JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY DECEMBER 199 1 VOL. 6 627 Cold Vapour Atomic Absorption Method for the Determination of Mercury in Iron(iii) Oxide and Titanium Oxide Pigments Using Slurry Sample Introduction lgnacio Lbpez Garcia Maria Jesus Vizcaino Martinez and Manuel Hernandez Cordoba* Department of Analytical Chemistry Faculty of Chemistry University of Murcia 30071 Murcia Spain A simple and rapid method for the determination of mercury in commercial iron(ii1) oxide and titanium oxide pigments combining a slurry sampling procedure with the cold vapour technique is described. The samples are suspended in water containing 0.02% m/v sodium hexametaphosphate and mercury vapour is generated from a hydrochloric acid medium by adding sodium tetrahydroborate(iii). A detection limit of 5 ng g-1 of mercury is achieved.Calibration can be performed using standard aqueous mercury solutions. Excellent agreement is found between the results of the slurry sampling procedure and those obtained for iron(iii) oxide pigments using lengthy conventional acid dissolution procedures. Keywords Mercury determination; iron(rrr) oxide pigment; titanium oxide pigment; slurry sampling; cold vapour In recent years the interest of many researchers has focused on the use of suspensions or slurries as a means of introducing solid samples into the atomizer of an atomic absorption spectrometer. Most of the papers to date have dealt with the determination of a number of elements in a wide range of materials including soils coal biological samples and inorganic pigments using electrothermal atom- ization.This is because for flame atomization purposes the atomization efficiency is markedly dependent on several factors for example a sufficiently low particle size is critical and comparatively few application~l-~ have been reported using this approach. Haswell e? aL5 have reported a procedure for the determination of cold acid-soluble arsenic in various matrices by slurrying samples in hydrochloric acid prior to hydride generation. Recently Madrid e? aP published a detailed paper on the determination of lead in foodstuffs and biological samples using hydride generation from slurried samples. Apart from the papers already mentioned as far as we know there are no reports dealing with the generation of mercury vapour from slurried samples.Interest in slurry- based procedures has increased as the length of time required for some dissolution procedures can be great and there is always the risk of analyte loss in pre-treatment of samples. The aim of this work was to study the determina- tion of mercury through the generation of mercuIy vapour from slurries prepared from iron(@ oxide or titanium oxide pigments. These commercial products meet the essential requirements for low particle size and their mercury contents are severely restricted by law. The procedure reported here gives a fast determination which is useful for routine analyses by avoiding dissolution steps. Experimental Apparatus A Perkin-Elmer Model 300 atomic absorption spectro- meter a Model MHS-10 hydride generation system and a Hewlett-Packard 3394-A integrator were used.Some exper- iments were performed using a Perkin-Elmer Model 1 lOOB atomic absorption spectrometer and the MHS-10 hydride generation system. All the measurements were made at the mercury wavelength of 253.7 nm using a conventional * To whom correspondence should be addressed. mercury hollow cathode lamp (Perkin-Elmer) operated at 4 mA with a band width of 0.7 nm. Background correction was not used. Reagents All reagents were of anlytical-reagent grade and were used without further purification. Doubly distilled water was used throughout. A stock solution containing 1000 pg ml-l of mercury was prepared by dissolving 1.3535 g of mercury(rr) chloride in 50 ml of concentrated hydrochloric acid and diluting to 1 1 with water.From this stock solution working standard solutions were freshly prepared by appropriate dilution with a solution containing 1% v/v nitric acid and 0.002% m/v potassium permanganate. Sodium tetrahydroborate was obtained from Fluka and used as a 4% m/v solution in water containing 1% m/v sodium hydroxide. This solution was prepared daily and filtered before use. Sodium hexametaphosphate (HMP) was obtained from Fluka. Slurry Procedure Although the pigments studied are sold as powders with low particle size in order to assure homogeneity the samples were ground using a ball mill for 10 min and then dry sieved using a 45 pm sieve (325 mesh). The small fraction (less than 0.5%) containing larger particles was discarded. A 0.5 ml volume of 2% m/v HMP solution was added to an accurately weighed amount of sieved sample (1-5 g) the solution was made up to 50 ml with water and stirred magnetically for a minimum of 10 min.While the suspen- sion was being stirred an aliquot (1-5 ml) was placed into the vessel of the hydride generation system. Concentrated hydrochloric acid [4 ml for iron(1n) oxide slurries or 1.5 ml for titanium oxide slurries] was added and the solution was made up to 10 ml with water. The plunger of the MHS-10 system was depressed for 8 s (this delivers 4 ml of the tetrahydroborate solution) and the peak height was mea- sured. Aqueous standard solutions of mercury (25-200 ng) and a reagent blank was used for calibration under the same experimental conditions. Aciddissolution Procedure For comparison purposes samples of iron(rxr) oxide pig- ments were analysed as follows 2-10 g of sieved sample628 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY DECEMBER 199 1 VOL.6 0.05 were weighed in a 100 ml covered conical beaker and 30 ml of a mixture of concentrated nitric and hydrochloric acids (1 + 3 v/v) were added. The samples were heated gently (50-60 "C) until total dissolution. It was necessary to add several 5 ml portions of the acid mixture in order to compensate for evaporation. After dissolution the liquid was made up to 100 ml using 6 mol dm-3 hydrochloric acid. Aliquots of this solution were placed into the reaction vessel the acidity was adjusted to 4.8 mol dm-3 hydro- chloric acid in a final volume of 10 ml and the mercury content was determined using the standard additions method.A reagent blank was also analysed to correct the results. - C n- I D m I I I I I I I Results and Discussion Previous experiments have shown that when tin(@ chloride was used as the reducing agent the profiles of the absorbance-time signals obtained from slurried samples were wider and lower than those obtained when sodium tetrahydroborate was used. Therefore all subsequent exper- iments were carried out using sodium tetrahydroborate solution. From an experimental point of view the effect produced by acids must be carefully considered as several reports have indicated significant differences in the generation of mercury vapour depending on the nature and concentra- tions of the acid media used. In practice sufficient acid must be present in the reaction vessel to neutralize sodium tetrahydroborate and to provide an excess as the reduction does not take place in an alkaline medium.A number of 2% iron(Ir1) oxide slurries were prepared as described under Experimental and the effect of sulphuric nitric and hydrochloric acids was studied. Different amounts of the acids were added to aliquots of the slurried samples in order to produce in a final volume of 10 ml the concentrations indicated on the x-axis of Fig. 1. Next 4 ml of the sodium tetrahydroborate solution were added using the hydride generation system and the absorbance-time profiles for mercury were obtained. When sulphuric or nitric acid was used very similar peak areas were obtained. However the peak heights obtained from samples in sulphuric acid media were higher than those found when nitric acid was used.This difference appears to be in agreement with the results of Koirtyohann and Khali17 who found that sulphuric acid but not nitric acid has a significant effect on the partition constant K defined as the ratio of the concentration of mercury in air to that in a liquid. Furthermore it has been suggested that the oxidiz- ing properties of nitric acid hamper the reduction of the determinant,* thus leading to wider and lower peaks. When hydrochloric acid was used the peak areas were also very similar to those found in sulphuric and nitric acid media but the peak height was the highest obtained. As can be seen from Fig. 1 the peak height is markedly dependent on the hydrochloric acid concentration and this behaviour is different from that shown by aqueous standard mercury(r1) solutions (curve D).The noticeable decrease in the peak height for the slurried samples at low hydrochloric acid concentrations could be attributed to the chemical form of mercury that is present within the iron(@ oxide samples. On the other hand the values for the reagent blanks were lower in hydrochloric than in sulphuric acid media (absor- bances of 0.018 and 0.071 respectively). Therefore a 4.8 mol dm-3 hydrochloric acid medium was selected as being adequate to generate mercury vapour from the iron(1rr) oxide slumed samples. A similar study was carried out using hydrochloric acid media for titanium oxide samples. In this instance 5% slurries were used and nearly constant peak areas were obtained within a wide range of acidities.As shown in Fig. 2 0.20 I I 0.15 z e 0.10 z 9 (0 5 I -7 * A 0 1 2 3 4 5 6 7 8 Hydrochloric and nitric acid concentration/mol dm-3 I I I I I I I 0 0.5 1 1.5 2 2.5 3 3.5 4 Sulphuric acid concentration/mol dm-' Fig. 1 Effect of acid concentration on generation of mercury vapour from slurried iron(@ oxide samples A hydrochloric acid; B sulphuric acid; C nitric acid; and D as for A but for aqueous mercury solution B 0 - - * w - I I 0 1 2 3 4 5 6 Hydrochloric acid concentration/mol dm-' Fig. 2 Effect of hydrochloric acid on generation of mercury vapour A slumed titanium oxide sample; and B reagent blank 0.20 I 0.15 Q C ; 0.10 9 z - / B A maximum peak heights were obtained for 1.5-2.5 mol dm-3 acid concentrations and so a 1.8 mol dm-3 hydrochloric acid medium was chosen.The effect of sodium tetrahydroborate concentration was studied using different solutions in the 0.4-6% m/v range. The solutions being studied were delivered to the reaction vessel which contained the slurried sample at the chosen acidity conditions in a total volume of 10 ml by depressing the plunger of the hydride generation system. When the tracing on the recorder indicated that the maximum signalJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY DECEMBER 199 1 VOL. 6 629 Table 1 Slopes of the standard additions calibration graphs obtained from iron(rr1) oxide sample Slurry Acid dissolution Sample P(%) Vt/ml Slope$/ 1 0-3 A ng-' m$/g Vt/ml Sl0pe$/10-~ A ng-I 1 2 5 1.78 f 0.05 2 5 1.78 f 0.04 2 2 2 1.82 k 0.08 2 5 1.82 f 0.03 3 2 2 1.85 k 0.04 2 5 1.88 ? 0.05 4 2 2 1.80 & 0.05 1 4 1.79 f 0.04 5 2 5 1.86 k 0.1 2 5 1.82 f 0.07 9 5 1.78 f 0.09 6 9 5 1.72 k 0.07 * Slurry percentage.$Mean of three valueskSD (standard deviation); slope for aqueous mercury 1.78 x $ Mass of iron(m) oxide taken in the acid-dissolution procedure. Volume of suspension or solution added to reaction vessel. A ng-I. Table 2 Slopes of the standard additions calibration graphs obtained from titanium oxide samples Original samples Calcined samples Sample P(%) I/t/rnl Sl0pe$/10-~ A ng-' P(%) Vf/ml Slope$/ 1 0-3 A ng-* 1 10 5 1.55k0.04 10 5 1.54 & 0.09 2 10 5 1.49 f 0.06 10 5 1.49 f 0.10 3 5 2 1.56 k 0.06 10 5 1.50k0.08 4 10 5 1 S O f 0.05 10 5 1.55f0.07 * Slurry percentage. t Volume of suspension added to reaction vessel.t Mean of three values k SD; slope for aqueous mercury 1.54 x 1 0-3 A ng-'. had been attained the plunger was released. As can be seen in Fig. 3 a 4% m/v solution of the reducing agent was necessary in order to obtain the maximum peak height. Similar results were obtained for slurried titanium oxide samples. Calibration Sensitivity and Repeatability In order to perform an optimum calibration it was necessary to ascertain whether mercury was entirely or partially removed from the slurried solids and the effect of varying the mass of solid introduced into the reaction vessel. Mercury vapour was generated using different volumes of a 2% m/v iron(rI1) oxide slurry prepared from a sample containing 0.91 ,ug g-l of mercury. As 4 ml of concentrated hydrochloric acid must be used in a final volume of 10 ml in the reaction vessel six experiments were performed using 1,2 3,4 5 and 6 ml of the slurry.A plot of the peak height against the volume of slurry gave a straight line showing a correlation coefficient r= 0.9995. Another set of experiments was performed using slurries prepared from an iron oxide sample with a very low mercury content (0.065 pg g-*). Five slurries covering the 2-10% range were prepared and 5 ml of each were placed in the reaction vessel. A straight line (r=0.9987) was again obtained suggesting the absence of a matrix effect due to the mass of solid present in the reaction vessel. Similar experiments were carried out using slurries prepared from titanium oxide samples and straight lines were also obtained. Because the maximum content of mercury in these samples is fixed by law at 2 fig g-l slurries containing more than 10% m/v of solid were not prepared as the procedures were considered sufficiently sensitive.These experiments suggest that a calibration using aque- ous solutions of mercury should be valid. To confirm this six iron(1n) oxide samples from two manufacturers were dissolved in acids and their mercury content determined as described under Experimental. It is important to point out that when the analysis was carried out on these solutions a practical problem was encountered for final hydrochloric acid concentrations of (3 mol dm-3 in the reaction vessel. Under these experimental conditions a dark precipitate was seen when the reducing agent was added and the slopes of the standard additions graphs were different to those obtained when aqueous standards of mercury were ana- lysed.The problem was avoided by using a 4.8 mol dm-3 hydrochloric acid medium. A number of slumes were prepared from these samples and standard additions Cali- bration graphs using several volumes of the slurried samples were again obtained. As can be seen in Table 1 where the results are summarized the slopes of the standard additions calibration graphs both for solutions and slurries are very similar and virtually identical to the slopes of calibration graphs for aqueous solutions of mercury proving that direct calibration with aqueous standards is valid. A different approach was followed in order to prove that a calibration for the analysis of titanium oxide pigments can also be made using aqueous standards.No reliable procedure with the exception of those based on the volatilization of mercury is available for the determination of the analyte. In order to overcome this the samples were calcined at 400 "C for 2 h. Slurries were then prepared from these freshly calcined samples and applying the procedure reported here no measurable amounts of mercury were found. Next standard additions calibration graphs were obtained by using both the previously calcined samples and the original titanium oxide pigments. The slopes of these graphs which are summarized in Table 2 demonstrate that as for iron oxide samples direct calibration with aqueous standards is also valid.Tables 3 and 4 show that the results for the determination of mercury content obtained using the proposed procedure agree with those found using acid dissolution for iron(xn) oxide samples and the indicated approach for titanium oxide.630 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY DECEMBER 199 1 VOL. 6 Table 3 Comparison of the results for mercury obtained from six iron(n1) oxide samples analysed by the acid dissolution and slurry procedures n = 3 Sample Acid dissolution/ Slurry 1 0.169 0.170 2 0.860 0.905 3 0.600 0.596 4 0.765 0.789 5 0.510 0.507 6 0.065 0.070 Pg g-' In order to evaluate the repeatability of the procedure four different iron(r1r) oxide samples were taken and five 4% m/v slurries were prepared from each one. Next ten successive measurements were obtained for each of the 20 slurries.The relative standard deviations (RSDs) were in the 1.6-6.4% range. The same method was followed for titanium oxide samples using 10% m/v slurries and RSD values in the 1.4-7.1 % range were obtained. The detection limit for mercury calculated on the basis of 20 for 5% m/v iron(rr1) oxide slurries using 4 ml of suspension in the reaction vessel was 5 ng g-l. A similar figure (7 ng g-l) was found when using 6 ml of 10% m/v titanium oxide slurries. Table 4 Comparison of the results for mercury obtained from four titanium oxide samples analysed by the proposed method n=3 Contenthg g-' Sample Aqueous calibration Standard additions 1 12 2 24 3 501 4 67 10 26 506 65 In spite of the fact that the results given in Tables 3 and 4 indicate very low mercury contents all of them below the limit imposed by law it is surprising that such levels can exist in commercially available samples which are submit- ted to thermal treatment during the industrial process used to obtain them.Serious risks of contamination by mercury during the storage of samples have been reported? and so additional experiments were carried out to ascertain whether any of the analyte is released during such a process. The experiments showed that when iron oxide pigments were vigorously shaken in cold 0.3 mol dm-3 nitric acid solution containing 0.0 1 mol dm-3 potassium permanga- nate about 75% of the total mercury was released into the solution suggesting that a great part of the analyte is easily available as a consequence of superficial contamination. The authors are grateful to the Spanish Comisi6n Intermin- isterial de Ciencia y Tecnologia (CICYT) (Project 87-0053) for financial support. 1 2 3 4 5 6 7 8 9 References Langmyhr F. J. and Wibetoe G. Prog. Anal. At. Spectrosc. 1985 8 193. Sample Introduction in Atomic Spectroscopy ed. Sneddon J. Elsevier Amsterdam 1990 ch. 3. Baxter D. C. and Frech W. Fresenius 2. Anal. Chem. 1990 337 253. Miller-Ihli N. J. Fresenius 2. Anal. Chem. 1990 337 271. Haswell S. J. Mendham J. Butler M. J. and Smith D. C. J. Anal. At. Spectrom. 1988 3 731. Madrid Y. Bonilla M. and CAmara C. J. Anal. At. Spectrom. 1989 4 167. Koirtyihann S. R. and Khalil M. Anal. Chem. 1976,48 136. Hon. P. Lau O. and Wong H. Anal. Chem. 1983 108 64. Welz B. Atomic Absorption Spectrometry VCH Weinheim Paper No. I /02 7200 Received June 7th 1991 Accepted August 5th 1991 1985 pp. 244-248.