AUGUST 1983 The Analyst Vol. 108 No. 1289 Combination of Flow Injection Analysis with Flame Atomic-absorption Spectrophotometry: Determination of Trace Amounts of Heavy Metals in Polluted Seawater Svend Olsen Luiz C. R. Pessenda," Jaromir RfiziCka and Elo H. Hansen Chemistry Department A The Technical University of 3enmark Building 207 DK-2800 Lyngby Denmark A simple flow injection system the FIAstar unit was used to inject samples of seawater into a flame atomic-absorption instrument allowing the deter-mination of cadmium lead copper and zinc at the parts per million level at a rate of 180-250 samples per hour. Further on-line flow injection analysis pre-concentration methods were developed using a microcolumn of Chelex- 100 resin allowing the determination of lead at concentrations as low as 10 parts per los (p.p.b.) and 1 p.p.b.for cadmium and zinc. The sampling rate was between 30 and 60 samples per hour and the readout was available within 60-100 s after sample injection; the sampling frequency depended on the pre-concentration required. Keywords Heavy metals determination ; flow injection analysis ; flame atomic-absorption spectrophotometry ; trace analysis ; polluted seawater Atomic-absorption spectrophotometry (AAS) is a well established extremely valuable tech-nique for the determination of trace amounts of metals. Since its introduction by Walsh,l the method has gone through a number of development stages aiming at obtaining an increase in reliability ease of operation and above all improvement in the limit of detection.Hence, atomisation by a flame has been supplemented or replaced by the graphite furnace technique, background correction has been introduced to compensate for non-specific absorption phenomena and most recently background correction has by exploiting the Zeeman effect, become more sophisticated by incorporating a polariser and a magnetic field into the commer-cial instruments. However despite all these improvements the assay of trace amounts of metals in samples with high salt content such as seawater still remains a time-consuming and difficult task. Because the high salt content in evaporated samples might result in clogging of the burner or poisoning of the inner surface of a graphite tube heavy metals generally have to be pre-separated from the salt matrix either by solvent extraction or by ion exchange.While diethyldithiocarbamate extraction into isobutyl methyl ketone is widely used this method is far from ideal because besides being laborious it also requires very careful work if blank values are to be kept low. Further analysis of a large number of samples yields along with the analytical results considerable volumes of used organic solvents that have to be disposed of in environmentally acceptable ways. Therefore there is a marked tendency to use another pre-concentration technique that is ion exchange on Chelex-100 resin,2 which is efficient and yields low blanks3 A recent paper by Danielssen et and an older yet much more detailed paper by Kingston et aZ.,5 have summarised the present state of seawater assays based on ion exchange.Thus a typical procedure requires 100-500 ml of seawater which is to be passed through a column containing 5 g of Chelex-100 at a rate of 2 ml min-l, followed by elution into 10 ml of 5 M nitric acid. Hence pre-concentration of a single sample by a factor of ten will require over 1 h while one 1-1 sample volumes have to be pre-concen-* Present address Centro de Energia Nuclear na Agricultura (CENA) Universidad de Sao Paulo, 13.40O-Piracicaba S.P. Brazil. 90 906 OLSEN et &. FIA WITH FLAME AAS DETERMINATION ANabySt vd. 108 trated overnight. Using this pre-concentration procedure and graphite furnace atomic-absorption spectrophotometry lead and cadmium concentrations down to the 0.5 parts per 109 (p.p.b.) level have been determined in seawater.Flow injection analysis (FIA)6 in combination with atomic-absorption spectrophotometry was first suggested and used by Zagatto et aZ.' as a means to dilute and to add lanthanum solution prior to sample introduction into the flame. As the reagent addition was done in the zone-merging mode both reagent and time economy were improved; indeed a sampling fre-quency of 300 samples h-l was achieved. Independently Wolf and Stewarts replaced aspira-tion by sampling via an FIA system achieving a sampling rate of 180 samples h-l. Standard addition and matrix effect compensation using FIA was recently suggested by Tyson and I d r i ~ ~ while the unique ability of FIA - AAS to handle samples with high salt contents (up to 25% magnesium chloride solution) was demonstrated by Mindel and Karlberg.lo The use of FIA as an on-line pre-concentration for flame-atomic absorption (FAA) has so far been based on low-flow solvent extraction and has been applied to assay for copper.ll However the flow system requiring solvent separation and re-injection seems to be complicated.The purpose of this work was to develop a rapid and reliable technique for screening of a large number of seawater samples while determining their lead cadmium zinc and copper contents. At levels of 10-4.1 p.p.m. this has been done by direct sampling of seawater into the flame using the FIAstar system.12 However samples with lower heavy-metal contents had to be pre-concentrated within a more sophisticated FIA system that was equipped with peristaltic pumps and microcomputer control.The method was to be used for the environ-mental control of an area around Marmorilik in Greenland where the water contamination, caused by extensive mining activity must be strictly controlled. This involved analysis of a large number of water samples collected at the mining plant itself and in the adjacent fjord system and therefore economy of time reagent consumption and acquisition of instrumenta-tion were besides the reliability of the method the important design parameters. Experimental Apparatus The atomic-absorption spectrophotometer (Varian Model AA-1275) was connected in parallel to a recorder (Radiometer Servograph REC 80 furnished with an REA 112 high-sensitivity module) and via a home-made interface to a computer (PET Commodore Model 3032 combined with a Commodore CBM Model 2040 dual-drive floppy disk) and a printer (Tractor Model 3022).For the FIA systems shown in Figs. 4 and 7 and for the automated system (Fig. 10) the contact to the computer was triggered by a microswitch in the injection valve of the FIA system which served to inform the computer of the sequence of events. The automated system in Fig. 10 additionally included a timer that controlled the timing of the stop - go intervals of the two attached peristaltic pumps (Ismatec Model Mini-S-840). The flow injection system consisted of a FIAstar unit details of which have been published elsewhere12 (for actual manifold designs used in this study see Results and Discussion). All connecting tubes consisted of 0.5 mm i.d. Microline.As the FIAstar valve is furnished with an external loop in order to regulate the injected sample volume change of volume was simply effectuated by changing the external loop. For small sample volumes (less than 200 p1) the sample loop consisted of 0.5 mm i.d. tubing while tubes of larger i.d. were used for larger sample volumes (up to 1.57 mm i.d. for sample volumes exceeding 1 ml). Reagents All chemicals were of analytical-reagent grade and re-distilled water was used throughout. All reagent and standard solutions were stored in clean polyethylene bottles. Standard solutions of lead(II) cadmium(II) zinc(I1) and copper(1I) were made up by suitable dilutions of 1000 p.p.m. aqueous standard solutions for AAS (BDH Chemicals), certified by the manufacturer to contain 1000 & 5 mg 1-1 of the metal.In order to simulate the matrix of the actual water samples all standards were prepared by dilution with water containing 31.3 g 1-1 of sodium chloride and 3 ml l-1 of concentrated nitric acid. The samples originated from a mining plant at Marmorilik in Greenland collected partly in the fjord system near the plant and partly in the plant itself where large volumes of seawate August 1983 OF HEAVY METALS IN POLLUTED SEAWATER 907 from the fjord are used in the flotation process for separating the minerals from the parent rock material. Prior to shipment from Greenland to Denmark all samples were preserved by adding 3 ml of concentrated nitric acid per litre of sample solution. The ammonium acetate buffer solutions used in the FIA system were prepared by aqueous dilutions of a 2 M stock solution made by mixing 55.5 ml of 99% acetic acid with 112.5 ml of 25% ammonia solution; re-distilled water was added to a total volume of 500 ml.Any re-adjustment of the pH in the diluted solutions (see Results and Discussion) was done by the addition of ammonia. The Chelex-100 cation-exchange resin 50-100 mesh (sodium form) was purchased from Bio-Rad Laboratories. Preparation of Chelex- 100 Microcolumn The microcolumn was made from the FIAstar gradient tube which is a Perspex block into which is drilled a 50 mm long channel of 2 mm i.d. furnished at both ends with threaded term-inals so that the connecting tubes can be securely attached by means of screws and gaskets as used for all FIAstar connections.12 Prior to packing the Chelex-100 resin was converted to the ammonium ion form by storing it in a large volumetric excess of 0.05 M ammonium acetate buffer solution for 2 d shaking the slurry intermittently and renewing the buffer solution 2-3 times each day.Being now ready for packing the resin in the ammonium ion form may be stored in the buffer solution for extended periods of time but it is recommended that the resin is washed with fresh buffer solution 2-3 times prior to packing. For packing the column the resin - buffer slurry was aspirated into a l-ml syringe the con-tents of which were then carefully emptied into the column. At the end of the column a small piece of polyurethane foam was inserted to entrap the resin particles mechanically within the column.The process was repeated until the column was completely filled with resin each time taking care that the introduction of air was avoided. Fully packed and closed at both ends by polyurethane foam the column contained about 25 mg of resin. As a routine all new columns were conditioned in the FIA system for about 10 min during which time ammonium acetate buffer solution and 2 M nitric acid solution were pumped through the column inter-mitt en tly . Results and Discussion Direct Injection of Sample Material The simplest approach to the water assays was tested by combining the Varian AA-1275 FAA spectrophotometer with the simplest FIA unit commercially available the FIAstar. The FIAstar unit originally developed for students teaching and research,12 is a robust modular system consisting of an injection valve (loop type so that the injected volume may be varied) assorted connectors coils and a carrier stream propulsion unit.The carrier stream is propelled by gas (air or nitrogen) and the very low pressure (0.1-0.5 bar) is maintained by means of a precision regulator. In this work the FIA system serves mainly as a means of Fig. 1. Single-line FIA - FAA manifold where the carrier solution (C) (5 x M sulphuric acid) is propelled by gas main-tained at a pressure P of 0.5 bar corresponding to a flow-rate of 4.9 ml min-l. A sample volume of 150 pl is injected by means of valve S and then passed through the shortest possible length of line (20 cm) to the FAA instrument. In order to obtain optimum performance of the system it is necessary to operate the pro-pulsion of the FIA system a t a higher flow-rate than the aspiration rate of the nebuliser of the FAA instrument (3.1 ml min-l).W = Waste 908 OLSEN ef at!. FIA WITH FLAME AAS DETERMINATION Analyst Vd. 108 transport and of exact timing and so a single line limited dispersion configuration6 was chosen and assembled (Fig. l) using tubing of 0.5 mm i.d. between the injection valve (S) and the FAA instrument. The useful calibration range for each of the four metal species was established by injecting a series of standards into the continuously flowing non-segmented carrier stream of 5 x 10-4 M sulphuric acid while the absorbance as measured by the FAA instrument was continuously recorded [Fig. 2 ( a ) ] . In all instances plots of the peak height zlenus concentration yielded strictly linear calibration graphs.Calculating the limit of detection (LOD) according to the recommendations in reference 13 a value of 0.010 p.p.m. was found for zinc while the maximum attainable sample frequency was 250 samples h-l with the readout available within 5 s after sample injection [Fig.- 2(6)j Copper cadmium and lead yielded equally satisfactory results (Table I). TABLE I PERFORMANCE DATA OF THE FIA - FAA SYSTEM WITH DIRECT INJECTION OF SAMPLE The sample volume was 150 pl. Element assayed Par am e ter Wavelength/nm . . LOD* (20),t p.p.m. . . LOD (20),$ p.p.m. . . LOD (34 p.p.m. . . LOQS (loo) p.p.m. . . Characteristic concentration tq p.p.m. . . Characteristic concentration $ p.p.m.. . . . Sensitivity,ll AA/Ap.p.m. . . Correlation coefficient. . Cadmium 228.8 0.007 0.001 5 0.009 0.030 0.01 1 0.01 0.220 0.9990 Copper 324.7 0.001 0.003 0.005 0.032 0.035 0.03 0.100 0.999 1 Lead 217.0 0.01 0.01 0.032 0.227 0.049 0.1 0.037 0.996 3 7 Zinc 213.9 0.007 0.000 8 0.010 0.026 0.011 0.008 0.498 0.999 2 * Limit of detection based on a 95.5% (20) and 99.7% (30) confidence level. t As found in this work. # As reported by the manufacturer. § LOQ limit of quantitation.13 7 Characteristic concentration defined as that concentration of the element which gives rise to 11 Sensitivity (the slope of the calibration graph) and correlation coefficient as found when operating 0.0044 absorbance unit (z.e.1 % absorption). the FIA - FAA system within the working range of the AA-1275. Whenever a flow injection system is being combined with a conventional instrumental method a natural question is asked “What is the trade-off between the increased sampling frequency and the loss of sensitivity caused by the readout of peak height rather than at the steady state?” The answer for this FIA - FAA combination is given in Fig. 2(b) where the recorder response is shown as obtained with a 1.5 p.p.m. zinc standard solution continuously aspirated in the conventional mode (B) and injected with a 15O-pl sample volume (A). It is obvious that the FIA peak is only 20% lower than the steady-state plateau but being much narrower with a standard deviation of ut = 1.5 s it allows theoretically a sampling fre-quency (Smax.) as high as 600 samples h-l [Sma,.= 3600/kut where k = 4 (reference S ) ] . In other words “the peak value for any given time f is as good a measure of the concentration of the analyte as the final steady state reached,” as already stated in the first paper on FIA.14 This feasible trade-off between sensitivity and sampling frequency is the result of careful design of the geometry of the flow channel of the injection unit and of the injected sample volume. Indeed the dispersion coefficient D = Co/Cmax. = 1.3,15 where C the concentra-tion [Fig. 2(b)] has a value confirming that a limited dispersion of the injected sampling plug has been achieved. Examination of the effects caused by the seawater matrix which in the samples originating from the coastal waters of Greenland may reach up to 3.3% salinity was the next step in th August 1983 OF HEAVY METALS IN POLLUTED SEAWATER 909 1 1.3 Q 2 Scan + Fig.2. Recordings obtained with the single-line FIA - FAA system of Fig. 1 and the flow charac-teristics detailed there. (a) Calibration run for zinc as obtained by the injection of 0.1 0.2 0.5, 0.75 1.0 1.5 and 2.0 p.p.m. zinc standards (sample volume 150 pl). (b) Recorder response for the 1.5 p.p.m. zinc standard as obtained by (A) injection via the FIA system and (B) continuous aspiration in the conventional mode. For the sake of comparison the aspiration rate in (B) was increased to 4.9 ml min-l corresponding to the propulsion rate used in (A) where S is the point of injection.Wavelength used in (a) and (b) 213.9 nm. D represents the dispersion number which in (B) is equal to 1. development of the method. Thus Fig. 3 shows a set of calibration recordings for a series of lead standards prepared with and without the addition of sodium chloride simulating the actual matrix of the seawater samples. Because the sampling period is only a fraction of each measuring cycle during the length of which the carrier stream flows uninterruptedly the nebuliser and the burner are effectively washed and therefore no salt deposit is being built up. Thus as originally observed by Mindel and Karlberg,lo large series of seawater samples may be analysed by FAA with no 2 0% T 3.3% Scan + b) 10.05 0.5 0.25 p.p.m. Fig. 3. Calibration runs for a series of lead standards (2 5 10 15 and 20 p.p.m.) as obtained with the FIA - FAA system of Fig.1 and the conditions rendered there recorded (a) 1 without and 2, with sodium chloride added to the standards to simulate the matrix of seawater. (b) With increased amplification the output for two lead standards containing 0.5 and 0.25 p.p.m. respectively, and the calculated values for the limit of detection (LOD) and the limit of quantitation (LOQ) attain-able with the FIA - FAA system. Wavelength used in both (a) and (b) 217.0 nm 910 OLSEN et al. FIA WITH FLAME AAS DETERMINATION Analyst VoZ. 108 adverse effect caused by the matrix. The limit of detection LOD (defined as 3 u where u is the standard deviation of the base-line fluctuation~l~) was 0.032 p.p.m.of lead and the limit of quantitation LOQ (defined as 100 of the base-line fluctuations) was 0.227 p.p.m. of lead [Fig. 3(c)]. Although these values are higher than the detection limits quoted occasionally in the literature on FAA they truly reflect the realistic limits obtained with an advanced instrument such as the Varian AA-1275 tuned to optimum performance. On-line Pre-concentration Using a Microcolumn of Chelex-100 The main advantage of using on-line pre-concentration is that all samples and standards are subjected to exactly the same treatment from the moment of injection to the moment of detection and that the same ion-exchanger column is used for all samples and standards. As all time events the geometry of flow and the ensuing chemical reactions are strictly controlled and reproducibly maintained in an FIA system; there is no need to achieve quantitative adsorption of metals from the injected sample zone although the subsequent elution must be completely effective otherwise carry-over from one sample to the next will occur.Because a number of critical parameters had to be investigated to optimise the conditions the method had to be developed in three stages as described under Single-line Two-valve System. Single-line Two-valve System This is a logical extension of the FIAstar system the components of which were used to build up the system described earlier as the direct single-line system (Fig. 1). For the FIA on-line pre-concentration procedure three units were added (Fig. 4) which were a second injection valve a mixing coil and a microcolumn (volume 150 pl 5.0 cm long) filled with Chelex-100.The carrier solution was 0.05 M ammonium acetate with which the sample zone (volume 1 .O ml) was mixed during the passage from the injection valve (S) through the coil and the bypass conduit of valve A to the Chelex-100 microcolumn. The trace amounts of metal to be assayed were adsorbed on the chelating group of the resin which has the selectivity sequence Cu 2+ > > Pb2+ > Fe3+ > AP+ > Cr3+ > Ni2+ > Zn2+ > Ag+ > Co2+ > Cd2+ > Fe2+ > Mn2+ > Ba2+>Ca2+>>Na+>H+.2 Thus metals adsorbed in neutral media can be desorbed by strong acid and by injecting 180 pl of 2 M nitric acid solution by valve A the metals originally present in the 1 ml sample volume can be released into the much smaller volume of the acid zone and then be transported into the flame where they absorb the light emitted by a hollow-cathode lamp and are measured in the usual way.Thus when samples containing 20-500 p.p.b. of lead were injected pre-concentrated and then eluted a calibration record was obtained (Fig. 5) which showed no evidence of carryover. This yielded a straight-line relationship between absorbance ( A ) and concentration (C) with a regression coefficient of 0.9999 a sensitivity of AA/Ap.p.m. = 0.517 LOD = 0.004 p.p.m. of lead and LOQ = 0.015 p.p.m. of lead. A closer examination of a single assay cycle recorded at a high chart speed [Fig. 5(b) (A)] revealed several interesting details. Each assay cycle consists of two periods the first and longer one (S-E 55 s) is the pre-concentration step and Fig.4. Single-line two-valve FIA - FAA system for on-line pre-concentration using a microcolumn of Chelex- 100. The metal-contain-ing samples (volume 1 ml) are injected by valve S into the carrier stream C (0.05 M CH,COONH,) propelled by a gas pressure P corresponding to a flow-rate of 3 ml min-l. The two components are mixed in the 1-m coil before being passed via the by-pass of valve -4 to the column (CH-100). After pre-concentration valve S is closed and a small zone of 2 M nitric acid (volume 180 p1) is injected by valve A the acid zone eluting the metal and transporting i t into the FAA instrument. During the latter sequence valve S is closed being ready for loading of the next sample.W = Waste Augzcst 1983 OF HEAVY METALS IN POLLUTED SEAWATER 911 0.25 -Q, C lu e s 2 0 - I B Scan -b Fig. 6 (a) Calibration runs for lead with a series of acidified standards (20 60 100 200 360 and 500 p.p.b.) prepared in a matrix of sodium chloride (31.3 g 1-l) using the system of Fig. 4. (b) High-speed recording of the 600 p.p.b. lead with-out (A) and with (B) background correction both sample injections started at S and the elution sequences started at E. Wavelength used in (a) and (b) 217.0 nm. the second and shorter one is the elution step (beginning at E and lasting 5 s). Two peaks appear during each assay cycle the first one low and wide the second one high and narrow the latter increasing with increasing content of lead. As the low peak appears during the pre-concentration period and can be completely eliminated by operating the FAA Varian AA-1275 spectrophotometer with background correction [Fig.5(b) (B)] it is obvious that this peak is due to non-selective absorption of light emitted by the hollow-cathode lamp by sodium atoms originating from the sample matrix (which passes through the column into the flame). The sodium interference and the influence of pH on the recovery of heavy metals on Chelex-100 has been discussed already in the first comprehensive work by Riley and Taylor,3 when it was shown how the uptake of metal increases with pH. The upper pH limit was set to 9.1, at which value calcium and magnesium precipitation was said to interfere. Although a number of papers have been published since (e.g.reference IS) Kingston et aL,5 in a more'recent work, still point out the difficulty of the complete separation of sodium which is present in high levels in seawater from heavy metals when a pH of 5.5 is used in their procedure. Having now at hand a tool to resolve the contribution of sodium to the peak height veysus that of heavy metal (Fig. 5) it was decided to investigate the influence of pH on the recovery. For this purpose solutions containing 10 20 50 and 100 p.p.b. of cadmium were prepared in 3.3% sodium chloride and subsequently injected (in duplicate) in the system depicted in Fig. 4. As already mentioned each sampling cycle consisted of a pre-concentration period, which commenced by the switching of the sample injection valve (S) followed by an elution period commencing by the injection of acid by means of valve A.Similarly to the system in Fig. 5 first the sodium and then the cadmium peak appeared during each cycle. The calibra-tion run was repeated four times (Fig. 6) using as carrier stream a 0.05 M ammonium acetate solution the pH of which was adjusted to 7.0 9.0 and 10.0. Obviously the peak height and the recovery of cadmium increased with increasing pH with an optimum at pH 10. Closer examination of the first smaller and broader (ie. sodium) peaks revealed that their height is constant only at pH 10 while at lower pH values most notably at pH 7 their height increases with increasing cadmium content showing that at low pH cadmium cannot be quantitatively retained on the Chelex-100 resin under the prevailing flow conditions and column geometry; hence the non-retained cadmium appears as part of the sodium peak.At pH 10 the recovery is quantitative and the high content of sodium in no way affects the second high and narrow peak of cadmium. This is best seen on the last experimental run where the four cadmium standards were injected in an increasing concentration sequence while the FAA spectrophoto-meter was operated without background correction. Afterwards another series of the stand-ards was injected in a decreasing concentration sequence while the FAA spectrophotometer was operated with background correction. The equal height of the comparable cadmiu 912 OLSEN et al. FIA WITH FLAME AAS DETERMINATION Analyst VoE. 108 Scan + Fig. 6. Influence of pH on the recovery of cadmium as obtained by using the system of Fig.4. The acidified cadmium standards injected in all experiments contained 10 20 50 and 100 p.p.b. respectively in a matrix of sodium chloride (31.3 g 1-I). (a) pH 7; (b) pH 9; and (c) pH 10 showing the influence of increasing pH on the cadmium peaks; and (d) also pH 10 depicting the Cali-bration series run without (left) and with (right) back-ground correction applied. Wavelength used in (a) (b), (c) and ( d ) 228.8 nm. peaks and the absence of any deviation on the base line when the background correction was applied confirm the complete recovery of cadmium from the carrier stream during the pre-concentration period and the complete elution of the cadmium from the column during the elution period.Thus two advances have been made at this stage (1) the LOD and the sensitivity of measurement have been increased compared with the direct injection of sample system; and (2) the sodium interference has been eliminated. The single line experimental set-up used so far has however certain shortcomings. The first is due to the fact that the ion exchanger undergoes a drastic change of volume any time the Chelex-100 changes from the NH,+ to the H+ form. This swelling has been reported to be more than 100% of the resin volume2 and causes tighter and tighter packing of the column because the flow in the single line system is uni-directional. Therefore the downstream end of the column eventually becomes blocked while the upstream end becomes void. The second shortcoming of the single-line system lies in its inappropriateness to mix the centre of the sample zone sufficiently with the carrier stream (unless a long mixing coil is used).While this was not a problem when synthetic seawater samples were injected the single-line system could not be used to handle seawater samples that have been acidified by nitric acid after collection, in order to preserve them (3 ml of 14 M nitric acid per litre of seawater). These shortcomings are eliminated in the system described under Two-line FIA - FAA System with a Directional Valve and Peristaltic Pumps. Two-line FIA - FAA System With a Directional Valve and Peristaltic Pumps There are two distinct features of the two-line system depicted in Fig. 7. The first is the confluence arrangement that allows continuous addition of the ammonium acetate buffer to the aqueous carrier stream in pre-selected proportions at the merging point M.This and the passage through the subsequent mixing coil ensure stabilisation of pH along the whole length of the injected sample zone even when acidified seawater samples are injected. The second feature is the use of an additional FIAstar valve used exclusively for diversification of the flow, which allows the flow to be directed through the Chelex-100 column (CH-100 Fig. 7) in one direction during the pre-concentration step and in the opposite direction during the subsequent elution step. Hence all directional functions symbolised by circles in Fig. 7 are executed by one standard FIAstar valve that has two positions (A) pre-concentration and (B) elution August 1983 OF HEAVY METALS I N POLLUTED SEAWATER 913 Fig.7. Two-line FIA - FAA system with directional valve and peri-staltic pump (a) pre-concentration cycle during which the sample (volume 2 ml) injected by valve S is merged with the ammonium acetate buffer (0.6 M) and then directed to the column; A the corresponding position of valve for cycle (a) ; and (b) elution cycle where the metal adsorbed on the Chelex-100 column is countercurrently eluted by 2 M nitric acid and subsequently measured in the FAA instrument; B the corresponding position of valve for cycle (b). M is the merging point and the lines propelling liquids into the FAA instrument in (a) or (b) are drawn thick. W = Waste. The pre-concentration cycle is shown in Fig.7 (a) where the thin line depicts the pre-concentra-tion line into which the sample has been injected; note that the flow line drawn solidly continuously washes the nebuliser and burner with pure carrier stream of eluting acid. When the rotor is turned [A + B Fig. 7 ( b ) ] the flow into the FAA (again drawn as a solid line) is diverted through the column in the former reverse direction and then into the FAA thus bringing the eluted metal into the flame for measurement; during this period the FIAstar sampling valve can be reloaded. It must also be noted that the flow arrangement never allows the seawater matrix to enter the flame because during the pre-concentration cycle the 30 min 500 20 50 1 500 Scan Fig. 8. Assay of lead in seawater samples using the FIA - FAA system of Fig.Shown are 6 samples bracketed by two series of standards the numbers All samples and stardards were 7. depicting the concentrations in p,p.b. of lead. injected in triplicate. Wavelength 217.0 nm 914 OLSEN et al. FIA WITH FLAME AAS DETERMINATION TABLE I1 COMPARATIVE RESULTS FOR ASSAY OF LEAD AS OBTAINED BY FIA - FAA Analyst VoZ. 108 AND BY POTENTIOMETRIC STRIPPING ANALYSIS (PSA) Lead p.p.b. Sample NO. ‘FIA - FAA* PSA’ 1 100 93 2 237 243 3 333 332 4 372 373 6 43 41 *As obtained with the FIA-FAA system of Fig. 7. sample plug enters the column and continues to waste [Fig. 7(a) top line] Therefore the two-line FIA - FAA system obviates the necessity for any background correction which in turn reduces the cost of the apparatus.The system was tested for lead zinc and cadmium assays in real seawater samples and performed so well that several hundred analyses have so far been executed with the same column, In Fig. 8 is shown a recording obtained for lead with a set of five seawater samples (pre-served by nitric acid) bracketed by two series of standards (20 50 100 200 and 500 p.p.b. of lead). All samples were injected in triplicate. The range encompassed satisfactorily covers the lead levels encountered in the samples from the mining process (where seawater is used for flotation). As the LOD was determined to be 10 p.p.b. the system could also be used for assay of the samples from the adjacent fjord. In a typical analysis much larger runs than the five samples shown in Fig.8 would of course be bracketed by standards the record shown here was merely shortened for graphical presentation. For the purpose of actual pollution control the FAA values were re-checked by another method i.e. potentiometric stripping analysis.17J8 A comparison of the results obtained by the FIA - FAA and by the potentiometric stripping analysis is shown in Table 11. The ion-exchange capacity of the miniaturised column was calculated and compared with a “breakthrough” graph obtained under prevailing experimental conditions with the aim of exploiting the limitation of the present technique. The microcolumn contains about 25 mg of Chelex-100 an amount much smaller than usually ~ s e d . ~ - ~ This amount is in a stoicheiometric excess of over 30 times that which would be needed to accommodate the amount of lead con-tained in 1 ml of a 500 p.p.b.lead standard. The experimental breakthrough graph was obtained by injecting increasing volumes of the 500 p.p.b. acidified lead standard and the 50 p.p.b. acidified cadmium standard in seawater matrix using the system shown in Fig. 7 and a carrier stream of 0.5 M ammonium acetate adjusted to a pH of 10 and is shown in Fig. 9. 1 .o al C Q e 8 0.5 9 0 5 10 15 Sample volume/ml Fig. 9. “Breakthrough” graphs for the miniaturised column of Chelex-100, containing 26 mg of resin as obtained by injecting increasing volumes of standard solutions containing (A) 500 p.p.b. of lead and (B) 60 p.p.b. of cadmium Azcgzcst 1983 OF HEAVY METALS IN POLLUTED SEAWATER 915 Sample volumes up to 2ml can readily be accommodated the breakthrough limit being reached at 4-5 ml.Thus further pre-concentration and an increase in the detection limit for the method can be achieved only by increasing the column size and the amount of Chelex-100 contained within it. Fully Automated System With several hundreds of seawater samples to be assayed the last task was the further automation of the system because exact timing during each cycling period was strenuous and labour demanding when performed manually. Therefore a single-valve two-pump system was designed. This was equipped with an electronic timer capable of sequencing pump 1 and pump 2 in a stop - go mode each pre-concentration - elution cycle being initiated by the turn of the injection valve (Fig.10). During the pre-concentration cycle pump 1 (Pl) was going while the sample was injected by turning the valve (S). Afterwards the injected zone was mixed with buffer in the coil and passed through the microcolumn. In the next sequence, pump 1 was stopped and pump 2 (P2) started permitting the eluting acid to move through the column in the opposite direction and thereby transport the eluted metal into the FAA spectro-photometer. Note that when pump 1 is stopped the liquids in the thus closed circuit cannot move in either direction. The sampling cycle is completed when the peak appears whereupon pump 2 may be stopped and pump 1 reactivated thus establishing the high pH inside the microcolumn and thereby making it ready for the next pre-concentration step. Note also that, as in the previous system the seawater matrix never enters the FAA spectrophotometer and that the microcolumn is carefully regenerated prior to each sampling cycle while being operated in countercurrent fashion.Typical calibration runs for series of lead and cadmium Fig. 10. Fully automated FIA - FAA system operated via two peristaltic pumps the stop and go sequences of which are controlled by an electronic timer (T). The stated pumping rates are in ml min-l with sample volume 2 ml. Similarly to the system of Fig. 7 the operation consists of a pre-concentration cycle and an elution cycle, during which the metal is desorbed from the column counter-currently. For further details see text 916 OLSEN et al. FIA WITH FLAME AAS DETERMINATION Analyst YoZ.108 Scan __+ Fig. 11. Calibration runs for (a) lead (217.0nm) (20, 50 100 200 350 and 500 p.p.b.); and (b) cadmium (228.8 nm) (2 5 10 20 35 and 50 p.p.b.) as executed in the FIA - FAA system of Fig. 10. standards are shown in Fig. 11. The slight increase in sensitivity as compared with the results of the manually operated FIA - FAA system shown in Fig. 7 is due to the improved flow geometry of the system. The manifold components shown within the shaded area of Fig. 10 were all incorporated into an integrated microconduit,lg which is a new way of fabricating flow-through systems. In the integrated microconduit all connectors mixing coil and the ion-exchange column are imprinted within a plastic block which is then permanently sealed by a second plate. Therefore con-necting tubing and Swagelock-type connectors are eliminated as the whole structure is em-bedded within a 7 x 4 x 1.5 cm thick plastic block with four inlets and two outlets.Such a micro-flow injection manifold can then be placed within and operated by a BIFOK - Tecator FIA 5020 instrument which has two pumps injection valve with variable volume and appro-priate timing and sequencing facilities. It is fitting to conclude this section by referring to a serious deficiency from which all methods for the determination of trace amounts of metals in natural waters and seawater suffer. At the parts per billion level and below there may be a substantial fraction of the lead or cadmium present in forms that have been described as “bound,”20-22 thus being inaccessible to determination because these metals are present in colloidal form insoluble in acid or in the form of strong complexes that are not dissociated.As only the “labile” and “free” metals can be electro-deposited extracted or adsorbed on Chelex-100 only these fractions are measurable by the present analytical techniques because the seawater matrix itself interferes in the direct measurement by flame or graphite furnace atomic absorption. It is of course questionable whether lead and cadmium concentrations below the parts per billion level present in either forms are objectionable from an environmental viewpoint as they approach the levels of natural abundance in ~eawater.~3,~4 This subject is however outside the scope of this paper, except that it allows us to put into perspective the usefulness of the FIA - FAA combination as a pollution-screening technique at realistic levels.Conclusion Seen from a practical viewpoint the FIA - FAA combination results in time saving because it allows an unprecedented sample throughput at the parts per billion level. As the analyti-cal readout is available within 5 s for the direct assay and the latest within 110 s for the system including pre-concentration (Fig. lo) smaller sample series can be treated expediently by manual injection. The saving in terms of equipment is not limited to a sample changer. Graphite furnace and background correction are no longer needed while the necessary FIA equipment may be acquired at a fraction of the cost of a graphite furnace. Truly the graphite furnace is still more sensitive than the FIA - FAA system by a factor of about 50 yet for sea-water assays a manual solvent extraction or Chelex separation nevertheless has to be performed.On the other hand it may still be possible to increase the sensitivity of the FIA - FAA system by a factor of five by increasing the injected sample volume to 10 ml and the size of the column about %fold while redesigning the geometry of flow in the microconduit August 1983 OF HEAVY METALS IN POLLUTED SEAWATER 917 Much effort has rightfully been made in the past to improve atomic-absorption equipment as far as computerisation electronics optics and sample atomisation are concerned. It is time now to turn attention to automation of the sample treatment prior to sample entry into the FAA instrument.While FIA as a simple means of transport is certainly useful the on-line FIA pre-concentration technique by ion exchange as described here or via hydride generation as developed by ~ h t r o m ~ ~ are amongst the significant future trends which are not limited only to FAA but naturally can be extended to ICP also. The authors express their gratitude to Greenex A/S and the Danish Academy of Technical Sciences for providing the funds for S. Olsen to DANIDA for a scholarship to L.C.R. Pessenda and to the Danish Council for Scientific and Industrial Research which together with the above-mentioned institutions financed part of the equipment used in these studies. 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18.19. 20. 21. 22. 23. 24. 25. References Walsh A. Spectrochim. Acta Part A 1955 1 108. Bio-Rad Laboratories Product Information 2020 March 1981. Riley J. P. and Taylor D. Anal. Chim. Acta 1968 40 479. Danielsson L.-G. Magnusson B. and Zhang K. At. Spectrosc. 1982 3 39. Kingston H. M. Barnes I. L. Brady T. J. and Rains T. C. Anal. Chem. 1978 50 2064. RhZiEka J. and Hansen E. H. “Flow Injection Analysis,’’ Wiley-Interscience New York 1981. Zagatto E. A. G. Krug F. J. Bergamin FO. H. Jergensen S. S. and Reis B. F. Anal. Chim. Wolf W. R. and Stewart K. K. Anal. Chem. 1979 51 1201. Tyson J. F. and Idris A. B. Analyst 1981 106 1025. Mindel B. D. and Karlberg B. Lab. Pract. 1981 30 719. Nord L. and Karlberg B. Anal. Chim. Acta 1981 125 199. Rbiitka J. Hansen E. H. and Ramsing A. U. Anal. Chim. Acta 1982 134 55. ACS Committee on Environmental Improvement Anal. Chem. 1980 52 2242. RbfiCka J. and Hansen E. H. Anal. Chim. Acta 1975 78 145. Hansen E. H. Rhiitka J. Krug F. J. and Zagatto E. A. G. Anal. Chim. Acta 1983 148 111. Florence T. M. and Batley G. E. Talanta 1975 22 201. Jagner D. and Graneli A. Anal. Chim. Acta 1976 83 19. Jagner D. and ArBn K. Anal. Chim. Acta 1978 100 375. ROZiEka J. Janata J. and Hansen E. H. in preparation. Florence T. M. and Batley G. E. Talanta 1976 23 179. Batley G. E. and Florence T. M. Anal. Lett. 1976 9 379. Abdullah M. I. El-Rayis 0. A. and Riley J. P. Anal. Chim. Acta 1976 84 363. Bruland K. W. and Franks R. P. Anal. Chim. Ada 19‘79 105 233. Jones P. G. W. Anal. Proc. 1982 19 565. Astrom O. Anal. Chem. 1982 54 190. Acta 1979 104 279. Received January 31st 1983 Accepted March 2nd I98