摘要:

JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY JANUARY 1994 VOL. 9 39 High-performance Flow Electrothermal Atomic Absorption Spectrometry for On-line Trace Element Preconcentration-Matrix Separation and Trace Element Determination Harald Berndt and Gerhard Schaldach institute of Spectrochemistry and Applied Spectroscopy Bunsen-Kirchhoff-Str. 7 7 D-44 739 Dortmund Germany With the aerosol deposition module (ADM) the samples to be analysed by electrothermal atomic absorption spectrometry (ETAAS) are carried as an aerosol into the graphite tube and deposited at about 140 "C. The ADM is normally equipped with a pneumatic nebulizer but in this work this is replaced by a hydraulic high- pressure (HHP) nebulizer. If HHP nebulization is used to produce the aerosol the efficiency of aerosol transport is substantially improved (maximum aerosol yield 60%).As a high-pressure flow system is required for this type of nebulization on-line matrix separations-trace element preconcentrations for ETAAS are also possible. The trace elements can be separated on-line from aluminium solutions (AICI,) and determined in 4 min. Detection limits are in the ng g-' range (as trace elements in the aluminium). In comparison with the original ADM for sample introduction the detection power is improved by about two orders of magnitude. Keywords Hydraulic high-pressure nebulization; electrothermal atomic absorption spectrometry; sample introduction; on-line matrix separation; preconcentration Electrothermal atomic absorption spectrometry (ETAAS) is one of the most efficient methods available for trace element analysis.Absolute masses in the low pg range can be deter- mined for many elements. The general limitations are the very small amounts of sample that can be introduced into the graphite tube and the large matrix interferences. Sample introduction into the graphite tube is usually per- formed manually using a microlitre pipette or automatically by a special autosampler. The whole sample is deposited in the graphite tube as a single drop of a precisely defined volume e.g. 20 p1. Sample introduction in the form of an aerosol [with an aerosol deposition module (ADM) Thermo Jarrell Ash] is less commonly used. However with this method as with flame AAS the solution is first converted into an aerosol by a pneumatic nebulizer.The aerosol is then transported from the nebulization chamber to the graphite tube which has been heated to 140°C. The aerosol droplets are precipitated on the hot wall of the graphite tube and the liquid evaporates. In this way the sample is collected in the graphite tube. The amount of sample is determined by the deposition As the solvent evaporates continuously during sample introduction large amounts of sample can be brought into the graphite tube in a short time. However the low efficiency of the pneumatic nebulization means that substantially more sample is used than by direct introduction of the solution with a microlitre pipette. Therefore this technique is not very suitable for microsamples. In general in ETAAS a large amount of matrix leads to strong interferences and a large spectral background.Hence the ADM is only advantageous when the sample is matrix free and large amounts are available (e.g. analysis of drinking water). In this case the detection power of ETAAS can be varied over a wide range through changes in deposition time. It could be expected that the efficiency of the aerosol deposition technique can be considerably improved by replac- ing the pneumatic nebulization with hydraulic high-pressure nebulization (HHPN) which has a much higher aerosol ~ i e l d . ~ . ~ Hydraulic high-pressure nebulization is well known and has been used for decades in technical fields where nebulization must be very efficient and is not dependent on a nebulization gas (e.g. diesel injection aircraft turbines and spray drying).This nebulization method is now available for sample introduction in atomic spectrometry with a high- performance flow-hydraulic high-pressure nebulization system (HPF-HHPN).4 The pneumatic nebulizer is replaced by a special HHPN nozzle with an opening of only 10-30 pm. A high-performance liquid chromatographic (HPLC) pump pro- duces a constant high-pressure carrier stream in which the sample is injected via a sample introduction valve and a sample loop and then transported to the nebulization nozzle. The nebulization of the liquid occurs only because of the high pressure [the HHPN principle i.e. pressure of 100-400 bar (1 x lo7-4 x lo7 Pa 1500-6000 psi)]. Using an HPLC pump for generation of the liquid pressure and an HPLC sample introduction valve the main components of the HPLC flow system become a functional part of an atomic absorption spectrometer.With an HPLC column additionally fitted in the high-pressure line prior to the HHPN nozzle an interface-free coupling of HPLC separation techniques with atomic spectrometric trace element determination is achieved. The outlet of the HPLC column is connected to the HHPN nozzle via a high-pressure capillary of low dead volume. The high-pressure flow system ends with the nebulization of the solution as the sample introduction method for the spec- trometer (HPF atomic spectrometry). This system is often confused with low-pressure flow systems using peristaltic pumps [e.g. flow injection atomic spectrometry (FIAS)]. The properties of low-pressure flow systems and high-performance flow systems (e.g.HPLC) are totally different. This is because FIAS is a technique involving coupling of a low-pressure flow system and pneumatic nebulization for sample introduction. Pneumatic nebulizers are two component nebulizers (liquid- air) working in self-suction mode. Therefore a pump is not necessary for their operation. The characteristic properties of HHPN have been described elsewhere (e.g. aerosol yield of about 50% in flame AAS nebulization of viscous solutions lowering of interferences and increased detection p ~ w e r ) . ~ ~ In comparison with conventional coupling of HPLC and flame AAS with pneumatic nebuliz- ation HPF-flame AAS with on-line separation for the determi- nation of Fe"-Fe"' shows an improvement in the detection power by a factor of approximately On-line trace element preconcentrations-matrix separations can also be carried out with HPF-flame AAS.* In this paper HHPN used for sample introduction and on-line sample preparation in ETAAS is described for the first time.Owing to the already existing ADM a relatively simple adaptation of the HPF-HHPN system for sample introduction could be expected. The replacement of pneumatic nebulization by HHPN leads directly to HPF-ETAAS.40 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY JANUARY 1994 VOL. 9 Separation- preconcentration column SarnDle-eluent I Aerosol sampling jet I High-pressure pump Furnace cuvette (at elevated temperature) jives TO vacuum input / \ I Solvent filter Protection filter HHPN nozzle (20 pm) (Ti sieve 3 pm) (15 prn) Fig.1 Schematic diagram of HPF-ETAAS (ADM with HHPN) The aim of this work was a fundamental examination of this mode of sample introduction. Aerosol transport into the graphite tube was compared with that of pneumatic nebuliz- ation. Whether the higher aerosol yield of HHPN offers an effective method of microsampling with the aerosol deposition system was also investigated. In addition to the investigations with pneumatic nebulization in the self-suction mode and HHPN aerosol generation with the original pneumatic nebul- izer but with a forced supply of the sample by an HPLC pump was studied. Using the high-pressure flow system on-line trace element preconcentration-matrix separation becomes possible followed by determination of trace elements by ETAAS.A matrix-free concentrate of the trace elements can be collected in the graphite tube and the elements can be subsequently determined almost interference free. Considering both factors matrix separation-trace element preconcentration and improved nebulization a strong improvement in the detection power could be expected. Trace element determi- nations in solutions of aluminium (A1C13) was selected as an example as it is known that small amounts of this matrix lead to strong interferences and a high spectral background in ETAAS. Experimental Apparatus Atomic absorption spectrometer. SH 1000 with a graphite furnace (aerosol deposition module) Thermo Jarrell Ash. HPF-HHPN system. With an adapter (plug) for the Thermo Jarrell Ash ETAAS nebulization chamber supplied by Dr.H. Knauer Wissenschaftliche Geratebau 14163 Berlin Germany. Arrangement for HPF-ETAAS and Mode of Operation A schematic diagram of the HPF-ETAAS system is shown in Fig. 1. An HPLC pump produces a liquid stream at a con- trolled flow rate. To keep the blank values as low as possible the liquid first passes through a 5cm long cation-exchange column [polyether ether ketone (PEEK) housing]. The sample to be analysed is brought into the high-pressure carrier stream by means of a six-port HPLC valve (PEEK) with a 10 p1-5 ml sample loop. For trace element preconcentration-matrix separ- ation an appropriate HPLC column [e.g. ion exchange or C18 reversed-phase (RP) column] is added behind the sample introduction valve. In this work a C18 RP precolumn (PEEK) with a bed length of 5 mm (5 pm packing material) was used.A 3 pm titanium sieve filter was used to protect the nozzle from larger particles. The Pt-Ir nozzle plate is in a titanium housing. A nozzle with a 15 pm opening was used for the experiments described. For changing over from pneumatic nebulization to HHPN a new stopper for the nebulization chamber was made from poly( tetrafluoroethylene) (PTFE) (Fig. 2). The air originally required for the pneumatic nebulizer is now transported to this stopper through a small needle valve and a flow meter flowing concentrically to the HHPN nozzle housing into the nebulization chamber. No nebulization gas is required for the nebulization itself but a carrier gas is necessary to transport the aerosol to the graphite tube. Owing to the special geometric form of the HHPN nozzle the aerosol forms as a very fine low-pressure aerosol jet.This aerosol jet is expanded to an aerosol cloud on a cloud converter the form of which is identical with a conventional bead. (The apparent similarity of this arrangement to impact jet n e b u l i z a t i ~ n ~ ~ ~ ~ can easily lead to confusion between the two nebulization techniques. With the same nozzle opening and flow rate HHPN operates with 30-50 times higher pressure.) Transport Gas Flow A transport gas (air) is necessary to transport the aerosol that has formed in the nebulization chamber to the graphite tube. Using the determination of copper (5 ng ml-') as an example the influence of the transport gas flow on the amounts of trace elements deposited in the graphite tube was examined.To produce continuously an aerosol containing trace amounts of copper for these fundamental experiments a solution of copper was carried through the HPLC pump but not injected via the sample introduction valve. With a sample flow rate of 1 ml min-' and a 15 pm nozzle the nebulization pressure obtained was approximately 200 bar (2 x lo7 Pa 3000 psi). The aerosol which formed in the chamber was transported to the graphite tube at various gas flow rates (0.2-4 1 min-l) for HHPN Carrier gas input (air) I nozzle (Ti housing Pt-lr nozzle plate) Circle of holes (12 holes 2 mrn) HPLC c ( 1 /I 6" (g I ass 1 /'" Fixing screw Fig. 2 Structure of HHPN adapter for the ADMJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY JANUARY 1994 VOL. 9 41 B 9 /- c- \ /x-x Q1 0.4 0.3 8 n 2 ] 0.2 z m - 0.1 6 iij n I I I I I 1 0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0- Transport gas flow rate/l min-' Fig.3 Relationship of signal height and reproducibility for determi- nation of Cu to transport gas flow (sample flow rate 1 mlmin-l 5 ng ml-' Cu2+ aerosol deposition time 10 s) A signal height; and B relative standard deviation 10 s each and deposited at 140°C. The graphite tube was then heated to 2100 "C and the atomic absorption measured (Fig. 3). This measurement was repeated 15 times for each gas flow rate tested. As expected the amount of copper deposited and therefore the sensitivity increased greatly with an increase in flow of the transport gas. The best reproducibility was achieved with a carrier gas stream in the range from 2.8 to 3.2 1 min-l. An increase in the gas flow rate of up to 4 1 min-' leads to a slight increase in the sensitivity but the reproducibility becomes distinctly poorer.All subsequent HPF-ETAAS experiments were therefore carried out with a transport gas flow of 3.2 1 min-'. Aerosol Yield and Transport to the Graphite Tube In this paper aerosol yield indicates the proportion of the sample that reaches the graphite tube in the form of an aerosol. To determine the aerosol yield distilled water was continuously nebulized using various sample uptake rates of the pneumatic nebulizer and various flow rates of the HPLC pump for 30 min each time. The amount of liquid drained off the nebulization chamber was ascertained by weighing. With pneumatic nebuliz- ation the best result was an aerosol yield of about 10% [Fig.4(a)]. The amount of aerosol transported to the graphite tube (aerosol flow rate) is strongly dependent on the sample uptake rate. With a sample uptake rate of 5 ml min-' 300 pl of sample are transported into the graphite tube in 1 min [Fig. 4(b)]. With HHPN a 10 pm nozzle was used for a sample flow rate of 0.8 ml min-'. A 15 pm nozzle was used for sample flow rates of up to 1.2 ml min-l and a 20 pm nozzle was used for up to 2.5mlminP1. An aerosol yield of about 60% is achieved at a flow rate of 0.8 ml min-' [Fig. 4(a)]. In HHPN the aerosol stream to the graphite tube is almost independent of the sample flow rate. A sample volume of 400-450 pl min-' - % . 2 300 !! 3 g 200 s - 0 100 0 1 2 3 4 5 6 0 1 2 3 4 5 6 Sample flow rate or uptake rate/ml min-' Fig.4 (a) Aerosol yield and (b) aerosol flow rate in relation to the sample uptake rate of the pneumatic nebulizer (A) and the sample flow rate (liquid flow rate) of the hydraulic high-pressure nebulizer (B) reaches the graphite tube [Fig. 4(b)]. The sample transport to the graphite tube is limited by the geometry of the nebulization chamber the supply lines and the transport gas flow rate. This also explains the decreasing aerosol yields of the whole system with increasing HPF sample flow rate [Fig.4(a)]. The large amount of aerosol generated by the HHPN nozzle is only partially transported from the chamber to the graphite tube up to a maximum value of about 450 pl min-'. With a pneumatic nebulizer it is also possible to establish a fixed flow rate by using an HPLC pump. Volumes of solution of 1 and 2 ml min-l were continuously supplied to the original nebulizer with this pump.The nebulized air was varied in the range from 2.2 to 51min-' using the adjustment on the nebulizer. The air flow was measured with a flow meter. In comparison with only pneumatic nebulization with a maximum aerosol yield of 10% [Fig. 4(a)] a maximum yield of 17% was achieved under these conditions (sample flow rate 1 ml min-' nebulization of air 5 1 min-') [Fig. 5(a)]. At best 210 pl min-' of sample solution reached the graphite tube (sample flow rate 2 ml min-l) [Fig. 5(b)]. Microsample Introduction Microsample introduction with the ADM was tested with three different nebulization modes whereby the optimal con- ditions ascertained in the previous sections were used for each mode of nebulization.Pneumatic nebulization (selfsuction mode) The sample uptake rate was 1.3 mlmin-' nebulization air flow 2.2 1 min-' and the aerosol flow was 130 pl min-' (aerosol yield 10%). Pneumatic nebulization In this mode the sample was supplied by means of an HPLC pump and a sample introduction valve. The flow rate of the HPLC pump was 1 ml min-'. The nebulizer settings were nebulization air flow 51min-'; and sample uptake rate 5 ml min-' (self-suction mode). The aerosol flow was 164 pl min-' (aerosol yield 16%). Hydraulic high-pressure nebulization The flow rate was 1 ml min-' and carrier gas (air) 3.2 1 min-'. A 15 pm Pt-Ir HHPN nozzle was used; working pressure about 200 bar (2 x lo7 Pa 3000 psi).The aerosol flow rate was 452 pl min-' (aerosol yield 45%). A deposition time of 30 s was used for all three nebulization modes. The determination of Cu by ETAAS was carried out at 2100°C. 2.0 2.5 3.0 3.5 4.0 4.5 5.0 2.0 2.5 3.0 3.5 4.0 4.5 5.0 Nebulization gas flow rate/l min-' Fig.5 (a) Aerosol yield and (b) aerosol flow rate in relation to the nebulization gas (air) flow rate. Transport of the liquid to the pneumatic nebulizer with an HPLC pump (sample flow rate A 1; and B 2 ml min-')42 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY JANUARY 1994 VOL. 9 Trace Element Preconcentration-Matrix Separation (Trace Elements in Aluminium Solution) Ammonium pyrrolidinedithiocarbamate (ammonium pyr- rolidin- 1 -yldithioformate APDC) forms very stable com- plexes13 with a group of metals that can be retained on c18 RP material.As aluminium does not react with APDC this reagent is suitable for separation of trace elements from an aluminium matrix. Preliminary experiments showed that this separation can be applied to a high-pressure flow system. Sample pre-treatmnt A 2ml stock solution containing 50mgml-' of A13+ (447.5 mg ml-I A1Cl3*6H20) was transferred into a 20 ml calibrated flask and diluted with distilled water to approxi- mately 10ml. To form the complex 200 pl of an APDC solution (0.2% in methanol) were added and the solution was shaken. The pH value was adjusted to 3 with 6OOpl of 2 mol 1-1 ammonia solution. To redissolve the resulting pre- cipitate the calibrated flask was placed in an ultrasonic bath for 3min and then diluted to 20ml with distilled water (5 mg ml-l of Al). For calibration (standard additions method) the samples were spiked with trace elements before diluting to volume.HPF-E TAAS on-line aluminium separation-trace element preconcentration and E TAAS determination Flow system. Carrier acidified water (70 pl I-' of 14 mol 1-1 HNO pH approximately 3) flow rate 1 ml min-l 1 ml sample loop (PEEK) 5 mm precolumn (PEEK housing) 5 pm Eurospher 100 (c18 RP material Knauer). Eluent methanol (0.05 mol 1-1 HN03). Working mode. The arrangement of the apparatus used for HPF-ETAAS is shown in Fig. 1. The sample (1 ml 5 mg of Al) is added to the sample loop with a plastic syringe. On switching the valve the sample passes through the C18 column which retains the complexed trace elements.The matrix is led to the waste outlet of the nebulization chamber. The sample loop is returned to the load position 120 s after sample injection and filled with eluent (the time required for this is approxi- mately 20 s). The temperature-time programme of the graphite furnace (aerosol deposition mode) is then started. The graphite tube is heated to 140 "C and the aerosol injection tip controlled by the furnace programme is positioned in front of the opening of the graphite tube. The carrier gas stream switches on and the waste outlet of the nebulization chamber closes simul- taneously. The aerosol is not transported to the graphite tube immediately but sucked off by means of vacuum for a few seconds (e.g. 6 s). This delay ensures a regular aerosol flow.When transport of the aerosol to the heated graphite furnace begins the trace elements collected on the column are eluted with methanol (switching of the valve) and carried into the graphite tube; 30s are required for the deposition. The trans- port gas flow then switches off the injection tip returns to its original position and the atomization programme of the fur- nace begins. During this step the sample loop can already be filled for the next cycle and the next matrix separation-trace element preconcentra tion started. Results Microsample Introduction If only small amounts of sample are available for pneumatic nebulization a microlitre pipette can be used for sample introduction. The nebulizer of the aerosol deposition system is connected via a short piece of tubing to a small PTFE funnel into which the sample is injected (injection method microsampling).11~'2 In accordance with Fig.4 a maximum of about 10% of the amount of sample that is sucked in is transported to the graphite tube using pneumatic nebulization. However this value is only reached with large amounts of sample (continuous suction). High sample losses occur with small discrete sample volumes (< 50 pl) during transportation of the aerosol to the graphite tube thus causing severe reduction in the sensitivity of the AAS measurements. These transport losses can be reduced for microsamples when the sample is transported to the graphite tube as a segment in a constant aerosol stream. This is achieved by coupling the pneumatic nebulizer with a flow system. For these experiments the nebulizer was connected to an HPLC pump.The sample was added to the carrier stream with an HPLC sample introduction valve-sample loop (forced sample supply). If an HPLC pump is already being used for sample introduction pneumatic nebulization can be replaced by the more effective HHPN (see under Experimental). Comparison of the three sample introduction techniques (pneumatic nebulization with microsampling HPLC pump controlled pneumatic nebuliz- ation HHPN) using the determination of trace amounts of copper as an example is shown in Fig. 6. Two very different concentrations of copper (20 and 200 ng ml-') were used. With pneumatic nebulization (injection method self-uptake mode) and 10 pl of sample no signals are obtained even with a high concentration of copper of 200 ng ml-l [Fig.6(b)]. When the same absolute mass of copper (2 ng) is nebulized in a substantially greater volume (100 pl 20 ng ml-I) the signal has an absorbance of 0.028 [Fig. 6(a)]. This means that sample volumes of < 10 pl are completely lost during transport to the graphite tube. When the samples are supplied to the pneumatic nebulizer as a liquid segment in a continuous liquid carrier stream (HPLC pump sample introduction valve with various sample loops 10-100 pl) sample volumes of only 10 pl can also be measured with pneumatic nebulization. Under these conditions a sample volume of 10 pl (200 ng m1-l) produces an absorbance signal of 0.073 the same amount of copper dissolved in 100 pl(20 ng ml-') produces the same absorbance (0.075). This means that by enveloping the various sample volumes in a constant aerosol stream the same amounts of copper are deposited in the graphite tube.However the increase in sensitivity (2.7-fold with a sample volume of 100 pl) results not only from the lower transport losses through the enveloping but also from the higher aerosol yield of the . 1.25 r - 2 1.00 0 .- 0.75 0.50 0 20 40 60 80 100 Sample volume/pl Fig.6 Comparison of sensitivity for the determination of Cu with microsample introduction [(a) 20 and (b) 200 ng ml-' Cu2+] using three different nebulization modes A HHPN; B pneumatic nebulization-HPLC pump; and C pneumatic nebulizationJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY JANUARY 1994 VOL. 9 43 pneumatic nebulization with the forced supply.The HHPN showed the highest sensitivity of the three sample introduction modes tested. With a sample of 100 pl and a copper concen- tration of 20ngml-' the signal reached an absorbance of 0.290. The sensitivity compared with pneumatic nebulization was increased by a factor of 10.2. This increased sensitivity is also due to the enveloping of the samples in the aerosol stream and the substantially more effective nebulization method (HHPN). Thus the use of HHPN makes an effective microsam- ple introduction method for the aerosol deposition system possible for the first time. Detection limits (3s values n=25) for the determination of copper are approximately 0.5 ng ml-' (100 pl sample) and approximately 5 ng ml-' (10 pl sample). Using larger sample volumes (0.5-5 ml) these values can be reduced further.In addition to the original fields of microanalysis (e.g. biological samples) an effective microsample introduction is also necessary for the combination of on-line flow systems with AAS. Because of the low dispersion in high-pressure flow systems the separated or preconcentrated trace elements e.g. after elution from the HPLC column are in a concentrated form in a small liquid segment which is enveloped in a carrier stream. If the high-pressure flow system is used simultaneously for HHPN the ADM becomes the effective interface for HPLC techniques and trace element determination with ETAAS. Trace Element Preconcentration-Matrix Separation (Trace Elements in Aluminium Solution) Fig. 7 shows the measured signal areas as a function of the deposition time. The maximum signal for all elements is reached after a deposition time of 30s. Also no further trace elements could be detected with a second additional elution with methanol.This indicates that the elution of the trace elements the transport and the deposition of the aerosol are completed within 30 s for all elements under investigation. I I 0 10 20 30 40 Deposition (elution) time/s Fig.7 Relationship between signal area and deposition time of the eluted element traces A gold; B cobalt; C copper; and D lead. Eluent is methanol (1 ml) at a flow rate of 1 ml min- ' Longer rinsing times (maximum time of 5 min) between sample injection and elution had no effect on the separation and preconcentration. The relative standard deviations (RSD) were/determined with a concentration of 2.5 ng ml-' for each/element (1 ml of sample 5 mg ml- ' of Al corresponding to 0.5 pgg-' of trace element in aluminium) from 12 HPF-ETAAS determinations. The following RSD values were obtained for specific elements [measured as signal height (signal area integrated absorbance)] Au 2.0 (1.7); Co 8.0 (9.2); Cu 1.8 (2.1); and Pb 3.1% (2.7%).The detection limits were determined by HPF-ETAAS measurement of 25 unspiked samples (Table 1). In order to compare the detection power the detection limits for the direct determination of the trace elements in an aluminium solution (0.5 mg ml-') were also established using the ADM in the standard mode of operation (suction of the solution by a pneumatic nebulizer).In this case the deposition time was limited by the spectral background of the AAS measurement caused by the concentration of the matrix. The spectrometer used was equipped with high current pulse background correc- tion (the Smith-Hieftje principle2) therefore it was possible to choose a relatively high background with an absorbance of 0.6-1. In comparison with the original system HPF-ETAAS improves the detection power by about two orders of magnitude. Discussion Using the aerosol deposition system for sample introduction in ETAAS the sample solution is first converted into an aerosol. The aerosol is transported by a flexible tube and a glass tip into the pre-heated (140 "C) graphite furnace. The non-volatile components (trace elements and the matrix) are thus precipi- tated on the wall of the tube or onto a platform.Using pneumatic nebulization a maximum of only 10% of the sample is transported into the graphite tube. Loss of sample during transport in the system is very high with small sample volumes (10-100 pl). This method of sample introduction is therefore only suitable for larger sample volumes. Sample amounts of < 10 pl are completely lost on the way to the graphite tube by impaction of the aerosol particles in the nebulization chamber and by adsorption on the surface of the transport tube. This sample introduction technique can be improved considerably if pneumatic nebulization is replaced by HHPN. The amount of aerosol transported into the graphite tube is increased to a maximum of 60%. For the first time also small sample amounts e.g.lOpl can be effectively brought into the graphite tube and determined with good sensitivity using the ADM. If particularly low detection limits are required larger amounts of sample e.g. 0.5-5 ml can be continuously nebulized and transported into the graphite tube using the HPF-HHPN technique for sample introduction. As with pneumatic nebulization in HHPN a polydisperse aerosol is generated in which the proportion of larger droplets Table 1 Detection limits (n = 25 3s values) for trace element determinations with ADM sample introduction. Improvement of the detection power by on-line preconcentration-matrix separation with HPF-ETAAS. Pneumatic nebulization 0.5 mg ml - ' A1 (background absorbance signal due to the element 0.6-1).HPF-ETAAS 5 mg ml-' Al trace element preconcentration from 1 ml of sample Signal height measurement Signal area measurement Aerosol deposition mode/ Improvement Aerosol deposition mode/ Improvement pg g-' of A1 in detection pg g-' of A1 of detection power power Element PN HPF-ETA AS (factor) PN HPF-ETAAS (factor) Au 3.4 0.019 179 1.8 0.038 47 c o 40 0.17 229 13 0.30 43 cu 6.8 0.021 324 6.4 0.0 17 376 Pb 7.8 0.050 156 4.8 0.012 400 * PN =pneumatic nebulization.44 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY JANUARY 1994 VOL. 9 is very small. With a 20pm HHPN nozzle the proportion (volume) of droplets with a diameter > 16 pm is only about l%.14 With a 15 pm nozzle as used here the calculated mean diameter (volume) of the droplets is 5 pm.15 In comparison with the aerosol produced by pneumatic nebulization (larger droplets) the HHPN aerosol is much more effectively trans- ported into the graphite tube.Up to 450pl of sample are deposited in the tube in 1 min at a flow rate of 1 ml min-l. In HHPN the sample to be nebulized is injected into a carrier stream by an HPLC sample injection valve. Loss of sample during transport is kept to a minimum when working with microsamples by enveloping the sample in the aerosol stream. The main advantage of the different nebulization technique is the facility of on-line sample pre-treatment for ETAAS. The improved sample transport by using HHPN instead of pneu- matic nebulization is fundamental for effective interface-free coupling of HPLC techniques and ETAAS for trace element determinations.If an HPLC column is inserted between the sample introduction valve and the HHPN nozzle on-line trace element preconcentrations-matrix separations for ETAAS can be carried out. In this case the HHPN and the aerosol deposition system act as interfaces between the HPLC and ETAAS techniques for the determination of trace elements. Aluminium is a strongly interfering matrix in ETAAS (e.g. large background absorption spinel formation and the matrix is difficult to vaporize). It could be shown that trace elements complexed with APDC were collected on a C18 column whereby the aluminium matrix was carried to the waste outlet via the nebulization chamber of the ADM. In an on-line working mode a matrix-free trace element concentrate can be deposited in the graphite tube by elution of the complexed trace elements from the column.With a 5 mg ml-' of alu- minium matrix and a 1 ml sample volume detection limits of 20-200 ng 8-l (trace element in aluminium) were achieved. In comparison with determinations of trace elements with the original aerosol deposition module the on-line matrix separa- tion-element trace preconcentration together with the more effective HHPN result in an improvement in the detection power up to a factor of 100. Use of a larger sample loop (larger amount of sample) allows even further reduction of the detection limits. The time required for a complete on-line separation and ETAAS determination is ca. 4min. An inert autosampler can be used for automation of the procedure. As it is a high-pressure flow system standard HPLC or ion chromatography columns with very fine grained packing mate- rial can be used (5 pm).Unlike the columns used in low- pressure FI systems (40-60 pm packing material) even very short HPLC columns have a high separation efficiency and a high capacity. Therefore in this work HPLC precolumns with a bed length of only 5 mm could be used. Conclusions With HHPN an HPF system becomes a functional component of the aerosol deposition system and thus also of ETAAS. In addition to the on-line matrix separation-trace element precon- centration for ETAAS already described speciation is another important application of HPF atomic spectrometry. After HPLC separation flow segments can be time controlled on-line to be deposited in the graphite tube. This offers the possibility to select small parts of the chromatogram characterized by their retention time.Work is currently being carried out on an automated and fast on-line determination of Cr"' and CrV1 species in drinking water with HPF-ETAAS. This work was carried out with the financial support of the Federal Ministry for Research and Technology Bonn and the Ministry for Science and Research of North Rhine-Westphalia. The authors are especially grateful to Thermo Instruments GmbH Germany (Thermo Jarrell Ash) for kindly making their equipment available for this work. 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 References Kantor T. Clyburn S. A. and Veillon C. Anal. Chem. 1974 46 2205. Sotera J. J. Cristiano L. C. Conley M. K. and Kahn H. L. Anal. Chem. 1983 55 204. Berndt H. Fresenius' 2. Anal. Chem. 1988 331 321. Berndt H. Chem. Labor Betr. 1989,40 299. Berndt H. and Muller A. Fresenius' J. Anal. Chem. 1993,345 18. Posta J. and Berndt H. Spectrochim. Acta Part B 1992,47,993. Weber G. and Berndt H. Chromatographia 1990 29 254. Berndt H. Muller A. and Schaldach G. Fresenius' J. Anal. Chem. 1993,346 711. Herrmann R. and Alkemade C. Th. J. Flammenphotometrie Springer Berlin 2nd edn. 1960 pp. 92-93. Doherty M. P. and Hieftje G. M. Appl. Spectrosc. 1984,38,405. Sebastiani E. Ohls K. and Riemer G. Fresenius' 2. Anal. Chem. 1973 264 105. Berndt H. and Jackwerth E. Spectrochim. Acta Part B 1975 30 169. Koch 0. G. and Koch-Dedic G. A. Handbuch der Spurenanalyse Part 1 Springer Berlin 1974,2nd edn. pp. 297-317 and 327-337. Posta J. Berndt H. and Derecskei B. Anal Chim. Acta 1992 262 261. Personal communication Bosch Research Centre Stuttgart Germany. Paper 310264 9 C Received May 11 1993 Accepted September 2 1993

ISSN:0267-9477    

DOI:10.1039/JA9940900039

出版商:RSC    

年代:1994

数据来源: RSC

摘要:

JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY JANUARY 1994 VOL. 9 45 Effect of Discharge Conditions on the Sputtering and Spatial Distribution of Atoms in a Radiofrequency Glow Discharge Atomizer for Atomic Absorption Spectrometry G. Absalan C.L. Chakrabarti* J.C. Hutton Department of Chemistry Carleton University Ottawa Ontario Canada K l S 566 M. H. Back Department of Chemistry University of Ottawa Ottawa Ontario Canada K1 N 964 C. Lazik R.K. Marcus Department of Chemistry Clemson University Clemson South Carolina 29634- 7905 USA The effects of discharge conditions on the sputtering spatial distributions and transport patterns of sputtered atoms in a demountable radiofrequency (r.f.) glow discharge (GD) atomizer were investigated. Results obtained for sputtering of oxygen-free hard copper indicate that the GD plume is constricted by an increase in the Ar gas pressure and disturbed by Ar gashacuum flow.It is shown that the glow is an inhomogeneous medium and that most of the sputtered Cu atoms are localized in the front of the sampling orifice. Independent control of flow rate and pressure of the sputtering gas (Ar) is critical in determining the spatial distribution of analyte atoms within the atomizer chamber. Both the flow rate and the pressure of the Ar gas affect the removal of ground-state atoms from the analysis volume. The plasma location at different discharge param- eters was also investigated in this study. The results show that the r.f. GD source is able to sustain a stable plasma which enables the sequential atomic absorption analysis and depth profiling of samples such as metal alloys.Keywords Atomic absorption spectrometry ; glow discharge; cathodic sputtering; radiofrequency glow discharge. The glow discharge (GD) sputtering source has proved to be an excellent atomization source for the direct analysis of bulk and trace analytes in solid samples such as metals semi- conductors and alloys,' as well as powder samples2 and solution residue^.^ The representative atoms molecules meta- stable species ions and photons4 in the plasma produce several analytical options. As the material ablated in a GD is to a large extent present as a vapour cloud of free GDs are suitable atom reservoirs for atomic absorption spec- trometry (AAS),1,3,"'3 atomic fluorescence spectr~metry,'~'~ atomic emission spectrometry2G26 and mass ~pectrometry.~~-~' One of the factors that has tended to limit the application of GDs has been the requirement that the sample be conductive in nature.An insulating surface such as that provided by NaCl matrix in solution residues subjected to positive ion bombardment will quickly build up a positive charge layer. This layer obtains a potential up to the floating potential,32 repelling subsequent discharge gas ions and thus slowing any further sputtering from the surface. This same process occurs for non-conductive powder samples. Although by mixing the non-conductive sample with a suitable conductive host matrix a conductive sample the mixing itself introduces certain problems. For example dilution of the sample decreases sensitivity and increases the likelihood of contamination.In addition many bulk solids are not easily transformed into powders and this process precludes any depth-resolved analy- sis. The r.f. GD circumvents these problems. To this end Anderson et ~ 2 2 . ~ ~ applied an r.f. potential to a metal plate behind a dielectric target which was immersed in the plasma of a low pressure d.c. gas discharge. Since the electron mobility is much greater than the ion mobility and the net current to the dielectric target must be zero the surface of the sample takes a negative d.c. self-bias and is sputtered. In a later extension of that work Davidse and M a i ~ s e l ~ ~ demonstrated that a self-sustained r.f.-induced gas discharge suitable for sputtering could be ignited using only the r.f.* To whom all correspondence should be addressed. electrodes. When the cathode is alternatively biased negatively and positively the positive charge accumulated during one half cycle will be neutralized by electron bombardment during the following half cycle. Ion bombardment of the insulator would occur for only half of each cycle at the most but the much greater mobility of electrons in a discharge enables almost continuous energetic ion bombardment to be achieved if high enough frequencies are used. Secondary electrons which have extremely high yield values in the case of insulators are accelerated across the positive ion sheath into the plasma and act as an additional supply of electrons for sustaining the At the same time the oscillatory nature of electrons in r.f.plasma fields and at plasma-sample and plasma-wall sheaths gives rise to efficient electron-impact ionization and discharge sustaining collisions and this enables sputtering to be carried out efficiently. As a result an r.f. discharge can operate at lower pressures (at mTorr uersus Torr level; 1 Torr = 133.322 Pa) than its d.c. counterpart. In order to avoid gas-phase scattering and consequently to prevent re-deposition of sputt- ered analyte atoms on the cathode surface r.f. systems are usually operated at the lower end of their operating pressure range when employed in sputter deposition application^.^^ The power source normally used to meet these requirements is an r.f. power generator working at 13.56 MHz. At low frequencies (<l M H z ) ~ ~ there will be a series of short-lived discharges with the electrodes successively taking opposite polarities (non- continuous plasma).Radiofrequency GDs can be powered in the constant power mode which strikes an advantageous compromise between current and voltage control. As cathodic sputtering occurs in the region of an abnormal GD where the voltage increases rapidly with the current the power is a more meaningful experimental parameter than either voltage or current. The objectives of this research are to evaluate a laboratory- built,22 demountable r.f. GD atomizer modified for use in AAS by finding regions in the discharge that are suitable for atomic absorption studies as it is known that the composition of the GD is not homogeneou~.~~ Such information will help in46 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY JANUARY 1994 VOL.9 understanding the re-deposition of sputtered analyte atoms the sputtering rate and diffusive losses to the atomizer walls and thereby promote any possible modifications to the GD source. Experimental Sputtering Atomizer The top view of the r.f. GD source is presented diagrammati- cally in Fig. 1. A similar source has been described and characterized previously by Winchester et ~ 1 . ~ ~ as an r.f. GD emission source.22 The source consists of a hollow stainless- steel atomizer chamber (9 x 5 x 5 cm). Although no distinctly defined ground exists in the r.f. discharge (the net flow of charge to either electrode is zero); sample is referred to as the cathode by analogy to the d.c.discharge. For experimental simplicity the atomizer chamber is the grounded electrode which is made larger than the cathode so that all of the bias voltage resides on the cathode and there is no chance of sputtering of the chamber-wall materials. By making the cathode area much smaller than the anode (chamber) area the cathode sheaths are optimal for sputtering application^.^^ The vacuum port is located in the top-centre of the atomizer chamber. Two circular fused silica windows (35 mm in diam- eter) are mounted on each side of the atomizer chamber to provide a light path for the hollow cathode radiation. An extra optical window is mounted directly opposite the sample location for optical emission purposes. The source also incor- porates an easily removable sampling orifice disc (304 stainless steel) equidistant from the two side windows.In the present work a sampling orifice diameter of 4.5mm with a disk thickness of 4mm was used. Placement of the sample into position for atomization is accomplished by using the brass torque bolt to compress the sample against a 6.5 mm id. Teflon O-ring forming the necessary vacuum seal. There is a recess region between the sample and orifice disk provided by the arrangement of the Teflon O-ring. In the case of the r.f. power source arcing across the sample and anode becomes a A Fused silica Vacuum port Copper conductor Male coax connector Femalecoax / ?llP connector RG-213N Coax cable (to matching network) Fig. 1 Diagrammatic representation of the r.f. glow discharge atom- izer source.View provided corresponds to top of cell with the 'Horizontal' sampling axis lying in the plane of the paper perpendicular to the sample surface problem with a recess thickness of less than about 0.2mm. The recess thickness in the cited source is small enough to operate at 13.56 MHz as the dark space thickness decreases with frequency. The width of the cathode dark space is typically 1 mm for usual operating conditions of the discharge so there is no chance of ion bombardment of the O-ring. This gap is also sufficiently small to prevent diffusion of the sputtered vapour into the region between the sample and the orifice disc thereby enabling the cathode to remain electrically isolated from the rest of the chamber. A ceramic spacer attached to the end of the bolt provides the electrical isolation of the bolt from the sample.The r.f. feedthrough attached to the end of the coaxial r.f. power cable (RG-213U Amphenol) is inserted into the bolt and the male type-N connector is mated to the female connector as shown in Fig. 1 until the copper rod (+ in in diameter) makes contact with the back of the sample. A disc of Cu (30 mm in diameter and 3 mm in thickness) is fastened at the tip of the Cu rod just between the Macor spacer and the sample to facilitate uniform r.f. power distri- bution over the sample cathode and to ensure that the pressure of the sample clamp is evenly distributed across the entire sample. In addition for sputtering of insulator materials it is generally accepted that an insulator backed by a conducting r.f.-powered electrode will not charge up to floating potential.32 The atomizer was mounted on a laboratory-constructed XYZ@-adjustable platform.On the bottom of the atomizer- mount was an A1 base which was securely attached to the instrument body by means of four bolts. The horizontal movement section was attached to the base on the top of which was the vertical movement section. The rotational movement section was attached to the top of the vertical section. The chamber was attached to the rotational movement section by means of two bolts. Adjustment of the rotational position altered only the position of the atomizer relative to the light path of the hollow cathode lamp without changing the horizontal or the vertical position of the atomizer. In this study a fixed rotational position of atomizer was used so that the central axis of the atomizer chamber was parallel to the hollow cathode lamp light beam.When the horizontal position was adjusted the vertical and the rotational movement sections also moved. When the height was adjusted only the rotational movement section and the atomizer body moved along the vertical axis. All three movements were controlled manually. Incremental linear markings at 1 k 0.5 mm and 2 & 1 mm inter- vals for horizontal and vertical movements respectively allowed reproducible placement of the atomizer chamber rela- tive to the optical path. The r.f. generator and the impedance matching network employed were Model RFX 1250 (Advanced Energy) and AMNS- 1500E (PlasmaTherm) respectively. Thorough electri- cal grounding was achieved for proper impedance matching.The impedance matching network was a LC circuit which was tuned so that the total impedance of the network and the plasma equalled the output impedance of the r.f. generator (50 a). To minimize r.f. interference and to avoid power losses due to the large reactive current flowing about the circuit the impedance matching network was mounted close to the dis- charge chamber as suggested by the manufacturer (specific to this system). Since the original components of the matching network (AMNS-1500E) did not permit matching of the output impedance of the generator to the atomizer the matching network was modified. For tuning the matching network in order to achieve maximum power transfer and to protect the generator a series of laboratory-built induction coils and different capacitors connected both in series and in parallel were used.A schematic diagram of the impedance matching network is shown in Fig. 2. Gas Control System Argon (99.997% purity) was used as the sputter gas. A moisture trap (Chromatographic Specialities Model No. MT120-4) wasJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY JANUARY 1994 VOL. 9 47 Matchina network Phase and magnitude x i -7 - -1 I detector ‘ I I 1 L ‘ I Matching network control unit generator I R-f. I Fig. 2 Modified matching network L 2.OpH CT 25-500 pF; T 300 pF; CL 7-1000 pF; C 500 pF; LM Loading motor; TM Tuning motor; X 50 Q coaxial cable CT varying blocking capacitor; T blocking capacitor; CL variable loading capacitor; L inductor; and C capacitor used to remove water vapour from the sputtering gas as water vapour has been shown to have a detrimental effect when present in the GD.37938 The sputtering gas was then passed through a particulate filter (Matheson Gas Products).For experiments in which independent control of Ar flow rate and pressure was required a laboratory-constructed gas control system12 was employed. The Ar flow rate was regulated on the inlet line using a Nupro fine-regulating needle valve and was measured using the flow rate transducer of a mass flow controller (Sierra Instrument Model No. 740-05-2). This trans- ducer was capable of measuring up to five standard litres per minute (1 min-’) of Ar gas. The Ar gas was then admitted into the atomizer chamber through a 6.3 mm compression fitting mounted on the top of the chamber.The pressure was moni- tored on the outlet side using a Pirani gauge (Granville- Phillips Convectron Gauge Model No. 275). The gauge was calibrated for both N2 and Ar gas. The pressure was controlled by throttling the vacuum line using a Whitey ball valve and a Whitey regulating valve in a parallel configuration. The ball valve was used for coarse control of the gas pressure while the regulating valve gave fine control. A two stage vacuum pump (Edwards High Vacuum Model 5) fitted with an oil mist filter (Edwards High Vacuum Model EMF10) and a fore-line trap for eliminating the back diffusion of pump oil vapour into the sputtering cell were used for pressure control. For measure- ments in which independent control of Ar pressure and flow rate was not essential the gas delivery/pumping system of an ATOMSOURCE (Analyte Corporation,) was employed.It should be noted that the gas flow rate studied here was not directed at the sample surface. Measurement System The discharge source was mounted in the usual position of the burner assembly in a Varian-Techtron Model AA-5 atomic absorption spectrometer and was operated as part of the spectrometer. The absorbance profiles were recorded with a Nicolet Model 201 dual channel digital oscilloscope. The hollow cathode lamp (Varian) current was 4 mA and was electronically modulated at 1 kHz with a time constant of 1 ms used on the lock-in amplifier. The light beam of the hollow cathode lamp was focused to 6mm in diameter as it passed through the atomizer chamber.The image of the light beam was 10mm on both entrance and exit optical windows of the atomizer chamber. The wavelength chosen was the Cu I resonance line 324.7 nm. The spectral bandpass was 1 nm. The r.f. generator was used in the constant power mode which made the parametric evaluation straightforward as there was a complex interdependence of the pressure gas flow rate current and voltage. Procedure Cathodes of oxygen free hard copper (OFHC) were used. The cathodes were sputtered as 23 x 23 x 12 mm blocks with the 23 x 23 mm surface applied against the sealing O-ring. The working surface was polished to 400 grit with an electric polishing machine then cleaned with ethanol and air dried. The cathodes were first sputter-cleaned for at least 3 min to ensure a reproducible atomization surface as large changes in reproducibility were found with changes in the cathode sur- face.39 Whenever the ATOMSOURCE control box was used it was programmed in the flushing mode.The discharge chamber pressure was maintained at 10Torr for lOs after which the chamber was evacuated to 100mTorr and then re-pressurized to the pre-selected operating pressure. Whenever the laboratory-built gas control system was used the gas solenoid on the vacuum line was opened evacuating the chamber to approximately 200 mTorr. When this pressure was attained the gas solenoid on the Ar inlet was opened and the fine-regulating needle valve was adjusted to give the desired flow rate. When the required flow rate was attained the valve on the vacuum line was adjusted to give the desired pressure.When the chamber pressure reached operating pressure the oscilloscope and spectrometer were triggered just prior to applying the r.f. power. The r.f. generator was powered for at least 3 min to clean the Cu surface. For measuring absorbances of Cu atoms sputtered from the cleaned surface the r.f. generator was powered for at least lOs which was sufficient for a steady-state sputtering. The r.f. generator output power was adjusted manually from the front panel controls of the generator which displayed forward and reflected powers. The impedance matching network allowed the reflected power to be minimized automatically or matched manually by changing the capacitance of the loading or the tuning parts of the network whenever needed.The r.f. generator output power can differ considerably from the actual power delivered to the discharge owing to power losses in the transmission lines electrical connections and the impedance matching network. The power losses have however been shown to be a relatively constant fraction of the total generator output power over a range of powers for at least one discharge system.40 Sample loss rates for Cu at different discharge parameters were obtained using pre-weighed Cu metal sheets (24 x 24 x 2 mm) that were sputtered for 15 min. The mass loss was determined by re-weighing the cathodes on a microbalance (Mettler AT250) after allowing them to cool to room temperature. Results and Discussion Preliminary Experiments Because there was some fluctuation in reflected r.f.power during the course of sputtering (less than approximately 1% of forward power) it was decided to investigate how changes in reflected r.f. power of this magnitude would effect absorbance signals. The reflected r.f. power displayed on the front panel of the power supply was set up manually by changing the capacitance of the loading and/or tuning components of the matching network. The reproducibility and stability of the results showed that there was no significant variation in the absorbance in the 1% range of reflected powers as the relative standard deviation (RSD) associated with the absorbance measurement was about 1.1% (n= lo) which was used as the criterion to indicate any significant variation in absorbance. Uncertainties associated with positioning adjustment of the atomizer in both the horizontal and vertical position propagate 1.7% RSD of uncertainty in the absorbance measurement.48 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY JANUARY 1994 VOL.9 This uncertainty was measured by taking replicate (n = 6) absorbance measurements for a specific location (horizontal position of 2mm and vertical position of 6mm) and set of discharge conditions (pressure of 3.7 Torr Ar flow rate of 0.1 1 min-' and r.f. power of 40 W) moving the atomizer position and re-setting it in between each measurement. The optimum pre-sputtering time (ie. the time required to sputter a new cathode to achieve a representative s~rface)~' was obtained by reading absorbance versus sputtering time at an Ar gas pressure of 3.7 Torr flow rate of 0.1 1 min-' and applied power of 40 W.Fig. 3 shows that the absorbance increased and then remained essentially constant as a new cathode layer was sputtered. After approximately 3 minutes the RSD of absorbance was about 2% (n= 1365 data points) over the following 12 min period. In general pre-sputter times of 1-3 min are required to clean the surface and to achieve stable atomization rates under these moderate discharge con- ditions. For the remainder of this work a uniform pre- sputtering time of 3 min was adopted. Fig. 3 also illustrates an additional desirable quality of the r.f. GD-AAS source in its ability to operate stably over extended periods of time. The results show that the source is able to sustain a stable plasma for at least 3-15 min which would enable the sequential atomic absorption analysis of samples such as metal alloys.This type of temporal stability is also important for in-depth profiling. The stability of the source is limited by the re-deposition of a thin film of conducting sputtered materials within the recess region within the circumference of the O-ring seal as noted previously by Lazik and Marcus,41 effectively short-circuiting the cathode to the anode across the O-ring. Spatial Distribution of Analyte Atoms In order to optimize atomic absorption signals it is desirable to find a region in the chamber where the concentration of sample atoms is highest. Furthermore the GD is an inhomo- geneous medium,"2 a fact which makes knowledge of the spatial distribution of sputtered neutrals critical for AAS.With this goal in mind absorbance signals were obtained at different regions within the chamber. Blocks of OFHC were sputtered under different discharge conditions and the absorbance of Cu (Cu I 324.7 nm) was measured. Horizontal profiles were meas- ured at distances of 11 11.5 12 13 14 15 17 19 21 23 26 29 32 35 38 and 41 mm relative to the Cu cathode surface. Vertical profiles were measured in 2mm intervals over the distances of 0-12mm as well as 0 to -10mm relative to the 'horizontal plane' (i.e. the plane containing the central axis of the sampling orifice at which the vertical distance is equal to zero). 0.30 0.25 0) (D g 0.20 e a 2 0.15 0 0.10 0.05 0 1 1 1 1 1 1 1 2 4 6 8 10 12 14 16 Time/min Fig.3 Variation in absorbance of Cu atoms with sputtering time of surface layers at an argon gas pressure of 3.7 Torr (1 Torr= 133.322 Pa); flow rate of 0.1 1 min-' and applied power of 40 W (Cu I 324.7 nm).The horizontal and vertical distributions of atoms taken at different Ar gas flow rates pressures and applied r.f. powers are shown in Fig. 4(a)-(4. The atom concentration is seen to be highest near the orifice disc surface and to decrease rapidly at greater horizontal distances. Similar atom transport patterns were observed vertically on either side of the 'horizontal plane'. It was not possible to determine the concentration gradient if any existed from the cathode surface through the sampling orifice in this study because of the geometrical constraints of the chamber.Specifically it should be noted that the closest approach of the hollow cathode lamp beam to the cathode surface is about 11 mm so absorbance measurements closer than this to the cathode surface are not possible. It is not possible by atomic absorption measurements to know how atoms are distributed around the vertical plane (the plane which contains the central axis of the sampling orifice and is normal to the hollow cathode light beam). The fact that the concentration of sputtered atoms decreased further away from the orifice disc surface is most likely due to simple diffusion and convection processes. Aggregation of analyte atoms13 or other atom-consuming processes may also occur. Metal clusters in the GD have been observed e~perimentally~~ and could provide a possible explanation for decreasing the number of analyte atoms.The sputtered atoms in the central region of the chamber are subjected to two counter currents two Ar gas streams which direct the atoms out of this region and could make a rich reservoir of atoms above and below the 'horizontal plane'. The first stream is directed towards the vacuum port which pulls the atoms towards the top of the atomizer chamber while the second one results from the influx of Ar gas and acts to push the atoms towards the bottom of the atomizer chamber. In general Fig. 4 shows that at the lower Ar gas flow rate the sputtered atoms are more populated at the top-side region of the atomizer chamber where the vacuum port is located in comparison with the bottom-side. This study shows how the location of the vacuum port and the Ar gas inlet affects the concentration of sputtered materials in the optical path.Fig. 4(a) and (b) show the distributions of ground state Cu atoms at Ar gas flow rates of 0.4 and 0.1 1 min-' respectively but at a constant pressure of 8 Torr and an applied power of 40 W. The absorbance of the sputtered atoms in front of the sampling orifice is decreased at the higher Ar gas flow rate. Simple geometric considerations of the source design show that the Ar gas is neither directed at the sample (cathode) surface nor subjected to direct vacuum within the confines of the sampling orifice and consequently the supply of atoms to the atomizer chamber remains constant. This would suggest that the Ar gas flow rate does not show a distinct effect on the sputtering rate and on the transport efficiency of sputtered atoms within the sampling orifice.To verify this assumption sample loss rates for Cu were determined at Ar gas flow rates of 0.4 and 0.1 1 min-l at a constant pressure of 8 Torr and an applied power of 40 W. The sample loss rates were found to be 701 pg min-'(9% RSD n=5) and 694 pg min-' (5% RSD n = 5) respectively. This suggests that the flow rate of Ar gas does not appear to have a marked effect on the transport efficiency of atoms within the sampling orifice. The sputter gas flow rate could have two possible effects on the absorbances in different regions of the chamber. Firstly it could affect the sputtering rate if it alters the local pressure and consequently the d.c.self-bias potential. The dependence of d.c. self-bias potential on the gas pressure has been investigated by Winchester et aLZ2 Secondly it could affect the transport processes of analyte atoms through the atomizer volurne.l2 High Ar flow rates would act to decrease the absorbance by decreasing the residence time of sputtered analyte atoms in all regions of the chamber as is shown in Fig. 4(a) [in contrast to Fig. 4(b)]. This study seems to suggest the beneficial effect of the use of low gas flow at the surface.44 Such an experimental condition should help to sweep away sputtered atoms fromJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY JANUARY 1994 VOL. 9 49 10 5 0 - 5 E E 2 -10 \ 10 15 20 25 30 35 40 I I I I L 10 15 20 25 30 35 40 Horizontal distance/mm Fig.4 Spatial distribution of Cu atoms inside the atomizer. (a) Gas pressure 8 Torr; gas flow rate 0.4 1 min-'; applied r.f. power 40 W; and maximum absorbance 0.29. (b) Gas pressure 8 Torr; gas flow rate 0.1 1 min-'; applied r.f. power 40 W; and maximum absorbance 0.43. (c) Gas pressure 3.7 Torr; gas flow rate 0.1 1 min-'; applied r.f. power 40 W; and maximum absorbance 0.38. ( d ) Gas pressure 3.7 Torr; gas flow rate 0.1 1 min-'; applied r.f. power 55 W; and maximum absorbance 0.39 the sample (cathode) surface minimizing re-deposition and maximizing transport efficiency of atoms through the sampling orifice thus increasing the sensitivity of AAS. Fig. 4(b) and (c) reflect the spatial distribution of ground state Cu atoms at the pressures of 8 and 3.7 Torr respectively but at a constant Ar flow rate of 0.1 1 min-l and applied power of 40 W. The Cu atoms are more extended from the orifice disc surface towards the central region of the chamber at the lower pressure.However most of the sputtered atoms are still localized around the orifice disc surface as was observed in Fig. 4(a). In addition side-on viewing of the GD plume through the optical windows showed that the GD plume was more restricted at the pressure of 8Torr towards the orifice disk surface. Greater transport of Cu atoms and the extension of the discharge plume towards the central region of the chamber at the lower pressure could be due to the reduction in the number of energy-attenuating collisions as a result of increasing the mean free-paths of Cu and Ar atoms at low operating gas pressures.The high kinetic energy of sputtered atoms at low pressures definitely promotes the diffusion pro- cess. Greater transport of Cu atoms towards the central region (> 20 mm) of the chamber at the pressure of 3.7 Torr [Fig. 4(c)] is responsible for the higher observed absorbance in this region. The definite skewing of the spatial distribution due to the vacuum pumping can also be seen more clearly here. The maximum absorbance observed near the sampling ori- fice at a pressure of 8 Torr [Fig. 4(b)] is due to a number of factors. At the pressure of 8 Torr the plasma resides within the restricting orifice and probably provides for more efficient Ar ionizing collisions which consequently increases the amount of sputtered atoms.Additionally since the current is the dominant channel for increasing discharge powerzz and the root mean square ion current at the cathode increases with pressure for a given power then the rate of sample loss should increase with increasing pressure. Another explanation for the maximum absorbance observed near the sampling orifice at the higher pressure could be increased localization (higher residence time) of Cu atoms in this region. Fig. 4(c) and ( d ) show the effect of different applied r.f. powers of 40 and 55 W on the spatial distribution of ground state atoms at the constant pressure of 3.7Torr and an Ar flow rate of 0.1 1 min-l. There is not a great difference between absorbances or spatial distribution of sputtered atoms at both applied powers suggesting that the sputtering efficiencies are more or less equal for both conditions.Actually the amount of sputtered material would be expected to increase with increasing discharge power for a given set of pressure/flow conditions. Fig. 5 shows the absorbance of Cu as a function of applied r.f. power. In this experiment the argon gas pressure and flow rate were 3.7 Torr and 0.1 1 min-' respectively. It can be seen from Fig. 5 that there is a threshold energy of about 5 W at a pressure of 3.7 Torr above which the sputtering rate increases in a near-linear fashion with increases in dis- charge power levelling off at the highest discharge powers. The levelling effect is probably the result of a combination of factors sample heating at high discharge powers; spatial re-orientation of the plasma; and possible discharge power losses through radiation or contact heating.Increases in dis- charge power are known to affect all of these factors. In terms of what is directly observable sample heating seems to be the50 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY JANUARY 1994 VOL. 9 0.32 0.24 Q) u C 0.16 2 a 9 0.08 0 15 30 45 60 75 90 Power/W Fig. 5 Absorbance of sputtered Cu atoms as a function of applied r.f. power at a flow rate of 0.1 1 min-' and pressure of 3.7 Torr most likely factor. The target temperature of hundreds of degrees which are quite likely given the temperature of the bulk sample are long known to have deleterious effects in secondary ion mass spectrometry type sputtering of Cu target^.^' Excessive surface heating would also be likely to contribute to the observed increased error in the measurements through sample volatilization and variation in the secondary electron yields.General observations from Fig.4 are that the Ar gas flow produced less turbulent conditions for the glow itself at high pressures with low Ar flow rates as Fig. 4(b) shows that the absorbance decreases monotonically from the orifice disc sur- face. Higher flow rates of Ar gas cause a physical disruption of the glow restricting it towards the orifice disc area. A similar observation was reported by Banks and Blades21 for a d.c. jet-assisted GD source. It can also be clearly seen that the distribution pattern of atoms is more or less symmetrical around the 'horizontal plane'. The population of sputtered atoms is higher at the top- side region of the chamber than at the bottom-side.Fig. 4(b)-(4 precisely show the general trends of atoms being transported towards the top-side region of the chamber where the vacuum port is located. A decrease in absorbance develops in front of the sampling orifice at the vertical position of 2mm below the 'horizontal plane' for different parametric conditions as is shown for simplicity in Fig. 6. As Cu exists in the sample primarily as metallic atoms Cu clusters may also be sputtered and sub- sequently dissociated collisionally in the negative glow. We do not expect this observation to be due to an inconsistency in measuring-time because absorbance is measured at the plateau of the signal obtained in the steady-state mode of atomization.An equation fitted to the curves (shown in Fig. 6) was derived from absorbance versus horizontal distance. A constant applied power of 40 W and Ar gas flow rate of 0.1 1 min-l were employed while pressure was varied from 8 to 3.7Torr. This equation has the exponential form of A = A exp [ - k (d - 0.5 )] where A is absorbance A, is the maximum absorbance of the curve observed for each condition d is the distance from the orifice disc surface inmm and k is the concentration gradient constant which is pressure-dependent and has the dimension of nun-'. At pressures of 8 and 3.7 Torr the values of the concentration gradient constants are 0.20 and 0.11 mm-' respectively. These data indicate that the concen- tration gradient and/or any process which removes atoms under these conditions is pressure-dependent and this depen- dence is such that an increase in pressure brings about a sharp decrease in the number of atoms.One of the possible processes that is responsible for the concentration gradient of atoms is 0.4 0.3 Q) c CU -e g 0.2 2 0.1 0 10 15 20 25 30 35 40 Horizontal distance/mm Fig. 6 Spatial distributions of Cu atoms along horizontal distances from the orifice disc surface while atomizer is set at the vertical position of 2 mm bellow the horizontal plane. A Pressure 8 Torr; flow rate 0.4 1 min-'; and applied power 40 W. B Pressure 8 Torr; flow rate 0.1 1 min-'; and applied power 40 W. C Pressure 3.7 Torr; flow rate 0.1 lmin-'; and applied power 40 W. D Pressure 3.7 Torr; flow rate 0.1 1 min-l; and applied power 55 W the aggregation process which is directly promoted by increas- ing pressure and its effect is more pronounced as atoms are transported from the negative glow farther from the orifice disc surface into regions where there are fewer dissociating collisions.The observation of an initial maximum in each of the spatial profiles is not entirely unexpected. Early studies by Stirling and W e s t ~ o o d ~ ~ in the modelling of low pressure (0.1 Torr) sputter deposition systems revealed such a maximum in atomic absorption data for Al. The position of this maximum coincided with the edge of the negative glow/dark space (NG/DS) interface which existed = 1 cm above the cathode surface for this lower pressure system. These workers46 suggested that this was evidence for a mechanism whereby fast ions in the dark space dissociate molecular species ejected from the cathode surface.Ferreira and Human4' came to a similar conclusion in mapping the sputtered atom population within a Grimm- type discharge source. In this case the higher operating pressure of the Grimm device ( ~ 3 Torr) resulted in maxima located z 1.5 mm above the cathode surface again correspond- ing to the NG/DS interface. Obviously the maxima absorbances illustrated in this paper do not correspond to the NG/DS region. Atomic absorption studies by McDonald43 have shown that appreciable amounts of agglomeration occur at distances as much as 2 cm from the cathode surface in a flow-assisted GD atomization source. In fact spatially and temporally-resolved measurements for Cu cathode sputtering showed maxima of the type illustrated in the present paper at distances of < 2 cm from the sample (cathode) surface.48 No single atomic absorp- tion experiment has in fact covered the entire spatial distance up to 3 cm with resolution of the order of 1 mm.Studies in both d.c. and r.f. GDMS have indicated that substantial populations of dimeric species (M,+and MAr') do indeed exist at regions removed from the interfacial region with their relative intensities decreasing at distances farther into the p l a ~ m a . ~ ~ * ~ * This would suggest that collisions within the NG may have a cumulative effect in dissociating such species. While more detailed study of these types of spatial distribution is in order it would appear that a sequence whereby two maxima occur may be possible; the first resulting from dis- sociation of sputtered molecules in the region of the NG/DS interface and the second from dissociation due to successive collisions with discharge gas atoms.JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY JANUARY 1994 VOL.9 51 0.40 0.30 a t m e g 0.20 a < 0.10 0 B J? 3 4 5 6 7 8 9 1 0 Ar gas pressureflorr Fig. 7 Absorbance of Cu as a function of Ar gas pressure and flow rate at a constant applied r.f. power of 35 W. A flow rate 0.1 1 min-'; B flow rate 0.21min-'; C flow rate 0.31 min-'; D flow rate 0.4 1 min-'; E flow rate 0.5 1 min-'; F flow rate 0.6 1 min-'; G flow rate 0.7 1 min-'; and H flow rate 0.8 1 min-' Fig. 7 shows the absorbance of Cu atoms sputtered from a Cu sheet as a function of both Ar gas pressure and flow rate at a constant applied power of 35 W while the atomizer was set up at the vertical distance of zero and at the horizontal distance of 12 mm.At constant pressure the absorbance values reach a maximum at a flow rate of approximately 0.2 1 min-' and then decrease. The occurrence of flow rate maxima indicate a trade off between atom transport uia diffusion processes and convection by the pumping system. Higher flow rates act to decrease the number of atoms in the analysis volume. No results for high flow rates at low pressures could be obtained because of limitations in the capacity of the vacuum pump associated with the apparatus. But it is clear that there is always a tendency for absorbance to decrease with increasing flow rate at a constant pressure and power beyond the point of the maximum (z 0.2 1 min-I).In addition Fig. 7 shows the dependence of absorbance on Ar gas pressure. A decrease in the pressure at a constant flow rate brings about a maximum in absorbance. This occurs approximately at a pressure of 5 Torr for flow rates of less than 0.4 1 min-'. Decreases in the absorbance at pressures in excess of 5 Torr is probably a result of increased re-deposition of sputtered atoms as shown by the Cu sample loss rates presented in Table 1. At pressures less than the optimum the amount of discharge current for a given power drops substantially. Thus optimum pressure provides a compromise between ionizing collisions (ie. discharge cur- rent) and re-deposition of analyte atoms. Ideally this optimum pressure should be the lowest possible operating pressure at the highest feasible current. Thus the pressure that provides the maximum number of sputtered atoms in the GD must be determined experimentally.The data presented in this study illustrate the existence of Table 1 Mass loss rates of Cu at different sputter Ar pressures. Copper atoms were sputtered at applied r.f. power of 30 W. ATOMSOURCE control box was used for controlling the pressure Pressure*/ Sputtering rate/ Torr pg min-' 2 4 5 6 10 380 685 670 430 400 ~~~~ ~ ~ *1 Torr= 133.322 Pa. spatial inhomogeneities within the chamber and energetic regions within the glow which indicate the necessity of the adjustable optical mount. It is shown that the most useful region for AAS of Cu atoms and ions is around the sampling orifice disc.The results of this study are consistent with those of many d.c. G D systems and seem to support the notion that the r.f discharge acts as a d.c. discharge.32 In the future it is planned to study the geometrical effects of the sampling orifice (e.g. the diameter and thickness of the sampling orifice disc) as well as to investigate the effect of a pulsed applied r.f. power. By applying a pulsed mode of atomization it will be possible not only to obtain higher sensitivity as a result of increases in the sputter-atomization rate but also to study the residence times of sputtered Cu atoms within the chamber. Pulsed operation has the advantage of applying higher average discharge power thereby avoiding significant increases in sample surface temperature while still maintaining a reasonably short analysis time.For AAS a pulsed mode of atomization not only increases sputtering rate but also cancels the background emi~sion.~' Winchester and Marcuss2 have studied the effects of a pulsed GD on the emission intensities of Cu lines. They have reported anomalies in the temporal emission intensity waveforms for some analyte transitions. Investigation of the spatial profiles of the average electron energy and the electron temperatures may provide us with greater verification of the observed results. This study seems to support the desirability of having the vacuum port directly opposite the orifice disc location. Such a design will help to pull away sputtered atoms from the sampling orifice increasing the population of sputtered atoms by controlling the convection process increasing it is hoped the number density of analyte atoms and the sensitivity by AAS.Conclusions The spatial distribution of Cu atoms sputtered from an OFHC sample by an r.f. (13.56 MHz) powered GD atomization source has been investigated by AAS. The data were taken from inside and outside the negative glow at different discharge parameters of pressure sputter gas (Ar) flow rate and applied r.f. power. The results show that the GD plasma is an inhomogeneous medium so that the sampling position is critical for any information detected by AAS and by other atomic spectro- metric methods. It was seen that the distribution pattern of atoms is more or less symmetrical around the horizontal plane which is a plane containing the central axis of the sampling orifice.It is not feasible by atomic absorption measurements to know how atoms are distributed within the sampling orifice and around the vertical plane the plane containing the central axis of the sampling orifice normal to the hollow cathode light beam. Side-on viewing of the GD plume through the optical windows showed that the GD plume is less restricted towards the sampling orifice at the 3.7Torr than at the 8Torr. The population of sputtered atoms is higher at the top-side region of the chamber than at the bottom-side with a general trend for atoms to be transported towards the top-side of the chamber where the vacuum port is mounted. Independent control of the Ar flow rate and pressure was found to be critical in determining the spatial distribution of Cu atoms within the chamber volume.The results suggest future simul- taneous spatial measurements of sputtered atoms ions meta- stable species and clusters as well as investigation of the average electron and ion energies and the electron temperature in order to get more information about the processes which occur in the GD. The surface history of the cathode seems to be of great significance. Large changes in reproducibility appeared to correlate with the changes in the cathode surface. This study will aid in modification of the geometry of the atomizer to enhance its performance for AAS while still rendering it applicable for other spectrometric methods.52 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY JANUARY 1994 VOL.9 The authors are grateful to the Natural Sciences and Engineering Research Council of Canada for financial support. C.L. and R.K.M. gratefully acknowledge the National Science Foundation of USA (grant # CHE-911752) for partial support in this collaborative project. 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 References Gough D. S. Anal. Chem. 1976,48 1926. Chakrabarti C. L. Headrick K. L. Bertels P. C. and Back M. H. J. Anal. At. Spectrom. 1988 3 713. Chakrabarti C. L. Headrick K. L. Hutton J. C. Bicheng Z. Bertels P. C. and Back M. H. Anal. Chem. 1990 62 574. Harrison W. W. and Bentz B. L. Prog. Anal. Spectrosc. 1988 11 53. Broekaert J. A. C. J. Anal. At. Spectrom. 1987 2 537. von Hippel A. Ann. Phys. 1926 80 672.Wehner G. K. Adv. Electron. Electron Phys. 1955 7 239. Woodyard J. R. and Cooper C. B. J. Appl. Phys. 1964,35,1107. Chakrabarti C. L. Headrick K. L. Hutton J. C. Marchand B. and Back M. H. Spectrochim. Acta Part B 1989 44 385. Chakrabarti C. L. Headrick K. L. Hutton J. C. Bertels P. C. and Back M. H. Spectrochim. Acta Part B 1991 46 183. Kim H. J. and Piepmeier E. H. Anal. Chem. 1988 60 2040. Hutton J. C. Chakrabarti C. L. Bertels P. C. and Back M. H. Spectrochim. Acta Part B 1991 46 193. Hannaford P. and Walsh A. Spectrochim. Acta Part B 1988 43 1053. Gough D. S. Hannaford P. and Walsh A. Spectrochim. Acta Part B 1973 28 197. Human H. G. C. Ferreira N. P. Kruger R. A and Butler L. R. P. Analyst 1978 103 469. Smith B. W. Omenetto N and Winefordner J. D. Spectrochim. Acta Part B 1984 39 1389.Patel B. M. and Winefordner J. D. Spectrochim. Acta Part B 1986 41 469. Travis J. C. Turk G. C. Watters R. L. jr. Yu L. J. and Blue J. L. J. Anal. At. Spectrom. 1991 6 261. Lunyov 0. S. and Oshemkov S. V. Spectrochim. Acta. Part B 1992 47 71. Banks P. R. and Blades M. W. Spectrochim. Acta Part B 1989 44 1117. Banks P. R. and Blades M. W. Spectrochim. Acta Part B 1991 46 501. Winchester M. R. Lazik C. and Marcus R. K. Spectrochim. Acta Part B 1991 46 483. Paschen F. Ann. Phys. 1916 50 901. Boumans P. W. J. M. Anal. Chem. 1972 44 1219. 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 Bengtson A. Eklund A. Lundholm M. and Saric A. J. Anal. At. Spectrom. 1990 5 563. Winchester M. R. and Marcus R.K. J. Anal. At. Spectrom. 1990 5 575. Coburn J. W. and Harrison W. W. Appl. Spectrosc. Rev. 1981 17 95. Harrison W. W. Hess K. R. Marcus R. K. and King F. L. Anal. Chem. 1986 58 341A. Harrison W. W. in ‘Inorganic Mass Spectrometry ’ eds. Adams F. Gijbels R. and Van Grieken R. Wiley New York 1988 p.85. Shao Y. and Horlick G. Spectrochim. Acta Part B 1991,46 165. Jakubowski N. Stuewer D. and Vieth W. Anal. Chem. 1987 59 1825. Chapman B. ‘Glow Discharge Process’ Wiley New York 1980. Dogan M. Laqua K. and Massmann H. Spectrochim. Acta Part B 1972 40 65. Anderson G. S. Mayer Wm. N. and Wehner G. K. J. Appl. Phys. 1962 33 2991. Davidse P. D. and Maissel L. I. J. Appl. Phys. 1966 37 574. Ferreira N. P. Strauss J. A. and Human H. G. C. Spectrochim. Acta Part B 1983 38 899. Gough D. S. Hannaford P. and Lowe R. M. Anal. Chem. 1989 61 1652. Larkins P. L. Spectrochim. Acta Part B 1991 46 291. Bruhn C. G. and Harrison W. W. Anal. Chem. 1978,50 16. Horwitz C. M. J. Vac. Sci. Technol. 1983 Al 1795. Lazik C. and Marcus R. K. Spectrochim. Acta Part B 1992 47 1309. Duckworth D. C. and Marcus R. K. Anal. Chem. 1989,61,1879. McDonald D. C. Anal. Chem. 1982 54 1057. Lazik C. and Marcus R. K. Spectrochim. Acta. Part B 1993 48 863. Carter G. and Colligan J. S. Ion bombardment of Solids Elsevier NY 1968 ch. 7. Stirling A. J. and Westwood W. D. J. Appl. Phys. 1970 41 742. Ferreira N. P. and Human H. G. Spectrochim. Acta Part B 1981 36 215. McDonald D. C. Anal. Chem. 1982 54 1052. King F. L. McCormack A. L. and Harrison W. W. J. Anal. At. Spectrom. 1988 3 883. Duckworth D. C. and Marcus R. K. J. Anal. At. Spectrom. 1992 7 711. Winchester M. R. Hayes S. M. and Marcus R. K. Spectrochim. Acta Part B 1991 46 615. Winchester M. R. and Marcus R. K. Anal. Chem. 1992,64,2067. Paper 3/01 267K Received March 3 1993 Accepted September 30 1993.

ISSN:0267-9477    

DOI:10.1039/JA9940900045

出版商:RSC    

年代:1994

数据来源: RSC

摘要:

JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY JANUARY 1994 VOL. 9 53 Matrix Effects of Potassium Chloride and Phosphoric Acid in Argon Inductively Coupled Plasma Atomic Emission Spectrometry Bojan BudiC and Vida Hudnik National Institute of Chemistry Hajdrihova 79 POB 30 67 7 75 Ljubljana Slovenia Changes in analyte emission intensity in the presence of low concentrations of potassium chloride and phosphoric acid were studied in inductively coupled plasma atomic emission spectrometry. It has been shown that both matrices can cause significant depressant effects on the relative emission intensities of Sr Mg Mn Fe Cu and Zn under normal analytical operating conditions and that the lowering of intensity was more pronounced for KCI than for H,PO at the same analyte to matrix concentration ratio.The matrix effects can be corrected by increasing the power supplied to the plasma which was found to correlate with the excitation energy of the emission line. Keywords Inductively coupled plasma; atomic emission spectrometry; matrix interference; radiofrequency forward power; matrix effects correction Although chemical matrix effects are less severe in inductively coupled plasma atomic emission spectrometry (ICP-AES) than in other spectrometric techniques they can influence the intensity of the emission signal and thus introduce errors into the analytical procedure. It has been shown that such effects occur not only for samples with relatively high concentrations but even for samples with relatively low concentrations of concomitant elements or mineral acids.'-7 From the extensive studies in this field it has been found that generally in the normal analytical zone of the plasma matrix effects cause depression of analyte emission signals for both easily ionizable elements (EIEs) or non-EIEs as concomitant species8 To understand the mechanisms by which these matrix effects occur investigations concerning transport processesg-" and those in the plasma it~elfl"'~ were carried out.Since emission from the ICP source is heterogeneous and the processes that produce emitting species are still not completely understood no one model is able to predict exactly which conditions are optimum for analyte excitation. To compensate for salt/acid effects different methods can be used including matrix match- ing standard additions internal reference or the method based on the measurement of the intensity of the hydrogen (HP) emission line at 486.13 nm.14 However these methods are either time consuming or do not eliminate matrix effects completely because the analyte emission signal seems to be dependent both on changes in transport processes and changes in the plasma which are not a linear function of the matrix concentrat ion.' Matrices containing phosphate or alkaline elements are frequently encountered in the analysis of different minerals such as silica after dissolution with phosphoric acid or after fusion with alkaline salts or hydroxides. In the present study the effects of potassium chloride and phosphoric acid concen- tration on emission signals of elements with different ionization potentials and excitation energies (Sr Mg Mn Fe Cu Zn) were investigated.Power supplied to the plasma was varied from 1.0 to 1.6 kW while other experimental conditions were kept constant and the possibility of correcting for salt/acid effects by increasing forward power is presented. Experimental Apparatus An Applied Research Laboratories ARL 3520 OES sequential vacuum spectrometer equipped with an SAS 11 automation system for instrument control data acquisition and data manipulation was used. The instrumentation and operating conditions used which remained virtually constant with the exception of the forward power are listed in Table 1. Samples Table 1 Instrumental and operating conditions Spectrometer Grating R.f. generator Plasma torch Nebulizer Argon flow rate Observation height Solution uptake rate ~~ Monochromator with 1 m radius concave grating in Paschen-rounge mounting Linear dispersion 0.926 nm mm-' Quartz-controlled 27.12 MHz and automatic network.Operating power between 1.0 and 1.6 kW Fassel type Glass Meinhard type TR-30-3A Inner 1.0; intermediate 0.8; outer 12 15 mm above the load coil 2.2 ml min - (unforced) 1 min-' were nebulized without the use of a peristaltic pump and six replicate measurements were made. The spectral characteristics of the analyte lines studied are presented in Table 2. Reagents Analytical-reagent grade salts were used to prepare all solu- tions. The analyte concentration was lOpgml-' and the matrix concentrations were 0.0313 0.0625 0.125 0.25 and 0.5 moll-l. For temperature measurements iron was added at a concentration of 50 pg ml-' to KCl or &PO,.Table 2 Wavelengths excitation potentials (Ed and ionization poten- tials (Ei) of the spectral lines considered Spectral line A1 I1 Sr I Sr I1 Mg I1 Mn I Mn I1 Fe I Fe I Fe I1 Fe I1 c u I c u I1 Zn I Zn I1 Mg 1 Wavelength/nm 260.92 460.73 407.77 285.2 1 280.27 279.48 257.61 385.99 360.67 256.69 275.33 324.75 224.70 213.86 202.55 EJeV 4.62 2.69 3.04 4.35 4.42 4.44 4.81 3.21 6.13 5.91 7.77 3.82 8.24 5.80 6.12 EJeV 5.98 5.69 7.64 7.43 7.90 - - - - - - 7.72 9.39 - -54 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY JANUARY 1994 VOL. 9 Excitation Temperature and Electron Number Density Measurements In order to determine the excitation temperature a set of iron lines reported by Blades and Caughlin16 was used by applying the Boltzmann plot method.For the determination of the effect of matrix concentration and power supplied to the plasma on the excitation temperature the method described by Houk et which does not require transition probabilities was used T,dE*ij AE* - kT, 1nR T, = where Ii/Ij =ratio of line intensities at matrix concentration or forward power other than at the reference point (Ii/Ij)ref= ratio of line intensities at the reference point xef = reference value of the excitation temperature E*i(i) =excitation energy of state i( j) above the ground state and I =net line intensity of state i or j respectively. Electron number density was measured using the HB 486.13 nm line." A correction for instrumental and Doppler broadening was applied to the H/J profile before calculation of the electron number density.Results and Discussion Influence of Matrix Concentration on Analyte Emission Intensities Measured emission intensities in different matrices are depen- dent on the concentration of matrix element. It was found that matrix effects are strongly spatially dependent and therefore enhancement or depressant effects can occur at different spatial positions of the plasma.' Ramsey et d3.' observed a depression of emission intensities in the presence of matrices and found that the magnitude of the depression was dependent on both the matrix and the analyte emission line. A decrease in excitation temperature in the presence of a matrix was found to be relationed to the lowering of the analyte emission sensitivity.' Similar conclusions were reached by Yoshimura et aL2 for the presence of mineral acids.Olesik and Williamsen7 found by measuring both fluorescence and emission intensities in the presence of different matrices that the number of ground-state ions decreased in the presence of matrix elements but more analyte ions were excited. Therefore matrix effects can counteract each other. The influence of KCl and H3P04 concentration on the relative intensities of Sr Mg Mn Cu and Zn atom and ion emission lines was measured. The analyte concentration was constant while the matrix concentration was varied between 0 and 0.5 moll-'. For these measurements forward power was kept constant at 1.3 kW. The matrix effect M is defined as the percentage difference in the net line signals between solu- tions of the analyte with and without matrix.For the matrices studied a decreasing effect is observed (Figs. 1 and 2). At the same molar ratio of KC1 and H3P04 to the analyte the matrix effect in the presence of KC1 [Fig. l(a) and (b)] was greater than when H3P04 [Fig. 2(a) and (b)] was added. Also the difference in matrix effect between analytes studied was larger for KC1 than for H3PO4 under the same experimental con- ditions. From Figs. 1 and 2 it is also obvious that the depressing effect on signals for atomic lines of Mn Mg Cu and Zn is only slightly greater than that on the ionic lines of the same elements. The matrix effect (M,%) in the presence of 0.5 mol I-' H,P04 is in the range 7-10% while at the same concentration of KCl this effect is more pronounced (12-15%) with the exception of Sr atomic and ionic emission lines [Fig.l(a) and (b)]. Yoshimura et aL2 obtained similar depressant effects when 0 -5 - 10 - 15 - 20 I I I I I -25 I I I I I I 0 0.10 0.20 0.30 0.40 0.50 [KCll/mol I-' Fig. 1 Matrix effect (M%) on (a) atomic and (b) ionic emission lines as a function of KCl for A Mn; B Cu; C Mg; D Zn; and E Sr. Concentration M(%) is defined as [(Zm-1,,)Zn] x 100 where I and I, refer to the net line emission intensity in water solutions and in the presence of matrix respectively Zn Ca and Mg emission intensities were measured in the presence of HNO and H2S04. To find a correlation between a change in the physical state of the plasma and observed depressant effects in the presence of matrix the electron number density ne and excitation temperature T,, were measured.Electron number density was established to be 2.15 x lo1' cm-3 under the experimental conditions applied. No difference was observed when either KCl or H3P04 were added at concentrations of up to 0.5 mol I-'. Because of the lack of local thermodynamic equi- librium in the ICP,I9 various species and states can show a considerable difference in T,, and emission lines having different excitation potentials could be considered to have their own excitation temperature. However for analyte excitation studies in the ICP this parameter is often used." In the present experiments regardless of the addition of KCl or H3P04 T, was found to be fairly constant ie. 6490 K. The uncertainty in the temperature measurement was estimated to be about +300 K.In addition the method described by Houk et which does not require transition probabilities was used. For a comparison of T, for atomic and ionic spectral lines two sets of iron line pairs were chosen Fe I 360.67/Fe I 385.99 nm and Fe I1 275.33/Fe I1 256.69 nm. As a reference value an excitation temperature of 6490 K for the blank solution was chosen. The changes in excitation temperature determined in this way did not exceed a value of & 90 K. These findings disagree with the observations of Yoshimura et a!.,' which could be owing to somewhat different operating conditions. Since no change in n and T, in the plasma was observed when KC1 or H3P04 were added it seems more probable that in this case transportation and vaporization effects predomi- nate.For a detailed study of this phenomenon it will beJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY JANUARY 1994 VOL. 9 55 0 -5 - 10 2 5500' % 7000 r I I 1 I !!? 1 -I5 t -20 t -25 I I I 1 I 0 0.10 0.20 0.30 0.40 0.50 IH,PO,l/mol I - ' Fig.2 Matrix effect (M%) on (a) atomic and (b) ionic lines as a function of H,PO concentration for A Mn; B Cu; C Mg; D Zn; and E Sr. M(%) as in Fig. 1 necessary to measure the spatial distributions of the emission and fluorescence intensity of analyte lines n and T,,,. Effect of Forward Power on Relative Emission Intensities and Excitation Conditions Experimental conditions such as carrier gas flow rate or power supplied to the plasma can significantly effect the plasma conditions and thus the relative analyte emission intensit- ies.16*20 Forward power is one of the experimental parameters that can be used for ICP optimization and the effects of forward power on emission intensities and signal-to-back- ground ratios have been widely described in the literature.21 In order to assess the possibility of correcting matrix effects by changing the power supplied to the plasma in the presence of KCI and H3P04 forward power was varied between 1.0 and 1.6 kW in increments of 0.15 kW while other operating conditions and matrix concentration remained constant.The elements Al Sr Fe Cu Mg Mn and Zn were considered. 3.5 3.0 2.5 2.0 1.5 1 .o 0.5 I I I 0.95 1.10 1.25 1.40 1.55 Forward power/kW Fig. 3 Influence of increasing forward power on the electron number density (n,) Fig.4 Relationship between forward power and T, calculated by using a reference temperature of 6490K for (a) atomic and (b) ionic Fe emission lines Electron number density and T, were measured in the same increments as analyte emission intensities and are pre- sented in Figs. 3 and 4. Since the main objective was to find a relationship between changes in T, on one hand and forward power on the other the method on the relative basis without transition probabilities mentioned above was applied. From Fig. 3 it is evident that n increases almost linearly when forward power increases and a similar observation was made for T, (Fig. 4). As a reference point a temperature value of 6490K determined by the Boltzmann plot method with a forward power of 1.3 kW was chosen.The difference between T, calculated for Fe I and Fe I1 lines (intensities were measured for six replicates) is evident indicating the lack of local thermodynamic equilibrium conditions and the tendency for T, for atomic lines to be more affected by changes in forward power. The effect of the forward power on relative emission intensity for Sr Mg and Zn is illustrated in Figs. 5 and 6. From Fig. 5 it can be seen that relative atomic emission intensities (each value was normalized to that of the highest value) behave quite differently depending mainly on the excitation energy (Ex) of the analyte emission line. It is obvious that at lower values of Ex relative intensity decreases [Sr 460.73 nm line with Ex=2.69 eV Fig. 5(a)] a similar effect was observed at the Fe 385.99 nm line (Ex= 3.21 eV) whereas emission intensity for atomic lines of Mg Mn and Zn having Ex values between 4.35 and 4.44 eV increased when forward power was increased from 1.0 to 1.6 kW.Plots for Mg and Zn are presented in Fig. 5(b) and (c). The relative emission intensity of the Cu atomic line at 324.75 nm (Ex = 3.82 eV) remains quite unaffec- ted when forward power is increased from 1.0 to 1.6 kW. It is most likely that both processes i.e. a decrease in the analyte atom population and an increase in the analyte excitation contribute to the changes in relative emission intensities when forward power is increased. For analyte ion emission lines (illustrated in Fig. 6 for Sr Mg and Zn) increasing the forward power causes a significant increase in signal for all the lines considered with the exception of the Sr I1 407.77 nm line [Fig 6(a)].In the latter case similar behaviour for Sr as for Ba ionic emission can be assumed as in the investigation of Olesik22 who found that the increase in the Ba ionic emission signal was due to increased excitation56 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY JANUARY 1994 VOL. 9 0.6 0.4 0.2 A ( b ) 1.1 1.0 - 0' 1 I I I 1.1 1 1 .o 0.9 0.8 0.7 0.6 0.5 0.4 1.00 1.15 1.30 1.45 1.60 Forward power/kW Fig. 5 Relative atomic emission intensities as a function of the forward power A without matrix; and in the presence of B KCl and C H,PO for (a) Sr (b) Mg and (c) Zn rather than to a change in the Ba ion number density which decreased when forward power was increased from 0.5 to 1.0 kW.From Figs. 5 and 6 it is obvious that in the presence of KC1 or H3P04 the depressant effect occurs to a similar extent for both atomic and ionic emission lines between 1.0 and 1.6 kW forward power and that lowering of relative emission intensit- ies is slightly greater in the presence of a KCl than in a H3P04 matrix. For correction of the matrix effects for ionic emission lines an increase of the forward power is needed. In Fig. 7 the correlation between Etot (which is the sum of excitation and ionization potential) and AP the power necessary to correct for the matrix effect of 0.5mo11-1 H3P04 using an initial forward power of 1.1 kW is presented. From Fig. 7 it is evident that in the range of E, between 12 and 16 eV (where most of the analytically useful lines are found) an increase in AP to about 25 W to correct for the H3P04 matrix effect is necessary.At lower values of E, much higher forward power for the correction of matrix effects is needed however for emission lines (such as the Sr 407.77 nm ionic line) which are subject to a large matrix effect (M>20%) and have low E,, matrix effects cannot be eliminated completely in this way. It is also shown (Fig. 7) that the behaviour of the Cu ionic line (E,,,= 15.96eV) differs from other analytes. It is interesting to note U 1.00 1.15 1.30 1.45 1.60 Forward power/kW Fig. 6 Relative ionic emission intensities as a function of the forward power A without matrix; and in the presence of B KC1; and C H,PO for (a) Sr (b) Mg and (c) Zn 17 cu Fe 14 < 1 3 - Q2 12 - 11 10 - - 9 - 8 ' I 1 I I 0 50.00 100.00 150.00 200.00 250.00 AP/W Fig.7 Relationship between E, of Sr 430.55; Sr 407.77; A1 260.92; Fe 256.69; Cu 224.70; Mn 257.61; Mg 280.27; and Zn 202.55 nm ionic emission lines and forward power increase needed to compensate for the H,PO (0.5 moll-') acid effect. A value of 1.1 kW was chosen for the initial forward powerJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY JANUARY 1994 VOL. 9 57 Table3 Results of the determination of Zn Fe Mn Mg and Sr in the presence of H,PO (0.4 mol I-'). Results given in pg ml-' Element Wavelength/nm A* Bt c D4 Zn 202.55 130 120 130 131 Fe 256.69 38 35 38 38 Mn 257.61 115 110 116 115 Mg 280.27 155 147 157 155 Sr 407.77 265 232 234 262 Sr 430.55 240 247 267 - * A In solution. 7 B Without matrix effect correction.1 C Matrix effect correction by forward power compensation. 0 D Matrix effect correction by sample matching. that the intensity of the same Cu emission line deviated from some other elements when the relationship between relative emission intensity for ionic lines was plotted against excitation potential for different concentrations of mineral acids.2 It is possible since the Cu 224.7 nm line lies near the ionization limit of Ar (16 eV) that this line is not excited by the same process as other lines. The relationship between relative atomic emission lines and forward power (Fig. 5 ) shows that for the correction of matrix effects for these spectral lines much larger changes (e.g. a decrease for the Sr 460.73 nm line) of power supplied to the plasma are needed than for the ionic lines.Applications The procedure described above was also used in the practical analysis of H,P04 samples containing Mg Mn Fe and Zn at lower concentration. In Table 3 a comparison of the results of the analysis are presented when no power correction was applied (using calibration curves for water solutions) when there was an increase in AP of 25 W for all the elements analysed and when calibration curves obtained with sample matching were used. The forward power was set up initially at 1.1 kW. The analyte and synthetic solutions contained 0.4 moll-' H3P04. A relative standard deviation of less than 3% (n = 6) was obtained. Good agreement between results obtained with power correction and matrix matching except for the Sr 407.77 nm line was achieved.Power correction is probably preferable because it is easy controlled and the preparation of synthetic solutions for matrix matching is time consuming. Thus for practical analytical problems when ana- lysing elements in the presence of matrix elements at higher concentrations the ionic emission lines with higher values of Etot could be corrected more easily for matrix effects by changing the power supplied to the plasma than could ionic emission lines having low values of Etot or atomic emission lines. However correction could be effectively performed for the emission lines whose intensities are influenced to a similar extent by the matrix. Conclusions Both KC1 and H3P04 matrices have a significant depressant effect on analyte emission intensities in ICP-AES.The matrix effects were enhanced when the concentration of KC1 and H3PO4 was increased up to 0.5 moll- Atomic lines appear to be more subject to changes in matrix concentration and the depressant effect was slightly greater when KC1 was added to the analytes in the same molar ratio as H3P04. No change in n and T, in the ICP was observed in the presence of matrix. Under the operating conditions applied the magnitude of the depressions of the analyte emission signals does not appear to correlate with the ionization potential or the excitation energy. It seems more probable that concomitant elements have a complex influence on analyte emission intensities which is a function of a greater number of parameters. Radially resolved emission and fluorescence data should be useful for the detailed study of matrix effects.By changing the power supplied to the plasma from 1.0 to 1.6 kW it was shown that the matrix effect does not change significantly. A correlation between the total excitation energy in a range of about 12-16eV for ionic emission lines and forward power required to correct for matrix effects was found. In this range of excitation energies for ionic emission lines a relatively small increase in forward power (about 20-30 W) was found to be needed to achieve similar performance to that found without a KCl or H3PO4 matrix. 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 References Shen X.-e. and Chen Q.4 Spectrochim. Acta Part B 1983 38 115. Yoshimura E. Suzuki H. Yamazaki S.and Toda S. Analyst 1990 115 167. Ramsey M. H. Thompson M. and Walton S. J. J. Anal. At. Spectrom. 1987 2 33. Thompson M. and Ramsey M. H. Analyst 1985 110 1413. Ramsey M. and Thompson M. J. Anal. At. Spectrom. 1986 1 185. Ramsey M. H. and Thompson M. J . Anal. At. Spectrom. 1987 2 497. Olesik J. W. and Williamsen E. J. Appl. Spectrosc. 1989,43 1223. Sun D.-h. Zhang Z.-x. Qian H.-w. and Cai M.-x. Spectrochim. Acta Part B 1988 43 391. Farino J. Miller J. R. Smith D. D. and Browner R. F. Anal. Chem. 1987,59 2303. Bates L. C. and Olesik J. W. J. Anal. At. Spectrom. 1990,5 239. de Loos-Vollebregt M. T. C. Peng R. and Tiggelman J. J. J. Anal. At. Spectrom. 1991 6 165. Blades M. W. and Horlick G. Spectrochim. Acta Part B 1981 36 881. Rybarczyk J. P. Yester C. P. Yates D. A. and Koirtyohann S. R. Anal. Chem. 1982 54 2162. Botto R. I. Spectrochim. Acta Part B 1985 40 397. Olesik J. W. and Den S.-J. Spectrochim. Acta Part B 1990 45 731. Blades M. W. and Caughlin B. L. Spectrochim. Acta Part B 1985 40 579. Houk R. S. Shoer J. K. and Crain J. K. Spectrochim. Acta Part B 1987 42 841. Griem H. R. Spectral Line Broadening by Plasmas Academic Press New York 1974. Caughlin B. L. and Blades M. W. Spectrochim. Acta Part B 1984,39 1583. Olesik J. W. and Bradley K. R. Spectrochim. Acta Part B 1987 42 377. Boumans P. W. J. M. and de Boer F. J. Spectrochim. Acta Part B 1977 32 366. Olesik J. W. Spectrochim. Acta Part B 1990 45 975. Paper 3/04528E Received July 28 1993 Accepted September 21 1993

ISSN:0267-9477    

DOI:10.1039/JA9940900053

出版商:RSC    

年代:1994

数据来源: RSC

摘要:

JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY JANUARY 1994 VOL. 9 CUMULATIVE AUTHOR INDEX JANUARY 1994 Absalan G. 45 Back M. H. 45 Berndt Harald 39 Branch Simon 33 Briand Alain 17 BudiE Bojan 53 Chakrabarti C. L. 45 Chartier FrCdCric 17 Dadfarnia Shayessteh 7 Dahl Kari 1 Dams Richard 23 Ebdon Les 33 Geertsen Christian 17 Hoult Gavin 7 Huang Zhuoer 11 Hudnik Vida 53 Hutton J. C. 45 Lacour Jean-Luc 17 Lazik C. 45 Marcus R.K. 45 Martinsen Ivar 1 Mauchien Patrick 17 Mermet Jean-Michel 17 O’Neill Peter 33 59 Radziuk Bernard 1 Salbu Brit 1 Schaldach Gerhard 39 Sjostrom Sten 17 Thomassen Yngvar 1 Thompson K. Clive 7 Vanhoe Hans 23 Versieck Jacques 23

ISSN:0267-9477    

DOI:10.1039/JA9940900059

出版商:RSC    

年代:1994

数据来源: RSC