Determination of Silicon in Biological Tissue by Electrothermal Atomic Absorption Spectrometry Using Slurry Sampling of Original and Pre-ashed Samples M. HORNUNG AND V. KRIVAN* Sektion Analytik und Ho� chstreinigung, Universita�t Ulm, D-89069 Ulm, Germany Two methods for the determination of silicon in biological step and the use of complex modifier mixtures represent serious tissue by electrothermal atomic absorption spectrometry using sources of contamination limiting the performance of the the slurry sampling technique are described.In one of them, a method. slurry of the tissue and, in the second, a slurry of the ash The above contamination problems can be avoided by direct obtained by separate thermal pre-treatment in an ashing analysis of solid samples. However, the dc arc atomic emission furnace was introduced into the graphite furnace. The second spectroscopy used by Indraprasit et al.2 as well as direct method proved to be superior regarding the elimination of neutron activation methods reported by Velandia and matrix interferences.Optimum sensitivity was obtained by Perkson26 and Ward and Mason27 suVer from low achievable using a mixture of palladium nitrate and magnesium nitrate as LOD. Guzzi et al.28 reported a radiochemical neutron actimodifier. The silicon contents determined were between about vation (RNAA) method ensuring a low contamination risk 3 and 14 mg g-1 and they were compared with results obtained too, but it requires a complex step for the selective separation by other methods and laboratories.The limits of detection of of the indicator radionuclide 31Si from all other radionuclides the direct method and the method involving pre-ashing were produced by irradiation of the sample. found to be 0.2 and 0.03 mg g-1, respectively. In recent years, the slurry sampling technique has increasingly been applied to the analysis of various sample types, Keywords: Silicon determination; biological tissue; slurry including biological matrices, by ETAAS.29,30 DiVerent aspects sampling; electrothermal atomic absorption spectrometry determining the performance of slurry ETAAS were studied comprehensively by Miller-Ihli.31–34 Considering the problems In recent years, the role of trace concentrations of silicon in connected with silicon determination by solution methods, biological systems, especially in the human body, has become especially the high risk of contamination and loss by volatilof increasing research interest.1 Some investigations have ization, slurry sampling ETAAS seemed to us to be very shown that patients with chronic renal failure have elevated promising because it avoids the sample digestion step and silicon levels in blood plasma and various tissues as well as at the same time oVers the advantages of a solution technique decreased silicon excretion through urine.2–9 In the brain tissue regarding sample introduction. of patients with Alzheimer’s disease, silicon has been found in In this work, two slurry sampling techniques for the determithe form of aluminosilicates in senile plaques and colocalised nation of silicon in biological tissue by ETAAS have been with aluminium in neurofibrillar tangles.10–14 In both cases, developed.The first uses slurries of powdered tissues while the however, the mode of silicon action has not yet been clarified. second technique involves ashing of the samples and prep- Furthermore, silicones are frequently used as breast prostheses.aration of slurries of the remaining ash. The results obtained Patients with such implants show increased concentrations of by these two methods are compared with those obtained by silicon in blood15,16 and in tissues.17 In view of these obser- other methods within an interlaboratory collaborative study. vations, there exists an urgent need for reliable methods for the determination of silicon levels in biological fluids and tissues.Several methods have been reported for the determination EXPERIMENTAL of silicon in serum and urine18–23 and in biological tissues.2,24–28 Krushevska and Barnes24 determined silicon in food by ICP- Instrumentation AES. After sample dissolution using HNO3, H2O2 and HF, A Perkin-Elmer (U� berlingen, Germany) Model 4100 ZL atomic the surplus acid was neutralized by addition of water-soluble absorption spectrometer, equipped with a THGA graphite tertiary amines in order to eliminate the attack on the quartz furnace, an AS-70 autosampler and a USS-100 slurry sampler, parts of the ICP-AES apparatus and to minimize the contamiwas used.Background correction was performed using the nation risk. They achieved an LOD for silicon of 75 ng g-1 longitudinal inverse Zeeman eVect. Perkin-Elmer THGA and checked this method by comparing the results with those graphite tubes with and without end caps were used. obtained in their laboratory by ICP-AES after a complex Slurry preparation was carried out in a clean bench sample fusion with LiBO2.Zhuoer25 proposed an ETAAS PBO/H13/6 (Meissner & Wurst, Stuttgart, Germany). method for the determination of silicon in bones and soft Suspensions were pre-treated in a Sonorex RK 255H ultrasonic tissues involving sample digestion with concentrated nitric bath (Bandelin Electronic, Berlin, Germany). For the determi- acid. As chemical modifier, a mixture of La(NO3)3, CaCl2, nation of sampling eYciency, an electronic micro-balance NH4H2PO4 and Na4EDTA was used for soft tissue digests, (Sartorius, Go� ttingen, Germany) was used.Sample ashing was whereas La(NO3)3 and tartaric acid disodium salt were applied performed using 15 ml nickel crucibles and an ashing furnace to the analysis of bone digests. However, due to the extraordinarily high overall concentration of silicon, both the digestion (Heraeus, Karlsruhe, Germany). Journal of Analytical Atomic Spectrometry, October 1997, Vol. 12 (1123–1130) 1123Samples and Reagents throughout. Calibration standards were prepared by dilution of a stock solution (1000 mg l-1 Si in 5 mol l-1 aqueous The pork liver samples L1 and L2 and the pork fillet sample NaOH, Merck, Darmstadt, Germany) with 5% HNO3. The F1 were prepared and distributed to the participants in the nitric acid (65%, pro analysi, Merck) was purified by subinterlaboratory collaborative study by Novartis (Basel, boiling distillation.Magnesium, palladium and calcium nitrates Switzerland). Samples (1 kg each) of meat minced by a butcher were of Suprapur quality (Merck). Triton X-100 (Merck) was were divided into two 0.5 kg portions, spread onto two stainless used as a surfactant for preparation of slurries of original steel trays and moistened using 20 ml of high-purity water for samples. samples L1 and F1 and using a freshly shaken mixture of 2 ml of polysiloxane spike in toluene and 18 ml of water for sample L2.After 3 days of freeze-drying using the Edwards Procedures ‘Supermodulo’, 18 l ice capacity (delivered by N. Zivy & Cie., Preliminary procedure for direct slurry sampling Oberwil, Switzerland), the two portions of each sample were combined and pre-cut into small pieces using the Cut-O-Mat, For the preparation of slurries, between 20 and 70 mg of pork Type H4 (Kneubu� hler AG, Luzern, Switzerland). Then they liver (samples L1 and L2), and between 20 and 60 mg of pork were powdered in a Retsch-Mill, type MS03, which has milling fillet (sample F1) were mixed with 10 ml of 5% nitric acid components made of ZrO2 (Retsch KG, Haan, Germany).The containing 0.004% Triton X-100 in a 15 ml polystyrene vessel. particle size distribution of the samples is shown in Fig. 1. It Before use, the vessels were rinsed twice with 5% nitric acid was determined by the Fraunhofer diVraction method by and ultrapure water and dried under a clean bench.The Novartis. suspensions were pre-treated in an ultrasonic bath for about Ultrapure water, obtained by using the Milli-Q System 30 s to accelerate wetting of the particles. (Millipore GmbH, Neu-Isenburg, Germany), was used The beakers containing the slurries were used directly for autosampling. Prior to pipetting, homogenization was performed by ultrasonic agitation with the USS-100 for 20 s at about 7 W. Fitres of the aqueous modifier solution (containing 20 mg of palladium as nitrate, and 20 mg of magnesium nitrate) and 20 ml of the sample slurry were introduced sequentially into the atomizer. For calibration using the standard additions technique, the slurries were spiked twice sequentially with 1 mg of silicon (10 ml of a standard solution containing 100 mg l-1 of silicon).The optimized temperature programme and the instrumental parameters used are summarized in Table 1. Developed procedure based on sample pre-ashing The nickel crucibles (15 ml ) used for sample ashing were cleaned by standing them for about 30 min in concentrated hydrofluoric acid. Then they were rinsed three times with Fig. 1 Particle size distribution of the samples analysed. ultrapure water, kept for 15 min in nitric acid (151) and rinsed Table 1 Temperature programmes and instrumental parameters used in the ETAAS analysis of untreated and pre-ashed samples T emperature programme 1 used for slurries of untreated samples— Ramp Hold Argon flow/ Step Temperature/°C time/s time/s ml min-1 Drying 110 1 60 250 130 60 60 250 Charring 300 60 10 50 450 60 10 50 750 60 10 50 1300 10 20 250 Atomization 2400 0 5 0 Cleaning 2600 1 5 250 T emperature programme 2 used for slurries of pre-ashed samples and for standard silicon solutions— Ramp Hold Argon flow/ Step Temperature/°C time/s time/s ml min-1 Drying 110 1 20 250 130 5 30 250 Charring 1100 10 20 250 Atomization 2400 0 5 0 Cleaning 2600 1 5 250 Instrumental parameters— Wavelength 251.6 nm Slit width 0.2 nm Source Hollow cathode lamp Lamp current 40 mA Read 5 s Signal mode Peak area Sample volume 20 ml Modifier volume 5 ml 1124 Journal of Analytical Atomic Spectrometry, October 1997, Vol. 12again three times with ultrapure water. Finally, they were dried subsequently. The maximum slurry concentration applicable for the samples L1 and L2 was about 0.7% m/v and for the under a clean bench. Portions of about 200–500 mg of the samples L1 and F1, sample F1 about 0.6% m/v.It is well known that in analyses involving slurry sampling and 200 mg of the sample L2, were placed in the crucibles and mixed carefully with 10% m/m Mg(NO3)2 6H2O. After being the particle size may aVect both the accuracy and the precision. As can be seen in Fig. 1, the particle size of the samples covered with a nickel lid, the crucibles were heated in an ashing furnace using the temperature programme given in investigated is distributed over a wide range.Therefore, the sampling eYciency had to be investigated thoroughly. For this Table 2. The duration of the final step depends on the sample portion, ranging from 10 min for 200 mg to 25 min for 500 mg. purpose, 20 ml of the slurry were pipetted 25 times by the autosampler into pre-freeze-dried and weighed autosampler The resulting ash (20–50 mg) was transferred quantitatively into a 15 ml polystyrene vessel and mixed with 10 ml of 5% cups. Then the cups were freeze-dried for 3 days and weighed again.The sampling eYciency was calculated by comparison nitric acid. The suspensions were pre-treated in an ultrasonic bath for about 5 min to disintegrate large particle agglomerates. of the sample mass weighed with that corresponding to sample mass in 500 ml slurry. For a kind of ‘blank control’, empty The beakers containing the slurries were used directly for autosampling. Homogenization was performed by magnetic cups as well as cups manually charged with 500 ml of the slurry medium and about 3 mg of sample, respectively, were processed stirring.The aqueous modifier solution (5 ml containing 20 mg of palladium as nitrate) and the sample slurry were introduced simultaneously. It was found that the mass of the empty cups and the cups charged with sample was constant. The sampling sequentially into the graphite furnace. The volume of the sample slurry injected was 20 ml for the samples L1 and F1 eYciency was calculated from the diVerence in the mass of the cups loaded with the slurry and with the pure slurry medium.and, due to its higher silicon content, 10 ml for the sample L2. The optimized temperature programme and the instrumental For the samples L1, L2 and F1, sampling eYciencies of 101±2, 99±1 and 97±6%, respectively, were determined from six parameters used are summarized in Table 1. Before use, the polystyrene vessels were cleaned by the replicates. Thus, it is rather surprising that, in spite of the extraordinarily broad particle size distribution of the materials procedure described above for the first method.To clean the magnetic stirring bars, they were kept for about 30 min in (see Fig. 1) and the relatively small id of the Perkin-Elmer standard pipette of approximately 300 mm, excellent sampling concentrated hydrofluoric acid and then rinsed twice with 5% nitric acid and ultrapure water. Finally, they were dried under eYciencies were obtained.A possible reason might be a decomposition eVect of the nitric acid leading to reduction of a clean bench. the particle size. The much broader particle size distribution of the sample F1 is obviously the reason for the higher standard RESULTS AND DISCUSSION deviation of the sampling eYciency compared with the two other samples. Optimization of the Procedures Poor sensitivity was obtained when silicon was atomized Direct slurry sampling from the sample without addition of a chemical modifier.Calcium nitrate,21,35 magnesium nitrate,36 palladium as Neither water nor 5% nitric acid was suitable for preparation nitrate,37,38 as well as its metallic form39,40 and a mixture of of slurries of untreated samples, because the particles were not palladium nitrate and magnesium nitrate41 have been used as wetted suYciently. For this reason, the suitability of several modifiers for silicon in various matrices. From a comparison suspension media with the sample L2 (containing 1.2 ng Si in of the integrated absorbances and signal shapes in Fig. 2, the 20 ml of slurry) was tested. Using ethanol, dioxane and a superiority of a mixture of palladium and magnesium nitrate mixture of 5% nitric acid and 0.004% Triton X-100 as suspenas a chemical modifier for the determination of silicon in sion media, blank values of 0.010, 0.022 and 0.008, respectively, biological tissue is apparent. In this work, palladium pre- and blank corrected integrated absorbances of 0.079, 0.076 reduced in the graphite tube at 1000 °C and palladium nitrate and 0.094, respectively, were obtained.The increase in the were tested. As there was no significant diVerence in the integrated absorbance using the mixture of nitric acid and sensitivity for silicon obtained with pre-reduced palladium and Triton X-100 as suspension medium can be explained by the with palladium nitrate, the latter was chosen for subsequent attack of the acid on the tissue, i.e., the partial digestion of the use as it is superior with respect to time consumption. sample reduces matrix eVects.For aqueous silicon standard, addition of magnesium nitrate Within a concentration of nitric acid of 1–5%, the integrated to palladium nitrate did not influence the sensitivity for silicon. absorbances showed no significant diVerences. However, the However, in processing sample slurries, the mixture of 20 mg maximum applicable slurry concentration for the sample L2 of palladium and 20 mg of magnesium nitrate proved to be the was reduced by about 30% when the acid concentration was optimum modifier.The addition of magnesium nitrate decreased from 5 to 3%. In order to achieve lower LOD, the improved the peak shape and increased the sensitivity by mixture of 5% nitric acid and 0.004% Triton X-100 was used about 20%. Using end-capped tubes, a further improvement of the sensitivity for silicon by a factor of about 2.5 could be Table 2 Temperature programme for ashing the samples in an achieved for both the aqueous standards and the sample ashing furnace slurries as compared with tubes without end caps.The silicon standard solutions could be processed by using Ramp Hold the short temperature programme 2 (see Table 1). However, Temperature/°C time/min time/min applying this short temperature programme to sample slurries 100 10 10 led to a considerable reduction of the silicon absorbance signal 150 10 10 compared with that obtained by a longer step-wise pyrolysis, 200 10 45 although application of short one-step pyrolysis was reported 230 10 45 for the determination of various elements in biological tissues 270 10 45 300 10 45 by slurry sampling ETAAS.42 We assume that the observed 350 10 10 signal depression using the short programme 2 is caused by 400 10 10 losses of silicon during the violent pyrolysis process and its 450 10 10 incomplete atomization.Fig. 3 shows the dependence of the 700 45 10–25 integrated absorbances of silicon for the samples L1 and F1 Journal of Analytical Atomic Spectrometry, October 1997, Vol. 12 1125Fig. 2 Influence of various chemical modifiers on the absorption signal of silicon (black line) and the background signal (grey line) in processing 96 mg of sample L2 as slurry (1.4 ng Si) (a) without modifier, (b) with 20 mg calcium nitrate, (c) with 20 mg magnesium nitrate, (d) with 20 mg palladium (as nitrate), (e) with 20 mg pre-reduced palladium and ( f ) with 20 mg palladium (as nitrate) and 20 mg magnesium nitrate. In all cases, the respective blank value was subtracted from the integrated absorbance value.PA=peak area; PH=peak height. temperature programme could be elaborated. Therefore, the influence of oxygen as an ashing aid on the pyrolysis behaviour of this material was tested. However, including an oxygen ashing step at 300 and 450 °C (ramp 1 s, hold 30 s) in temperature programme 1 led to no improvement of the integrated absorbance (see Table 4).The oxygen ashing at 450 °C caused a significant worsening of the absorption signals after five runs (see Fig. 4), indicating that an unfavourable modification of the platform surface was taking place. Therefore, a further increase of the oxygen ashing temperature was not considered appropriate. When programme 1 was applied, the graphite tubes could be used for up to about 150 atomization cycles.The increase of the id of the end-caps led to a decrease of sensitivity, indicating the end of the graphite tubes’ lifetime. Fig. 3 Dependence of the integrated absorbances of silicon in the samples L1 (0.5 ng Si) and F1 (1.0 ng Si) on the mode of pyrolysis. The capitals A–F stand for the applied temperature programmes listed Procedure Based on External Ashing of Samples in Table 3. Because of the limited application of the direct slurry sampling discussed above to liver samples, eVorts were directed to the on the mode of pyrolysis.It can be seen that the use of temperature programmes with more and more stepwise pyro- development of a method applicable to determination of silicon in all biological tissues. The addition of magnesium nitrate to lysis led for sample L1 (sample L2 behaved similarly) to a stabilization of the integrated absorbance, indicating the the sample before ashing was necessary to obtain ash quality suitable for preparation of slurries; otherwise the ash was too achievement of optimized pyrolysis conditions.Thus, the eVects causing interferences in the determination of silicon could be coarse-grained and stable slurries could not be obtained. When slurries were prepared from the ashed samples, in contrast to minimized by appropriate drying and pyrolysis of the sample and by reduction of the argon flow during the pyrolysis. For slurry sampling applied directly to the powdered tissue, the presence of Triton X-100 in the suspension medium had no the thermal pre-treatment of the liver samples, the temperature programme D in Table 3 and Fig. 3 (identical with programme influence on either the sensitivity or the reproducibility. Consequently, for the preparation of these slurries, 5% HNO3 1 in Table 1) proved to be the optimum one, with respect to the sensitivity as well as to the analysis time. However, with alone was used. Furthermore, the mode of slurry homogenization had only a slight influence on the sampling error.With the sample F1, no satisfactory stabilization of the integrated absorbance could be achieved within a meaningful pro- magnetic stirring the integrated absorbances were about 5% higher compared with ultrasonic agitation, presumably due to longation of the pyrolysis, and therefore no well optimized 1126 Journal of Analytical Atomic Spectrometry, October 1997, Vol. 12Table 3 Temperature programmes used in optimization of the experimental conditions for ETAAS determination of silicon T emperature programme A As in temperature programme 2, Table 1 T emperature programmes B and C— Ramp Hold Argon flow/ Step Temperature/°C time/s* time/s* ml min-1 Drying 110 1 30/60 250 130 30/60 30/60 250 Charring 300 30/60 10 250 450 30/60 10 250 750 30/60 10 250 1300 10 20 250 Atomization 2400 0 5 0 Cleaning 2600 1 5 250 * Temperature programme B, 30 s, and C, 60 s.T emperature programme D As in temperature programme 1, Table 1 T emperature programmes E and F— Ramp Hold Argon flow/ Step Temperature/°C time/s* time/s* ml min-1 Drying 110 1 60/99 250 130 60/99 60/99 250 Charring 150 60/99 10 50 250 60/99 10 50 350 60/99 10 50 450 60/99 10 50 550 60/99 10 50 700 60/99 10 50 850 60/99 10 50 1300 10 20 250 Atomization 2400 0 5 0 Cleaning 2600 1 5 250 * Temperature programme E, 60 s, and F, 99 s.Table 4 Integrated absorbances for processing 100 mg of F1 slurry (1.0 ng of Si) without and with oxygen ashing (n=4) Mode of treatment Temperature programme 1 +O2 at 300 °C +O2 at 450 °C Integrated absorbance 0.022±0.002 0.024±0.003 0.022±0.002 more continuous slurry mixing during pipetting and crushing medium is evidently 100%.Thus, due to the leaching eVect, the actual sampling eYciency for all slurries was higher than of the coarse ash particles by the magnetic stirring bar. The slurry sampling eYciency with magnetic stirring was 99%. Applicability to both tissue materials investigated seems to determined for all three samples by comparing the amount of sample actually dispensed with the calculated value.For this be the main advantage of the slurry sampling technique using ashed samples in comparison with that using untreated purpose, 30 consecutive pipetting and drying steps were performed with a slurry of known concentration and the weight samples. For chemical modification, 20 mg of palladium (as nitrate) diVerence between the empty tube and the tube containing the dry sample was determined by using a micro-balance.For four per atomization were used. Addition of magnesium nitrate to the modifier solution was not necessary because this reagent replicates of the samples L1, L2 and F1, sampling eYciencies of 92±3, 92±4 and 99±5%, respectively, were obtained. This was added to the sample in the sample ashing step. The same temperature programme was used for the sample ash slurries procedure was applicable to the pre-ashed samples as these, in contrast with the untreated samples, did not decompose at the and the matrix-free standard solution (see Table 1).For slurries of pre-ashed samples, it was not necessary to use the long drying temperature used. However, due to the leaching eVect, the actual sampling temperature programme as it was for slurries of the untreated samples as the ashing of the samples was a separate analysis eYciency related to the analyte for the samples L1 and L2 is much higher than that determined by the above procedure.stage. With the short temperature programme, the diameter of the orifice in the end caps of the tubes started to increase only For the determination of the leaching factor, sample ash slurries were prepared and the particles were allowed to sink after about 450 runs. Thus, omission of ashing the tissue samples from the temperature programme by processing on the ground. From the determination of silicon in the liquid fraction and of the total silicon in the slurry, it was found that pre-ashed samples considerably increased the lifetime of the graphite tube as compared with processing untreated samples. 87±6, 84±4 and 89±7% (n=4) for samples L1, L2 and F1, respectively, were extracted from the ash particles into the liquid phase. From the partial eYciency for the particles and Calibration and Sample Analysis the liquid phase and the percentage distribution of silicon in these two phases, the total eYciency was calculated.The Both calibration via a calibration curve using aqueous standards as well as via standard additions was tested. Although sampling eYciency of silicon extracted in the suspension Journal of Analytical Atomic Spectrometry, October 1997, Vol. 12 1127Fig. 6 Absorption signals of silicon for (a) L2 slurry (1.5 ng of Si) of untreated sample: (A) without addition of silicon standard, (B) spiked with 2.0 ng of silicon, and (C), spiked with 4.0 ng of silicon; and (b) L2 slurry (3.1 ng of Si) of pre-ashed sample: (A) without addition of silicon standard, (B) spiked with 1.0 ng of silicon, and (C) spiked with 2.0 ng of silicon.In all cases, the respective blank value was subtracted from the integrated absorbance value. PA=peak area. Table 5 Silicon concentrations (mg g-1) in the samples L1, L2 and F1 obtained by slurry ETAAS Untreated samples Pre-ashed samples (n=6) (n=4) L1 3.3±0.4 2.5±0.2 L2 14±2 14.2±0.7 F1 5±1 10±4 For the liver sample L2, the results of the two methods are in excellent agreement and for the liver sample L1, taking into Fig. 4 Silicon absorption signals (black line) and background signals account the low silicon content, the agreement can be con- (grey line) obtained in processing 100 mg of sample F1 as slurry (1.0 ng sidered good. For the 12.5 mg g-1 of silicon spiked to the Si); (a) with temperature programme 2 and (b) with temperature sample L1 to prepare the sample L2 (see the section Samples programme 2 extended by ashing with oxygen at 450 °C, fifth replicate.and Reagents), recoveries of 11±2 and 12±1 mg g-1 were PA=peak area; PH=peak height. obtained using untreated and pre-ashed samples, respectively, for slurry sampling ETAAS. Thus, these results are in satisfacthe charring curves (see Fig. 5) and the absorption signals (see tory accordance, too. However, the contents of silicon in the Fig. 6) indicate similar behaviour of silicon in aqueous standard sample F1 determined by using the two slurry sampling solution and in unspiked and spiked slurries of untreated and techniques diVer considerably.Obviously, for the kind of tissue pre-ashed samples, the characteristic masses of silicon atomized represented by sample F1, the temperature programme 1 in from aqueous solution (47±3 pg) and from a slurry of the Table 1 as well as the programmes E and F in Table 3 do not sample L2 used as an example (56±2 pg) diVer significantly.provide suYcient thermal pre-treatment of this material prior Therefore, the method of standard additions had to be used to the atomization step. Thus, with this programme, the matrix for the calibration. interferences are not suYciently eliminated. The results pre- The silicon content in all three samples determined by the sented in Fig. 3 support this assumption: whereas within the two slurry sampling ETAAS methods is compared in Table 5. programme modes D–F, no further increase of the absorbance was achieved with the sample L1, a still increasing absorbance could be observed with the sample F1. Besides this, for the sample F1, according to the ratio of the concentrations determined in the pre-ashed samples L1 and F1 (see Table 6), a higher absorbance is to be expected under interference-free conditions than the maximum one in Fig. 3. We did not attempt further optimization of the temperature programme which would allow elimination of the matrix interferences, as such a programme, if at all possible, would be very complicated and time-consuming and it would cause a further shortening of tube lifetime.Therefore, the technique involving sample preashing seems to be generally preferable, as for both types of samples under investigation, the interferences in the determination of silicon can be suYciently minimized. We assume that this technique might be applicable also to analysis of other types of biological tissue for silicon whereby the same tempera- Fig. 5 Charring curves of silicon obtained: (A) for aqueous silicon ture programmes for sample ashing and sample analysis can standard (2.0 ng of Si); (B) for slurry of the untreated sample L2 be used. (2.0 ng); (C) for slurry of the untreated sample L2 spiked with aqueous In Table 6, the results obtained for all three tissue materials silicon standard (3.0 ng); (D) for slurry of the pre-ashed sample L2 by slurry sampling ETAAS using pre-ashed samples are com- (3.1 ng); (E) for slurry of the pre-ashed sample L2 spiked with aqueous pared with those obtained in an interlaboratory collaborative silicon standard (4.1 ng) [for (A), (B) and (C), 20 mg of Pd–20 mg of study organized by J.Pavel43 using direct wavelength dispersive Mg(NO3)2, and for (D) and (E), 20 mg of Pd were added per atomization as chemical modifiers]. X-ray fluorescence (WDXRF), solution ETAAS and ICP-AES, 1128 Journal of Analytical Atomic Spectrometry, October 1997, Vol. 12Table 6 Contents of silicon in the samples L1, L2 and F1 obtained by ETAAS using slurry of ash and by diVerent methods in other laboratories Sample Silicon concentration/mg g-1 Silicon concentration/mg g-1 This work Other methods and laboratories Other methods and laboratories Ash slurry ETAAS Direct WDXRF Sol-ETAAS Sol-ICP-AES 1 2 3 4 5 6 7 89 10 11 12 13 14 L1 2.5±0.2 7.7±0.8 5.8±0.8 7.7±1.3 9.9 9.9 4.4 2.6 4.7 1.7±0.6 5.0 <5 2.0±0.5 2.2 [2.3, 2.7] [7, 9] [5, 7] [6, 9] [9, 13] [4.0, 4.7] [1.1, 2.5] [1, 5] [1.5, 2.5] [2.1, 2.3] L2 14.2±0.7 20.5±2.9 17.0±0.1 17±2 22.4 21.2 11.1 4.1 7.7 11.9±0.7 14.3 12 11±2 12.3 [13.5, 15.1] [18, 25] [17, 17] [15, 18] [20, 23] [11.0, 11.2] [11.0, 12.7] [11, 13] [9.8, 13.5] [11.2, 13.3] F1 10±4 13±3 12±8. 11±7 9.8 11.2 24.4 9.4 15.1 3.5±1.9 14.5 8 16±3 14 [5.3, 13.2] [9, 16] [7, 28] [<5, 28] [8, 13] [1.9, 5.8] [7, 18] [14, 20] [7, 20] In brackets: [extreme values]. Methods: 1 slurry sampling ETAAS of pre-ashed samples.Sample size 200–500 mg (n=4). 2–5 direct analysis by WDXRF. Sample size for 3: ~5 g (n=6), for 4: ~100 mg (n=4), for 5: ~2.5 g (n=3) and 6: ~4 g. 6 digestion in an open PTFE vessel with HNO3–H2O2, followed by ETAAS. Sample size ~40 mg. 7 pressurized microwave digestion in a PFA-bomb with HNO3, followed by ETAAS. Sample size ~350 mg (n=2). 8 extraction of silicon using tetramethylammonium hydroxide, followed by ETAAS after dilution with water.Sample size ~100 mg. 9 digestion in a capped PTFE-tube at 120 °C with HNO3–HF (251), followed by ETAAS. Sample size ~200 mg. 10 pressurized microwave digestion in a closed PFA-bomb with HNO3–H2O2, followed by ETAAS. Sample size 60–100 mg (n=6). 11 fusion with LiBO2, followed by ICP-AES as described24 for low-silicon food samples. Sample size ~4 g. 12 pressurized microwave digestion in a PFA-bomb with HNO3–H2O2, followed by ICP-AES. Sample size ~500 mg. 13 pressurized digestion in a PTFE-bomb at ~180 °C with HNO3–HF (151), followed by ICP-AES.Sample size ~120 mg (n=3). 14 fusion with NaOH suprapur (500 °C), followed by ICP-AES. Sample size ~5 g (n=2). both involving diVerent sample handling procedures in 12 for silicon is presently impossible but preparation for its future use is in progress. laboratories. From this comparison, it is evident that the determination of silicon in biological tissue represents one of the most diYcult analytical tasks which still cannot be solved Limits of Detection satisfactorily.Even the results obtained by only one of the two LOD were calculated as three times the standard deviation of types of solution methods, i.e., ETAAS and ICP-AES, show the replicates from the blank measurements. For this purpose, considerable disagreement. For example, the silicon contents in the direct slurry sampling technique, the suspension medium determined in the individual laboratories by solution ETAAS was processed, and in the ash slurry technique, the whole in the samples L1, L2 and F1 were in the range 1.7–9.9, procedure including pre-ashing was performed with 25 mg of 4.1–21.2 and 3.5–24.4, respectively.Only the results obtained magnesium nitrate. For the direct and the ash slurry sampling, for the samples L2 and F1 by WDXRF and for the sample L2 a blank value of 0.8±0.07 mg g-1 (n=10, assuming a sample by ICP-AES show reasonable consistency. Nevertheless, for amount for slurry of 70 mg) and 0.1±0.01 mg g-1 (n=8, the samples L2 and F1, the results of WDXRF and ICP-AES, assumed sample amount: 500 mg), was obtained leading to respectively, are in good agreement with our result.For the LOD of 0.2 and 0.03 mg g-1, respectively. samples L2 and F1, the agreement with the results of WDXRF and of ICP-AES, respectively, is within 60%. However, in no one case could satisfactory agreement of our results with the CONCLUSION results of one method and laboratory for all three samples be Slurry sampling ETAAS applied to untreated and to pre-ashed achieved.The highest degree of agreement was achieved with samples seems to be an advantageous and promising method lab 11 and lab 14 using sample fusion with LiBO218 and with for the determination of silicon in biological tissue materials. NaOH, respectively. Unfortunately, the method could not Compared with methods involving sample digestion with acids, provide any more exact results for the sample L1, because the an essential minimization of the risk of contamination and content was near to the LOD.For the samples L2 and F1, the volatilization loss seems to be the main advantage of the two agreement was within 20%. The agreement of our results with developed slurry sampling techniques. Ash slurry sampling is those of laboratory 14 was very good for the samples L1 and superior to direct slurry sampling regarding LOD, sample L2 and it was within 40% for the sample F1.From the homogeneity, precision and its applicability to both biological minimum and maximum values of the replicates, it can be seen materials processed in this work. However, owing to the lack that for the sample F1, the replicates of all methods and of a biological standard reference material with a certified laboratories show a much larger scatter compared with the silicon content as well as of a reliable reference method, the samples L1 and L2. accuracy check could be performed only to a limited degree.Thus, a conclusive comparison, which would provide a more Thus, further improvement of the state of development of the reliable judgement of the degree of accuracy of our results, is methods for determination of trace silicon in biological tissue presently barely possible. An essential improvement of this is necessary. unfavourable situation could be expected from an RNAA method. However, it requires the availability of a reactor The authors thank J.Pavel for making available the results of providing an extremely high ratio of thermal neutron flux to other methods and for fruitful discussion. fast neutron flux which would minimize the primary interference reaction induced on phosphorus by fast neutrons. In REFERENCES addition, sensitive detection of the indicator radionuclide 31Si can be achieved exclusively by non-specific counting of its beta 1 Cavic-Vlasak, B. A., Thompson, M., and Smith, D. C., Analyst, radiation.This makes selective separation of the indicator 1996, 121, 53R. radionuclide in the radiochemically pure form necessary. For 2 Indraprasit, S., Alexander, G. V., and Gonick, H. C., J. Chronic Dis., 1974, 27, 135. these reasons, application of RNAA to analysis of the samples Journal of Analytical Atomic Spectrometry, October 1997, Vol. 12 11293 Mauras, Y., Riberi, P., Cartier, F., and Allain, P., Biomedicine, 23 Zhuoer, H., Spectrochim. Acta, Part B, 1995, 50, 1383. 24 Krushevska, A. P., and Barnes, R. M., J. Anal. At. Spectrom., 1980, 33, 228. 1994, 9, 981. 4 Adler, A. J., and Berlyne, G. M., Nephron, 1986, 44, 36. 25 Zhuoer, H., J. Anal. At. Spectrom., 1994, 9, 11. 5 Hosokawa, S., and Yoshida, O., Int. Urol. Nephrol., 1991, 23, 281. 26 Velandia, J. A., and Perkson, A. K., J. Radioanal. Chem., 1974, 6 Gitelman, H. J., Alderman, F. R., and Perry, S. J., Am. J. Kidney 20, 715. Dis., 1992, 10, 140. 27 Ward, N. I., and Mason, J.A., J. Radioanal. Chem., 1987, 113, 515. 7 Marco-Franco, J. E., Torres, V. E., Nixon, D. E., Wilson, D. M., 28 Guzzi, G., Pietra, R., and Sabbioni, E., J. Radioanal. Chem., 1976, James, E. M., Bergstrath, E. J., and McCarthy, J. T., Clin. 34, 35. Nephrol., 1991, 35, 52. 29 Miller-Ihli, N. J., Anal. Chem., 1992, 64, 964A. 8 Wrobel, K., Blanco Gonzalez, E., and Sanz-Medel, A., J. Anal. 30 Bendicho, C., and de Loos-Vollebregt, M. T. C., J. Anal. At. At. Spectrom., 1993, 8, 915. Spectrom., 1991, 6, 353. 9 Wrobel, K., Blanco Gonzalez, E., and Sanz-Medel, A., J. Anal. 31 Miller-Ihli, N. J., Fresenius’ J. Anal. Chem., 1993, 345, 482. At. Spectrom., 1994, 9, 281. 32 Miller-Ihli, N. J., J. Anal. At. Spectrom., 1994, 9, 1129. 10 Candy, J. M., Oakley, A. E., Klinowski, J., Carpenter, T. A., 33 Miller-Ihli, N. J., Spectrochim. Acta, Part B, 1995, 50, 477. Perry, R. H., Atack, J. R., Perry, Z. K., Blessed, G., Fairbairn, A., 34 Miller-Ihli, N. J., J. Anal. At. Spectrom., 1997, 12, 205. and Edwardson, J. A., L ancet, 1986, 1(8477), 354. 35 Hauptkorn, S., Schneider, G., and Krivan, V., J. Anal. At. 11 Carlisle, E. M., in Silicon Biochemistry-CIBA Foundation Spectrom., 1994, 9, 463. Symposium 121, Wiley, Chichester, 1986, p. 123. 36 Hauptkorn, S., and Krivan, V., Spectrochim. Acta, Part B, 1994, 12 Birchall, J. D., and Chapell, J. S., Clin. Chem., 1988, 34, 265. 49, 221. 13 Perl, D. P., and Brody, A. R., Science, 1980, 208, 297. 37 Fuchs-Pohl, G. R., Solinska, K., and Feig, H., Fresenius’ J. Anal. 14 Henderson, V. W., and Finch, C. E., J. Neurosurg., 1989, 70, 335. Chem., 1992, 343, 711. 15 Peters, W., Smith, D., Lugowski, S., McHugh, A., and Baines, C., 38 Zhuang, Z., Yang, P., Wang, X., Deng, Z., and Huang, B., J. Anal. At. Spectrom., 1993, 8, 1109. Ann. Plast. Surg., 1995, 34, 343. 39 Grobenski, Z., Erler, W., and Voellkopf, U., At. Spectrosc., 1985, 16 Teuber, S. S., Saunders, R. L., Halpern, G. M., Brucker, R. F., 6, 91. Conte, V., Goldman, B. D., Winger, E. E., Wood, W. G., and 40 Bulska, E., and Jedral, W., J. Anal. At. Spectrom., 1995, 10, 49. Gershwin, M. E., Biol. T race Elem. Res., 1995, 48, 121. 41 Welz, B., Schlemmer, G., and Mudakavi, J. R., J. Anal. At. 17 Thomson, J. L., Christensen, L., Nielsen, M., Brandt, B., Breiting, Spectrom., 1992, 7, 1257. V. B., Felby, S., and Nielsen, E., Plast. Reconstr. Surg., 1990, 85, 38. 42 Miller-Ihli, N. J., J. Anal. At. Spectrom., 1988, 3, 73. 18 Lo, D. B., and Christian, G. D., Microchem. J., 1978, 23, 481. 43 Pavel, J., Novartis Inc., Central Analytical Department, 19 Berlyne, G. M., and Caruso, C., Clin. Chim. Acta, 1983, 123, 239. R-1055.402, CH-4002 Basel, Switzerland, private communication. 20 Gitelman, H. J., and Alderman, F. R., J. Anal. At. Spectrom., 1990, 5, 687. 21 Holden, A. J., Littlejohn, D., and Fell, G. S., Anal. Proc., 1992, Paper 7/01313B 29, 260. 22 Pe�rez Parajo� n, J. M., and Sanz-Medel, A., J. Anal. At. Spectrom., Received February 25, 1997 1994, 9, 111. Accepted June 25, 1997 1130 Journal of Analytical Atomic Spectrometry, October 1997, Vol