JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY MARCH 1994 VOL. 9 267 Automatic Sample Handling and Calibration Methods for Analysis by Flame Atomic Absorption Spectrometry Using Discontinuous Flow Analysis* Brian L. Krieger Graham M. Kimber Mark Selby,? Fraser 0. Smith and Elvan E. Turak Analytical Spectroscopy Group Centre for Instrumental and Developmental Chemistry Queensland University of Technology G. P.O. Box 2434 Brisbane Queensland 400 7 Australia John D. Petty and Russell M. Peachey lonode Pty. L td. P. 0. Box 52 Holland Park Queensland 4 12 I Australia The method of discontinuous flow analysis (DFA) is presented as a rapid automated method of sample handling and analysis in flame atomic absorption spectrometry (FAAS). Three DFA arrangements for combined DFA-FAAS are described.The first provides automatic calibration in approximately 30 s using a single solution the second provides rapid automated standard additions and the third provides an automated version of the sample bracketing procedure. The last two methods allow for concurrent calibration and determination of samples in a single operation lasting around 30 s. This continual re-calibration minimizes the effects of instrument drift and ensures accurate and precise measurement. The DFA-FAAS combination provides a great deal of information that could be used to advantage for quality assurance purposes. However this extra information gathering does not require additional labour on the part of the operator because only one or two standard solutions need to be employed to construct complete analytical calibration curves.In contrast to other automated methods of sample handling the DFA technique uses high-precision cam-driven piston pumps and a valve arrangement to manipulate flowing streams containing reagents diluents standards and samples in a controlled and rapid fashion. Keywords Discontinuous flow analysis; flame atomic absorption spectrometry; automatic calibration stan- dard addition and dilution; sample bracketing Flame atomic absorption spectrometry (FAAS) is widely regarded as an 'aging' analytical method,' an old war horse still fulfilling an essential role in mining agriculture manufac- turing health services environmental studies and a variety of other important applications. However FAAS is fast becoming a non-developing technology; publication of research papers on FAAS has declined and fundamentally new developments are considered to be unlikely.'** The application of new sources such as laser diodes could yet generate a resurgence of scientific interest in FAAS but considerable technological development is necessary before such instruments are commercialized.' We believe that the accepted routine of calibration and sample delivery an area of analytical practice that has changed little since the early days of FAAS,3 still offers considerable scope for fundamental improvement.Methods for automated sample handling and calibration in FAAS reported previously have included exponential dilution chambers435 and various methods of controlled dispersion in flow injection (FI) as reviewed recently by Valcarcel.6 Exponential dilution chambers and FI have also been ~ombined.~.~ Significant recent works on automatic sample presentation and calibration for FAAS are those of Sperling et aL9 for FI methods and Starn and Hieftje" who used a high-performance liquid chromatography (HPLC) pumping arrangement to control the mixing of stan- dards and diluents in a series of discrete ratios. A further new methodology which is described in this paper is to combine FAAS with the new method of discontinuous flow analysis (DFA).''-l3 Conventional FAAS might be described as a 'zero-order' method14 because only a single result is generated for each measurement.As suggested by Sperling et one area for improvement of FAAS is to extend it from a zero-order method to a more information rich first-order methodology.The dilution chamber technique can be considered a first-order * Presented at the XXVIII Colloquium Spectroscopicum Internationale (CSI) York UK June 29-July 4 1993. t To whom correspondence should be addressed. method because the normal outcome is a vector of absorbance values collected during the exponential decay of analyte con- centration according to eqn. (1) rather than the usual single scalar measurement C=Co exp(-ut/v) (1) where C is the concentration at any given instant after introduc- tion of the sample volume C is the initial concentration of the sample u is the flow rate of diluent u is the volume of the chamber and t is time This first-order data vector contains sufficient information to calibrate the instrument response from the initial concentration down to the detection limit.The FI techniques that rely on the controlled dispersion of a plug of sample after injection into a flowing carrier stream are also first-order method^.^,'^.'^ A single transient FI signal contains information on the concentration range from zero to the concentration of the peak maximum. If the dispersion and transport characteristics of the FI system are reproducibly known then a single FI standard injection can be used to construct a complete analytical calibration curve using the so called 'electronic dilution' m e t h ~ d . ' ~ . ' ~ Another FI first-order approach is the 'zone-penetration' technique that disperses the sample within standard and diluent solution^.'^ In the zone-penetration method the sample-to- standard ratio varies across the FI peak permitting optimiz- ation of the sample-to-standard ratio for standard additions ~a1ibration.l~ The sequential injection (SI) analysis technique of Baron et al." is a more recent development of the zone analysis concept where a holding loop is used to store a concentration gradient formed from a single calibrant.A further FI method is based upon zone re-sampling where a standard solution is injected into a first carrier stream and allowed to disperse for a given time t; a zone from the dispersed standard is then re-injected into a second carrier stream and merged together with a sample injected at the same time.19-21 In the zone-resampling technique the concentration of stan- dard added to the sample is controlled by varying At which selects differing zones from the original standard injection.The268 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY MARCH 1994 VOL. 9 zone-resampling technique is a zero-order procedure because after merging of the sample with the standard only a single peak height or area is used in subsequent calculations and because a separate injection of standard and sample is required for each standard In this paper three new first-order methods for sample delivery and calibration for FAAS are presented all based on the new technique of DFA."-l3 The DFA system allows the proportions of samples calibration standards and diluents in a flowing stream to be manipulated rapidly resulting in the measurement of samples and calibration standards concur- rently.Also described in combination with FAAS is how DFA can be used for automation of ( i ) instrument calibration (ii) the 'standard additions' procedure and (iii) a form of the 'sample bracketing' method. Items (ii) and (iii) above include instrument re-calibration each time a new sample is presented for analysis. After the initial optimization of the spectrometer every measurement produces a calibration and a result; the distinction between measurements made for calibration purposes and those made for analytical purposes is no longer necessary. Although each measurement cycle takes a little longer (approximately 30 s) this combination of calibration with measurement means that every cycle produces a result for a determination thus ensuring that productive throughput is not compromised.The DFA-FAAS combination is a first-order method in which rapid manipulation of the analyte concentration in the stream delivered to the spectrometer is achieved by the differential actions of precisely controlled pumps. It is a combination that offers advantages over all previously reported methods for automated calibration and sample handling. A full discussion of these advantages becomes easier if preceded by a detailed description of the technique and is therefore presented at the end of this paper. Experiment a1 Discontinuous Flow Analyser A prototype discontinuous flow analyser (DFA) was used (Ionode Tennyson Queensland Australia). The DFA contains a number of major components that are integral to its oper- ation (Fig.1). The precision machined cam is central to the instrument and its shape dictates the movement of the pistons. Fig. 1 DFA system cut away to illustrate the central drive shaft (A) and encoder disc (B). Also shown are the cam (C) which is split into two sections X and Y the cam follower (D) and piston pumps E-G. While pump E is dispensing pumps F and G are in suction. These pumping modes are reversed in the other half of the cycle (see text). This figure shows only half of the DFA another cam follower is in horizontal opposition to D driven by the other side of the cam A central driveshaft locates both the cam and the encoder disc. The optical encoder disc generates 2500 pulses per revolution and has two functions precise identification of the rotational position of the cam (by counting pulses from an initialization marker) and setting the rate of data acquisition.For each encoder pulse one data point is taken from the detector (in this instance an FAAS instrument) thus for a full 360" rotation 2500 data points are acquired. The rotation speed of the cam is variable between 0.25 and 4 revolutions per minute. The cam is split into two asymmetric parts X and Y (as shown in Fig. l) part X being exchangeable. The profile of X dictates the flow regime for the particular experiment while the profile of Y provides a constant flow regime. In the experiments described below two different X sections were used to produce incremental or gradient flow profiles. The cam is able to drive two sets of horizontally opposed piston pumps (of up to five pistons in each set).Each set moves independently. (Only the left-hand side of the DFA is shown in Fig. 1.) The 360" rotation of the cam is divided into half cycles a measurement half cycle and a refill/dispense-to-waste half-cycle. The piston pumps are poly(methy1 methacrylate) (PMM) with stainless-steel plungers and Viton 0 ring seals. Associated with each piston is a three-way valve (ASCO-Angar Florham Park NJ USA Part No. 368132430) used to direct the discharge or recharge of the pistons. The position of the cam dictates the states of the valves; during the measurement half- cycle the valves are open to the measurement device and during the refill/dispense-to-waste half-cycle the valves allow the piston pumps to refill with reagents or dispense to waste.Poly(tetrafluoroethy1ene) (PTFE) tubing of 1 mm i.d. (Gradko International Winchester UK) is used for all fluid connections between valves reservoirs pistons and the detector. Data acqusition was via a 386-class personal computer (Total Peripherals Brisbane Queensland Australia) operating at a clock speed of 16 MHz and equipped with a PC-74LC data acquisition expansion board (Boston Technology Perth Western Australia Australia). Operation conditions for the analogue-to-digital (A/D) sub-system are shown in Table 1. Signals from two atomic absorption spectrometers were acquired through the chart recorder output and preamplified before A/D conversion via the PC-74LC board. Data acqui- sition software has been written using QuickBasic version 3.5 (Microsoft Redmond CA USA).Experimental conditions and instrumental parameters for the FAAS experiments are listed in Table 2. Atomic Absorption Spectrometry Stock calcium solutions (1000 mg 1-I) were prepared from analytical-reagent (AR) grade calcium carbonate dissolved in AR grade hydrochloric acid and distilled de-ionized water. Table 1 Data acquisition system Resolution 12 bit Non-linearit y Input range Number of analogue inputs Conversion rate G0.75 least significant bit k 5 V (bipolar) 8 differential channels 30 KHz Table 2 Instrumentation and operating parameters Instrumental parameter Varian 1275 Varian AA6 Wavelength/nm 422.7 422.7 Solution uptake rate/ml min-' 6 6 Ca lamp (Varian Techtron)/mA 5 5 Oxidant (air)/l min-l 7 7 Fuel (acetylene)/l min- 2.5 3.5 Slit-width/nm 0.50 0.2JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY MARCH 1994 VOL.9 269 Fresh working solutions were prepared from this stock solution daily using distilled de-ionized water. The calcium signal was monitored at 422.7 nm and a stoichiometric air-acetylene flame was used throughout. The incremental self-calibration and incremental standard additions methods (Methods 1 and 2 below) were carried out with a Varian 1275 flame atomic absorption spectrometer (Varian Techtron Springvale Victoria Australia). The uptake rate of the nebulizer was controlled with an Alitea (Stockholm Sweden) peristaltic pump using 0.89 mm i.d. poly(viny1 chlor- ide) (PVC) bridged tubing (Gradko International). The uptake was maintained at a rate slightly lower than the normal aspiration rate of the nebulizer.Other operating conditions for the Varian 1275 are shown in Table2. Experiments with the gradient self-calibration method (Method 3 below) were carried out with a Varian Techtron AA6 spectrometer (Varian Techtron) and operating conditions are shown in Table 2. Results and Discussion The DFA combines samples standards diluents and reagents by a process of pre-programmed mixing in the flowing stream under the control of the cam-driven pumps. Thus the analyte concentration in the stream reaching the detector varies in a precise way with time during a cam cycle. The variation of analyte concentration during a cycle is described as either gradient or incremental. In the former case (gradient) analyte concentration increases or decreases linearly with time during the cycle.In the latter case (incremental) the analyte concentration variation with time features several steps and plateaus (Figs. 3 and 4). The experiments described are classified in two ways firstly according to the programme of analyte dilution during the cycle (viz. incremental or gradient) and secondly depending on whether a ‘standard addition’ method is being used. Method 1 Incremental Self-calibration Consider the simple one-piston configuration as schematically illustrated in Fig. 2. Two solutions only are involved one standard and one diluent. The cam design is such that during the measurement portion of the cam cycle an incremental (stepped) concentration profile with four plateaus is generated. The peristaltic pump of the spectrometer ensures that the total flow rate reaching the nebulizer remains constant.However the relative contribution to this constant flow rate from the standard (delivered by the piston pump) and the diluent (aspirated from the reservoir) varies in a stepwise manner during the cycle such that F (constant) = F + Fd (2) where F is the flow rate of the peristaltic pump F is the flow rate of the standard and Fd is the flow rate of the diluent. This is illustrated in Fig. 3(a) plateau 1 represents 100% diluent 0% standard; plateau 2 represents 70% diluent 30% standard; Cam f---. Pe r i st a I t i c . . To FAAS Diluent Discontinuous flow analyser Fig. 2 Single piston DFA-FAAS configuration for Method 1 a con- stant flow rate to the FAAS is governed by the peristaltic pump.However the relative flow rates of standard and diluent reaching the FAAS depends upon the dispensation of standard by the pump which varies as a function of the cam rotation and the complementary uptake of diluent Cam angle -. I Time -+ Concentration - Fig. 3 Illustration of the stepped self-calibration method (Method 1) (a) stepwise increase of analyte concentration during a cycle; (b) resulting change in absorbance during the cycle; and (c) derivation of an analytical working curve from data collected during a single cycle (b) plateau 3 represents 29% diluent 71% standard; and plateau 4 represents 100% standard. Figure 3(b) is an idealized illustration of how these changes in analyte concentration in the nebulizer stream are tracked as changes in absorbance during the cam cycle.Fig. 3(c) shows how knowing the concentration of undiluted standard a calibration curve of absorbance versus concentration can be derived. This can be carried out using only the information gathered during a single cam cycle and with only one standard solution. At present cycle times of the order of 30 s are typical. This can be expected to decrease with further improvement in instrumentation. A plot of the absorbance uersus encoder pulses resulting from an experiment using a single 20 mg 1-’ calcium standard is shown in Fig.4(a) with the resulting calibration plot in Fig. 4(b). The plot is linear with a squared product-moment correlation coefficient (r’) of 0.989. Method 2 Incremental Standard Additions In Method 1 a rapid calibration of the FAAS is achieved.In Method 2 quantification of an unknown can be achieved as part of the DFA cycle. This incremental technique also allows automation of a form of the normally labour-intensive method of standard additions. A schematic representation of the DFA configuration for Method 2 is shown in Fig. 5. The standard is dispensed by the DFA piston pump increasing incrementally during the cycle. The sample and diluent are aspirated into the system from separate containers. Again the rate at which the standard-sample-diluent blend reaches the nebulizer is constant controlled by the speed of the peristaltic pump. The diluent and sample pass into the stream through T-pieces in a fixed proportion to each other. On the highest plateau (Fig.6 )270 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY MARCH 1994 VOL. 9 Encoder pulses 0 5 10 15 20 Concentration (ppm) Fig.4 Representative signal traces for Method 1 (a) plot of absorbance versus encoder pulses for a single cycle with a standard solution of 20 mg 1-1 of Ca and (b) the analytical working curve for Ca using data from (a) Per i st a It i c Standard To FAAS Discontinuous flow anal p e r Diluent Sample Fig. 5 Configuration for the incremental standard additions method (Method 2) showing the constant suction provided by the peristaltic pump the piston pump dispensing standard at a rate that varies according to the cam profile the constant uptake of sample and the complementary uptake of diluent the flow rate of the dispensing piston pump is matched to the pumping rate of the nebulizer and consequently the stream at this point ideally consists of standard only.On the lowest plateau the piston pump is stationary and hence the stream at this point contains no standard. The intermediate plateaus represent incremental mixtures of standard with sample-diluent. Fig. 6 is a typical example of the resulting traces. In this case two traces (A and B) have been acquired in sequence using a standard solution of 10 mg 1-l of calcium. Trace A is a diluent only trace in this case the reservoir marked 'sample' in Fig. 5 contained a blank solution. Trace B is a sample plus diluent trace where the sample consisted of a 10 mg 1-l solution of calcium. Trace A is acquired to ascertain a zero absorbance value. It also serves as a check on the validity of the standard additions assumption by allowing the slope of the absorbance 4.0 3 3.6 3.2 4 2.8 0 (5 .- 0 C 9 2.4 1 I 1 1 I I I Encoder pulses 0 200 400 600 800 1000 Fig.6 Plots of absorbance versus encoder counts for replicate pairs of cam cycles using Method 2. In the lower traces A the sample consisted of a Omg 1-' of Ca blank whereas in the upper traces B the sample consisted of a 10 mg 1-' of Ca solution versus concentration plot derived from this trace to be com- pared with the slope of a similar plot based on trace B. The concentration of the unknown is determined by a conventional standard additions calculation. The concen- trations of the additions are related to the known mixing ratios provided by the cam. A regression line is fitted to the absorbance versus addition data and the intercept (hence the unknown concentration) is calculated.Results for synthetic samples (10 mg I-' calcium solutions) using this method show high repeatability [less than 1% relative standard deviation (RSD)] but demonstrate an analyt- ical bias (typically about 2%). This bias appears to be related to the difficulty of exactly matching the flow rates of the peristaltic pump and the piston pump. Method 3 Gradient Self-calibration In this method the cam cycle is divided into two half-cycles encompassing firstly a calibration half-cycle and secondly a sample measurement half-cycle. In the two methods described above the calibration-measurement is performed during a single half-cycle. No analytically useful data are available in the second half-cycle when the piston pump is refilling.However in the method outlined below the full 360" rotation of the cam is used to obtain analytical data by reciprocal action of the piston pumps i.e. when the sample pump discharges other pumps are refilling or expelling and vice versa. In the calibration half-cycle a complete analytical calibration curve is generated. The concentration limits of this calibration plot are defined by two standards of high and low concen- tration. During the sample measurement half-cycle valves are activated to introduce the sample solution to the spectrometer. Fig. 7 shows a typical trace acquired in less than 30s. The two half-cycles include short wash-out phases (Fig. 7). The experimental arrangement for this method is shown schematically in Figs.8 and 9. Understanding of these figures is aided by consideration of the action of the pumps during the calibration half-cycle (Fig. 7). During this half-cycle the composition of stream S5 (the stream arriving at the spec- trometer) and consequently the measured absorbance varies with time according to the simultaneous actions of pumps P1 P2 and P3. The operation of these pumps is described below. Pump PI This is a constant flow-rate dispenser that sets the rate at which the analysis stream reaches the spectrometer. Pump P1 delivers a low concentration standard (stream Sl). Stream S1 is equal in flow rate to stream S5 the stream feeding the nebulizer. The flow rate is somewhat less than the natural aspiration rate of the nebulizer.This is a gradient flow suction pump that with- draws stream S2 at a linearly increasing rate. This is a gradient flow dispensing pump that Pump P2 Pump P3JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY MARCH 1994 VOL. 9 27 1 0.45 1 High -n Wash-out standard 0.38 c 0) a 0.31 5 a 2 0.24 0.17 measurement \ Blank 1 measurement a 0.10 I I I I -1 0 300 600 900 1200 1500 1800 Encoder pulses Fig. 7 Replicate traces of the full cycle for the gradient self-calibration method (Method 3 ) showing the generation of a linear calibration gradient defined by low and high concentration standards followed by measurement of the sample. The calibration and measurement half cycles include short wash-out periods. The top two traces are for a sample solution of 10 mg I-’ of Ca and the bottom two traces are for a sample consisting of a 0 mg 1-l of Ca blank solution.The high and low concentration standards are 20 and 2 mg l-‘ respectively P2 P3 P4 P Fig. 8 Pumping arrangement (top view) for Method 3. Pump P1 is discharging at a constant rate which sets the flow rate to the nebulizer. Pumps P2 and P3 are of equal capacity and both are driven by the same cam frame but in opposite directions. Pump P2 is withdrawing while pump P3 is dispensing. Pumps P2 and P3 have no net effect upon the overall flow rate. During the time that pumps P1 P2 and P3 are refilling or expelling valves (not shown) are activated so that P4 dispenses sample to the FAAS at a constant rate P3 P4 s3 s4 To FAAS s1 PI r 1 I J P2 Fig. 9 Fluid-flow diagram for Method 3.During the first half cycle the piston pumps P1 P2 and P3 act so as to provide a linear gradient in concentration between a standard of low concentration in P1 and a standard of high concentration in P3. Pump P2 acts to withdraw solution at the same rate as it is added by P3 ensuring a constant overall flow rate. At the end of the half cycle valves (not shown) activate to present a sample solution in P4 to the FAAS using the same constant rate as before delivers a high concentration standard (stream S3) at a linearly increasing rate. The actions of pumps P2 (gradient suction) and P3 (gradient dispensing) are complementary. These two pumps are of equal capacity and are driven by the same cam frame (see Fig. 8) but in opposite directions. The additive contribution of P3 to the over-all flow rate is exactly balanced by the subtractive effect of P2 and the total flow rate defined by the constant dispensing rate of P1 remains invariant.Fig. 10(a) and (b) shows the contribution of stream S1 (low standard) and stream S3 (high standard) to the composition of stream S5 during the calibration half-cycle of the rotation of the cam. The result of this simultaneous action of pumps P1 P2 and P3 on the analyte concentration at the nebulizer is shown in Fig. lO(c). At the beginning of the gradient stream S5 is made up entirely of stream S1. The analyte concentration is that of the low standard dispensed by pump P1 (constant flow dispenser). At the completion of the gradient stream S5 is made up entirely of stream S3. The analyte concentration is now that of the high standard dispensed by pump P3.The intermediate variation in composition is linear owing to the linear cam gradient. Thorough mixing before the stream reaches the nebulizer is ensured by a vibrating reed mixerg in the mani- fold (Fig. 9). Fig. 10(d) shows the predicted trace as the spectrometer absorbance measurements track the compositional changes in stream S5 as described by Fig. lO(c). If Beer’s law holds over this concentration range the resulting plot is linear with encoder position and hence with concentration. The calibration half-cycle just described is followed by a sample measurement half-cycle. At the beginning of this phase the valves switch so that the solution now presented to the nebulizer comes only from pump P4.Pump P4 has been refilling during the calibration segment and now dispenses the sample at a constant rate which is equal to the over-all flow rate to the nebulizer during the previous half-cycle. This results in the absorbance plateau region labelled ‘sample measurement’ in Fig. 7. Clearly the absorbance measured on this plateau can be related to the absorbance uersus concentration curve pro- duced in the calibration half-cycle and thus to the determi- nation of the concentration of the unknown. 0 180 360 0 180 360 Cam angie/degrees Cam angle/degrees 1 .o 1 .o 0) c $ 0.5 0.5 8 2 1 500 1000 2 10 20 Encoder pulses Concentration (ppm) Fig. 10 Contribution of the piston pumps to the over-all flow rate during the gradient calibration half cycle in Method 3. (a) proportion of the over-all flow due to the combined effects of pumps P1 and P2 (this is the flow rate at which the low standard is added to the stream); (b) proportion of the over-all flow due to pump P3 (this is the flow rate at which the high standard added to the stream); (c) resulting variation of analyte concentration with encoder pulses; and ( d ) resulting plot of absorbance uersus concentration (assuming adherence to Beer’s law)212 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY MARCH 1994 VOL.9 Both Method 2 and Method 3 have the advantage of providing a calibration with each sample measurement. This continual re-calibration is a potentially rich source of quality assurance data on instrumental drifts and also has the practical result that for actual sample measurements the distinction between short-term and long-term precision is no longer meaningful.Method 3 despite its more complex mechanical arrange- ment has several advantages over Method 2 (stepped standard additions). (a) The resulting trace is readily understandable being similar to the conventional 'multi-point calibration fol- lowed by sample measurement' procedure. (b) Common prob- lems associated with peristaltic pumps namely pulsations and degradation of pump tubing with time are eliminated. The cam-driven piston pumps are highly reproducible over several years of use. (c) The uptake rate of the spectrometer is solely and precisely controlled by a cam-driven piston pump (either pump P1 or pump P4). The difficulty of matching the flow rate from the DFA to nebulizer uptake (as in Method 2) is overcome.( d ) Separate analytical calibration curves consisting of approximately 400 data points are determined for every sample. Because of the abundance of data no assumption of linearity is necessary a linear or higher order regression calculation is applied as statistically appropriate. (e) This close definition of the calibration curve means that the technique known as sample bracketing can be automatically applied to each sample measurement. In a conventional application of the bracketing technique a preliminary measurement establishes the approximate concen- tration of the unknown then standards with concentrations closely bracketing the unknown are prepared and measured and finally the sample absorbance is re-measured.Although the reliability of measurements is improved by this method it is not routinely used because it is clearly very time consuming. With DFA however use of Method 3 means that sample bracketing is automatically effected for all concentrations within the calibration range. A disadvantage of the present gradient calibration arrange- ment is that the sample fills and is dispensed from piston pump P4. This requirement increases wash-out times and carryover between samples and necessitates larger sample volumes than is the case with Methods 1 and 2. It is anticipated that with future improvements in the instrumentation for DFA-FAAS that this problem will be overcome. A replicate set of analyses of a synthetic 10 mg 1-' calcium sample using a 2 mg 1-l calcium solution as the low standard and 20 mg I-' calcium solution as the high standard showed an analytical bias of less than 0.5% and repeatability of better than 0.5% RSD.Comparison With Other Automatic Calibration Methods Of the automatic sample handling and calibration methods described earlier the exponential dilution chamber is the least flexible and extensible option. The FI and DFA methods offer a wide range of sample management options. The primary difference between the FI and DFA methods is that FI relies on controlled dispersion to produce gradients in concentration whereas the DFA technique uses cam-driven piston pumps to deliver precise volumes of samples diluents standards and reagents into a well-mixed stream according to pre-determined flow profiles.The three automatic calibration methods pre- sented are intended to enhance the precision and accuracy of the FAAS method Methods 1 and 2 increment the standard to diluent or standard to sample to diluent ratios over several fixed compositions and the cam profile is cut so as to dwell at each new composition for a period of time long enough to gather statistics at that composition. An added advantage for Method 3 is that because the calibration curve is so finely determined the method provides automatic bracketing of the sample concentration. In conventional FAAS the bracketing method is known to produce favourable precision and accu- racy.3 On the other hand the FI-FAAS combination is subject to limitations in precision as described by T y ~ o n . ~ ~ ~ ~ In comparison with FI techniques there are disadvantages with DFA.Firstly greater care must be exercised in order to minimize the effects of dispersion. Best results with DFA are obtained with laminar flow through short low-volume fluid connections. Secondly the relative complexity and cost of the instrumentation is greater with DFA. The incremental DFA experiments described above (Methods 1 and 2) are similar to those described recently by Starn and Hieftje" using an HPLC pumping arrangement. However the DFA experiments described here demonstrate a number of significant advantages over those of Starn and Hieftje:" (i) fast response time (less than 30 s for an analytical run instead of around 10 min); (ii) no pH dependent pulsation noise has been encountered with the DFA system; (iii) sample introduction is much simpler and far more convenient with the DFA system because there is no need to fill a pump reservoir with sample; and (iv) sample volumes can be as small as a few hundred microlitres with the DFA arrangement.This work was supported by the Commonwealth Department of Employment Education and Training through an Australian Research Council Grant (File No. AE9130183) and an Australian Postgraduate Research Award (Industry). The authors thank Sally Watson for assistance with the artwork and Dennis Sweatman for technical advice. 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 References Hieftje G. M. J. Anal. At. Spectrom. 1989 4 115. Sturgeon R. E. J. Anal. At. Spectrom 1992 1 13N. Schrenk W. G. Flame Emission and Absorption Spectrometry VoZume 2 Components and Techniques ed.Dean J. A. and Rains T. C. Marcel Dekker New York 1971 ch. 12. Booher T. R. Elser R. C. and Winefordner J. D. Anal. Chem. 1970 42 1677. Posta J. and Lakatos J. Spectrochim. Acta Part B 1980,35,601. Valcarcel M. Sample Introduction in Atomic Spectroscopy ed. Sneddon J. Elsevier Oxford 1990 ch. 11. Carbonell V. Sanz A. Salvador A. and de la Guardia M. J. Anal. At. Spectrom. 1991 6 233. Beinrohr E. CsCmi P. and Tyson J. F. J. Anal. At. Spectrom. 1991 6 307. Sperling M. Fang Z. and Welz B. Anal. Chem. 1991 63 151. Starn T. K. and Hieftje G. M. J. Anal. At. Spectrom. 1992,7,335. Arnold D. P. Peachey R. M. Petty J. D. and Sweatman D. R. Anal. Chem. 1989 61 2109. Petty J. D. Peachey R. M. and Sweatman D. R. US Pat. 5,080,866 1992. Sweatman D. R. Petty J. D. and Peachey R. M. US Pat. 5,040,898 1992. Sanchez E. and Kowalski B. R. J. Chemom. 1988,2,247. Olsen S. RBiiCka J. and Hansen E. H. Anal. Chim. Acta 1982 136 101. RBiiEka J. and Hansen E. H. Flow Injection Analysis Wiley New York 2nd edn. 1988. Fang Z. Harris J. M. RBBiEka J. and Hansen E. H. Anal. Chem. 1985 57 1457. Baron A. Guzman M. RBiiCka J. and Christian G. D. Analyst 1992 117 1839. Jacintho A. O. Zagatto E. A. G. Bergamin H. Krug F. J. Reis B. F. Bruns R. E. and Kowalski B. R. Anal. Chim. Acta 1981,130 243. Zagatto E. A. G. Jacintho A. O. Bergamin H. Krug F. J. Reis B. F. Bruns R. E. and Arajo M. C. U. Anal. Chim. Acta 1983 145 169. GinC M. F. Krug F. J. Bergamin Filho H. Friere dos Reis B. Zagatto E. A. G. and Bruns R. E. J. Anal. At. Spectrom. 1988 3 673. Tyson J. F. Spectrochim. Acta Rev. 1992 14 169. Tyson J. F. Analyst 1985 110 419. Paper 3/04998A Received August 17 1993 Accepted October 15 1993