ANALYST, FEBRUARY 1992, VOL. 117 117 On-line Microwave Digestion of Slurry Samples With Direct Flame Atomic Absorption Spectrometric Elemental Detection Stephen J. Haswell and David Barclay School of Chemistry, University of Hull, Hull HU6 7RX, UK A flow injection (FI) system for on-line microwave digestion of slurried samples with direct elemental determinations by flame atomic absorption spectrometry is described. Organically based elemental reference samples were prepared as slurries in 5% v/v HN03 and the system was optimized for slurry mass, acid strength and tube and microwave cavity geometry. Bubble formation during digestion was controlled by post-digestion cooling and pressure regulation. Comparison of direct and FI calibrations indicated no apparent loss in sensitivity. Various samples were examined and elemental recoveries for Car Fe, Mg and Zn were typically found to be in the range 94-107% with precisions of less than 4.5% relative standard deviation.The major source of error was found t o be in the dispersion of solids (<I80 pm) as slurries in dilute HN03. The throughput of samples in the system developed was found t o be 1-2 min per sample. Keywords: Microwave digestion; on-line digestion; flame atomic absorption spectrometry; elemental analysis; flow injection The decomposition of samples by microwave digestion prior to trace elemental determination has now become a popular method with many analysts for a wide range of matrix types.' To date, the majority of digestions described have been based upon a batch technique, in which the microwave digestion replaces traditional wet or dry ashing methodology.2.3 The move to a microwave digestion approach offers many advan- tages over the conventional methods including reduction in digestion time, digestion of difficult matrices and dissolution in what is essentially a closed environment , which reduces volatile analyte loss and atmospheric Contamination.' Despite these obvious advantages, the batch type of approach to sample digestion is still prone to contamination problems associated with the sample, reagent and containment together with analyte loss and potential errors from volumetric transfers.Many of these problems can be overcome or controlled by adopting a flow injection (FI) methodology.4 It would seem therefore that the advantages of a microwave digestion approach to sample dissolution could be further improved by incorporating the digestion into an FI manifold.There have been only a few reports of systems based upon this approach54 to date, and none of the methodologies describes a continuous-flow system with on-line microwave digestion. This paper reports the development of such an FI system with on-line microwave digestion coupled directly to an atomic absorption (AA) spectrometer for elemental determination. Experimental Apparatus A schematic diagram of the on-line digestion system is shown in Fig. 1 and was assembled as follows: (i) an Ismatec pump MV-Z; (ii) a Rheodyne injection valve (Anachem 5020) with a 1 ml sample injection loop; (iii) a CEM MDS81 microwave oven containing 20 m of 0.8 mm i.d.poly(tetrafluoroethy1ene) (PTFE) tubing; (iv) a 5 m cooling loop in an antifreeze bath cooled by six Peltier devices (MI 1069T-O3AC, Marlow Industries, Tadworth, UK); (v) a Rheodyne injection valve (Anachem 5020) with an in-line back-flush filter fitted in the place of the injection loop; (vi) a CEM pressure sensor; and (vii) a 516.75 kPa (75 psi) back-pressure regulator (Anachem P736). The outlet from the back-pressure regulator was coupled directly to the nebulizer of an AA spectrometer (Thermoelectron 357) using 0.8 mm i.d. PTFE tubing (Anachem 33-1331). The same tubing was used throughout the system for all couplings and loops. A magnetic stirrer for slurry preparation and a chart recorder (Linseis LS52) for data collection were also used.Reagents Concentrated nitric acid and all 1000 ppm standards were of AnalaR quality supplied by Merck (Poole, Dorset, UK) and the water used was distilled, de-ionized. The following Certified Reference Materials (CRMs) were used: Chlorella, Mussel, Sargasso and Pepperbush supplied by National Institute for Environmental Studies (NIES), Japan, together with a National Institute of Standards and Technology (NIST) Standard Reference Material (SRM) 1577a Bovine Liver. Procedure Calibration was carried out by filling the 1 ml sample loop of the injector with standards prepared in 5% v/v HN03 within the linear calibration range for each of the elements studied as follows: Ca, 0-2 ppm; Fe, 0-4 ppm; Mg, 0-0.4 ppm; and Zn, 0-0.8 ppm. In addition, calibration was performed directly reservoir Sample loop (1 mi) Injection valve Microwave Back-flush filter Injection valve Spectrometer Fig.1 system Schematic diagram of the on-line microwave digestion FI118 ANALYST, FEBRUARY 1992, VOL. 117 with the AA spectrometer, i.e., without the FI system in place. Slurries of the reference materials were prepared by accurately weighing an appropriate mass (50-500 mg) of the solid (t180 pm) into a beaker and adding dilute HN03 (5% v/v; 25-250 ml) using a pipette to obtain a slurry in the range 0.005-0.5% d v . The selection of a suitable percentage slurry was dictated by the elemental concentration in the particular sample but was chosen to give an acceptable minimum mass (not less than 25 mg) without producing a slurry of greater than 1%.It was observed that for some samples a slurry of >1% could cause blockages in the small apertures within the switching plate of the Rheodyne valve. The slurries were agitated by a magnetic stirrer for approximately 30 s, and an aliquot of approximately 2-3 ml was taken using a syringe (10 ml) to fill the 1 ml sample loop. The slurry was found to be stable for at least 10 h. The flow rate throughout the system was adjusted to match the optimum nebulizer uptake rate of the AA spectrometer. The operating parameters of the AA spectrometer were in accordance with the manufacturer’s recommended instrument conditions for each element using an air-acetylene flame. Absorption signals for both standards and slurries were recorded on a chart recorder and peak height measurements were taken.Replicate injections were perfor- med for standards and samples in the same experimental run for the appropriate element. Results and Discussion Calibration Calibration using a series of five standards in 5% v/v HN03 was carried out within the linear calibration range for each element by using the AA spectrometer in a conventional manner. The results from this direct nebulization were compared with those obtained by processing the same standards through the FI system with the microwave oven at 90% power. This established that no loss in instrumental performance occurred from analyte dispersion or mixing owing to the total length of tubing used and heating in the microwave cavity. The results in Table 1 indicate that, for standards in dilute acid (5% v/v HN03), no apparent loss in linearity or sensitivity occurs following the passage of a 1 ml slug of standard through the F1 system compared with direct nebulization.During the passage of the standards through the microwave cavity, out-gassing was observed as bubbles but these completely recondensed in the cooling loop. The flow rate through the system was maintained during this out- gassing period. The relevance of gas evolution is discussed in detail in the following section. Optimization of Digestion Conditions Given that the flow rate for the FI system was more or less limited to a narrow range (4-6 ml min-1) by the nebulizer uptake rate of the AA spectrometer, the digestion conditions or degree of dissolution in the microwave cavity were governed by three variables, namely, microwave power, tube length and acid slurry strength.As one of the objectives of this work was to minimize the time taken for the digestion whilst maintaining sensitivity, power ratings from 0 to 90% were evaluated for both signal sensitivity and digestion. Signal sensitivity was evaluated by comparing the absorbance response for a standard solution of 4 ppm of Cu in 5% v/v Table 1 Comparison of the calibration for Mg of the AA spectrometer with nebulization and an FI system Correlation System Slope Intercept coefficient Nebulizer 32.7 0.95 0.9994 FI 32.02 1.04 0.9999 HN03, over a range of microwave powers (Fig. 2). At microwave powers of below lo%, very little or no out-gassing occurred in the 20 m digestion loop and this led subsequently to a high degree of sample dispersion and a corresponding loss in sensitivity.In the range 10-55% power (stage 2), an increase in gas formation was observed that, by stage 3, led to the sample slug undergoing optimum fragmentation into discrete cells or sample fractions, which on cooling recom- bined to produce the original sample with minimal dispersion (Table 1). This rather unexpected consequence of bubble formation together with the efficiency of digestion observed for solid samples during the out-gassing process clearly represents an important process in the method described (a point returned to later). The results for slurry samples indicated that, as expected, a high power rating was preferable and that at 90% power (525 W) maximum digestion and sensitivity were achieved for the samples investigated.Selec- tion of 100% power was not found to give any improvement in results over those for 90% and so, in order to minimize heating of the oven components by continuous use, 90% power was selected for the work described. Having selected a fixed microwave power, the geometry of the tubing in the micro- wave cavity was evaluated. Various lengths (10,20 and 30 m) and internal diameters (0.3, 0.5 and 0.8 mm) of tubing were examined and these were either knotted into a ball or wrapped around the conventional 12 bomb holder as supplied by CEM for conventional microwave digestion. A satisfactory diges- tion was identified by observing the passage of the digested 1 ml slug of slurry (for this chlorella was used) for particulates following centrifugation, collected after the back-pressure 160 E 140 E 2 120 E 100 0, .- Y ca 80 2 60 40 ca - - -- - - Stage Stage * 2 3 Stage 1 1 I I I I I I I 0 10 20 30 40 50 60 70 80 90 Microwave power (%) Fig. 2 Plot of microwave power versus signal response for a Cu standard (4 ppm). Stage 1, insufficient power to cause out-gassing therefore little signal improvement; stage 2, increasing power gives increasing out-gassing therefore signal improves; and stage 3, no further increase in out-gassing therefore no further signal improve- ment, i.e., maximum sensitivity above 55% I m* Stage Stage ’ 1 0 5 10 15 20 25 30 Length of tubingh Fig.3 Percentage recoveries of Fe (chlorella 0.2% slurry in 5% v/v HN03) versus length of digestion tubing (0.8 mm i.d.) at a flow rate of 6 ml min-I.Stage 1, increasing time in microwave up to a point where maximum out-gassing occurs and digestion is fully underway; stage 2, increasing time in cavity and hence digestion without increasing dispersion; and stage 3, full digestion is accomplished for sampleANALYST, FEBRUARY 1992, VOL. 117 119 regulator with no in-line filter in place. The 0.3 mm i.d. tubing was found to block readily with slurries. The 0.8 mm i.d. tubing was found to give a longer residence time in the cavity over the 0.5 mm i.d. tubing for the same length with no obvious blocking effects or loss in sensitivity. By using 0.8 mm i.d. tubing it was found that, at lengths greater than 20 m and flow rates of up to 6 ml min-1, total digestion was achieved, i.e., negligible particulates remained (Fig. 3). Some particu- lates (2-5, visible by eye) were always present probably owing to poor dissolution associated with an unavoidable dilution or 3 4 5 6 7 HN03 in slurry (% v/v) Fig. 4 Percentage recoveries of Fe (chlorella 0.2% slurry) versus acid strength of the slurry. Stage 1, increasing digestion as out-gassing develops; stage 2, optimum digestion with out-gassing sufficient for digestion without disrupting the flow; and stage 3, flow disrupted and erratic owing to excessively violent out-gassing giving loss in repro- ducibility and increase in dispersion and sampling time dispersion effect of the 1 ml slug of 5% v/v HN03 at the trailing interface with the water carrier, characteristic of FI peaks. It was decided to place a back-flush filter in the system to protect the back-pressure regulator from possible block- ages, and to prepare for the likelihood of residual material with matrices that might be studied at a later date.This filter was constructed in-house from 4 mm diameter stainless-steel mesh (50 pm) housed in a modified zero dead volume coupling packed with acid-washed glass wool to minimize dispersion. As a 20 x 0.8 mm i.d. length of tubing was found to be adequate for digestion of all the samples analysed in this work, it only remained necessary to establish the best geometry for the tubing in the cavity. Knotting of the tubing, in an attempt to minimize dispersion, was found not to be advantageous because it led to local heating of the solid material and eventual blockage of the tubing, as it became physically stuck in the tight curves of the knot.The best geometry for the 20 m tubing was found to be when it was wound around the carousel designed conventionally for holding 12 digestion bombs; a modified version of the carousel was used in subsequent work. During the digestion of slurries various acid strengths were evaluated ((1-70% v/v HN03) by investigating the percentage recoveries of Fe from a 0.2% m/v slurry of chlorella in 5% v/v HN03 (Fig. 4). At acid strengths below 5% v/v HN03, the production of bubbles or a gas phase in the digestion loop was low which led to poor digestion and consequently lower signal recoveries. As indicated previously, the presence of a gas phase has been found to be essential for the digestion of samples and early attempts to remove the bubbles completely by increasing the back-pressure of the system was found to give poorer digestions of the slurry material.As yet it is not clear what the actual mechanism of digestion is but the Table 2 Percentage recoveries and precisions for reference materials, n = 10 Mg Ca Recovery RSD Recovery RSD (Yo) (Yo) (Yo) (Yo) Chlorella 96.3 2.3 99.0 3.1 Reference value 0.33% 0.49% CRM Mussel 97.2 1.4 100 7.6 Reference value 0.21% 0.13% CRM Sargasso 102 0.89 93.8 1.9 Reference value 0.67% 1.41% CRM Reference value 0.408% 1.38% SRM 1577a Pepperbush 96.5 0.8 103 1.9 - Bovine Liver 94.0 4.5 ND Reference value 600 pg g- 1 120 pg g- 1 *ND = Not detectable in a slurry of S0.5%. Zn Fe Recovery RSD Recovery RSD ND* - 97.5 1.5 20.5 pg g-1 0.185% (Yo) (%) ( Y O ) (Yo) 102 7.0 96.8 4.0 106 pg g- 158 pg g- ND - 95.4 3.3 16.4 pg g- 187 pgg-l 93.6 0.7 98.0 3.0 340 pg g- * 205 pg g- 1 95.9 0.7 107 4.3 123 pg g-1 194 pg g- Table 3 Relative standard deviations and percentage slurries for reference materials, n = 10 Mg Ca Zn RSD Slurry RSD Slurry RSD Slurry (%) (% m/v) (YO) ('7; m/v) (%) (YO m/v) CRM CRM CRM No.9 CRM SRM 1577a - - Chlorella 2.3 0.01 3.1 0.01 Mussel 1.4 0.01 7.6 0.01 7.0 0.5 Sargasso 0.89 0.005 1.9 0.005 - - Pepperbush 0.8 0.01 1.9 0.01 0.7 0.2 Bovine Liver 4.5 0.01 - - 0.7 0.5 Fe RSD Slurry (%) (YO m/v) 1.5 0.2 4.0 0.5 3.3 0.4 3.0 0.5 4.3 0.5120 ANALYST, FEBRUARY 1992, VOL. 117 presence of gas, liquid and solid phases in narrow bore tubing under the influence of heat from microwave power is significant and should form the basis of a more fundamental study.If, however, gas production is allowed to become too excessive then controlling the bubble size and flow charac- teristics through the digestion loop will lead to erratic flows and dispersion effects. These problems, which can be asso- ciated with increasing acid strength, led to a loss in signal sensitivity and as a consequence the apparent low recoveries observed in Fig. 4 for the higher acid concentrations. Obviously the degree of gassing will be a function of sample type and volume; the experimental conditions described in this paper were found to be adequate for the range of sample types studied. It may be necessary, however, to review the acid slurry strength, microwave power and pressure and length of the digestion loop in the microwave cavity for alternative sample types.Analysis of Samples The results, expressed as percentage recoveries and precision [relative standard deviation (RSD)], for the four elements determined by atomic absorption spectrometry (AAS) are summarized in Table 2. It was necessary to select different slurry masses or dilution factors for the various samples to facilitate obtaining a signal in the linear working range of the spectrometer. It was considered that sample masses below 25 mg would lead to unacceptable precision, and the percentage slurries (m/v) that were selected are summarized in Table 3. The elemental concentrations in some samples, for example Zn in chlorella/Sargasso and Ca in Bovine Liver, were found to be too low, using the maximum slurry concentration of 0.5%, to give a measurable signal.What is apparent from observations taken during experimental work is that it is not just the mass of the sample taken that influences precision but also the wettability or dispersion of the solid as a slurry that affects the results. For example Bovine Liver and Mussel were difficult to disperse in the 5% v/v HN03 and gave correspond- ingly higher RSD values. Surfactants were not used in this work but this is one possible area to investigate further. In general the results obtained were found to be acceptable in terms of recoveries and precision. Conclusion The method described offers rapid and efficient sample preparation using on-line microwave digestion of slurries with direct elemental detection by flame AAS. The sample preparation and analysis time for ten replicate samples was approximately 0.5 h for the method described compared with 2 h for the same number of samples prepared by the microwave bomb digestion technique.The elemental levels that can be determined are at present governed by the limited calibration range of the AA spectrometer and the current percentage slurry range used. Clearly this limitation could be overcome by the use of a wider dynamic calibration range such as that offered by inductively coupled plasma atomic emission spectrometry. Problems were experienced with the dispersion or wettability of some of the powdered (<180 pm) organic reference material used but at worst these gave precisions of 6 5 % RSD. Samples were processed in approximately 2 min and elemental recoveries for the samples studied were typically in the range 96107%. The authors gratefully acknowledge the contribution of Dr. P. Riby and L. Neville to the progress of this work. References Introduction to Microwave Sample Preparation, eds. Kingston, H. M., and Jassie, L. B., American Chemical Society, Washing- ton, DC, 1988. Fisher, L. B., Anal. Chem., 1986,58, 261. Millward, C. G., and Kluckner, P. O., J. Anal. At. Spectrom., 1989, 4, 709. RfiiiEka, J., and Hansen, E. H., Flow injection Analysis, Wiley, New York, 2nd edn., 1988. de la Guardia, M., Salvador, A., Burguera, J. L.. and Burguera, M., J. Flow Injection Anal., 1988, 5 , 121. Burguera, M., Burguera, J. L., and Alarcon, 0. M., Anal. Chim. Acta, 1986, 179,351. Paper ll03046I Received June 20, 1991 Accepted September 10, 1991