Fast Environmental Impact Assessment Through ICP-MS : Application to Bivalves From a Tropical Estuary *

The use of semi-quantitative analysis in environmental impact assessment studies was evaluated through a comparative study using quantitative and semi-quantitative operational modes in ICP-MS. Twenty-one isotopes, namely, 7 Li, 11 B, 27 Al, 48 Ti, 51 V, 52 Cr, 55 Mn, 58 Ni, 59 Co, 63 Cu, 64 Zn, 69 Ga, 88 Sr, 90 Zr, 93 Nb, 98 Mo, 114 Cd, 181 Ta, 137 Ba, 205 Tl, and 208 Pb were analyzed in both methods. Sample digestion was performed in closed microwave Teflon vessels using nitric acid and hydrogen peroxide. The semi-quantitative analyses were performed using Rh as an internal standard; a solution containing Be, Ge, In, and Re was used to calibrate the instrument. Accuracy studies for CRM samples using the semi-quantitative mode demonstrated that all the elements considered were within the certified range, except for Cu and Pb, which gave higher values. In order to verify the applicability of the semi-quantitative method to environmental assessment studies, mollusk samples from a tropical estuary (Pina Bay, Pemambuco, Brazil) were analyzed. The results show that some species concentrate some elements relative to others, probably as a consequence of the feeding habits of each species. Even though there is no specific legislation regarding metal concentration in seafood in Brazil, the results show that metal concentrations do not exceed international limits, except for V, which exceeded the EPA risk level. Pina Bay is highly impacted by sewage discharges, but the metal concentrations in the mollusk populations do not seem to pose a threat for human consumption. The results also suggest that the semi-quantitative method can be used as a screening method in environmental impact assessment studies.


A tomic S pectroscopy
Atomic Spectroscopy serves as a medium for the dissemination of general information together with new applications and analytical data in atomic absorption spectrometry.
The pages of Atomic Spectroscopy are open to all workers in the field of atomic spectroscopy.There is no charge for publication of a manuscript.
The journal has around 3000 subscribers on a worldwide basis, and its success can be attributed to the excellent contributions of its authors as well as the technical guidance of its reviewers and the Technical Editors.
A manuscript can be submitted to the editor by mail or e-mail in the following manner: 1. Triplicate, double-spaced.2. Include abstract for articles.3. Number the references in the order they are cited in the text.4. Include bibliography.

Submit original drawings
or glossy photographs and figure captions.6. Consult a current copy of Atomic Spectroscopy for format.
7. E-mail text in .docfile and graphics in .tiffiles to the editor: anneliese.lust@perkinelmer.comAll manuscripts are sent to two reviewers.If there is disagreement, a third reviewer is consulted.
Minor changes in style are made in-house and submitted to the author for approval.
A copyright transfer form is sent to the author for signature.
If a revision of the manuscript is required before publication can be considered, the paper is returned to the author(s) with the reviewers' comments.
In the interest of speed of publication, a typeset copy is faxed to the corresponding author for final approval.
After publication, the senior author will receive 50 complimentary reprints of the article.
Additional reprints can be purchased, but the request must be made at the time the manuscript is approved for publication.

INTRODUCTION
The demand for chemical analysis in environmental studies in a great variety of matrices is increasing exponentially.The use of semiquantitative analysis using inductively coupled plasma mass spectrometry (ICP-MS) in environmental assessment studies is very attractive because it is faster and less expensive than the corresponding method of quantitative analysis.Another advantage of semi-quantitative analysis is the concomitant determination of a large number of elements (up to 80) with a precision and accuracy better than 30% for most elements of environmental interest when appropriate calibration procedures are performed.This type of screening technique would indicate areas where a more detailed and specific study should be undertaken, therefore saving time and money.
To improve the analytical accuracy using the semi-quantitative mode, the pre-calibrated internal response table can be updated using only a few selected elements (1).Even though this procedure does not account for variations due to matrix effects, the use of an internal standard can minimize such effects and compensate for any drift which eventually occurs during the analysis.
In this work we compare the quantitative and semi-quantitative analysis using ICP-MS in order to verify the applicability of semiquantitative multi-elemental analysis to environmental impact assessment studies; thus reducing data acquisition time and analysis cost.We also used a modified scheme of the microwave-assisted digestion method described by El Azouzi and co-workers (2).This methodology was applied to four different mollusk species from a Brazilian tropical estuary (Pina Bay, Pernambuco) in order to verify its applicability in environmental studies.Determination of the metal content in different mollusk species with diverse feeding habits can provide an overall picture of the environment they live in.The samples were collected in March, June, and September 1999 in order to look for any change due to seasonal (dry and rainy season) variations.

EXPERIMENTAL Instrumentation
The analyses were performed using an ELAN ® 5000 ICP-MS (PerkinElmer SCIEX, Concord, Ontario, Canada), equipped with a Ryton ® Scott-type spray chamber and GemTip™ cross-flow nebulizer (PerkinElmer Instruments, Shelton, CT USA).The solutions were aspirated using a peristaltic pump.Sample introduction conditions were: plasma argon flow rate, 12 L/min; auxiliary argon flow rate 12 L/min; and nebulizer flow rate, 0.8 L/min.The quantitative analyses were performed using multi-elemental calibration curves and the semi-quantitative analyses were done using PerkinElmer TotalQuant™ II software, which uses pre-programmed corrections for common isobaric interferences.Before analysis, the ICP-MS was optimized using a solution containing Mg, Rh, and Pb at a concentration of 100 µg/mL each.The parameter conditions for the TotalQuant II analysis were those recommended by Soldevilla and coworkers (2): dwell time of 250 ms, 2 replicates, 1 reading per replicate and 1 sweep per reading, which

ABSTRACT
The use of semi-quantitative analysis in environmental impact assessment studies was evaluated through a comparative study using quantitative and semi-quantitative operational modes in ICP-MS.Twenty-one isotopes, namely, 7 Li, 11 B, 27 Al, 48 Ti, 51 V, 52 Cr, 55 Mn, 58 Ni, 59 Co, 63 Cu, 64 Zn, 69 Ga, 88 Sr, 90 Zr, 93 Nb, 98 Mo, 114 Cd, 181 Ta, 137 Ba, 205 Tl, and 208 Pb were analyzed in both methods.Sample digestion was performed in closed microwave Teflon vessels using nitric acid and hydrogen peroxide.The semi-quantitative analyses were performed using Rh as an internal standard; a solution containing Be, Ge, In, and Re was used to calibrate the instrument.Accuracy studies for CRM samples using the semi-quantitative mode demonstrated that all the elements considered were within the certified range, except for Cu and Pb, which gave higher values.In order to verify the applicability of the semi-quantitative method to environmental assessment studies, mollusk samples from a tropical estuary (Pina Bay, Pernambuco, Brazil) were analyzed.The results show that some species concentrate some elements relative to others, probably as a consequence of the feeding habits of each species.Even though there is no specific legislation regarding metal concentration in seafood in Brazil, the results show that metal concentrations do not exceed international limits, except for V, which exceeded the EPA risk level.Pina Bay is highly impacted by sewage discharges, but the metal concentrations in the mollusk populations do not seem to pose a threat for human consumption.The results also suggest that the semiquantitative method can be used as a screening method in environmental impact assessment studies.
gives a total analysis time of 103 seconds.To increase the accuracy, the pre-calibrated intensities per concentration unit were updated using spiked samples containing Be, Ge, In, and Re.Rh was used as an internal standard.

Reagents
The solutions were prepared using Nanopure water, with a resistivity of 18.2 MΩ, obtained from a Milli-Q™ water purification system.Nitric acid (65%) and hydrogen peroxide (30%) were of analytical grade.Single-and multi-element solutions were used to calibrate the instruments and to prepare the calibration standards.Quantitative analysis calibration curves were prepared using solutions containing 1000 µg mL -1 of Al, B, Ba, Cd, Co, Cr, Ga, In, Li, Mn, Ni, Pb, Rh, Sr, Tl, and Zn, and 100 µg mL -1 of Ti, Zr, V, Nb, Ta, and Mo.

Sample Collection, Preservation, and Preparation
Samples of four different species, Anomalocardia brasiliana (mussel), Protothaca pectorina (mussel), Tagelus plebeius (mussel), and Crassostrea rhizophorae (oyster) were carefully hand-picked in order to avoid contamination.The main characteristics of the water body during the sampling periods were: mean surface water temperature 29.1ºC, the dissolved oxygen concentration 2.3 mL L -1 O 2 , the oxygen saturation 25.8%, pH 7.61, Biological Oxygen Demand (BDO) 1.63 mg kg -1 and Secchi disc water transparency 0.57 m.The individual samples were washed with local water, packed in plastic bags, and kept in ice until arrival at the laboratory where they were frozen at -18ºC.The sampling took place in March, June, and September 1999.Sampling sites were selected according to the species distribution in the bay (Figure 1).However, local conditions did not permit a homogeneous sampling distribution throughout the basin.
As soon as possible, the samples were defrosted and after valve removal, the soft tissue was ovendried at 60ºC.Each analyzed sample consisted of a composite sample of 30 individual samples.The dried samples were ground and homogenized in a glass mortar and a sample aliquot of 500 mg was microwave digested in 120-mL Teflon ® vessels using 3 mL of concentrated HNO 3 (65%) and 2 mL of H 2 O 2 (30%) at 650 W power.The samples were heated for two periods of 3 minutes each with a cooling interval of at least 5 minutes.After complete cooling, the digested samples were transferred to a 50-mL volumetric flask.Prior to analyses, Rh was added to the samples for a final concentration of 50 ppb, and the samples were brought to volume with deionized H 2 O. Certified Reference Material (CRM) MA-MEDPOL-1 (Fish homogenate) and blanks were prepared the same way as the composite sample in order to verify procedure accuracy.Moisture content of lyophilized MA-MEDPOL-1 was found to be 9.4%.For the Pina bay mollusks, the moisture content in dry samples varied from 10.4% (oyster) to 16.6% (mussel).

Analysis
The samples previously digested were analyzed quantitatively and semi-quantitatively using ICP-MS.The quantitative analyses were obtained using external calibration.Weighted linear regression and internal standards were used.Two sets of standards were used for calibration and 21 elements plus In and Rh were determined.Semi-quantitative analysis was obtained using the TotalQuant II software.A solution containing Be, Ge, In, and Re was used to update the software response table (2,3).Rh was used as an internal standard.Even though the semi-quantitative method could determine up to 80 elements, only 21 elements were selected for comparison between the two methods.The isotopes measured were 7 Li, 11 B, 27 Al, 48 Ti, 51 V, 52 Cr, 55 Mn, 58 Ni, 59 Co, 63 Cu, 64 Zn, 69 Ga, 88 Sr, 90 Zr, 93 Nb, 98 Mo, 114 Cd, 181 Ta, 137 Ba, 205 Tl, and 208 Pb.Isobaric interference corrections on Ti (Ca), Ni (Fe), Zn (Ni), Mo (Ru), and Cd (Sn) were automatically made by the software.2 show the analytical results using semi-quantitative (TotalQuant II) and conventional quantitative mode analyses for the CRM MA-MEDPOL-1 (fish homogenate).Accuracy figures for CRM elements, except for Cu, are within the certified range in both methods and for Pb in the semiquantitative mode.These elements give values higher than the upper certified limit.The majority of elements determined are within a ±30% relative error between both methods.Elements in the concentration range 1-10x10 -3 mg kg -1 (d.w.), which in solution would be close to the method detection limits, gave relative errors of higher than 30%.Ti was the only element with concentrations higher than 1 mg kg -1 resulting in relative errors of higher than 30%.Figure 2 presents a plot of quantitative mode (x) against semi-quantitative (y) analyses for CRM MA-MEDPOL-1.The data show good statistical parameters.The data provide a regression line, y = 0.9256x -135.22,and a correlation coefficient of 0.978.

Table I and Figure
Considering these results, the semi-quantitative (TotalQuant II) and quantitative mode analyses were applied to mollusks of the Pina Bay area as a screening method for environmental assessment studies.This type of comparison has been done successfully in different types of matrices (2,3).
Table II shows the results obtained for the different mollusk species.The results are expressed on a dry weight basis (500 mg to a 50-mL final volume).Al, B, Cu, Fe, Mn, Sr, Ti, and Zn are expressed in mg kg -1 , whereas the other elements are expressed in µg kg -l .Accuracy relative errors between methods averaged ±30% (Table II). Figure 3 presents a plot of quantitative (x) against semi-quantitative (y) contents found for the mollusks in the Pina Bay area.These data are pre-    Q=Quantitative analysis; SQ=Semi-quantitative analysis; RD(%)= Relative difference between quantitative and semi-quantitative analysis.
sent at two different concentration ranges in order to give more realistic statistical parameters.A linear regression line through all the data is y = 1.365x -10.387, with a correlation coefficient of 0.9878 (Figure 3a).These data are influenced by the high Zn concentrations in C. rhizoporae.Linear regression using the data in two separate sets, according to the concentration level, gives more realistic statistical data.Data with concentrations higher than 5 mg kg -1 (Figure 3b) give a regression line, y = 1.308 + 0.967x, with a correlation coefficient of 0.955.A regression line through the data in the concentration range below 5 mg kg -l is y = 1.228x + 0.17 with a correlation coefficient of 0.96 (Figure 3c).Al, Ba, Co, Cr, Cu, Fe, Mn, Mo, Sr, V, and Zn, with a concentration in the range of 10 -1 -10 2 mg kg -1 , have relative errors better than ±30% (Figure 3a).Li (6-12x10 2 µg kg -1 ) was the only element that gave consistent differences in the 100% range (Figure 3c).Ga, Nb, Ta, Tl, and Zr, with a concentration range of 1-10 2 µg kg -1 gave very erratic relative errors.However, for the majority of elements of environmental interest the repeatability between methods was better than 30%.
Cadmium was below the method detection limit in most samples, and when detected, the results in the semi-quantitative mode were usually higher than the quantitative results.
Even though the samples were collected in distinct seasons (dry and rainy), it was not possible to detect a clear seasonal variation in metal assimilation for each species.March was the last month of the dry season (summer) and September the end of the rainy season (winter).Mn is the only element in the concentration range of 10 mg kg -1 -10 3 mg kg -1 which shows a significant increase when collected during the rainy season (see Figure 4).
As expected, some bivalve species showed preferential assimi-lation towards specific metals (Figure 4).We can divide the metals into two main groups according to their content in the mollusks: elements with concentration between 10 mg kg -1 -10 3 mg kg -1 and those in the concentration range between 10-10 3 µg L -1 .In the first group, elements in the highest concentration were Zn, Al, Fe, Mn, and Sr.The following considerations will be based on the results of semiquantitative analysis.
Tagelus plebeius showed preferential accumulation of Fe and Al in relation to the other species, with concentrations up to 316 mg kg -1 and 437 mg kg -1 and median values of 277 mg kg -1 and 375 mg kg -1 , respectively.The same phenomenom regarding Fe preferred relative accumulation by T. plebeius was observed by Souza and co-workers (4) with basically the same median value (276.4 mg kg -1 ).
The results of Zn concentrations in C. rhizophorae also suggest specific accumulation.Zinc is the trace element with the highest concentration in the mollusks studied.The overall median for the whole period in C. rhizophorae was 2.139 mg kg -1 , with a maximum value of 2.275 mg kg -1 during March.Souza and co-workers (4) detected a 10-fold increase in the Zn content in oysters collected in September.In this study, we did not observe this increase.

Anomalocardia brasiliana had
Mn concentrations almost 10-fold that of the other species.For this metal, the median concentration found was 63 mg kg -1 , with a maximum value of 94 mg kg -1 (Figure 4b).Within the first group, the metals less assimilated by the four species were Ba, B, Ti, and Cu.This may be the result of metabolism for these metals or that they were not bioavailable in the environment.However, it was observed within the group that Cu was more concentrated in C. rhizophorae and Ti in T. plebeius (Figure 4c), reaching up to 19.6 mg kg -1 and 15 mg kg -1 , with median concentrations of 17.9 mg kg -1 and 12 mg kg -1 , respectively.The second group includes Zr, Ta, Nb, Ga, Cd, Cr, Co, Li, V, Ni, and Mo in the concentration range 10-10 3 µg L -1 .From all metals determined, Ta presents the lowest concentration, with a maximum content of 20 µg L -1 in the Protothaca pectorina.Tagelus plebeius and Protothaca pectorina show the highest concentration in Ga, 326 µg L -1 and 294 µg L -1 , respectively, compared to the others.Tagelus plebeius presents a median Mo content, which is three times the median value of the other species (1521 µg L -1 and 513 µg L -1 , respectively) (Figure 5).All other elements are within the same concentration range in the species studied.
In general, the concentrations found in the soft tissues of bivalves at Pina Bay in the year of 1999 compare well with the literature (5,6,7).Even though V is above the U.S. EPA risk level (8), it is below the results reported in the literature for mussels (7,9).C. rhizophorae shows Zn values higher than those found for mussels in the literature (5,6).However, it compares well with the data reported for the same species in other areas of the Brazilian Northeast (10).Wallner-Kesarnach and co-workers (10) found the Zn and Cu content in C. rhizophorae in the Cotegipe Channel, Todos os Santos Bay (Bahia), which averaged Zn 3000 mg/kg (d.w.) and 1600 mg/kg (d.w.).The values are 1.5 and 10 times higher than the median value found for Zn and Cu, respectively, in the Pina Bay area.
In other areas, highly impacted by sewage discharges, such as in the Pina Bay, differences in metal content in the soft tissues of marine bivalves were found to be sporadic and did not seem to necessarily correlate to the presence of sewage (11).These variations can also be linked to storms and intense urban runoff conditions (12).The lack of cyclical patterns in the data varia-tions suggests that these are not a function of the biology of the test organisms in response to changes in water quality due to seasonal changes (12).

CONCLUSION
The metal content in the studied bivalve populations of the Pina Bay area is probably a consequence of its general degradation scenario.Important factors include water quality, total suspended solids, and bottom sediment quality which directly influence the quality of the feeding habitat of the species studied.
Even though the Pina Bay estuary is a highly impacted area as a consequence of the sudden and unplanned increase in human population, in addition to other types of anthropogenic activity, the metal content in the species studied does not exceed the limits approved for human consumption.The only exception is V, which exceeded the U.S. EPA risk level (8).
The results obtained in the present study also show the applicability of semi-quantitative analyses (TotalQuant II) in environmental impact assessment studies as a reliable alternative to quantitative analysis by the use of appropriate calibration.They also show that the practical quantification limit, considering a 30% relative variation compared to quantitative analysis, is of the order of 1-2 µg L -1 in the semi-quantitative mode.However, this limitation does not preclude the use of TotalQuant II as a screening method because the risk level for most metals of environmental interest is higher than this limit.

INTRODUCTION
The W-Rh permanent chemical modifier has been successfully employed in electrothermal atomic absorption spectrometry (ETAAS) for the determination of Cd and Pb in aqueous solutions (1), biological materials and sediment digests (2,3), biological materials and sediment slurries (4).Several advantages over the conventional modifiers were reported (1-4) such as less time required for sample dispensing, leading to simpler and faster heating programs for ETAAS; lowering modifier blanks due to the elimination of volatile impurities during graphite treatment resulting in better detection limits; longer signal term stability, reducing the number of recalibrations during routine analysis; remarkable improvement in tube lifetime (1-4); and lowering analytical costs (2).
Although permanent chemical modifiers provide attractive advantages, they have scarcely been applied to the analysis of real samples in ETAAS for the successful determination (1-6) of elements such as Cd, Pb, Ag, Sn, As, Bi, and Sb.However, the determination of Cd and Pb was thoroughly investigated using permanent modifiers for the analysis of different sample matrices employing sample digests or slurry sampling ETAAS (1)(2)(3)(4).In this context, a more detailed investigation of permanent chemical modifiers is required for the ETAAS analysis of real samples.
Permanent modifiers applied in slurry analysis could simplify the already existing methods.Instead of adding the chemical modifier to the platform at each firing, which requires high-purity chemical modifier reagents in order to avoid elevated blank values, the graphite surface can be treated with a carbide-forming element followed by a noble metal.This process eliminates volatile impurities in the modifier solution during its pretreatment cycle, resulting in a decrease in the detection limits (2,4).
The aim of the present work was to study the application of the W-Rh permanent modifier for the determination of Cu in digested sediment samples as well as slurries.This work broadens the application of W-Rh permanent modifier for other elements and also shows that the performance attained is similar to the most-employed Pd + Mg(NO 3 ) 2 conventional modifier (22).The main goals of employing a permanent modifier in the analysis are improvement of the already existing methods by diminishing measurement time, decreasing detection limits, extending tube lifetime which leads to a decrease ABSTRACT A tungsten-rhodium (W-Rh) coating on the integrated platform of a transversely heated graphite atomizer was used as the permanent chemical modifier for the determination of Cu in sediments by using digested and slurry samples for electrothermal atomic absorption spectrometry (ETAAS) analysis.The W-Rh permanent modifier was as efficient as the Pd + Mg(NO 3 ) 2 conventional modifier for obtaining good Cu thermal stabilization in the digested and slurry samples.The W-Rh permanent modifier remained stable for approximately 300 or 250 firings when 20 µL of digested sample or 20 µL of slurry was delivered into the atomizer, respectively.In addition, the permanent modifier increased the tube lifetime up to 1360 and 730 analytical measurements for the digested and slurry samples, respectively.With the W-Rh permanent modifier, there was less variation in the slope of the analytical curves during the total atomizer lifetime, which decreases the need for recalibration during routine analysis, thus increasing sample throughput and consequently diminishing analytical cost.
The atomizer lifetime was limited to the THGA wall durability, due to a non-efficient protection provided by the sheath gas, but the W-Rh treated platform was found to be intact after more than 1360 and 730 analytical firings for the digested samples and slurries, respectively.
The results for the determination of Cu in the samples were in agreement with those obtained with decomposed sample solutions by using Pd + Mg(NO 3 ), since no statistical differences were found after applying the paired t-test at the 95% confidence level.

EXPERIMENTAL Instrumentation
A PerkinElmer Model 4100ZL atomic absorption spectrometer (PerkinElmer Instruments, Shelton, CT USA) with a longitudinal Zeeman-effect background correction system furnished with a transversely heated graphite atomizer (THGA™) and an AS-71 autosampler were used together with a USS-100 controller for the Vibracell VC 50 ultrasonic processor with a titanium probe (Sonics and Materials, Danbury, CT USA).The standard PerkinElmer THGA (Part No. B050-4033) with integrated platforms was used either without previous treatment (referred to as the pyrolytic carbon platform) or after pre-treatment first with W and then with Rh (referred to as the W-Rh treated platform) (1).
All measurements were based on integrated absorbance and performed at 327.4 nm (slit 0.7 nm) by using a PerkinElmer hollow cathode lamp with an operating current of 15 mA.
The graphite platform with the W-Rh permanent chemical modifier was prepared automatically by using the facilities provided by the original software for the autosampler and graphite furnace, then 250 µg W and 200 µg Rh were thermally and sequentially deposited onto the integrated platform, as described elsewhere (1).
The furnace programs for the direct determination of Cu in sediments using digested samples and slurries are shown in Table I.All measurements were made with at least five replicates and are based on integrated absorbance.Argon (AGA, Campinas, Brazil) was used as the protective gas throughout.
All solutions were stored in highdensity polypropylene bottles.Plastic bottles, autosampler cups, and glassware were cleaned by soaking in 10% (v/v) HNO 3 for 24 hours, rinsing five times with Milli-Q water, and drying and storing in a class 100 laminar flow hood.
The water used to clean the sampling capillary was replaced with a solution containing 0.1% (v/v) HNO 3 + 0.01% (v/v) Triton X-100 to avoid clogging of the autosampler pipette, thus improving slurry dispensation onto the platform.
Preparation of the tungsten and rhodium solutions for graphite surface treatment are described elsewhere (1).
A 1000-mg L -1 Cu stock standard solution was prepared by dissolution of 1.000 g metallic Cu (Johnson & Matthey, Royston, U.K.) in dilute HNO 3 .Analytical reference solutions were prepared within the range of 25.0-300.0µg L -1 Cu by suitable serial dilution of stock solutions in 0.5% (v/v) HNO 3 .The calibration was periodically checked after 25 measurements with 150.0 µg L -1 of Cu solution.

Samples
The following reference materials were used for checking the accuracy of the proposed method: SL-1 Lake Sediment, SD-M-2/TM Marine Sediment, and SD-N-  tute for Environmental Studies (NIES), Ibaraki, Japan.
Ordinary sediment samples were collected from a lake located in São Mateus do Sul, Parana State, Brazil.The sediments were collected with an acid-washed plastic scoop and put into polyethylene bags.The sediments were oven-dried at 60°C and large aggregates were broken up.Stones and large shell fragments were discarded and the material that passed through a 2-mm mesh nylon sieve was retained for analysis.
The reference materials and ordinary sediment samples were ground in a ball mill (Tecnal, Piracicaba, Brazil) to obtain samples with a diameter of < 30 µm using a scanning microscope.Sample sieving was not used.Copper contamination for the reference materials was not observed during this procedure.

Slurry Preparation
Slurries were prepared by accurately weighing 0.1-1.0g of the sample into 50-mL glass calibrated flasks and diluting to the mark with 0.04% Triton X-100 containing 0.5% (v/v) HNO 3 .The resulting slurry was homogenized in an ultrasonic bath (Thornton, Vinhedo, Brazil) for 10 minutes in order to break up particle agglomerates.Under sonication, the slurries were transferred into acid-cleaned poly(propylene) autosampler cups.All operations were carried out on a clean bench.For each sample, five slurries were prepared and all measurements were carried out in at least five replicates.
Prior to pipetting, the slurry was homogenized by sonication at about 10 W for 20 seconds.In order to avoid possible sedimentation errors (23), the autosampler capillary was immersed 10 mm below the surface of the slurry.Then, 20 µL of slurry was taken up and delivered into the atomizer.For untreated platforms, 10 µL of the chemical modifier was subsequently deposited onto the atomizer.The heating program of Table I was employed throughout.Calibration was performed by using aqueous reference solutions.

Wet Digestion
For comparative purposes, a microwave-assisted wet decomposition was performed.Sediments were decomposed in triplicate in a Model PMD microwave oven (Anton Paar, Austria) according to the following procedure (24): 0.2-g samples were accurately weighed into PFA decomposition vessels, then 1.5 mL HNO 3 , 0.5 mL HCl, plus 1.5 mL HF were added to the vessels.The bomb was placed inside the microwave oven and the decomposition carried out at 550-700 W for 10-15 minutes.After cooling, the bomb cap was removed, and the open vessels placed onto a hot plate.In order to eliminate excess HF, 1 mL of H 2 SO 4 (Suprapur®, Merck, Darmstadt, Germany) was added to the bomb, the samples boiled to near dryness, and subsequently transferred to 100-200 mL volumetric flasks, adjusting the final acidity with 0.5% (v/v) HNO 3 .
The resulting solutions were analyzed by ETAAS by delivering 20 µL of decomposed sample material into the atomizer and using the heating program listed in Table I.When a conventional modifier was employed, 10 µL of its solution was delivered into the atomizer after delivering the sample solution and using the heating program listed in Table I.

Determination of Sampling Efficiency
An empty THGA tube was weighed and then 40-100 consecutive pipettings of 20 µL of slurry and drying steps were performed with a series of slurries of various concentrations.The mass difference between the tube con-taining the dry sample and the empty tube divided by the theoretical sampled mass provides the sampling efficiency (21).This procedure was performed at least three times for each slurry concentration.

RESULTS AND DISCUSSION
The Cu concentration in the digested samples and slurries ranged from 40-300 ng mL -1 (keeping a dilution factor within the range of 500-1000 mL g -1 ).Due to its high concentration in the samples, an alternative wavelength (327.4 nm, slit 0.7 nm) was employed to decrease the Cu analytical signal.Gas stop conditions were used during the atomization step.Argon mini-flow in the atomization was avoided in order to decrease the sensitivity, since low analyte recoveries in slurry analysis were observed due to a non-isothermal atomization condition caused by the mini-flow (25).

Modifiers
Copper is an element that is routinely determined in sediment samples.Use of a chemical modifier is typically required for the determination of this element, mainly due to a high concentration of concomitants present in the samples.Suitable coverage of the THGA platform with the W-Rh permanent modifier was accomplished with 50-µL aliquots of the modifier solution.This amount of solution is sufficient to spread over the entire platform during the graphite surface treatment, as reported in the literature (1).A better performance of the permanent modifier was obtained with 250 µg W deposited onto the graphite platform.The use of larger W masses was studied, but double and broader peaks were observed for Cu, which then requires a high atomization temperature to avoid memory effects.
Pyrolysis curves for Cu in 20 µL digested CRM 320 River Sediment sample solution (dilution factor of 500 mL g -1 , 1.76 ± 0.04 ng of Cu per injection) and in 20 µL of 0.25% (m/v) slurry (2.21 ± 0.05 ng Cu per injection) without and with W-Rh permanent modifier are shown in Figure 2. As can be seen, Cu present in the sediment slurry is stabilized up to 1200°C with and without modifier.This indicates that Cu was probably stabilized in the sediment matrix, most probably by occlusion in the sample particles such as reported by Bendicho and Loos-Vollebregt (26).It should be noted that in digested sample solutions without modifier, the Cu signal kept stable up to 800°C, then the analytical signal decreased abruptly.It was also observed that the Cu maximum attainable pyrolysis temperature in the digested sample solution was 1100°C by using 250 µg W + 200 µg Rh permanent modifier.
The stabilization of Cu in sediment slurries can be attained without the use of chemical modifiers; however, under these conditions, Cu recovery in river sediment slurry was only 80%.In addition, when no modifier was used, the relative standard deviation (%RSD) for 10 measurements was above 10.0%for all samples analyzed.However, when the W-Rh permanent modifier and 5 µg Pd + 3 µg Mg(NO 3 ) 2 were employed, the RSD was below 3.0% (n=10) and a quantitative recovery was observed.For this reason, 250  µg W + 200 µg Rh were chosen as the permanent modifier throughout this work.For comparison purposes, the recommended Pd + Mg(NO 3 ) 2 conventional modifier (22) was also employed.

Recoveries in Digested Samples
In order to verify whether sample matrices interfere on Cu atomization, recovery experiments (Table II) were carried out by spiking the digested samples with known amounts of Cu.
No significant interference of the sediment sample digests on Cu atomization was observed when 250 µg W + 200 µg Rh permanent modifier and 5 µg Pd + 3 µg Mg(NO 3 ) 2 were employed, which shows robustness of the proposed method.In addition, there was a linear relationship between absorbance and increasing amounts of Cu in the digests.
With 250 µg W + 200 µg Rh permanent modifier, slightly better recoveries and lower standard deviations were achieved as compared with Pd + Mg(NO 3 ) 2 .It can be assumed that the graphite coating process produces a very thin layer of the permanent modifier on the graphite surface.This may allow faster, and more efficient, release of the analyte compared with the situation where it has to migrate out of relatively large Pd droplets of the conventional modifier (2)(3)(4).

Slurry Optimization
The success of slurry sampling depends on the diameter of the sample particle, the number of particles present in the injected volume, analyte homogeneity, suspension medium, slurry concentration, stirring method, and sampling depth, which in turn may affect the accuracy and precision of the analysis (7-9).The choice of particle size depends on the density of the material and analyte homogeneity.The higher the density of the material, the smaller the parti-cles required.In order to limit sampling errors, it was established that at least 50 particles should be introduced into the atomizer when 20 µL of slurry is sampled (7,8).In the present work, the estimated number of particles varied from 1300 to 6700 on dispensing 20 µL of slurry into the atomizer, containing 0.2-1.0%(m/v) (40-200 µg) of sample.In the calculations, the particle was considered to be spherical (diameter 30 µm) and the average density of the sediment, estimated as recommended (9), was 2.1 g cm -3 .The large number of particles introduced results in low sampling error.
Although the number of particles of sediment slurries introduced into the atomizer was reasonably higher than 50 for 0.05 and 0.1% (m/v) slurries, i.e., from 337 to 675 particles, the precision of the measurements (n=10) was impaired (RSD 12%), which indicates that the sample was not as homogeneous for masses lower than 50 mg (final volume 50 mL).On the other hand, for slurries containing up to 1.0% (m/v), there was a linear relationship between the integrated absorbance and the amount of sample introduced into the atomizer.For slurry concentrations higher than 1.5% (m/v) (>300 µg sample), the RSD for the measurements was higher than 6.5% (n= 10).A problem associated with the introduction of large inorganic sample amounts into the graphite tube is the buildup of inorganic residue that can encrust upon the optical path, resulting in drastic deterioration of the signal-to-noise ratio.Taking into account the mean Cu concentration in the samples, it was decided to work with sediment slurries within the 0.2-1.0%(m/v) range.
Another variable to avoid sedimentation error was the sampling depth (23).The slurry was pipetted 10 mm below the sample surface as recommended by Majidi et al. (23).Under these conditions, the sampling efficiency varied from 97.2 ± 2 to 102.9 ± 1% (n=3) for slurries ranging from 0.2 to 1.0% (m/v) when the samples were sonicated for 20 seconds before being sampled.When the ultrasonic probe was turned off, a sampling efficiency as low as 52.6 ± 2 to 60.9 ± 3% was obtained for a 0.2-1.0%(m/v) slurry.As can be seen, agitation of the slurry prior to pipetting avoids settlement of the particles and also breaks up sample agglomeration, thus minimizing sampling error (7)(8)(9)23).

Analytical Characteristics
Calibrations were carried out against aqueous reference solutions with a linear range extending up to 300 µg L -1 Cu.The Cu characteristic masses for the W-Rh permanent modifier and for 5 µg Pd + 3 µg Mg(NO 3 ) 2 , based on integrated absorbance, were 30 ± 1 pg and 31 ± 1 pg, respectively (uncertainty based on 10 average results obtained on different days), by using the Cu atomic line of 327.4 nm.The experiments were performed in a 0.5% (v/v) HNO 3 medium, although concentrations as high as 5.0% (v/v) could be tolerated by the permanent modifier without significant variations in tube lifetime, as already reported (1-4).Detection limits when using the W-Rh permanent modifier for a digested solution (dilution factor of 500 mL g -1 ) and 1.00% (m/v) slurry concentration, were 1.6 µg g -1 Cu and 0.33 µg g -1 Cu, respectively.The DLs were calculated from 20 consecutive measurements of the blank solution [0.5% (v/v) HNO 3 for digested solutions and 0.5% (v/v) HNO 3 + 0.04% (v/v) Triton X-100 for slurries samples] according to IUPAC (27).
Use of the W-Rh permanent modifier, in comparison to untreated pyrolytic carbon platforms, increases tube lifetime up to 100% for a typical 20-µL direct sample solution and/or slurry Variations of sensitivity are based on the integrated absorbance of the first 20 firings obtained in each platform (digested solution -A = 0.258 W-Rh treated platform, A = 0.238 untreated platform; slurry -A = 0.388 W-Rh treated platform, A = 0.376 untreated platform).The W-Rh platform was recoated at each 350-300 analytical firings when delivering 20 µL of digested solution and 20 µL of slurry, respectively.introduction, as already described (1)(2)(3)(4).The treatment remains stable for approximately 300 and 250 analytical firings when 20 µL of digested sediments and sediment slurries are delivered onto the platform, respectively.The used treated platform can be reconditioned by performing 4-5 firings at 2500ºC for a 5-second cleanout, and subsequently carrying out the treatment procedure described elsewhere (1).
Long-term stability and repeatability using the pyrolytic carbon platform with W-Rh treated platform vs. 5 µg Pd + 3 µg Mg(NO 3 ) 2 for the determination of Cu in CRM 320 River Sediment digested solution (dilution factor 500 mL g -1 ) and in a 0.3% (m/v) slurry are presented in Table III.As can be seen, repeatability of the analytical signals obtained with the W-Rh coating is superior to using the Pd + Mg(NO 3 ) 2 modifier throughout the atomizer lifetime for digested solutions as well as slurry samples.Variations of sensitivity during the analytical firings are much lower for the permanent modifier as compared to the conventional modifier using untreated platforms.There also is less need for recalibration during routine analysis when using the permanent modifier, compared to the Pd + Mg(NO 3 ) 2 modifier.It was observed that the slope of the analytical calibration curve remains stable up to about 300 and 250 measurements with each single coating for the digested solutions and slurry samples, respectively.It is also important to note that use of the permanent modifier reduces analytical time, because the heating program is shortened, and thus results in increased sample throughput.
However, direct introduction of the inorganic matrix into the atomizer cuts tube lifetime in half.For 0.3% (m/v) CRM 320 River Sediment slurry, the tube lifetime was limited to 730 and 325 analyti-cal firings by using the W-Rh permanent modifier and Pd + Mg(NO 3 ) 2 , respectively.On the other hand, for digested river sediment solutions (dilution factor 500 mL g -1 ), the tube lifetime was extended to 1360 and 620 analytical firings when the W-Rh permanent modifier and Pd + Mg(NO 3 ) 2 , respectively, were used.

Analysis of Samples
Six sediment certified reference materials and four ordinary sediment samples were analyzed by employing digested sample solutions as well as slurries with the proposed W-Rh permanent modifier (Table IV).For comparison purposes, the samples were also digested employing the Pd + Mg(NO 3 ) 2 modifier.In all cases, the calibration was run against aqueous reference solutions.
The obtained analyte concentrations for the different methods were in agreement with the certified values (Table IV).The results are presented as an average ± confidence interval (at 95% confidence level -t student = 2.776).
The performance and accuracy of the method employing the W-Rh platform treatment with different coatings and tubes for the determination of Cu in the chosen samples was found to be better than with the conventional modifier.It should be pointed out that the standard deviations of the results for all samples using 250 µg W + 200 µg Rh permanent modifier were always lower than those related to the conventional modifier for digested samples.The lower blank values and longer signal stability in function of time when using the proposed permanent modifier account for this achievement.

CONCLUSION
In this work it has been shown that the analytical improvements obtained with the W-Rh permanent modifier for the determination of Cd and Pb as reported in the literature (1-4) can also be obtained in the ETAAS determination of Cu in digested sediment sample solutions and slurries.
It was observed that direct introduction of the inorganic slurry material into the graphite tube halved tube lifetime (730 vs. 325 firings) in comparison to using digested sample solutions (1360 vs. 620 firings) with the W-Rh permanent modifier and Pd + Mg(NO 3 ) 2 , respectively.On the other hand, the confidence interval of analyte concentration for slurry sampling was always lower (highest confidence interval value 1.4 µg g -1 ) than that obtained for digested sample solutions (highest confidence interval value 2.1 µg g -1 ).The detection limits for the slurry samples were at least four times lower than for the digested solutions.
The W-Rh permanent modifier can be successfully used for routine analysis in environmental analytical chemistry laboratories for the determination of copper in sediment samples.3) have estimated that as a result of coal production during the last two decades, the As and Hg amounts that could potentially pollute the environment are 700,000 and 14,000 short tons, respectively.The fact that As and Hg are present in coal has received great attention since these elements are known to be harmful to life at relatively low concentrations.Thus, legislation in some countries imposes atmospheric discharge limits (4).It has therefore become imperative to develop simple, accurate, and sensitive methods to determine As and Hg in coal.
Analytical methodologies most often used for As and Hg determination in coal and coal combustion products (coal fly ash and slag) are those that do not need a previous sample pre-treatment such as INAA (5-8), RNAA (9-12), XRF (13,14), XAFS (15,16), and LA-ICP-MS (17).These methods result in slight sample contamination but offer highspeed analysis.However, these methods require instrument calibrations using reference materials with a composition very close to those of the samples being analyzed, which is a very time-consuming process.Furthermore, these techniques require instrumentation that measurement is quick.In addition it must be considered that coal requires dissolution methods capable of decomposing large quantities of organic matter.Slag and coal fly ash require methods capable of dissolving acid-resistant minerals and glasses.
Methods used for the digestion of coal (30,31) and coal fly ash (30,32) samples have been reviewed.Different acid mixtures (nitric, sulphuric and hydrochloric, hydrofluoric and perchloric acids) were used in open vessels (18,33).Some of the disadvantages related to acid extraction methods include incomplete extraction of some chemical forms of the element present in the sample, possible contamination during sample manipulation and, most importantly, loss of volatile elements such as As and Hg (34).Over the last several years, microwave oven procedures have been developed for sample digestion (24)(25)(26)28,29,(35)(36)(37), because this method is rapid, safe, and requires low acid volumes, eliminates loss of volatile elements, and shows less contamination risk than traditional digestion methods or alkaline fusion (38,39).However, the sample amount is a limiting factor because 1.0 g is the maximum amount of sample that can be subjected to microwave energy.Thus, after microwave acid extraction, pre-concentration methods such as liquid-liquid extraction, co-precipitation, or adsorption on sorbents of different materials can be carried out to reach the sensitivity required.The sensitivity can also be improved by ashing the sample prior to acid extraction.This procedure allows metal pre-concentration and removal of the organic ABSTRACT Methods for the determination of As and Hg in coal, coal fly ash, and slag samples involving microwave acid extraction and cold vapor/hydride generation AAS have been developed.Two microwave acid extraction procedures using aqua regia and aqua regia + hydrofluoric acid were studied.For samples with high organic matter content (coal samples), a low temperature ashing pre-treatment (LTA) (below 150°C) was carried out prior to microwave acid extraction.The results were compared to those obtained without a LTA pre-treatment.Satisfactory analytical recoveries were obtained for As, while around 18% of Hg was removed during the LTA pretreatment.Mercury cold vapor and arsenic hydride were generated from hydrochloric acid medium using sodium tetrahydroborate in a batch hydride generation system coupled to AAS.The LODs obtained were 0.20 and 0.15 µg L -1 for As and Hg, respectively.The accuracy was studied using NIST-1633b coal, NIST-1635 coal, and NIST-1632b coal fly ash reference materials.The methods were applied to several coal, coal fly ash, and slag samples.
The purpose of this paper is to optimize a microwave acid digestion procedure for As and Hg extraction from coal, coal fly ash, and slag samples prior to cold vapor/hydride generation atomic absorption spectrometry (CV/HG-AAS) determination.Two microwave acid extraction procedures using aqua regia and an aqua regia/HF mixture are compared.In addition, for coal samples, a low temperature ashing procedure is also optimized to remove the organic matter content.

EXPERIMENTAL Instrumentation
A Technique Plasma GmbH 200-G system was used for low temperature ashing of coal samples.A MLS 1200 Milestone microwave oven (Sorisole, Italy), programmable for time and microwave power, was used for the metal extraction from the samples.An Advanced Mercury Analyzer AMA-254 (Altec Ltd, Praha, Czech Republic) was used for the Hg determination in untreated coal samples.A PerkinElmer MHS-10 hydride generation system (PerkinElmer Instruments, Shelton, CT USA) was employed for Hg and As vapor generation.As and Hg absorbances were measured with a PerkinElmer Model 2380 atomic absorption spectrometer equipped with hollow cathode lamps as the radiation sources, using the peak height measurement mode.The spectrometer operating conditions are shown in Table I.

Low Temperature Ashing Pre-treatment
Low temperature ashing was performed in an oxygen plasma asher according to the following procedure.About 1.0 g dried coal samples were placed and smoothly spread inside ceramic boats (1.0 mm in depth).In order to increase the surface area, some waves on the sample surface were made by using a ceramic spatula.The samples were ashed at temperatures below 150°C using the low temperature ashing (LTA) conditions listed in Table II.This LTA procedure involved two ashing cycles of three hours each.Between the two ashing cycles, the samples were stirred and new waves developed on the surface; thus, the new material was again exposed to the oxygen plasma (surface regeneration).

Microwave Acid Extraction Procedures Aqua Regia Extraction
A 0.4-g amount of coal (previously ashed at low temperature) or 0.6 g of coal fly ash or slag samples were placed into the PTFE bombs: 6 mL and 2 mL of HCl(c) and HNO 3 (c), respectively, were   III).After microwave extraction, the bombs were cooled in a water bath.Then the acid liquid phase was filtered using Whatman Nº 40 paper filter (Whatman International Ltd, Maidstone, UK).The solutions were made up to 25-mL volume (for coal and slag) and 50-mL volume (for coal fly ash) with ultrapure water, then stored in polyethylene bottles at 4°C before measurements.

Aqua Regia + Hydrofluoric Acid Extraction
A 0.4-g amount of coal samples (previously ashed at low temperature) or 0.6 g of coal fly ash or slag samples were weighed and placed into the PTFE bombs.Then, 6 mL, 2 mL, and 3 mL of HCl(c), HNO 3 (c), and HF(c), respectively, were added, and the mixture subjected to microwave energy (Table III).The bombs were cooled, then 1 g of boric acid was added in order to avoid damage of the glassware by the hydride generation system.The mixture was subjected again to microwave energy (Table III).The bombs were cooled and the acid liquid phase was filtered by using Whatman No. 40 paper filter as stated above.The solutions were finally made up to 25-mL volume (for coal and slag) and 50-mL volume (for coal fly ash) with ultrapure water and stored in polyethylene bottles at 4°C before measurements.
Cold Vapor/Hydride Generation Procedure 5 mL of acid extract from coal, coal fly ash, and slag samples in each case were placed in the reaction vessels.3.0 or 0.7 mL of HCl (37%, m/v) was added for As or Hg, respectively, (Table I) and made up to 10-mL volume with ultrapure water.An adequate amount (~6 mL) of NaBH 4 (3%, m/v) was added.The vapors formed were transferred with argon gas (argon flow rate of 1.1 mL min -1 ) to the quartz cell.For As determination, the quartz cell was heated with an acetylene/air flame; for Hg determination, a glass wool column was used to dry the cold vapor, avoiding water condensation into the quartz cell.

Optimization of Operating Conditions Low Temperature Ashing Conditions for Coal Samples
To optimize the LTA conditions, 1.0 g of NIST-1632b bituminous coal and NIST-1635 sub-bituminous coal reference materials were subjected to different ashing times (3, 6, 9, and 15 hours).For an ashing time of 6 hours, two ashing cycles of 3 hours were carried out and a sample surface regeneration was performed between each cycle.Thus, an ashing time of 9 hours (three ashing cycles of 3 hours) involved two sample surface regeneration steps; an ashing time of 15 hours involved five ashing cycles and four sample surface regeneration steps.Sample surface regeneration for each cycle is necessary to improve sample ashing.The percentage of mass loss achieved (in the 3-15 hour range) for the different reference materials studied is shown in Figure 1.The ashed reference materials were subjected to microwave acid extraction.Arsenic and Hg analytical recoveries were obtained analyzing the digested NIST-1635 and NIST-1632b coal reference materials, respectively.As

Microwave-Acid Extraction Program a
Step Time (min) Power (W)  can be seen in Table IV, satisfactory As analytical recoveries were achieved for the ashing times studied.Thus, a pre-concentration of 87% (for an ashing time of 15 hours) could be reached without As losses, resulting in good sensitivity increase.For an ashing time of 6 hours, a pre-concentration of 75% was reached.The Hg analytical recoveries obtained were 82 and 75% for the 3-and 6-hour ashing times, respectively.Thus, Hg losses are obtained with LTA pretreatment; however, it must be taken into account that the Hg analytical recoveries were calculated using an information Hg value (value give by NIST as additional information, but not certified) offered by the NIST-1632b coal reference material.To achieve the Hg loss percentage, a coal sample was analyzed using the proposed procedure (with different LTA times) and the results obtained were compared with those obtained with the AMA-254 advanced mercury analyzer.The AMA-254 manufacturer's suggested conditions were applied for the Hg determination.The results show Hg loss percentages of 15, 18, 30, and 45% for ashing times of 3, 6, 9, and 15 hours, respectively.These results are in agreement with published arsenic data.Finkelman et al. (39) and Gentzis et al. (40) reported that As is not removed during the LTA procedure with ashing temperatures of less than 200°C and 120°C, respectively.These authors also reported Hg losses between 25-80% and 95%, respectively.Figure 1 shows the lost mass percentage relative to ashing time.
As can be seen, an ashing time of 6 hours was enough to reach 73.2 and 68.5% of mass removal for NIST-1635 sub-bituminous coal and NIST-1632 bituminous coal, respectively.The high percentage of mass lost was achieved by using low ashing times.
An ashing time of six hours with two ashing cycles was selected as the optimum conditions for LTA pre-treatment.The sensitivity achieved with CV/HG-AAS using the optimum conditions was adequate (see Table II).In addition, foam formation during vapor generation due to the organic matter reaction with NaBH 4 was reduced.

Acid Microwave Extraction Conditions
The microwave energy (150-700 W), exposure time (2-8 min), and volume of aqua regia (4-12 mL) and aqua regia/hydrofluoric acid (4-12 / 1-5 mL, respectively) used were investigated for the determination of As and Hg in coal, coal fly ash, and slag samples.The optimum conditions used are shown in Table III.Different amounts of coal and coal fly ash reference materials (0.2, 0.4, 0.6, and 0.8 g) were also investigated.The results show low As and Hg analytical recoveries with sample amounts of 0.8 g.In addition, foam formation during vapor generation was observed.Low As and Hg absorbance signals were obtained by using 0.2 g of samples.Thus, 0.4 g for coal samples and 0.6 g each for coal fly ash and slag samples were selected.
Using the conditions shown in Table II and employing aqua regia, the results obtained for As (Figure 2a) show adequate analytical recoveries for NIST-1632b and NIST-1635 coal and coal fly ash reference materials.For Hg, adequate analytical recovery was achieved for coal fly ash.Thus, we can assume that there are no Hg losses during the extraction procedure.The use of aqua regia + HF mixture results in low As and Hg analytical recoveries (Figure 2b) for the reference materials studied.In addition, the aqua regia + HF mixture results in foam  formation during vapor generation as well as in high blank signals.Thus, aqua regia was selected for this study.

Analytical Figures of Merit
Calibration and addition equations were obtained for coal, coal fly ash, and slag samples, spiked with 2.0, 4.0, 6.0, and 8.0 µg L -1 of As or Hg, in the determination of As and Hg.For Hg (see Table V), the slopes of the calibration and standard addition curves are different (ttest for a confidence level of 95.0%).Therefore, Hg presents an important matrix effect.For As, matrix effects were obtained for the coal samples.The limit of detection (LOD) and limit of quantification (LOQ), defined as 3 Sd/m and 10 Sd/m (Sd is the standard deviation of 11 measurements of a blank and m is the slope of the calibration curve) were 0.20 and 0.67 µg/L for As and 0.15 and 0.50 µg/L Hg, respectively.The LODs and LOQs (expressed in mg/kg) relative to mass of sample used in the microwave acid extraction for each kind of sample are shown in Table VI.Due to the fact that As losses do not occur during the LTA pre-treatment, the arsenic detection limit (relative to coal samples) could be reduced to 0.13 mg/kg using a LTA time of 15 hours.The within-batch precision obtained (relative standard deviation for 11 replicate measurements at different levels) was good with RSDs (%) lower than 10% for all concentrations studied.
The reference materials NIST-1632b (bituminous) and NIST-1635 (sub-bituminous) Trace Elements in Coal, and NIST-1633b Trace Elements in Coal Fly Ash were analyzed in order to assess the accuracy of the methods proposed.The results obtained, in agreement with the certified values for As, are shown in Table VII.For Hg, adequate accuracy was achieved for coal fly ash reference material, while for NIST-1632b coal reference material the value obtained was 18% lower than the Hg information value.This fact is due to the Hg losses during the LTA pre-treatment.Finally, for slag samples (reference materials are not available), the study of the analytical recovery was carried out.Concentrations of 5.0, 10.0, and 20.0 µg L -1 of As and Hg were added to the slag samples and then subjected to the acid extraction procedure.The results show adequate analytical recoveries (98-105%) for As and Hg at all concentrations studied.

APPLICATION
The optimized procedures were applied to several coal, coal fly ash, and slag samples.Each sample was subjected twice to the microwaveassisted acid digestion.For the coal samples, the LTA procedure was also carried out before microwave extraction.The results are shown in Table VIII where the concentrations of each element (N   = 3) are given together with the SDs.As can be seen, the Hg levels are lower than the As concentrations in all samples studied.The highest As concentrations were obtained in the coal fly ash samples.This is probably due to the high volatility of Hg; Hg concentrations in the coal fly ash appear to be lower than in the coal samples.

CONCLUSION
The use of the LTA pre-treatment procedure allows element pre-concentration in coal samples due to the removal of the organic matter.Thus, elements such as As and Hg (present at low levels in these samples) can be determined using a simple technique.In addition, the LTA pre-treatment procedure is necessary for the removal of organic matter and thus avoids foam formation during vapor generation.Using the optimized LTA conditions, As losses did not occur, while around 18% of Hg is removed.The use of an aqua regia microwave acid extraction is adequate to obtain a complete As and Hg extraction from coal, coal fly ash, and slag samples as evidenced by the results of the accuracy studies.The use of the aqua regia/hydrofluoric acid mixture results in foam formation during cold vapor/hydride generation and high blank signals.In addition, the analytical recovery from coal and coal fly ash reference materials, achieved using this mixture, are very low.Using the method described, As and Hg are determined with adequate sensitivity in the analysis of these kind of samples.

INTRODUCTION
Norfloxacin (1-ethyl-6-fluoro-1,4dihydro-4-oxo-7-(1-piperazinyl)-3quinoline carboxylic acid) is a low toxic fluoro-quinolone antibiotic which is highly effective against a broad range of bacterial infections.It is active against many gram-positive and gram-negative bacteria and has found widespread use in the treatment of cystitis, dysentery, and other infectious diseases.
It is believed that the mechanism of the action of fluoroquinolone antibiotics lies in the fact that they can hinder the duplication of DNA.It is the metal ions that act as intermediary in the combination of the drug with DNA.Literature (3,27,28) has demonstrated acid medium, the complex of norfloxacin-Fe(III) was adsorbed on a micro column filled with cation exchange resin for a certain period of time, followed by the elution with HNO 3 , then sent to the nebulizer for measurement by FAAS.The concentration of norfloxacin in the samples was determined by measuring the absorbance of Fe(III).The whole process is very simple, resulting in high sample throughput and low cost, and therefore suitable for routine analysis of norfloxacin pharmaceuticals.

EXPERIMENTAL Instrumentation
A Model IFIS-C flow injector (Xi'an Ruike Electron Equipment Corporation, P.R. China) was used (30).A Model WFX-1C atomic absorption spectrometer (Beijing Ruili Analytical Instrument Plant, P.R. China), equipped with an iron hollow cathode lamp, was used to measure the absorbance of Fe(III).The wavelength and lamp current used was 248.3 nm and 6.0 mA, respectively.1.80 L min -1 of acetylene and 7.4 L min -1 of air were employed to obtain a steady flame.A Model 3066 recorder (Sichuan Instrument Plant, P.R. China) was used to record the absorbance peaks.

Reagents
All chemicals were of analytical reagent grade and doubly distilled water was used throughout.
Stock standard solution of norfloxacin (1000 µg mL -1 ), provided by dissolving 0.1000 g norfloxacin in water and diluting to 100 mL, was stored in the refrigerator.Working standard solutions were prepared fresh daily by appropriately diluting the stock solution.

ABSTRACT
A flame atomic absorption spectrometry (FAAS) method is proposed for the indirect analysis of norfloxacin, based on the complexation reaction of norfloxacin with Fe(III).In a flow injection on-line preconcentration and separation system, a norfloxacin solution was mixed with a stream of Fe(III) solution to produce the cation complex norfloxacin-Fe(III).It was adsorbed on a micro column filled with a cation exchange resin, while the excessive Fe(III) reacted with NH 4 F to form anion (FeF 6 ) 3-, which then exited through the micro column as waste.The adsorbed norfloxacin-Fe(III) complex compound was eluted reversely with HNO 3 to the nebulizer and measured by FAAS.The absorbance of Fe(III) is proportional to the concentration of norfloxacin.When the reaction and adsorption time was 60 s and 100 s, the calibration curve was linear over the range of 8 -180 µg mL -1 and 3 -150 µg mL -1 , with a relative standard deviation of 1.4% and 2.9%, at an analytical frequency of 30 and 22 samples per hour, respectively.This method was applied to the analysis of norfloxacin in capsules and the results compared well with those provided by perchloric acid titrimetry.

Flow Injection Flame Atomic Absorption Spectrometry for the Indirect Analysis of Norfloxacin
Zhi-Qi Zhang* and Yu-Cheng Jiang Department of Chemistry, Shaanxi Normal University, Xi'an 710062, P.R.China that norfloxacin can form a sort of 1:1 complex compound with various transition metals, such as Cu(II) and Fe(III), through its 3-carboxyl group and 4-carbonyl.Based on the complexation reaction of norfloxacin with Fe(III), two flow injection (FI) methods for the analysis of norfloxacin have been reported (28,29).In this work, a FI-FAAS method for the analysis of norfloxacin is proposed.In a weak Fe(III) standard solution (1000 µg mL -1 ) was prepared by dissolving the required amount of ammonium iron alum in water and adjusting the acidic concentration of 0.2 mol L -1 by nitric acid.Nitric acid solution: 3.0 mol L -1 .Ammonium fluoride solution: 0.2 mol L -1 .
Amberlite IR-120 resin (Xi'an Electric Power Resin Plant, particle diameter 0.3-1.2mm) was first purified by soaking in saturated solution of sodium chloride for 20 hours; washing with water, 1 mol L -1 of hydrochloric acid, 1 mol L -1 of sodium hydroxide, and then again with water; then soaking in 10% acetic acid solution for 24 hours.The resin, rinsed with water, was filled into a micro-column.

Flow Injection System
A schematic of the flow injection system used in this work is shown in Figure 1.The instrumental parameters are listed in Table I.
When the injection valve was in the sampling position, the stream of the sample solution (S) passed first through a cation exchange resin micro column (B) (2 mm i.d and 10 cm in length) to separate the interference.(Based on the amount of interference ions in the sample, column B should be regularly cleaned).It then merged and reacted with a stream of Fe(III) solution (R) in a reaction coil (L 1 ) (2 mm i. d. and 100 cm in length ) to form the cation complex compound norfloxacin-Fe(III), which merged again with a stream of NH 4 F solution (F).Excessive Fe(III) reacted with (F) to form anion (FeF 6 ) 3-in another reaction coil (L 2 ) (2 mm i. d. and 100 cm in length).The cation norfloxacin-Fe(III) was adsorbed on another cation exchange resin micro column (A) (2 mm i. d. and 4cm in length), which was connected to the injection valve by PTFE tubes (0.5 mm i.d.) as a sample loop, while anion (FeF 6 ) 3-passed through micro column (A) to waste.
A washing process of 30 s using water instead of sample and Fe(III) solution was used after a fixed reaction and concentration time.
When the injection valve was in the injection position, the norfloxacin-Fe(III) compound adsorbed on the micro column (A) was eluted reversely by the eluent (E) (3 mol L -1 nitric acid ) to the nebulizer at a flow rate of 5 mL min -1 , then measured by FAAS.

Optimization of Reaction and Adsorption
The variables studied for reaction and adsorption were the pH value, Fe(III) concentration, ammonium fluoride concentration, length of reaction coil, length and inner diameter of micro-column, the sam-ple and Fe(III) solution flow rate, and sampling time.
The effect of the pH value ranging from 0.5 to 7 was studied on the analysis of 50 µg mL -1 norfloxacin.The influence of the pH mainly lies in the formation of norfloxacin-Fe(III) compounds and the adsorption of norfloxacin-Fe(III) on the microcolumn (A).Experimental results showed that absorbance (A) increased with an increase in pH up to 2, then remained constant at pH 2 -6, but decreased slightly at pH above 7.For optimum reaction and adsorption conditions, a pH value of 4 with nitric acid for the sample solution was chosen.
In order to examine the influence of the concentration of Fe(III), a series of different concen-  trations of Fe(III) in solution were tested in the analysis of 50 µg mL -1 of norfloxacin.The experimental results showed that the absorbance was basically constant when the concentration of Fe(III) solution was higher than 50 µg mL -1 .An Fe(III) concentration of 200 µg mL -1 was chosen.
The concentration of the screening agent NH 4 F is of importance.Low concentrations of NH 4 F could not ensure the complete screening of excessive Fe(III), but extremely high amounts of NH 4 F would affect the adsorption of norfloxacin-Fe(III) compound on microcolumn (A).The experimental results showed that when the concentration of Fe(III) was 200 µg mL -1 , 0.2 mol L -1 of NH 4 F did ensure the complete screening of excessive Fe(III) and did not affect the adsorption of norfloxacin-Fe(III) compound on the micro column (A).
The absorbance increased with a reaction coil (L 1 ) up to 100 cm in length; above that length the absorbance did not increase significantly.Thus, a reaction coil (L 1 ) length of 100 cm was chosen.
The effect of the length of the micro column (A) was examined ranging from 30 -60 mm, with an inner diameter of 2.0 and 3.0 mm, respectively.The results indicated that the exchange capacity increased with an increase in length and diameter, but the broadening and lowering of the absorbance peaks caused a decrease in sensitivity.Thus, a micro column of 4.0x2.0mm was chosen.
The effect of sample and Fe(III) solution flow rate was tested from 1 -7 mL min -1 .It was found that the faster the flow rate, the better the sensitivity.Therefore, the maximum value of 7 mL min -1 as provided by the instrument was selected.
The absorbance increases proportionally with an increase in reaction and adsorption time, which enhances the sensitivity of the analysis, but slows the analytical frequency.It is suggested that reaction and adsorption times should be chosen based on the concentration of norfloxacin in the different samples.

Optimization of Elution Condition
The elution variables studied were concentration and flow rate of eluent, elution time and mode in the optimized reaction, and adsorption condition.
The experimental results based on using ammonium nitrate, sodium chloride, hydrochloric acid, and nitric acid showed that nitric acid was the best eluent.The concentration of nitric acid in the range of 0.5-5 mol L -1 was tested.The absorbance increased with an increase in nitric acid concentration up to 3.0 mol L -1 ; above that amount it remained constant.The eluent chosen was 3.0 mol L -1 nitric acid.
Quickening the elution flow rate would be advantageous to the elution process, but not beneficial to the FAAS analysis.The optimum flow rate chosen was 5 mL min -1 .
Changing the acidity in the adsorption and elution process could cause swelling and shrinking of the resin and would result in an irregular filling of the micro column (A) (31).A reverse elution process, in which the flow direction of the eluent stream was opposite to the reaction and adsorption streams, was chosen which not only quickened the elution process but also avoided an irregular filling of the micro column (A).

Analytical Performance for Norfloxacin Measurements
Using optimum experimental conditions, two calibration lines were prepared with reaction and adsorption times at 60 s and 100 s, respectively.The analytical performance for the norfloxacin measurements is listed in Table II.The detection limits are expressed in characteristic concentrations (1% absorption) and the relative standard deviations (RSD) were obtained for seven samples containing 50 µg mL -1 norfloxacin.

Interference Studies
The effect of other substances was investigated on the analysis of 50 µg mL -1 norfloxacin.The results showed that large amounts of foreign substances usually used as excipients and additives in the production of medicine did not interfere (error less than 5%) such as 10 mg mL -1 of starch, dextrin, sodium carboxyl methyl cellulose, carboxyl propyl cellulose, glucose, fructose, or lactic acid.* A is the absorbance; C is the concentration of norfloxacin expressed in µg mL -1 .
The cations in the sample did not interfere in the analysis of norfloxacin because they have already been separated by cation-exchange resin microcolumn (B) before reacting with Fe(III) .

Application
To investigate the applicability to real samples, the proposed method has been applied to the analysis of norfloxacin capsules produced by different pharmaceutical manufacturers.
The content of the norfloxacin capsule was dissolved in 2 mL glacial acetic acid and diluted to 100 mL with water in a calibrated flask.After filtering, an aliquot of the solution was diluted to an appropriate concentration and analyzed directly.
The results obtained for the analysis of four norfloxacin capsule samples from different manufacturers using the proposed method and the pharmacopoeia method (1,2) are summarized in Table III.The results of both methods are in good agreement.

CONCLUSION
The results presented in this paper clearly demonstrate that the complex reaction of norfloxacin with Fe(III) can be used for the indirect analysis of norfloxacin by means of a flow injection on-line cation exchange resin column adsorption and separation system, followed by flame atomic absorption spectrometry analysis.The proposed method provides a number of advantages over the titration method recommended by pharmacopoeia and includes higher sensitivity, lower sample consumption, higher precision, higher sampling frequency, and is very suitable for the routine analysis of norfloxacin in pharmaceutical preparations.

INTRODUCTION
Monitoring of contaminants in soil has become a very important aspect of environmental chemistry.The analysis of environmental materials, including soil samples, often requires preparation of the solutions.The analytical procedures used in the determination of elements need to be modified depending on speciation form and matrix character of the sample (1) and the final results of the analysis depend on the sample preparation and the interference of other metals.
For the determination of Mn, flame atomic absorption spectrometry (FAAS) is the commonly used technique, which requires that the samples are in solution.Thus, the method used for sample preparation, e.g., mineralization, is an important part of the analytical procedure and influences the final analytical results after digestion (2).Sample mineralization methods have been modified in recent years with the aim to reduce error of analysis.Different digestion methods for the determination of Mn are suggested in the literature .For this study, the methods of reference soil mineralization were chosen according to recommendations by the Office of Measures, Poland -Methods 1-3 and 7 (see Table I); those recommended by "The methods of analysis and estimation of properties of soil and plants" (21), and literature data (20,23).Concentrated acids usually used for the digestion of soil samples suggested in the literature include: HCl + H 2 SO 4 (3); HF + H 2 SO 4 + HClO 4 (3,4); HNO 3 + H 2 O 2 + H 3 PO 4 + with concentrated acids HNO 3 , HF, HClO 4 or aqua regia (7), or by means of the ready set of reagents (Nanocolor set).NanOX N: solid reagents for oxidative decomposition of samples containing nitrogen compounds, NanOX Metal : solid reagents for oxidative decomposition of samples containing heavy metals.Mineralization of soil through carbonate alkali metals fusion was performed as well (3,4,7,8).
The basic conditions for the mineralization of soil samples before the determination of elements (Mn) are available as standard procedures; however, the accuracy of the results obtained are often unsatisfactory.The determination of Mn after digestion is often interfered with by elements such as calcium, magnesium, titanium, or iron (3)(4)(5)(21)(22)(23).Masking reagents used to eliminate interferences are lanthanum and strontium (3,5,22,23); EDTA or DTPA in the HNO 3 medium (3,5); APDC (PDCA, pyrrolidine dithiocarbamic acid) in the HCl medium (3); and sulfosalicylic acid (19) which eliminates cation and anion interferences.Cation interferences are considerably eliminated using flame air/N 2 O, while with flame air/acetylene, interferences are not suppressed (5,23).
In this paper, the optimum conditions for the preparation of soil samples for the FAAS determination of Mn are described.We performed a comparative analysis to determine total Mn in certified reference material (CRM certified obtained from Central Office of Measures -GUM, Poland) soil samples (CRM Soil PL-1) by FAAS analysis after wet digestion, using different qualitative compositions of the digestion solution.Optimum digestion conditions were established and the depen-

ABSTRACT
The aim of this research was to study the influence of sample preparation conditions for total Mn determination in soil by FAAS.The influence of the composition of digestion solutions, initial mass and particle size of the soil sample, and addition of masking substances that eliminate the interference of other elements in the determination of Mn have been studied.On the stage preparation of samples for analysis three mineralization methods of soil were applied: (1) wet method (digestion with concentrated acid: HCl, HF, HClO 4 , HNO 3 ); (2) dry method (digestion with concentrated acids HClO 4 and HNO 3 or HF after initial incineration of the organic matter, fusion with Na 2 CO 3 and K 2 CO 3 ); and (3) microwave method (using mineralization setting Nanocolor or digestion with concentrated acids HCl and HNO 3 ).The Nanocolor system offers two rapid and convenient methods for sample preparation with solid NanOx N and NanOX Metal reagents: a) regular decomposition in a heating block, b) vigorous pressurised decomposition in a microwave oven.

EXPERIMENTAL Instrumentation
A PerkinElmer Model 3100 air/acetylene flame atomic absorption spectrometer (PerkinElmer Instruments, Shelton, CT USA) was used for the analysis at a wavelength of 279.5 nm and fuel flow rate of 0.8 -1.0 dm 3 /min.Also used were an electric hot plate with temperature control, sand bath, electric furnace RK 44 (Vebwetron, Weida, Germany), microwave oven Model AKL 535 (IGNIS, Italy) and closed vessel mineralization (Macherey -Nagel GmbH, Germany).
Standard manganese solutions were prepared from standard solutions for atomic absorption (manganese concentration 1.000 µg/cm 3 , Aldrich).The environmental soil taken from the surface layer in Rzeszów, Poland and CRM Soil PL-1 (obtained from the Central Office of Measures -GUM, Poland) were used for this study.The soils were airdried before use.The samples were placed in disposable polyethylene (PE) vessels.

Preparation of Soil Samples for Digestion
The soil was digested using the eight different methods listed in Table I.Mineralization was performed on a hot plate (methods 1, 2, 4-6), in a microwave oven (methods 7 and 8), or directly in a flame (method 3).Methods 1, 2, 6, and 7 are examples of wet mineralization, in which the organic matter of the sample In method 3, the soil sample was fused with sodium and potassium carbonate, and the alloy obtained was digested with hydrogen chloride.
Methods 4 and 5 are examples of a dry and wet methods combination.A soil sample was incinerated at temperatures from 450-500ºC over a period of 6-8 hours (before analysis).The residue was digested with HClO 4 /HNO 3 (mixture) or HF to decompose the organic matter, then transferred the complexed metal ions into solution.
Microwave methods 7 and 8 shorten the mineralization time.
The FAAS determination of Mn in solution was performed after digestion and dilution of the stock solution to concentrations of 0.2-10.0µg/cm 3 .
In preliminary tests, the amount of solution necessary for mineralization of the soil sample was established at 1.00 g/sample.The conditions of the mineralization process of the soil samples are given in Table I.Three soil samples were studied with triple manganese determination each.The blanks were prepared for all methods described.An overall meanx, a standard deviation overall mean s - x = S -͌n and a confidence limit overall meanx ± t 0.05 • s - x were estimated (24).The results of the determination of Mn are presented in Tables II (influence of composition of digestion solution on total Mn determination by FAAS in CRM PL-1 soil sample), Table III (influence of grain diameter of environmental surface soil on total Mn determination by FAAS), and Table IV (influence of substances to eliminate interferences of other elements on total Mn determination by FAAS in environmental surface soil).

Influence of Mineralization on Mn Determination in Soil Samples
For comparison, all soil samples were digested using the methods listed in Table I.The results obtained (Table II) show that the least difference between reference values, according to the Analysis Certificate of Soil (obtained from the Central Office of Measures -GUM, Poland): Mn = 394.5 (± 17.5) µg g -1 and the experimental value  for the given mineralization method, was obtained for method 2: Mn = 391.0(± 38) µg g -1 , where the soil was digested with the HClO 4 and HF acid mixture.This method was found to be most accurate ( = 99 %) and is recommended for the determination of total Mn in environmental soil.
It was also found that the lowest value for Mn concentration (157.1 µg g -1 ) was obtained by digestion of the soil samples with concentrated acid mixtures of HNO 3 and HClO 4 after dry ashing (method 6, Table II).The results obtained with digestion methods 5-8 are similar to each other and show lower values than expected.Methods 1 and 3 result in high analytical accuracy (99% > > 81%) but the low value of Mn obtained renders these methods not applicable for the analysis of soil by FAAS.

Influence of Grain Diameter on the Analysis of Soil Samples
An air-dried surface soil sample was crushed in an agate mortar and screen-classified using a sieve of mesh size 0.25, 0.12, 0.102, and 0.06 mm.The soil sample was digested by method 2 (Table II) using an acid mixture of HClO 4 and HF.The results (Table III) show that the grain diameter between 0.06-0.25 mm does not influence the results of the determination of Mn in soil and is effectively homogeneous.It was concluded that the optimum soil grain diameter for FAAS analysis is р0.25 mm.

Influence of Masking Solutions to Eliminate Interferences
According to the literature (3)(4)(5)(21)(22)(23), there are substantial interferences between manganese and other elements found in the solution after digestion, which decreases (Fe, Ti, Ca) or increases absorption (Mg).In order to eliminate interferences from these metals, the following correction solutions were used: lanthanum (1% La in analyzed solution); strontium (1% Sr in analyzed solution); NH 4 Cl (1% in analyzed solution).The results in Table IV show that the addition of masking substances eliminates the influence of interferents on the Mn determination by FAAS.The addition of correction solutions such as 1% La or 1% NH 4 Cl (calculated for the mass of solution after digestion) considerably increases the Mn determination results in the analysis of soil.The 1% Sr solution showed no quantitative masking,resulting in an accuracy of Mn determination at 94% (Table IV).

CONCLUSION
The determination of manganese in mineralized soil samples leads to the determination of total or closeto-total manganese.The techniques of mineralization of soil and the use of correction solutions in FAAS measurement are not unified and sometimes lead to inconsistent results.For the determination of Mn in soil by FAAS, the following con-ditions are recommended: soil grain diameter < 0.25 mm, digestion solution of HClO 4 and HF (1:10), and addition of substances that eliminate the effect of other elements, like 1% La or 1% NH 4 Cl (calculating the mass of solution after digestion).The results of this study show that the method proposed can be successfully applied by environmental laboratories in the determination of Mn in soil.

INTRODUCTION
Separation and preconcentration of trace elements with flow injection (FI) techniques and their subsequent determination with atomic absorption spectrometry (AAS) have made great progress in recent years.With these continuous procedures, the time of analysis, consumption of sample and reagents, and risk of contamination have been dramatically reduced.
Several methods have been developed for continuous systems such as liquid-liquid extraction, coprecipitation, ion exchange, and sorption (1-7).Fang et al. (8) developed a system for the separation and preconcentration of trace metals using a knotted reactor (KR) made of PTFE tubing.At first, this system was used for coprecipitation methods (8)(9)(10)(11) collecting the precipitate on the walls of the KR.This precipitate was subsequently dissolved with a solvent, eliminating the need for a filtration unit.In this case, the function of the KR provided radial mixing of the sample and the reagents led to reproducible conditions for the precipitation step.Moreover, the knots caused changes in the flow direction and created a secondary flow with some centrifugal force in the stream that carried the particles towards the tube walls, acting as a collector of the precipitate.12) also evaluated this system for the adsorption of a soluble complex in the inner walls of the KR, since a particular interaction of soluble complexes with the hydrophobic tubing material was observed during the development of the coprecipitation and chelate sorption on RP-Si-C18 tech-absorption spectrometry (ETAAS) (15)(16)(17)(18)(19), and recently to inductively coupled plasma optical emission spectrometry (ICP-OES) (20) and ICP mass spectrometry (ICP-MS) (21,22).
To obtain satisfactory results using this methodology, several operational conditions must be carefully optimized in order to guarantee a reproducible complex adsorption and an efficient complex elution.Among these conditions, the sample and reagent flow rates and the loading times are particularly related to reproducibility and sensitivity.
The aim of this study was to investigate the relationship of the sample loading time with the complex adsorption process in a knotted reactor during the preconcentration of Cd where sodium diethyldithiocarbamate (Na-DDC) is used as the chelating reagent.For this purpose, several assays were performed varying the sample loading time, and the inner walls of the KR were studied using scanning electron microscopy and X-ray fluorescence.These experiments also showed whether preconcentration occurs only by sorption of the soluble complex Cd-DDC on the inner walls of the KR or whether some coprecipitation takes place.In addition, two eluents (ethanol and IBMK) were tested to establish which would achieve an efficient elution of the Cd-DDC complex.

ABSTRACT
Several experiments were performed to establish whether the preconcentration of cadmium as Cd-diethyldithiocarbamate (Cd-DDC) using a knotted reactor (KR) occurs only by sorption or by coprecipitation.Different sample loading times were assayed and the inner walls of the KR were examined by scanning electron microscopy and X-ray fluorescence.When high sample loading times were used, a reduction of the analyte response and waste flow rate was observed.It can be concluded that this is due to a possible "saturation" effect of the KR.The microscopic examination revealed the presence of solid deposits which could cause such effects.The analysis of these particles indicated that they did not correspond to Cd-DDC precipitate, rather that this might be generated as a result of excess reagents and their impurities.The use of iso-butyl methyl ketone (IBMK) during elution, removed the solid deposits from the tube walls more efficiently than ethanol and eliminated extra clean steps between sampling.Today, this on-line preconcentration procedure based on the adsorption of a metal chelate on the inner walls of a KR has already been applied to the determination of several trace elements in a wide variety of samples.The analytical techniques used in conjunction with this procedure include coupling the FI system to flame atomic absorption spectrometry (FAAS) (12,14), electrothermal atomic Vol.22 (6), November/December 2001 tubes were used to propel the sample and standard solutions and the complexing reagent, whereas silicon tubes were employed to propel ethanol.The knotted reactors and all connections consisted of 0.5-mm i.d.PTFE tubing.

Reagents
All reagents were of the highest purity available or at least of analytical reagent grade.A Cd stock standard solution (1000 mg/L) for atomic absorption analysis was obtained from PerkinElmer Instruments (Shelton, CT USA) and working Cd aqueous standards were prepared by diluting the appropriate aliquots of the commercial stock solution in a 0.008 M HCl medium.Hydrochloric acid (37.8%, Baker Instra-Analyzed) was used for trace metal determination.Diethyldithiocarbamate (DDC) solution was prepared daily by dissolving an adequate amount of sodium diethyldithiocarbamate (Sigma Chemical Co.) in a pH 9.2 aqueous buffer solution (ammonium chloride/ammonia) in order to avoid its decomposition.Ethanol (ACS-ISO, Merck) and iso-buthyl methyl ketone (IBMK) (PA, ACS, Panreac) were utilized as eluents.Ultrapure water, 18 MΩ-cm specific resistivity, was obtained from a Milli-Q™ system (Millipore Corporation).All material was kept in 10% (v/v) nitric acid for at least 24 hours and then subsequently washed three times with ultrapure water before use.

Procedure
The separation and preconcentration of cadmium was carried out with the KR-based FI on-line system using the experimental conditions listed in Table I.All assays were made with a 30-µg/L Cd standard solution in 0.008 M HCl.The inner walls of the KRs were examined after the preconcentration step using two different sample loading times (60 and 80 seconds).The inner walls were also studied after the elution step for both eluents considered in this work, ethanol and IBMK.Several preconcentration-elution cycles were performed with each reactor to reproduce the usual working conditions as much as possible.Afterwards, eight fragments, equally separated, from each reactor were cut and longitudinally opened to explore both sides of the inner walls by scanning electron microscopy.The residues found in these fragments were subjected to elemental analysis by X-ray fluorescence.

RESULTS AND DISCUSSION
Preconcentration of the trace elements using a KR-based method involves a molecular adsorption of the metal complexes on the inner surface of the PTFE reactor.When optimizing the conditions for preconcentration of Cd in a KR using Na-DDC as the complexing reagent, it was observed that the efficiency of this adsorption was affected by several factors, but the sample flow rate was the most important due to its influence on the complex formation.
The inner walls of the KR were examined by scanning electron microscopy after preconcentration of Cd with Na-DDC.Using a 60-second sample loading time, the surface of the tube looked very similar to that of the original Teflon ® tube (Figure 1 a-b), although isolated particles were found in some fragments (see Figure 1 c-d).However, with an 80-second sample loading time, the group of particles appeared in almost all portions examined (see Figure 2 a) and some From these observations, it can be concluded that (a) high sample loading times produce an excessive amount of complex and (b) a high concentration of complexing reagent for the KR length used (300 cm) caused the formation of precipitates and results in partial blockage of the reactor.This phenomenon could be defined as a 'saturation' effect which diminishes the adsorption efficiency of the Cd-DDC complex due to both the solid particles and the molecules of the reagent.They inhibit the molecular adsorption of the complex on the inner walls of the reactor since they partially occupy the surface of the tube.Moreover, a reduction of the waste flow rate was observed with the 80-second sample loading time, which is another sign of this 'saturation' effect.Therefore, to establish the appropriate sample loading time, it is important to take into account not only the maximum analytical response but the possible reactor 'saturation' which could lead to inefficient complex adsorption.
In order to clarify the nature of the particles formed, they were examined by X-ray fluorescence.The majority of them presented a similar elemental composition based on Cl, K, Ca, and Si (Figure 3).Only some particles contained Na and S. The presence of some of these elements such as Cl is easy to justify since it comes from hydrochloric acid and also from the buffer solution (ammonium chloride/ammonia).Sodium should be present in all particles, because the chelating reagent is used as sodium salt, but the absence of this element could be explained in terms of its small atomic mass which places it in the limit of detection of the technique.Sulphur is also a constituent of the chelating reagent but its signal is sometimes masked by the corresponding one from gold (which is present in great excess because it is used to cover the fragments of the reactor for the scanning electron microscopic observation).The origin of the other elements found (K, Ca, and Si) is difficult to explain.For example, K could be an impurity of the chelating reagent, but more studies need be performed to understand the process.
The inner walls of the reactor were also examined after the elution step, using the eluents ethanol and IBMK.When the reactor was loaded during the 60-second sampling time and elution was performed with ethanol, the observation of the inner surface revealed that it was almost completely clean, i.e., only a few very small particles were found in the portions examined (Figure 4 a).Using the same eluent and a sample loading time of 80 seconds, some isolated particles as well as several groups of particles were found (Figure 4 b-c).For these groups, two kinds of structures were observed: one structure was similar to that found in previous assays (Figure 4 b) and the other structure had a drop-like appearance (Figure 4 c).When IBMK was used as the eluent, the situation was only slightly different.With a sample loading time of 60 seconds, fewer particles of smaller size were found when eluting with IBMK rather than with ethanol (Figure 5 a).Increasing the sample loading time to 80 seconds, the particles found were slightly larger compared with those observed during the 60-second sample loading time (Figure 5 b-c).In brief, taking into account the efficiency of the elution, fewer particles appeared when elution was performed with IBMK than with ethanol, but some of the particles were larger.
The elemental analysis revealed that the composition of the particles found after elution was equal to those that appeared before elution.This fact indicates that the solid particles formed during the metallic complex formation were only partially washed away or dissolved by the eluent, independent of the solvent used (ethanol or IBMK).However, IBMK removed these particles more efficiently.The elemental analysis of the drop-like particles (see Figure 4 c) did not give any results and it can therefore be assumed that these structures are of an organic nature; however, further studies are required to elucidate their origin.To submit articles for publication, write or fax: Editor, Atomic Spectroscopy PerkinElmer Instruments 710 Bridgeport Avenue Shelton, CT 06484-4794 USA

Fig. 2 .
Fig. 2. Comparison between quantitative and semi-quantitative data obtained by ICP-MS in CRM MEDPOL-1A.*Ti is the only element that gives very low concentration in semi-quantitative mode compared to quantitative mode analysis.

Fig. 3 .
Fig. 3. Comparision between quantitative and semi-quantitative data obtained by ICP-MS in mollusks samples from Pina Bay, Recife, Northeastern Brazil.(A) All data presented in Table II.; (B) Data in the concentration range 5 to 500 mg kg -1 ;(C) Data for elements with concentration < 5 mg kg -1 .

Fig. 4 .
Fig. 4. Trace element distribution in mollusks from the Pina Bay area.Numbers 1, 2 and 3 refers to samples collected in March, June and September, respectively.Subfigures A to C correspond to semi-quantitative analysis for different elements according to their concentration.

Fig. 5 .
Fig. 5. Trace element distribution in mollusks from the Pina Bay area.Numbers 1, 2 and 3 refers to samples collected in March, June and September, respectively.Subfigures A to C correspond to semi-quantitative analysis for different elements according to their concentration.
Coal (bituminous), NIST-1635 Trace Elements in Coal (sub-bituminous), and NIST-1633b Trace Elements in Coal Fly Ash (U.S.Department of Commerce National Institute of Standards and Technology, Gaithersburg, MD USA) were employed to study the accuracy of the methods.Acetylene C-26 was used to heat the quartz cell, and argon N-50 purity (99.999%) was used to transfer the vapor to the atomizer (Carburos Metálicos, Barcelona, Spain).

a
Aqua regia + HF extraction: after bombs were cooled, 1 g of boric acid was added; then steps 3-4 from the microwave-acid extraction program were run.

Fig. 1 .
Fig. 1.Schematic of flow injection system used for the analysis of norfloxacin.P 1 and P 2 = Peristaltic pumps; S = sample; R = Fe(III) solution; F= screening agent NH 4 F solution; E = eluent; V = injection value; FAAS = flame atomic absorption spectrometer; A and B = cation exchange resin micro columns; L 1 and L 2 = reaction coils; W = waste.

Fig. 3 .
Fig. 3. Elemental analysis by X-ray fluorescence of the particles observed on the inner walls of the KR when Cd-DDC complex is formed.

TABLE I ICP-MS Results for CRM MA-MEDPOL Using Semi-quantitative and Quantitative Mode (expressed in µg kg -1 ) (Data expressed in a dry weight basis.)
*RD= Relative difference between semi-quantitative and quantitative results.

TABLE II Metal Concentration (expressed in µg kg -1 ) in Mollusks From Pina Bay and Relative Differences Using Quantitative (Q) and Semi-quantitative (SQ) ICP-MS Mode Analyses. All data expressed in dry weight basis (d.w.). *Note: these data are expressed in mg kg -1 (d.w).
Q=Quantitative analysis; SQ=Semi-quantitative analysis; RD(%)= Relative difference between quantitative and semi-quantitative analysis.

TABLE I Heating Program for the Determination of Cu in Digested Sediment Samples and Slurries (W-Rh permanent modifier and Pd + Mg(NO 3 ) 2 modifier were employed throughout)
a Sediment samples digested.b Sediment slurry.

TABLE II Recoveries (%) of Cu in Digests of Sediment Samples*
*The samples were spiked with 100, 300 and 500 pg of Cu, respectively.The value given is the mean ± standard deviation (n=5).Final acidity 0.5% (v/v) HNO 3 .Pd + Mg:

TABLE VII Accuracy of Methods Obtained for Different Reference Materials (n = 11)
a Non-certified value.b Information value.