 1842-1843. Humphrey, R. E.; Hinze. W. Anal. Chem. 1971, 43, 1100. Bark, L. S.; Rixon, A. Analyst (London) 1970, 95, 786-790. Kirowa-Elsner, E.; Talmor, D.; Osteryoung, J. Anal. Chem. 1981, 53, 581-583. Canterford, D. R. Anal. Chem. 1975, 47. 86-92. Davison, W.; Garbutt, C. D. J. Nectroanal. Chem. 1979, 99, 311- 320, and references cited therein. Turner, J. A.; Abel, R. H.; Osteryoung, R. A. Anal. Chem. 1975, 47,

Hasebe, K.; Kambara, T. Bull. Chem. Soc. Jpn. 1970, 43,

Humphrey, R. E.; Sharp. S. W. Anal. Chem. 1978. 48, 222-223. Hanagatta, G.; Ohzeki, K.; Kambara, T. Bull. Chem. Soc. Jpn. 1980,

"Standard Methods for the Examination of Water and Waste Water", 13th. ed.; American Public Health Association, New York, 1971. Small, H.; Stevens, T. S.; Bauman, W. C. Anal. Chem. 1975, 47,

Smith, F. C., Jr.; Chang. R. Crlt. Rev. Anal. Chem. 1980, 9 (3).

Mulik, J. D.; Sawicki, E. Envlfon. Scl. Technol. 1979, 13, 804809. Girard, J. E. Anal. Chem. 1979, 51, 836-839. Rabestein, D. L.; Suetre, R. Anal. Chem. 1977, 49, 1036-1039. Anderson, J. E.; Bagchi, R. N.; Bond, A. M.; Greenhill, H. B.; Hender- son, T. L. E.; Walter, F. L. Am. Lab. (FalrfMl, Conn.) 1981, 13 (2), 21-32. Anderson, J. E.; Bond, A. M.; Heritage, I. D.; Jones, R. D.; Wallace, G. G., Deakin Universtty, Waurn Ponds, Vlc. 3217, Australia, 1980-1981, ynpubiished work. Handbook of Chemistry and Physics", 57th. 4.; C.R.C. Press:

Cleveland, OH, 1976-1977. Mekes, L.; "PolaroeraDhic Techniaues"; Interscience: New York.

CA, 1969; p 162-199.

367-370.

111-113.

1343-1347.

2110-2114.

53, 3025-3026.

1601-1809.

197-21 7.

_ . 1965; p 432. Hanekamp, H. B.; Van Nieuwkerk, H. J. Anal. Chlm. Acta 1980, 121, 13-22. Kolthoff, I. M.; Miller, C. S. J. Am. Chem. Soc. 1941, 63, 1405-1411. Newman, L.; Cabral, J. U.; Hume, D. H. J. Am. Chem. SOC. 1958,

Tanaka, N.; Murayama, T. 2. Phys. Chem. (Lelpzig) 1973, 1 1 , 366. Hanekamp, H. B.; Bos, P.; Brinkman, U. A.; Frei, R. W. Fresenlus' 2. Anal. Chem. 1979, 297, 404410. Dieker, J. W.; van der Linden, W. E.; Poppe, H. Talanta 1979, 26.

Bond, A. M.; Wallace, 0. G. Anal. Chem. 1981, 53, 1209-1213.

80 , 1614-1819.

511-516.

RECEIVED for review review July 2,1981. Accepted November 12, 1981.