SciELO - Scientific Electronic Library Online

 
vol.47 número2PREPARATION OF PALLADIUM THIN FILMS AND THEIR USEFULNESS AS MODIFIED ELECTRODESCOPOLYMERIZATION OF STYRENE BY DIPHENYLZINC-ADDITIVE SYSTEMS: PART III. COPOLYMERIZATION OF STYRENE/para- METHYLSTYRENE USING CpTiCl3-MAO and Ph2Zn-CpTiCl3-MAO INITIATOR SYSTEMS índice de autoresíndice de materiabúsqueda de artículos
Home Pagelista alfabética de revistas  

Boletín de la Sociedad Chilena de Química

versión impresa ISSN 0366-1644

Bol. Soc. Chil. Quím. v.47 n.2 Concepción jun. 2002

http://dx.doi.org/10.4067/S0366-16442002000200008 

Bol. Soc. Chil. Quím., 47, 123-135 (2002) ISSN 0366-1644

 

DETERMINATION OF Cd IN MUSSELS AND NON-FAT MILK
POWDER BY FLOW INJECTION - FLAME ATOMIC ABSORPTION
SPECTROPHOTOMETRY (FI-FAAS) WITH ON-LINE EXTRACTION
BY A CHELATING RESIN

Carlos G. Bruhn*1, Víctor H. Campos1, Víctor P. Díaz1,
Hernán J. Cid2 and Joaquim A. Nóbrega3

1Depto. de Análisis Instrumental, Facultad de Farmacia, Universidad de Concepción, P.O.
Box 237, Concepción, Chile. FAX: 56-41-226382. E-mail: cbruhn@udec.cl
2Centro EULA-CHILE, Universidad de Concepción, P.O. Box 156-C, Concepción, Chile
3Departamento de Química, Universidade Federal de Sao Carlos, Caixa Postal 676, 13565-
905, Sao Carlos, SP, Brazil.
(Received: August 20, 2001 - Accepted: January 21, 2002)

ABSTRACT

A simple low-cost FI-FAAS methodology was developed for determination of Cd traces in food samples using preconcentration by on-line extraction in a chelating resin, a strategy usually applied for analytical measurements in water samples. In the case of food samples, a more complex matrix medium, the pH of acid digested solutions (pH 1) was raised first to pH 3-5 off-line with 0.5 mol/L NH3, and then was on-line adjusted to pH 8.0 with ammonium acetate buffer (1.15 mol/L CH3COOH / 2.0 mol/L NH3) to proper retention of Cd as Cd-8-hydroxiquinoline complex, minimizing interferences. Cadmium was preconcentrated in a minicolumn filled with 8-hydroxiquinoline azo-immobilized on controlled-pore glass (80 mg). A multivariate approach was adopted to establish the main parameters involved in the FI system. Subsequently, the sample, eluent, and buffer flow rates were optimized using an univariate approach. An enrichment factor of 27 was achieved for 3.0 mL of reference solution and the sampling frequency was 63 h-1. The detection limit (3sBL / slope) was 0.7 and 0.4 mg/L for a preconcentration time of 30 and 60 s, respectively. The developed methodology was validated employing certified reference materials (Oyster tissue, Mussel, and Spiked skim milk powder) and the determined and certified values are in agreement (95% confidence level). The methodology was applied for trace levels determination of Cd in lyophilized mussel samples from the Chilean coast and non-fat milk powder, and a comparison made with graphite furnace electrothermal atomization (ETAAS) showed no statistically significant differences in results (95% confidence level).

Keywords: Flow injection, flame atomic absorption spectrophotometry, cadmium, chelating resin, food samples.

RESUMEN

Se desarrolló una metodología simple y de bajo costo por EAALL-IF para la determinación de trazas de Cd en muestras de alimentos empleando preconcentración por extracción en línea con una resina quelante, una estrategia aplicada normalmente en mediciones analíticas realizadas en muestras de agua. En muestras de alimentos, siendo esta una matriz más compleja, el pH de las soluciones de digeridos ácidos (pH 1) fue aumentado primero fuera de línea hasta pH 3­5 con 0.5 mol/L NH3, y luego, en el sistema en línea se ajustó a pH 8.0 con tampón de acetato de amonio (1.15 mol/L CH3COOH / 2.0 mol/L NH3) para una retención apropiada de Cd como complejo Cd-8-hidroxiquinoleina minimizando las interferencias. El cadmio fue preconcentrado en una minicolumna rellena con 8-hidroxiquinoleina azo-inmovilizada sobre perla de vidrio de poro controlado (80 mg). Se adoptó un procedimiento multivariado a fin de establecer los principales parámetros involucrados en el sistema de IF. Posteriormente, se ajustaron los flujos de la solución muestra, el eluente, y el tampón empleando un procedimiento univariado. Se logró un factor de enriquecimiento de 27 para 3.0 mL de solución de referencia y la frecuencia de muestras fue 63 h-1. El límite de detección (3sBL / pendiente) correspondió a 0.7 y 0.4 mg/L para un tiempo de preconcentración de 30 y 60 s, respectivamente. La metodología fue validada empleando materiales de referencia certificados (tejido de ostras, moluscos, y leche en polvo desnatada con recarga de Cd) obteniéndose resultados concordantes con los valores certificados (95% de nivel de confianza). Se aplicó en la determinación de trazas de Cd en muestras de moluscos liofilizados obtenidos en la costa chilena y de leche en polvo descremada, y una comparación efectuada con atomización electrotérmica por horno de grafito (ETAAS) no presentó diferencias estadísticamente significativas en los resultados (95% de nivel de confianza).

Palabras claves: Inyección en flujo, espectrofotometría de absorción atómica por llama, cadmio, resina quelante, muestras de alimentos.

INTRODUCCION

Cadmium is a toxic element present at low concentrations in nature1) and one of the most dangerous trace elements in the food and environment of man2). Its occurrence stems from anthropogenic sources such as mining operations, waste incineration and combustion of fossil fuels, while it occurs naturally in the environment as a result of volcanic emissions3). In humans it is known to accumulate mainly in the kidneys and lungs during lifetime, and the most common sources of exposure to Cd are environmental contamination, diet and smoking. The daily oral minimal risk level is set at 0.2 mg/kg body weight4) and the typical exposure level through the diet and/or smoking is about 3 times lower than the Provisional Tolerable Daily Intake (PTDI) calculated from the Provisional Tolerable Weekly Intake (PTWI) proposed by the FAO/WHO (57 - 72 µg/day)5-7).

The contamination from Cd has increased rapidly in recent years and it is commonly found in aquatic and terrestrial environments. One of the main concerns is due to its long environmental persistence. Therefore, its presence in some coastal waters as contaminant originating from industrial pollution can lead to contamination of seafood. Bivalve mussels are filter-feeders which are used as bioindicator organisms to assess bioavailable contaminant concentrations in coastal waters, and are typical seafood components included regularly in the Chilean diet. Furthermore, non-fat milk powder is another important diet component in which the levels of trace metals are unknown in milk powder sold in the local market. Hence, the determination of Cd in these samples is relevant considering nutritional and environmental aspects.

Atomic absorption spectrophotometry (AAS) with flame (FAAS) or electrothermal atomization (ETAAS) is the predominantly applied method for Cd determination. Although FAAS is less sensitive than ETAAS, its robustness regarding to matrix interferences enhances its utility in particular for routine analysis. Because of the very low concentration of Cd in these food samples, a preliminary preconcentration step is usually necessary before their determination8). As an additional benefit, this step also promotes a removal of interferents.

Flow injection (FI) on-line preconcentration and separation coupled to atomic spectrometric techniques has been shown to be very effective in enhancing the sensitivity and selectivity, and extending the detection limits for trace metals in samples with complex matrices9). In addition, this technique requires low consumption of reagents, sample and time, involves less risk of sample contamination and losses, increased sampling frequency and throughput, can be easily automated, and presents higher efficiency and better reproducibility than batch procedures.

Cadmium has been determined in some biological, food and environmental samples by FI preconcentration / separation techniques involving coprecipitation10), and particularly sorbent extraction11-13) and ion-exchange,1415). These techniques have shown efficiency and effectiveness in enhancing the sensitivity and selectivity in the determination of Cd and Pb in water samples by FAAS12,16). Fang et al.12,17) and Sperling et al.18) determined Cd and Pb in sea-water and drinking water with a preconcentration system based on on-line sorbent extraction. Recently, a FI system with a minicolumn packed with reverse-phase bonded silica with octadecyl groups (C18) was used and evaluated by FAAS in this laboratory to preconcentrate Cd as diethyldithiocarbamate complex in acid-digested hair and blood solutions19). With this system the detection limit was 0.7 µg/L for Cd. However, the enrichment factor obtained was low (e.g., 19) attributed to the relatively large amount of metals present in the sample solutions of both matrices19).

Among the most frequently used packing materials for preconcentration columns are chelating ion-exchangers, such as 8-hydroxiquinoline azo-immobilized on controlled-pore glass (CPG/8-Q)9,20). This material is one of the most often used due to its excellent mechanical properties. Most applications have dealt with trace elements determination in natural and sea waters, probably because its exchange capacity is rather low.

In this work, a FI-FAAS extraction procedure by CPG/8-Q chelating resin was developed and evaluated for Cd determination in acid digests of bivalve mussels and non-fat powdered milk. Besides a peristaltic pump, the FI system was built with simple, low-cost components including a 3-way multiport connector, a manual 5-channel multifunctional valve, a laboratory-made manual commutator and minicolumn, Tygon tubing, polyethylene tubing and a short piece of Tefzel ETFE tubing. It was optimized by a multivariate method based on a factorial experimental design in 2 levels and divided in 3 blocks for establishing the most important parameters21), which were further optimized in univariate mode. The configuration chosen rely on a time-based approach to select the sample volume passing through the column.

EXPERIMENTAL

Apparatus

A Perkin-Elmer (PE) 3110 atomic absorption spectrophotometer equipped with deuterium arc background corrector and standard air - acetylene burner system was used. The spectrometer was connected to a PC computer supplied with the instrument handling software and coupled to an IBM printer. Hollow-cathode lamp (PE) for Cd, operated at the manufacturer's recommended conditions, was used at its primary resonance line (228.8 nm). The acetylene flow rate was 2.0 L/min and the air flow rate was 17.0 L/min to ensure suitable flame conditions. The burner height was adjusted for optimum sensitivity and the nebulizer uptake rate was optimized (6 - 7 mL/min) to provide optimum absorbance signal in conventional sample aspiration. The spectrometer was operated with a time constant of 0.2 s.

A PE Model 1100B atomic absorption spectrometer equipped with deuterium arc background corrector, a PE Model HGA-700 graphite furnace, a PE Model AS-70 autosampler and hollow cathode lamp (PE) for Cd operated at the manufacturer's recommended conditions was used in ETAAS. The spectrometer was operated with a time constant of 0.02 s. Pyrolytic graphite coated tubes (Part N° B3001254) and pyrolytic graphite platforms (Part N° B3001256) were used throughout and background corrected absorbance signals were measured in peak area. The HGA-700 heating conditions (temperature/ ramp time/ hold time/ gas flow) were as follows: dry 1, 90°C/ 5 s/ 5 s/ 300 mL/min; dry 2, 150°C/ 20 s/ 5 s/ 300 mL/min, pyrolysis: 800°C/ 20 s/ 30 s/ 300 mL/min; atomize: 1200°C/ 0 s/ 5 s/ 0 mL/min; clean-out: 2300°C/ 1 s/ 3 s/ 300 mL/min; cool: 20°C/ 1 s/ 10 s/ 300 mL/min. The integration time was 6 s, and 10-µL aliquots were introduced with the autosampler.

The flow injection (FI) system (Figure 1) consisted of an Ismatec MV-MS/CA peristaltic pump (variable speed drive, 1 - 100 rpm) with 4 channels furnished with Tygon pump tubing to propel sample, buffer, wash and eluent solutions into the system. The manifold was mounted on with polyethylene tubing of 0.8 mm i.d. A 3-way multiport connector (Cole Parmer, N° G-06473-02) was used to mix on-line the sample and buffer solutions. One 5-channel multifunctional valve (V) and a manual commutator (C)22) were associated to arrange the flow system. A 5 cm polyethylene tubing (0.8 mm i.d.) was used as linear reactor (R) before the valve manifold (V), followed by a 43 cm polyethylene tubing (0.8 mm i.d.) between the valve V and commutator C. The laboratory-made ion-exchange minicolumn (MC) was a cylindrical column made with polypropylene tubing (3.0 mm i.d.). The optimized minicolumns were packed with 80 mg of CPG/8-Q. The packing was sealed at both ends by small 2 mm thick plugs of white plastic foam followed by Tygon tubing (5 mm x 3.0 mm o.d.) and push-fit connections made with Teflon FEP tubing (45 mm x 0.8 mm i.d.). The chelating ion-exchanger beds were 30 mm long for total volume of solid substrates of 210 µL. To minimize analyte dispersion in the column, the buffered sample and eluent solutions passed through the column in counter-flow. Dispersion was further restricted by using the shortest possible length of 0.3 mm i.d. Tefzel ETFE tubing (4.0 cm) to connect the injection valve to the nebulizer (TT). The subsequent use of a 2 mol/L HNO3 solution was sufficient to recover the active material in the column.

A WTW pH meter (Weilheim, Germany) model pH 522 with an Ingold 405 electrode was used.

Microwave digestions were carried out in Milestone MLS-1200 MEGA microwave system (Bergamo, Italy) with the MDR-300-S/10 TFM rotor, TFM vessels and standard cooling system for MDR rotor. Acid digested samples were subsequently evaporated in a MCR-6-E rotor using a 6 steps evaporation program (36 min).

Reagents, materials and samples

All reagents were at least analytical-reagent grade from Merck (Darmstadt, Germany), except for HNO3 and HCl which were further purified in a quartz sub-boiling still (H. Kürner, Rosenheim, Germany) and stored in quartz bottles until use. Hydrogen peroxide (30% m/m), NH4H2PO4, (NH4)2HPO4, NaNO3, Mg(NO3)2 (hexahydrate) and NH3 (25% m/m) were Suprapur. The 8-hydroxiquinoline/CPG-550 ion exchanger (quinolin-8-ol azo-immobilized on controlled-pore glass) was 177 - 840 µm in particle size, 55-nm pore diameter, with a surface area (typ) of 70 m2/g (Pierce Chemical Co., USA). Ultra pure (u.p.) water (18 MW/cm) was used throughout this work. Reference solutions containing 2.5 - 50 µg/L Cd were prepared in 0.2% v/v HNO3 by appropriate dilution of the stock standard solution (1000 mg/L Cd) immediately prior to use. Certified reference materials (CRM) of bivalve mussels Oyster tissue (SRM-1566a, National Institute of Science and Technology, NIST, USA) and Mussel (GBW-08571, National Research Centre for Certified Reference Materials, NRCCRM, China); non-fat milk powder (SMR-1549, NIST) and spiked skim milk powder (CRM 151, Community Bureau of Reference, BCR, Belgium) were used to validate the analytical procedure. The CRMs were dried according to the manufacturer's instructions before use, and were kept in a dessicator. Extra-pure acetylene and argon (99.998%) were used as the flame and purge gases in FAAS and ETAAS, respectively.

The chemical modifier for the ETAAS determination of Cd in mussel was a solution containing 0.02 mg PO43- + 0.1% v/v HNO3, and for non-fat milk powder samples, 10 µg Mg(NO3) 2 + 15 µg Pd + 0.1% v/v HNO3, added to 10 µL aliquots of acid digested solutions of both sample matrices.

Glass and plastic materials were cleaned as described elsewhere23).

The samples of mussels corresponded to homogenized, freeze-dried, powdered samples prepared from fresh samples of the bivalve mussel "navajuelas chilenas" (Tagelus dombeii) and "almejas" (Semelle sólida) collected in natural banks of the Chilean coast and classified by size24); and the samples of non-fat milk powder were obtained from local markets. All samples were kept in clean, dry containers. The samples were acid digested under pressure in a Milestone MLS-1200 MEGA Microwave system using a MDR-300-S/10 TFM rotor following a five steps program. The sample amounts were 0.5 g of freeze-dried mussel powder and 1.0 g of milk powder. Both samples were digested in 6 mL HNO3 + 2 mL H2O2 following the manufacturer's suggested program of 5 steps (25 min): 2 min- 250 W; 2 min - 0 W; 6 min - 250 W; 5 min - 400 W; 5 min - 650 W, 5 min - vent. The acid digests were subsequently evaporated in a MCR-6-E rotor using an acid scrubber unit and an evaporation program optimized in this laboratory (36 min). The dry residue was carefully dissolved in 7 mL of 1 mol/L HNO3 solution, by placing first the PFA vessel in a double boiler to warm it, and subsequently the solution was transferred quantitatively into a volumetric flask (50 mL) and made up to volume with u.p. water25). The sample solutions (in 1% v/v HNO3) were further diluted before quantification by the standard additions method, and its acidity was lowered prior to introduction in the preconcentration system (see next section). Reagent blank solutions were prepared likewise. For ETAAS measurements, the sample masses used in acid digestion were 0.10 g of mussel and 0.25 g of milk powder. The final solution volume was 25 ml.

On-line preconcentration - elution

The FI manifold used for on-line preconcentration and elution is shown in Figure 1 and was operated in a time-based mode.

In the preconcentration step (Step 1, Fig. 1), the reference solutions containing between 2.5 - 50 µg/L Cd in 0.2% v/v HNO3 or sample solutions were continuously pumped through the manifold at a loading rate of 4.5 mL/min during 30 s for digested mussel samples (60 s for digested milk sample solutions). Previously, the sample solutions were adjusted off-line within pH 3.0 - 5.0 with 0.5 mol/L NH3 to minimize interference effects as will be discussed later. Ammonium acetate buffer solution (1.15 mol/L CH3COOH / 2.0 mol/L NH3) at pH 8 was introduced at a rate of 0.5 mL/min and mixed with the acidified sample at a confluence point before entering the column to adjust pH to 8.0 for proper retention of Cd as Cd-8-Q complex. Cadmium was extracted by chelation on the CPG/8-Q minicolumn ion exchanger and the sample matrix was discarded. During the wash step (Step 2, Fig. 1) the valve V was switched to flow u.p. water (3.6 mL/min) for 20 s through the minicolumn to remove the residual sample matrix. During the wash step, the eluent solution (2.0 mol/L HNO3) was pumped to the spectrometer to record the baseline. In the elution step (Step 3, Fig. 1), the commutator C was switched to flow the eluent solution for 7 s through the minicolumn and release the analyte directly into the nebulizer of the spectrometer. Immediately after switching commutator C, the valve V was switched to fill the reactor with the next sample solution mixed with the buffer. The Read function of the spectrometer was activated simultaneously with the commutator C switching and background corrected absorbance was measured (7 s) in peak height and recorded. Blanks of reference solutions (HNO3 0.2% v/v) and blanks of sample acid digestions (adjusted to pH 3.0 - 5.0 with 0.5 mol/L NH3) were on-line mixed with the appropriate buffer solution and pumped through the minicolumn following the same sequence and time periods to establish the respective blank absorbances prior to standard or sample preconcentration. No further flushing or conditioning of the minicolumn was necessary between one sample preconcentration and the next one, because the buffer solution mixed on line with each sample solution recovered adequately the quelating capacity of the column material. The optimized operating parameters (flow and time) of the preconcentration-wash-elution sequence and the valve (V) and commutator (C ) positions are given in Table I.

Procedure

Preliminary studies were performed to establish the pH for retention of Cd with 8-Q and the eluent type and concentration using a FI system configured according to the previous discussion. Thereafter, the FI system was optimized using a multivariate method, based on a factorial experimental design in 2 levels and divided in 3 blocks for establishing the most important parameters, which were further optimized in univariate mode. In addition, the effect of the sample acidity was studied separately with the optimized FI conditions.

The analytical capability of the FI-FAAS methodology with optimized conditions was assessed by systematic evaluation of the linear working range, the characteristic concentration, the repeatability and reproducibility (i.e. short-term and long-term variation coefficient, % CV), the detection limit (3sblank / slope), the column retention efficiency, the enrichment factor (E.F.)(calculated as the ratio between the slopes of the linear sections of the calibration curves obtained with the eluted analyte after preconcentration from standard solutions and with the same solutions before preconcentration using conventional sample introduction),9) the concentration efficiency (E.F. x samples/min.), the accuracy and precision. The selectivity of the system was studied based on the effects of concomitant cations and anions [Ca2+ (150 mg/L); Mg2+ (50 mg/L); Na+ (100 mg/L); K+ (200 mg/L ); Fe3+ and Zn2+ (10 mg/L); Cu2+ (1 mg/L); SO42-,(100 mg/L S); Cl- (150 mg/L) and PO43- (150 mg/L P)] present as nitrate and ammonium salts, respectively in Cd solutions. The FI on-line preconcentration approach was validated by determination of Cd in CRMs of mussels and milk powder. Recovery studies were also performed in acid blanks, for Cd (100 µg/L) in digested solution of BCR 151 (spiked skim milk powder) 6-fold diluted before quantification by the standard additions method. The methodology was applied to the determination of Cd in 5 samples of Chilean mussels and 1 commercial sample of non-fat milk powder.

RESULTS AND DISCUSSION

Optimization of the FI preconcentration-wash­elution sequence

Preliminary studies were performed on chemical variables in order to establish the effects of the pH of the buffer solution, the eluent composition and concentration on the chelation of Cd. The effect of the buffer pH was investigated over the range 3.0 - 10.0. The results which are presented in Figure 2 show that retention of Cd occurred in basic medium and the optimum pH is 8. It is recognized that 8-hydroxyquinoline (HQ) is amphiprotic containing both weakly basic heterocyclic nitrogen as well as a weakly acidic phenol group. The probable species in relationship to pH are as follows: H2Q+ (acid medium) « HQ (neutral) « Q` (basic medium). According to the dissociation constants of HQ in aqueous solution (pKa1 = 5.02, referred to the pyridine nitrogen; pKa2 = 9.81, referred to the hydroxyl group)26) the retention of Cd is ascribed to the quinolinate form (Q`) which is present at pH 8. An acid strengthening effect due to the azo-immobilization on CPG could lower somewhat the pKa values as was reported previously27) and account for the higher retention at pH 8. Hence, the ammonium acetate buffer of pH 8 was selected for subsequent work. The optimum pH retention for Cd is quite consistent with the stability constant of the Cd-8-Q complex (log K = 7.8)28)

The elution was studied with 20 µg/L Cd by using HNO3 solutions at different concentrations between 0.5 - 2.0 mol/L, and also with an acid mixture containing HCl 1.0 mol/L + HNO3 0.1 mol/L reported in previous work with a silica immobilized HQ.29) The results showed a slight sensitivity increase of 20% between 0.5 and 2.0 mol/L HNO3; the maximum sensitivity was attained at 2.0 mol/L and the acid mixture provided 90% of this level. Since Cd has a tendency to form chloro complexes with chloride ions28) the use of only 2.0 mol/L HNO3 as eluent was preferred.

As mentioned, the FI conditions were optimized by a multivariate method using a factorial experimental design in 2 levels (high and low) and divided in 3 blocks for establishing the relative weight of the most important parameters21). Based on preliminary data, 8 experimental parameters were considered in the multivariate study: the sample flow rate (SF), the eluent flow rate (ELF), the buffer flow rate (BF), the eluent concentration (ELC), the preconcentration time (PT), the minicolumn ion-exchanger filling (MC), the reactor tube length (R) and the transference capillary length (TT). The SF and the PT were changed simultaneously to keep constant the sample volume (2.2 mL).

Figure 1. Flow diagram for Cd determination by FI-FAAS. Step 1: Preconcentration; Step 2: Wash; Step 3: Elution and Prefill. (P = peristaltic pump; CF = confluence point; R = reactor tube; V = 5-channel multifunctional valve; C = manual commutator; MC = minicolumn; TT = transference capillary; W = waste; AAS = atomic absorption spectrometer)

Figure 2. Effect of the pH on the retention of Cd on the CPG/8-Q ion exchanger. Preconcentration - elution conditions: 20 µg/L Cd in HNO3 0.2% v/v; minicolumn filling = 80 mg CPG/8-Q; sample flow rate = 3.8 mL/min; buffer flow rate = 1.5 mL/min (ammonium acetate pH 8); wash flow rate = 3.6 mL/min (u.p. water); eluent flow rate = 5.8 mL/min (1.0 mol/L HNO3); sample loading time = 60 s; wash time = 20 s; integration time = 15 s.

The chemometric approach showed that the most important parameters are eluent, buffer, and sample flow rates. Therefore, these parameters were optimized in univariate mode. The ELC was included in the preliminary study, and the optimum concentration established was in agreement with the results obtained in the multivariate method.

The ELF was studied between 2.0 and 6.0 mL/min and the enrichment factor (E.F.) was the figure of merit selected for assessment of optimum elution rate. As is shown in Figure 3 the highest enrichment factor (E.F.) was obtained at 4.5 mL/min, being inferior at lower rates probably due to inefficient nebulization. Beyond 4.5 mL/min the E.F. was slightly lower suggesting that E.F. reached a steady value at 6 mL/min, as this high rate seems to be not practicable for flow injection systems. Probably, the optimum nebulizer uptake rate for normal aspiration (6 ­ 7 mL/min) limited the nebulization efficiency attained with the upper ELF (6 mL/min) used in this study. Therefore, the ELF chosen was 4.5 mL/min and the approximate volume required for complete recovery of retained Cd was 300 µL (4 s) and no carry-over was observed.

The effect of the acetate BF was studied between 0.35 and 2.0 mL/min (Figure 4) for total sample volumes of 3.0 and 4.5 mL showing a similar trend independent of sample volume. The optimum rate for a more efficient Cd retention was 0.5 mL/min. An increase in BF between 0.75 and 2.0 mL/min showed a slight drop in Cd retention owing both to less contact time of the sample with the ion exchanger and to concomitant sample dilution at this relatively high BF rates. Therefore, the selected BF was 0.5 mL/min throughout this work.

The SF was studied between 0.9 and 6.0 mL/min for a solution containing 60 µg/L Cd, with PTs between 200 and 30 s, respectively to keep constant the sample volume in 3.0 mL, and showed small variations of the analytical signal within the range 1.8 - 4.5 mL/min. Above 4.5 mL/min the analytical signal decreased owing to the short residence time of the sample which results in somewhat lower retention of Cd. The selected sample flow rate was 4.5 mL/min rather than 1.8 mL/min to increase sample throughput. With this SF and with a MC filled with 80 mg ion exchanger the retention efficiency obtained for Cd (20 and 40 µg/L, 30 s preconcentration) was 97.8 ± 0.5%.

A wash step with u.p. water was included between the preconcentration and elution steps, and the selected flow rate and time were 3.6 mL/min and 20 s, respectively to insure no interference effects of residual sample matrix in basic medium on the transfer of retained Cd from the column material to the acid eluate during the elution step.

A study of the capacity of the MC with 80 mg ion exchanger was performed with a reference solution containing 60 µg/L Cd using a SF of 6.0 mL/min, BF of 0.5 mL/min and ELF of 6.0 mL/min. Quantitative recovery of Cd was obtained up to 6.0 mL (PT of 60 s) corresponding to 360 ng of Cd, which is satisfactory for the amount of ion-exchanger employed and for the upper PT used.

The effect of potential interferences occurring in mussel and non-fat milk samples on the determination of Cd was investigated using the optimized FI system. Cations and anions in concentrations according to the highest levels present in the sample matrices in study were added individually to solutions containing 20 and 80 µg/L Cd in 0.2 %v/v HNO3 and the resulting solutions were introduced in the FI system. The results are shown in Figure 5 and are presented as the mean % recovery of the analytical signal of Cd (e.g., the ratio between absorbance in presence of concomitant ion and absorbance in absence of concomitant ion x 100). The worst interference effects yield low Cd recoveries and were caused by Zn2+ (22% in 10 mg/L), Cu2+ (57% in 1 mg/L) and Fe3+ (56% in 10 mg/L) and somewhat less by Ca2+ (73% in 150 mg/L). The lower recoveries are thus probably caused by breakthrough of the analyte in the presence of the larger concentrations of the competing ions which is consistent with the stronger complexes formed with HQ by Fe3+ (log K1 = 13.7)30), Cu2+ (log K1 = 12.2)17) and Zn2+ (log K1 = 8.56)17) compared to Cd2+ (log K1 = 7.8)28). By increasing the preconcentration time (and thus the sample volume) no signs of breakthrough became apparent when the same Cd solutions were used in the absence of the concomitant ions. Though, these interferences were strongly minimized when the pH of the standard solutions containing these competing ions was raised and adjusted off-line between 3.0 and 5.0 with 0.5 mol/L NH3. Probably, this increase in pH (between 3 ­ 5) turned out to enable the formation of hydroxo- and ammonia- complexes of these competing cations in the ammonium acetate buffer17) during the mixing of the sample and buffer solutions to adjust pH to 8.0 before entering the column, and prevented their chelation in the ion-exchanger. For the other tested ions the method is relatively free from interferences, since signal variations larger than ±10% were not observed.

Figure 3. Effect of the eluent flow rate on the enrichment factor of Cd in FI-FAAS. Preconcentration - elution conditions: 0, 5, 10 and 20 µg/L Cd solutions in HNO3 0.2% v/v; minicolumn filling = 80 mg CPG/8-Q; sample flow rate = 4.5 mL/min; buffer flow rate = 0.5 mL/min (ammonium acetate pH 8); wash flow rate = 3.6 mL/min (u.p. water); eluent = 2.0 mol/L HNO3; sample loading time = 30 s; wash time = 20 s; integration time = 10 s.
Figure 4. Effect of buffer (pH 8.0) flow rate on Cd absorbance in FI-FAAS. Preconcentration - elution conditions: 60 µg/L Cd in HNO3 0.2% v/v; minicolumn filling = 80 mg CPG/8-Q; sample flow rate = 4.5 mL/min; wash flow rate = 3.6 mL/min (u.p. water); eluent = 2.0 mol/L HNO3 ; eluent flow rate = 6.0 mL/min (2.0 mol/L HNO3); sample loading time = variable; wash time = 20 s; elution and integration time = 7 s.

Figure 5. Interference effects of concomitant ions on Cd in FI-FAAS. Preconcentration - elution conditions: 20 and 80 µg/L Cd in HNO3 0.2% v/v; minicolumn filling = 80 mg CPG/8-Q; sample flow rate = 4.5 mL/min; buffer flow rate = 0.5 mL/min (ammonium acetate pH 8); wash flow rate = 3.6 mL/min (u.p. water); eluent flow rate = 4.5 mL/min (2.0 mol/L HNO3); sample loading time = 30 s; wash time = 20 s; elution and integration time = 7 s.

Analytical figures of merits and application

The analytical curve was run under the optimum chemical and flow conditions using the manifold shown in Figure 1. The linear working range was between 2.5 and 50 µg/L (sample volume 2.2 mL) and the correlation coefficient was 0.9983. The slope was 0.0046 and the intercept was ­0.0044. The analytical figures of merit are indicated in Table 2. The instrumental detection limit was 0.7 µg/L for 30 s of preconcentration time (2.2 mL sample volume). The detection limit was calculated as 3sBL / slope (n = 15 runs of blanks). These results are quite satisfactory considering the relatively short preconcentration time employed. The reproducibility was 5.0% (CV%) for Cd and could be improved if the system is automated. Sample volumes larger than 3.0 mL were used for determination of Cd in acid digested solutions of milk powder because the analyte concentration was lower than the limit of quantification. In this case, the sampling frequency was decreased to 41 h-1. The method detection limit, calculated for 3s (n = 15 analytical blanks) was 0.5 µg/L Cd (60 s preconcentration, sample volume 4.5 mL) in mussel and milk powder, corresponding to 0.05 and 0.03 µg/g, respectively.

The analytical procedure was validated by determination of Cd in CRMs of mussels and non-fat milk powder as shown in Table III. A recovery study was performed by adding 100 µg/L Cd to milk powder BCR 151 following the sample digestion procedure described above. The accuracy was < 3% in mussels and 8.9% in milk powder, expressed as the mean relative percent error (Erel%). The mean precision (RSD%) was 5% in mussel and 9% in milk powder. The percent recoveries were 97.5 ± 2.8% (n=3; 94.7-100.3%) for Cd in milk powder. These results obtained by the standard additions method are quite satisfactory considering the dilution factor applied to the sample solutions, and the interferences from Zn (content 830 µg/g in oyster tissue, 138 µg/g in mussel), Fe (content 539 µg/g in oyster tissue, 221 µg/g in mussel), Cu (66.3 µg/g in oyster tissue, 7.7 µg/g in mussel) and P (0.623% in oyster tissue, 1.35% in mussel). The off-line pH adjustment of sample solutions was effective to partial correction of interferences but for complex samples as those employed here the use of the standard additions method is still necessary to attain proper accuracy.

The methodology was applied to the determination of Cd in 5 samples of lyophilized Chilean mussels: 2 "almejas" and 3 "navajuelas chilenas" and in 1 sample of non-fat milk powder sold in local market (Table IV). Cadmium was also determined by ETAAS for comparison. No statistically significant differences were established by Student t-test between the results obtained in mussels by both analytical methods at a 95% confidence level. The mean Erel% for Cd in five mussel samples was 12.9% (1.7 - 28%) and the mean RSD% was 9.9% (6.9 - 14%). In the sample of non-fat milk powder, Cd content was under the method quantification level. The results obtained for Cd in samples of "almeja 2" and "navajuelas 2" were somewhat lower than those obtained by ETAAS. However, to the other 3 samples, Cd results were quite comparable. In these 3 mussel samples, the mean Erel% in comparison with ETAAS results was 4.9% (1.7 ­ 6.7) and the mean imprecision was 10.5% indicating that these results are within the imprecision obtained by FI-FAAS.

The proposed methodology could be easily implemented at low-cost by modest analytical laboratories with basic FAAS equipment which have no possibility to buy a graphite furnace ETAAS unit for trace level determination of Cd. Besides the choice of applying this methodology in samples with relatively simple matrix (e.g., tap water, groundwater), it is suitable too in more complex matrices as shown in this work.

CONCLUSIONS

The simple, low-cost developed FI on-line preconcentration procedure with FAAS can be applied reliably to the determination of Cd in lyophilized mussel samples with Cd levels > 0.18 µg/g, for 0.5 g of sample dissolved in 50 mL, and using a 60-s preconcentration time. In non-fat milk powder due to the relatively low levels present in this matrix, this procedure was not completely adequate. However, if the sample volume available is sufficient and the sampling frequency is not critical the detection limit can be again improved increasing the preconcentration time. The main aspect of the developed FI-FAAS procedure is the possibility of trace level determination of Cd in complex matrix samples without severe interferences.

ACKNOWLEDGEMENTS

The authors are grateful to the Fondo Nacional de Desarrollo Científico y Tecnológico (FONDECYT, Research Grant No. 1960664); to the European Commission (Contract CI1-CT94-014); and to Dirección de Investigación of the Universidad de Concepción, Concepción, Chile (DIUC Research Grant No. 95.71.01-4) for financial support; to the Facultad de Farmacia, Universidad de Concepción; and to the Centro EULA-CHILE of the Universidad de Concepción for allowing the use of a PE-1100B atomic absorption spectrometer with HGA-700 graphite furnace and AS-70 autosampler. The authors thank Professor Dr. Horacio A. Mottola of the Oklahoma State University, USA, for the gift of the CPG/8-Q exchanger.

REFERENCES

1. B.J. Aylett, Cadmium in Comprehensive Inorganic Chemistry, A.F. Trotman-Dickenson (Ed.), Pergamon Press, Oxford, 1973, p 254.         [ Links ]

2. M. Stoeppler, Hazardous Metals in the Environment. Techniques and Instrumentation in Analytical Chemistry, Vol. 12, Elsevier, Amsterdam, 1992, p 177.         [ Links ]

3. D.W. Fassett, in Metals in the Environment, H.A. Waldrom (Ed.), Academic Press, New York, 1980, pp 61-110.         [ Links ]

4. U.S. Department of Labor, Occupational Safety & Health Administration, 1910.1027 AppE-Cadmium in Workplace Atmospheres, Subpart Z, Method Number ID-189 (U.S. Department of Labor, Washington, D.C.)

5. M.J. Gartrell, J.C. Craun, D.S. Podrebarac, E.L. Gunderson, J. Asoc. Off. Anal. Chem., 69, 146 (1986).         [ Links ]

6. E.L. Gunderson, J. Asoc. Off. Anal. Chem., 71, 1200 (1988).         [ Links ]

7. Evaluation of Certain Food Additives and the Contaminants Mercury, Lead and Cadmium - 16th Report of the Joint FAO/WHO Expert Committee on Food Additives (1972) World Health Organization, Geneva.

8. M.C. Yebra-Biurrun, A. Bermejo-Barrera, and M.P. Bermejo-Barrera, Analyst, 116, 1033 (1991).         [ Links ]

9. Z. Fang, Flow Injection Separation and Preconcentration, VCH, Weinheim, 1993.         [ Links ]

10. B. Welz, Sh. Xu, M. Sperling, Appl. Spectrosc., 1433, 45 (1991).         [ Links ]

11. J. Ruzicka, A. Arndal, Anal. Chim. Acta, 216, 243 (1989).         [ Links ]

12. Z. Fang, M. Sperling, B. Welz, J. Anal. At. Spectrom., 5, 639 (1990).         [ Links ]

13. R. Ma, W. Van Mol, F. Adams, Anal. Chim. Acta, 293, 251 (1994).         [ Links ]

14. Z. Fang, B. Welz, J. Anal. At. Spectrom., 4, 543 (1989).         [ Links ]

15. M.F. Enríquez-Domínguez, M.C. Yebra-Biurrun, M.P. Bermejo-Barrera, Analyst, 123, 105 (1998).         [ Links ]

16. V. Porta, E. Abollino, E. Mentasti, C. Sarzanini, J. Anal. At. Spectrom., 6, 119 (1991).         [ Links ]

17. Z. Fang, J. Ruzicka, E.H. Hansen, Anal. Chim. Acta, 164, 23 (1984).         [ Links ]

18. M. Sperling, X. Yin, B. Welz, J. Anal. At. Spectrom., 6, 295 (1991).         [ Links ]

19. C.G. Bruhn, C. Vilches, H.J. Cid, Bol. Soc. Chil. Quím., 44, 321 (1999).         [ Links ]

20. Z. Fang, Flow Injection Atomic Absorption Spectrometry, Wiley, Chichester, 1995.         [ Links ]

21. B. Barros, Y. Spacino, R. Edward, Planejamiento e optimizacao de experimentos, Universidade Estadual de Campinas, SP, Brasil (1996).         [ Links ]

22. F.J. Krug, H. Bergamin F., E.A.G. Zagatto, Anal. Chim. Acta, 179, 103 (1986).         [ Links ]

23. C.G. Bruhn, F.E. Ambiado, H.J. Cid, R. Woerner, J. Tapia and R. García, Anal. Chim. Acta, 306, 183 (1995).         [ Links ]

24. I. De Gregori, D. Delgado, H. Pinochet, N. Gras, M. Thieck, L. Muñoz, C.G. Bruhn, G. Navarrete, Sci. Tot. Environ., 111, 201 (1992).         [ Links ]

25. C.G. Bruhn, J.Y. Neira, G.D. Valenzuela and J. Nobrega, J. Anal. At. Spectrom., 13, 29 (1998).         [ Links ]

26. R.C. West (Ed.), CRC Handbook of Chemistry and Physics, 61st Ed., CRC Press, Inc., Boca Raton, FL, USA (1980).         [ Links ]

27. M.A. Marshall and H.A. Mottola, Anal. Chem., 55, 2089 (1983).         [ Links ]

28. O. Abollino, E. Mentasti, V. Porta and C. Sarzanini, Anal. Chem., 62, 21 (1990).         [ Links ]

29. S. Nakashima, R.E. Sturgeon, S.N. Willie and S.S. Berman, Fres. Z. Anal. Chem., 330, 592 (1988).         [ Links ]

30. J. Malamas, M. Bengtsson and G. Johansson, Anal. Chim. Acta, 160, 1 (1984).         [ Links ]