摘要:

JOURNAL O F ANALYTICAL ATOMIC SPECTROMETRY APRIL 1994 VOL. 9 493 Gas-phase Reldistribution of Analyte Species in the Integrated Contact Cuvette Furnace Atomization Plasma Emission Spectrometry Source Shoji lmai Department of Chemistry Joetsu University of Education Joetsu Niigata 943 Japan Ralph E. Sturgeon* National Research Council of Canada Institute for Environmental Chemistry Ottawa Ontario Canada K7A OR9 Interactions of molecular and atomic analyte species released from the primary site of deposition of the sample into a heated integrated contact furnace have been studied with furnace atomization plasma emission spectrometry. With a 30 W forward power plasma double peaks are observed in atomic emission during atomization of Cd Pb Ag Mn Cu Fe and Co reflecting early release of analyte from both the wall (primary site of deposition) and subsequently the radiationally-heated centre electrode (secondary release).With increased r.f. power there is a corresponding early shift in all emission transients reflecting plasma induced heating of both the tube wall and the centre electrode. At higher r.f. power the centre electrode temperature is sufficient to prevent condensation of molecular species of Cd and Pb with the result that the second (late) peak for these elements is eliminated. Similarly when the primary site for analyte deposition is the centre electrode and the r.f. power is 50 W or more transfer of analyte occurs and secondary release from the tube wall is observed. A directed internal convective flow of plasma gas in the char stage has a significant effect on this process.Keywords Furnace atomization plasma emission spectrometry; analyte transfer; volatile oxide; centre electrode; radiofrequency power Furnace atomization plasma emission spectrometry (FAPES) has been shown to be a promising new multi-element emission technique for ultra-trace analysis based on a combined source a graphite furnace as a vaporizer/atomizer and an atmospheric pressure r.f. plasma as excitation medium.'*2 Although several studies have been aimed at characterizing the plasma and studying the effect of forward power on analytical response little attention has been focused on the release of analyte species from the graphite surface and the impact of these processes on performance. The vaporization characteristics of analytes and temporal response in FAPES is different from that in graphite furnace atomic absorption spectrometry (GFAAS) despite the same furnace being used in both tech- nique~.~ The presence of the plasma and centre electrode in the FAPES source provides a secondary input of thermal energy as well as a potential site for condensation and second- ary release of analyte species.Double peaks are frequently observed in the analyte emission transient^*^ and have been ascribed to second surface vaporization phenomena3 involving the cooler centre electrode. Similarly two unresolved peaks were reported for Cr (ref. 6) in a hollow-anode furnace atomiz- ation non-thermal spectrometer (HA-FANES) having a source geometry identical to that of the FAPES. The temporal response of the analyte signal serves as a useful diagnostic in GFAAS and in a similar manner emission transients should provide information on analyte release pro- cesses in the FAPES source.Falk et aL7 reported on the low temperature atomization of Cd in a hollow cathode FANES system and suggested that plasma induced dissociation of molecular Cd species occurred. Several molecular species have been identified in the FAPES source' and it is well known that there are significant gas-phase and heterogeneous reac- tions occurring while the analyte is transported within the furnace as a consequence of concentration and temperature gradient^.^ The centre r.f. electrode in the FAPES source induces an added complexity to the temporal thermal hetero- * To whom correspondance should be addressed.NRCC No. 37553. geneity of the system in that it may be hotter or cooler than the furnace wall early in the atomization stage depending on the forward r.f. power and its temperature lags behind that of the furnace wall during the latter stages of the atomization cycle." It is clear that the possibility of transfer of atomic and molecular analyte vapour between the furnace wall and centre electrode exists and that a convective flow of purge gas through the furnace will have a marked effect on response. This work was undertaken in an effort to study these phenomena. Experimental Apparatus All studies were conducted with a water-cooled integrated contact cuvette (ICC) pyrolytic graphite coated furnace of a design similar to that described by Ballou et aI.," housed within a 10 cm vacuum 6-way cross fitted with a feed-through for r.f.power. A coaxial pyrolytic graphite coated 1 mm centre r.f. electrode supported in a Ta holder was used to deliver power from a crystal controlled 13.56 MHz 1500 W r.f. gener- ator (Dionex Model PM 112-1500 Hayward CA USA). Impedance matching was achieved with a manually adjusted antenna tuner (Heathkit Model SA-2060A Benton Harbor MI USA). The furnace was powered by a Perkin-Elmer Model 2200 supply and fitted for maximum power heating uia an optical feedback circuit. The chamber could be evacuated to 25 Torr ( 1 Torr = 133.322 Pa) pressure with a rotary pump and backfilled and flushed with high-purity He gas. The FAPES workhead was interfaced to a Model SMI I11 echelle grating 0.75 m monochromator (Spectrametrics Andover MA USA).All optical components and the FAPES source were aligned with the aid of an 8 mW He-Ne laser. Analyte resonance lines were isolated with the use of appropriate hollow cathode lamps operated in the d.c. mode. Photocurrents were fed to a current- to-voltage amplifier with a gain of lo9 digitized with 12-bit resolution and stored to disk using an IBM AT processor. All data manipulations were performed using in-house software written in Turbo Pascal version 4 (Borland International Scotts Valley CA USA).494 1 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY APRIL 1994 VOL. 9 I Fig. 1 schematically illustrates the various gas flow con- figurations used in this study. As the ICC furnace is normally operated in a static atmosphere of inert gas [configuration (a)] it was necessary to introduce gas jets into the system designed to direct an independent flow of He through the furnace tube this being designated an 'open' [(b) and (c)] flow configuration.Open configuration (c) could not be used for Cd studies as this analyte tended to condense too easily on to the electrode support and produce spurious late emission peaks during atomization. Configuration ( b ) could not be used for study of Pb Ag and Mn because the higher char temperatures used for these elements resulted in melting of the plastic piping components. High-purity boron nitride tubing (grade HBC Union Carbide Cleveland OH USA) used as the ceramic (b) 10 1 I I W I U (c) 7 5 Fig. 1 Schematic diagram illustrating internal purge gas system.(a) Static configuration; (b) open configuration for Cd; and (c) open configuration for Pb Ag and Mn. 1 Boron nitride tube; 2 ICC furnace; 3 1 mm diameter centre electrode; 4 Ta support; 5 quartz window; 6 r.f. power; 7 He flow in; 8 He flow out; 9 polypropylene 't' connector; and 10 Teflon tube Table 1 Experimental conditions for analyte atomization nozzle penetrated the end of the graphite tube approximately 1 mm but was not in contact with the surface. Temperature measurements of the graphite surfaces were made with both an Ircon Series 1100 (Niles IL USA) auto- matic optical pyrometer and a Thermodot Model TD-6BH (Infrared Santa Barbara CA USA) optical pyrometer. The latter was calibrated to 1300 "C by focusing onto a hole drilled into a graphite block heated in a muffle furnace and compared to the output from a thermocouple.The lower temperature limit was 120 "C. Temperature measurements of the centre electrode were obtained by sighting the Thermodot pyrometer through the sample dosing hole of the furnace. Temperature measurements of the furnace wall during the atomization cycle were taken with the Ircon pyrometer (blackbody assumed) when the centre electrode was absent. The temperature of the furnace wall during the char stage was measured using a chromel-alumel thermocouple. Reagents High-purity He (Matheson) was used as the plasma gas and for purging of the source. Stock solutions of all elements were prepared by dissolution of the high-purity metals (Cd Pb Ag Mn Cu Fe and Co) in sub-boiling distilled HN03.Working standard solutions were prepared by dilution of the stocks with de-ionized distilled water acidified to 1% v/v with quartz sub-boiling distilled HN03. Procedures Different volumes of sample (5 and 1 pl) were pipetted by hand onto the furnace wall and centre electrode respectively using an Eppendorf pipette fitted with polyethylene tips. Samples were then dried for 30 s at a temperature of 80 "C under reduced pressure. Following a 30 s 'char' stage at 80°C for Cd 250°C for Pb and 400°C for Ag Mn Cu Fe and Co under reduced pressure He was admitted to the chamber to restore the pressure to atmospheric and adjusted to maintain a flow of 1 1 min-' during the remainder of the operation. The r.f. power was then applied and the plasma ignited spon- taneously.Following a further 5 s plasma stabilization period the atomization stage was activated the signal recorded and the r.f. power turned off. Table 1 summarizes the atomization conditions for each element. Data acquisition commenced with a trigger start pulse commensurate with the beginning of the atomization stage. The maximum power heating mode was used during atomization to upper pre-set temperatures ("C) of 1100 for Cd 1700 for Pb 2000 for Ag and 2400 for Mn Cu Fe and Co. All temperature values refer to the pre-set tempera- tures as read from the front meter panel of the HGA-2200 power supply. The actual measured temperatures ("C) of the 'char' stage were 86 240 and 330 for the pre-set values of 80 250 and 400 respectively. For the purposes of examining the effect of internal purge gas flow char temperatures were maintained at 80 "C.Element Cd Pb Ag Mn cu Fe c o IJnm 228.8 283.3 328.1 279.5 324.8 248.3 242.5 Char temperature*/"C 80 250 400 400 400 400 400 Atomize temperaturet/"C 1100 1700 2000 2400 2400 2400 2400 Heating rate/"C s - ' 1610 1600 1530 1530 1530 1530 1530 * Time = 60 s for all elements. There is a plasma stabilization period during the final 5 s of the char stage. t Time = 4 s for all elements.JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY APRIL 1994 VOL. 9 495 0 0.4 0.8 1.2 1.6 2.0 0 0.4 0.8 1.2 1.6 2.0 Time/s Fig. 2 Analyte transient emission signals for atomization of samples placed on the furnace tube wall (Wall) and centre electrode (CE) and atomic absorption of samples atomized from the wall in the absence of the CE.(a) Pb 0.5 ng 30 W; (b) Ag 1 ng 30 W; (c) Mn 1 ng 30 W; and ( d ) Fe 5 ng 30 W Results and Discussion Fig. 2 illustrates some typical emission transients obtained during the atomization of the elements when the primary site of deposition was either the furnace wall or the centre electrode. Signals shown for Pb are representative of those observed for Cd as well; similarly the signals for Fe exhibit the same features as those for Cu and Co. Consequently only Pb Ag Mn and Fe were selected for display purposes. A relatively low power plasma was used (30 W). Atomic absorption response from the same mass of each analyte produced in the ICC without a centre electrode is also given for comparative purposes. Samples initially deposited onto the centre electrode give rise to a single emission transient whereas those dosed onto the furnace wall produce two or more peaks.In all cases the late peak produced with wall deposition coincides tem- porally with that produced when the sample is deliberately placed on the centre electrode. It is evident that analyte species released from the wall contribute to the early emission response efficiently condense on the cooler centre electrode and sub- sequently re-desorb from this secondary surface as the system continues to heat thereby producing the second late peak. Hettipathirana and Blades3 have reported similar observations for Pb Ag and Mn in their FAPES system although the peaks are not as well resolved despite lower rates of heating.Whereas the appearance of atomic absorption and emission transients are essentially coincident for Ag Cu Fe and Co those for Cd Pb and Mn produce a significant emission response from the wall at temperatures lower than those necessary to produce a detectable free atom concentration for atomic absorption. This is indicative of plasma mediated dissociation of thermally stable molecular forms of these elements desorbed from the tube wall at low temperature during the atomization stage. Silver Cu Fe and Co undergo reduction prior to sublimation and no stable low temperature molecular oxide species of these elements are produced. The estimated appearance temperatures for emission from Cd and Pb are 520 and 590 K respectively. These are signifi- cantly lower than their corresponding atomic absorption appearance temperatures (620 and 1380 K respectively).Evidence for the low temperature volatilization of CdO (470 K) and its plasma induced dissociation in a low pressure FANES source has been given by Falk et a1.' as the corresponding appearance temperature of the atomic absorption signal was 570K. In the same manner low temperature plasma dis- sociation of volatile PbO can occur. Gaseous PbO has been detected in the G F by mass spectrometry'2 as well as in ab~orption'~ and emission spectrometry.' Emission response from Ag commences at 840K and a shoulder is observed at loo0 K. The appearance temperature for atomic absorption is 1000 K. Silver nitrate undergoes decomposition at 717 K.14 The close correlation between this temperature and the tem- perature of appearance of emission from Ag suggests that the initial early peak may arise as a result of plasma-induced decomposition/dissociation of Ago or other molecular species ejected into the gas phase during the 'explosive' decomposition of the nitrate salt and subsequent excitation of Ag.A similar process has been suggested to account for the observation of molecular oxide species in quadrupole MS experiments used to monitor the gas-phase composition within the G F during analyte atomizati~n.'~*'~ The second more diffuse peak from Ag probably results from the transfer of atomic vapour from the wall to the cooler electrode uia a simple desorption of the reduced metal. These observations and conclusions contrast with those of Hettipathirana and blade^,^ who report occurence of only two peaks for vaporization of Ag from the wall in atomic absorption and a single peak in emission.They imply that when Ag is dosed onto the wall it desorbs from the wall; when placed on the centre electrode it desorbs earlier from this surface because the presence of the plasma raises the temperature of the electrode above that of the wall (Fig. 4 in ref. 3). Transfer of Ag from the wall to the centre electrode in this system likely occurs as the result of crystal shattering and sublimation as noted above. The slight difference in the peak shapes noted by these workers (Fig. 4 in ref. 3) for the depos- ition of Ag on the wall and the centre electrode may be rationalized with the argument that the ultimate source for both signals is the centre electrode and the late shift associated with wall deposition is due to the difference in release rates496 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY APRIL 1994 VOL.9 for secondary uersus primary dep~sition.'~ In other words under the experimental conditions used Hettipathirana and Blades3 never observe release of Ag from the wall despite it being a primary site of deposition. During the initial decompo- sition of the nitrate salt it is also likely that molecular species of Ag adsorb onto the centre electrode. However such species must desorb from this surface only as atomic Ag since the temperature of the centre electrode is approximately 1200 K when this event occurs [cf. Fig. 2(b)] and Ago would have been thermally reduced below this temperature.For Fe Co and Cu there is a coincidence of the atomic absorption and emission appearance temperatures and the diffuse early peak observed in emission for each of these elements is probably due to their desorption from the wall to the centre electrode in atomic form. Decomposition of the nitrate salts of these elements occurs at temperatures lower than their appearance in atomic emission. In the case of Mn three partially resolved peaks are visible; the first might correspond to the plasma induced dissociation of molecular oxide species desorbing from the tube wall as this appears at a lower temperature than that at which atomic absorption commences. The second peak likely represents atomic Mn desorbing from the wall as its appearance temperature is coincident with that of the AA response.Both of these early species may condense on the cooler centre electrode which with subsequent heating causes their re-desorption and the observed late emission peak. Hettipathirana and Blades3 report only two emission peaks from Mn when the sample was placed on the tube wall of their FAPES system. These corresponded to the early release of atomic Mn from the wall and its subsequent condensation on the electrode followed by re-desorption from this surface at higher temperature. It is possible that the earlier peak obtained in this study is a result of the more efficient detection of analyte with the use of the higher heating rates [ 1.6 K ms-' as opposed to 0.2 K ms-' (ref. 3)J. At higher r.f. powers (40 W ref. 3) an early Mn peak is seen in emission in the system used by Hettipathirana and Blades (Fig.14 in ref. 3). The appearance temperature of this peak is approximately 1250 K suggesting that MnO rather than atomic Mn is the species actually being transferred from the wall to the centre electrode in their experiments. Fig. 3 illustrates the effect of forward r.f. power on emission transients for several elements when the primary site of depos- ition is the tube wall. As r.f. power is increased there is a corresponding early shift in the emission transients of all elements The time for the signals to return to baseline does not continuously decrease as the r.f. power increases. This arises because plasma volume expands with increased power and hence the residence time increases. The plasma contrib- utes to the thermal heating of the tube wall thereby altering the temperature-time characteristics of the furnace during the atomization cycle.This is illustrated in Fig. 4(u). The measured temperature-time profile of the tube wall obtained with the automatic optical pyrometer for maximum power heating to a pre-set temperature of 2400°C in the absence of a plasma is illustrated by the solid curve. Acompanying curves describing the tube wall temperature in the presence of a plasma were constructed from the measured atomic emission appearance times for Ag Cu Fe and Co obtained in r.f. plasmas of 30 50 and 80 W forward power. It was assumed that the appearance temperatures for these elements when desorbing from the wall remain constant as the power is varied (Le.no specific plasma mediated atomization processes). With the exception of Ag these elements should be well-suited for this purpose as they undergo atomization uiu a simple desorption process and there is good agreement between the measured appearance tempera- tures in both absorption and emission. This is summarized in Table 2 which presents appearance temperature data for atomic absorption in both a He and Ar atmosphere. The latter are in good agreement with values reported in the literature for GFAAS measurements whereas those obtained in a He atmosphere are generally higher. It can be seen that at 0.7 s when the tube wall has reached 1600 K the surface temperature can be increased by up to 100 K in the presence of a 50 W r.f. plasma and by 300 K at an r.f.power of 80 W. No significant increase can be measured for a 30 W plasma. The appearance times used to construct the surface temperature-time curves have a precision of replicate measurement better than k0.02 s. The magnitude of the temporal shift in emission response is thus dependant on the power and the element. At the lowest powers the effect appears minimal since further increases in power produce little effect. As an example the early shift in a' 100 (b) 80 100 0 0.4 0.8 1.2 1.6 2.0 0 0.4 0.8 1.2 176 2.0 Time/s Fig. 3 Etrect of r.f. power on transient emission signals for atomization of analyte from the tube wall. (a) Pb 0.5 ng; (h) Ag 1 ng; (c) Mn 1 ng; ( d ) Fe 5 ng; values on peaks given in WJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY APRIL 1994 VOL.9 v) E - 497 (c) 80 100 ; 50 i ; 30 I 1 I I Table 2 Appearance temperatures (in K) for atomic absorption and emission response GFAAS* Element Cd Pb Mn cu Fe c o Ag He Ar 620 670 1380 1090 lo00 990 1560 1450 1580 1400 1670 1580 1850 1660 FAPESt He 520 590 840 1200 1510 1540 1670 * No centre electrode present. t 30 W plasma. Pb response does not occur until more than 40 W forward power is used. Fig. 5 illustrates the effect of r.f. power on the emission response during atomization when the primary site of depos- ition is the centre electrode. Several points merit discussion. As forward power is increased there is an early shift in the appearance of emission which is directly attributable to a measured increase in the temperature of the centre electrode.This is quantitatively evident from the data presented in Fig. 4(b) which for several char temperatures of the system shows the change in the measured temperature of the centre electrode as the forward power is increased. For comparison a dynamic temperature-time heating curve characterizing the surface temperature of the centre electrode for a forward power of 30 W char temperature of 400 "C and atomization tempera- ture of 2400°C is also given in Fig. 4(a). This curve was constructed on the basis of measured appearance times for the desorption of Ag Cu Fe and Co from the centre electrode (primary site of deposition) and a knowledge of their appear- ance temperatures (Table 2). It is clear from Fig. 2 that desorp- tion of Ag from the centre electrode produces a single peak due to the release of atomic Ag into the gas phase (i.e.the early peak associated with the release of Ag from the wall does not occur since the initial heating rate of the centre electrode is low enough to prevent the explosive decomposition of the t a! 80 I ! 2000 1500 000 500 I I I 1 0 0.20 0.40 0.60 0.80 1.00 Tirne/s 700 A' 600 400 I 1 I I I 20 30 40 50 60 70 80 PowerW Fig. 4 Effect of r.f. forward power on the temperature-time character- istics of the graphite tube wall and centre electrode during atomization. (a) Tube wall temperature obtained for r.f. plasmas at A 80; and B 50 W forward power. C. Centre electrode temperature at 30 W forward power. Appearance times for A Ag; A Cu; 0 Fe; and 0 Co at a char temperature of 400°C. The solid line indicates the profile of the tube wall obtained in the absence or presence of a 30 W plasma using maximum power heating to 2400 "C.(b) Centre electrode temperature observed at various r.f. forward powers with char tempera- ture 0 room temperature; 0 80; A 250 and A 400°C 0 0.4 0.8 1.2 1.6 2.0 Time/s Fig. 5 (c) Mn 1 ng; and ( d ) Fe 5 ng Effect of r.f. power (given in W) on emission transients for atomization of analyte from the centre electrode (a) Pb 0.5 ng; (b) Ag 1 ng;498 JOURNAL O F ANALYTICAL ATOMIC SPECTROMETRY APRIL 1994 VOL. 9 salt) It was further assumed that this simple release process is identical to that occurring in the AAS experiments and for this reason an appearance temperature of 990 K (Table 2) was used as the data point in preparing this curve.Transfer of molecular Cd species from the centre electrode to the tube wall during the 5 s plasma stabilization period occurs for powers above 40W. Between 40 and 50W the electrode temperature increases from 490 to 600 K. Since the appearance temperature for emission from Cd off the wall is 520 K no late peak can be observed for Cd vaporizing from the centre electrode above forward powers of 40 W. No early peak is associated with atomization of Pb from the centre electrode when the forward power is 30 W. At 35 W an early peak appears and with 50 W power the late peak has been elrminated. This is consistent with the transfer of a volatile molecular Pb species from the electrode to the wall during the brief 5 s period that the plasma is stabilized before the atomiz- ation stage is initiated; the temperature of the centre electrode is 590 K at 35 W power in agreement with the appearance temperature for emission when the primary site of deposition of Pb is the tube wall.The early release peaks evident for Pb in Fig. 3 are not reported by Hettipathirana and blade^,^ possibly as a consequence of the low plasma power and low rate of heating of their furnace. A more plausible interpretation of their data for Pb suggests simple transport of volatile PbO and its plasma induced dissociation in all cases. There is no need to invoke complex changes in the gas-phase chemistry to describe vaporization/atomization of this element in their system. The late peak which arises when Pb is deposited on their centre electrode (Fig.9 in ref. 3) most likely occurs as a result of desorption of Pb from some other cooler surface within their source (e.g. the portion of the electrode extending outside the furnace). Such a late peak (in this case a third peak) can be obtained in our own system if the atomization temperature is set above 1700°C. Indeed the pronounced tailing of the Pb signal shown in Fig. 2(u) may be indicative of the slow desorption of Pb from such a site. The data presented by Hettipathirana and Blades (Fig. 9 in ref. 3) may be more consistently interpreted as being release of Pb from the centre electrode even when the primary site of deposition was the wall. Their Fig. 8 (ref. 3) clearly shows release of Pb from the centre electrode peaking at approximately 4 s in the absence of a plasma.Since the presence of a plasma heats the centre electrode it is to be expected that the FAPES response for Pb from the centre electrode should occur earlier than 4 s. Based on the reported 1.4 s shift noted for Ag with the plasma present it is to be expected that the emission transients for Pb should peak at approximately 2.6 s for deposition on the centre electrode. This is clearly evident in their data (Fig. 9 in ref. 3). Similar changes in signal shape are also evident for Ag at 50 W. At this power the temperature of the centre electrode is 670K close to the decomposition temperature of AgN03. Silver species thus scatter from the electrode upon rapid decomposition of the nitrate as the latter heats during the plasma stabilization period. At 80 W the surface temperature is high enough so that emission signals are observed for Cd Pb and Ag during the plasma stabilization period when the primary site of deposition is the electrode.When Mn Cu Fe and Co are initially placed on the centre electrode only single late emission transients are obtained at all r.f. powers. With the exception of Mn this is consistent with the earlier proposal concerning release of these elements as atomic vapour. For Mn data presented in Fig. 3 suggested early release of this element from the tube wall in molecular form during the atomization stage. In contrast it appears that Mn oxide(s) are reduced by graphite or plasma mediated processes (possibly due to increased CO partial pressure near the centre electrode3) when the sample is initially placed on the electrode since in this case.only a single late transient is generated. In order to verify that release of analyte species from the centre electrode occurs during the plasma stabilization period several experiments were undertaken to discern the effect of a convective flow of He gas through the tube during the char stage. Forward r.f. powers were used that were sufficiently high to ensure that the centre electrode was hot enough to cause analyte transfer. Using the 'open' flow configuration illustrated in Fig. 1 a 1260 ml min-' flow of He was directed through the tube during the 5 s plasma stabilization period with the tube wall heated to the char temperature. The gas flow was then stopped the atomization stage initiated and emission recorded. The presence of the gas flow during this time period decreased the intensity for Cd by 66% and that for Pb by 62% and eliminated the response from Ag.These figures reflect an upper limit to the decrease in efficiency of the transfer and re-deposition process caused by the gas flow. Because the convective flow of internal gas may produce an analyte re-distribution over the tube wall which may be different from that generated under static transfer conditions direct compari- son of responses is only approximate. If the newly deposited analyte lies closer to the extreme edge of the furnace reduced residence times contribute to a lowered emission response. In a subsequent experiment the sample was subjected to a 5 s plasma stabilization period at the char stage but with no internal convective gas flow.This was followed by a further 30s char stage with the plasma off but a He flow of 1260mlmin-' through the tube. The internal gas flow was then terminated the plasma re-ignited and allowed to stabilize for 2 s and the sample atomized. Analyte species remaining in the gas phase at the end of the first plasma stabilization period are thus convectively transported out of the furnace volume. By comparing peak height response from the early peak in each case it was determined that approximately 25 94 and 73% of the Cd Pb and Ag is retained reflecting the portion of the total analyte that is readily transferred from the heated centre electrode to the tube wall during the usual 5 s plasma stabilization period. The balance exists in the gas phase at the end of the stabilization period.Fig. 5 also shows that emission response generally increases with r.f. power but that optimum response does not necessarily correlate with greatest r.f. power for samples deposited on the centre electrode. As r.f. power increases the temperature of the electrode and plasma rise increasing the rate of diffusive loss from the system. The effect on peak intensity is most pronounced for the volatile analytes and decreased peak half-widths are evident for all analytes. Conclusions Double analyte emission peaks frequently observed in the FAPES source are the result of second surface adsorption-de- sorption phenomena. Data support the suggestion that volatile oxide precursors of Pb Cd and Mn are responsible for a re-distribution of these analytes from their primary site of deposition on the tube wall to that of the cooler centre electrode.This occurs during the early stages of the atomization cycle. In the case of Ag it is suggested that re-distribution occurs as a result of both violent crystal shattering during the low temperature decomposition of the nitrate salt and desorp- tion of the reduced metal. For Fe Cu and Co a simple condensation onto the cooler centre electrode of the desorbing reduced analyte occurs early in the atomization stage At sufficiently high r.f. powers Pb Cd and Ag initially deposited on the centre electrode are similarly re-distributed to the wall during the char stage as the plasma is being permitted to stabilize. This process is thus strongly dependant on the temperature of the electrode which is determined by the forward r.f.power. The authors thank S. Willie for assistance in setting up the source-spectrometer system. S.I. thanks the NRCC for finan- cial support while in Ottawa.JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY APRIL 1994 VOL. 9 499 References 1 Liang D. C. and Blades M. W. Spectrochim. Acta Part B 1989 44 1059. 2 Sturgeon R. E. Willie S. N. Luong V. T. Berman S. S. and Dunn J. G. J . Anal. A t . Spectrom. 1989 4 669. 3 Hettipathirana T. D. and Blades M. W. J . Anal. At. Spectrom. 1992 7 1039. 4 Smith D. L. Liang D. C. Steel D. and Blades M. W. Spectrochim. Acta Part B 1990 45 493. 5 Sturgeon R. E. Willie S. N. Luong V. T. and Berman S . S. Anal. Chem. 1990 62 2370. 6 Riby P. G. Harnly J. M. Styris D. L. and Ballou N. E. Spectrochim. Acta Part B 1991 46 203. 7 Falk H. Hoffmann E. and Ludke Ch. Prog. Anal. Spectrosc. 1988 11 417. 8 Sturgeon R. E. and Willie S . N. J . Anal. At. Spectrom. 1992 7 339. 9 Gilmutdinov A. Kh. Zakharov Yu. A. and Voloshin A. V. J. Anal. At. Spectrom. 1993 8 387. 10 11 12 13 14 15 16 17 Sturgeon R. E. Luong V. T. Willie S. N. and Marcus R. K. Specrrochim. Actu Part B 1993 48 893. Ballou N. E. Styris D. L. and Harnly J. M. J. Anal. At. Spectrum. 1988 3 1141. Sturgeon R. E. Mitchell D. F. and Berman S . S. Anal. Chem. 1983 55 1059. Ratcliff J. and Majidi V. Anal. Chem. 1992 64 2743. Weast R. C. Handbook of Chemistry and Physics 59th edn. Chemical Rubber Company Cleveland OH USA 1978. Bass D. A. and Holcombe J. A. Anal. Chem. 1987 59 974. T. McAllister Paper No. H-3 presented at the XXVIII Colloquium Spectroscopicum Internationale Post Symposium on Graphite Atomizer Techniques Durham July 1993. Lynch S. Sturgeon R. E. Luong V. T. and Littlejohn D. J . Anal. At. Spectrom. 1990 5 311. Paper 3104954J Receiued August 16 1993 Accepted December 16 1993

ISSN:0267-9477    

DOI:10.1039/JA9940900493

出版商:RSC    

年代:1994

数据来源: RSC

摘要:

JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY APRIL 1994 VOL. 9 50 1 Choice of Fluorescence Wavelengths for the Determination of Trace Amounts of Chlorine by Graphite Furnace Laser-Excited Molecular Fluorescence Spectrometry of Indium Monochloride* Evelyn G. Su and Robert G. Michelt Department of Chemistry University of Connecticut 275 GIenbrook Road Storrs CT 06269-3060 USA New excitation and detection wavelengths were explored in an attempt to improve the sensitivity of laser- excited molecular fluorescence spectrometry (LEMOFS) in a graphite tube furnace for the determination of chlorine through indium monochloride molecular fluorescence. Fluorescence spectra of the A- B- and C-band systems of indium monochloride excited at 267.21 nm were obtained and the sensitivities were compared.The most sensitive excitation/detection scheme was found to be 267.2 nm/269 nm where the detection wavelength was within the same C-band electronic transition as the excitation wavelength but with a different vibrational transition. To the knowledge of these workers this was the first recorded analytical use of a vibrational transition within the same electronic transition for LEMOFS. The vibrational transitions of the fluorescence spectra were spectroscopically well resolved. A detection limit of 10 pg for chlorine was achieved which was a 50-fold improvement over our experiments carried out at wavelengths used by other workers. Unfortunately the best detection limit of these previous researchers at their wavelengths could not be reproduced by us.As a result the detection limit reported here is no better than the best previously reported detection limit which remains an improvement of two orders of magnitude over molecular absorption spectrometry of indium monochloride. The effects of metal ions and non-metal ions were studied. Chlorine was determined in two biological standard reference materials obtained from the National Institute of Standards and Technology. Vaporization matrix effects were observed and removed by dilution. By use of simple aqueous calibration the results obtained for the two reference samples were in good agreement with the non-certified reference values of chlorine. Keywords Chlorine; laser-excited molecular fluorescence spectrometry; graphite furnace; indium monoch- loride; fluorescence wavelength Chlorine is widely distributed in the natural environment and in animal tissues and is used extensively in industrial processes.Chlorine is commonly added to drinking water at about 1 mg 1-' as a disinfectant to prevent the spread of waterborne diseases.' It is crucial for the human body to maintain a proper level of chloride. Chloride as the major extracellular anion is important in maintaining proper water distribution osmotic pressure and normal anion-cation balance.2 Low serum chlor- ide values are associated with chronic pyelonephritis Addison's disease renal failure and diabetic acidosis and high serum chloride values are associated with dehydration congestive heart failure and general kidney pathology.2 The determination of chloride is therefore of the utmost importance in medical diagnostics.The determination of chloride (as sodium chloride salt) in food products is also important for health reasons. A variety of analytical methods have been developed for the determination of chlorine. These include chloride ion-selective electrode ion chromatography photometric techniques titrimetric methods atomic spectrometry and molecular spectrometry . The use of chloride ion-selective electrodes is one of the most widely used approaches. This method which meas- ures only free chloride ion in solution has a detection limit of 3.5-35 pg 1-1,3,4 depending upon the membrane used. Disadvantages of this technique include poor accuracy for complicated samples and significant consumption of the chemi- cal reagents that are added to the sample as the total ionic strength adjustment buffer.This method is routinely used to measure ppm levels (about mmol 1-') of chloride ion in water. It is difficult to use this technique to determine trace amounts of chlorine due to problems associated with sensitivity selec- * Presented in part at The Sixth Biennial National Atomic Spectroscopy Symposium July 22-24 1992 Plymouth UK and the 19th FACSS meeting September 20-25 1992 Philadelphia PA USA. t To whom correspondence should be addressed. tivity interference and accuracy especially for complicated samples. Ion chromatography has been utilized for the determination of chloride ion. Kudermann and Blauful3' determined chlorine in high-purity aluminium by ion chromatography after pyrohy- drolytic separation and reported a detection limit of 0.1 pg g-'.Mehra and Frankenberger6 used ion chromatography with amperometric detection for the simultaneous determination of cyanide chloride bromide and iodide in environmental samples The limit of detection (LOD) for chloride was Photometric methods have been used for the determin- ation of chlorine7** and can determine ppm levels of chloride ion in solution with an LOD of 100-1OOOpg1-' or 2 x 10-'-3 x mol l-1.7.9*'0 Disadvantages of photometric methods include stability problems with the colouring com- pound long analysis time considerable consumption of chemi- cal reagents and short linear dynamic range. Generally photometric methods are not desirable for routine analysis. Titrimetric methods have been developed for the determi- nation of chlorine. Official methods recommended by the American Public Health Association and others include the use of N N-diethyl-p-phenylenediamine as an indicator in a titrimetric procedure with ammonium iron( 11) sulfate.The disadvantages of common titrimetric methods include poor sensitivity selectively interferences and time-consuming pro- cedures. Recently Midgley and Gatford" developed a known addition-dilution titrimetric potentiometric method for the determination of traces of chlorine in borax-nitrate solution. Also chloride concentrations of 75-250 pg 1- were determined in simulated power auxiliary coolant." All the methods discussed above are only responsive to free chloride ion and suffer from problems associated with sensi- tivity selectivity and interferences especially for real samples.The determination of chlorine at trace levels is still a difficult practical problem." Atomic spectrometry is widely used for the determination 0.2 pg 1 - I .502 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY APRIL 1994 VOL. 9 of trace and ultra-trace amounts of metals and metalloids due to its high sensitivity and ability to detect the total amount of all chemical forms of an element. However trace and ultra- trace non-metals cannot be readily determined by direct atomic spectrometric methods because their resonance lines often fall in the vacuum UV region. In addition non-metals are difficult to atomize and they tend to form stable diatomic molecules due to their high electronegativity. As a result indirect and molecular spectrometry of small molecules that contain the analyte have often been used to determine non-metals. A comprehensive review on this subject has been p~b1ished.l~ Among these indirect spectrometric methods molecular absorption spectrometry (MAS) of diatomic molecules in a graphite furnace has been widely used.In this technique diatomic molecules of a non-metal analyte and an added metal are formed in the gas phase during the vaporization step of the furnace programme. This process can be represented by the simple reaction M(excess)+X+MX (1) where M X and MX represent the metal the non-metal and the diatomic molecule respectively. Unlike the non-metal itself the diatomic molecule absorbs in the normal UV or visible region and thus can be detected readily.A salt that contains the metal is added in excess to shift the equilibrium toward the formation of the diatomic molecule. Molecular absorption spectrometry of diatomic molecules in a graphite furnace has been investigated for the determi- nation of chlorine. Table 1 summarizes the best detection limits for chlorine obtained for various diatomic molecule^.'^^'^ They are in the low nanogram range. Molecular absorption spec- trometry in a graphite furnace often suffers from interferences from the matrix which affect both the molecular absorption signal and the background. For halides including chloride liquid-liquid extraction with triphenyltin hydroxide followed by back-extraction into an aqueous solution of sodium or barium hydroxide has been used to separate the matrix interferent cations and anion^.'^.'^ The separation of the analyte from the matrix and the pre-concentration of the analyte by extraction and back-extraction improve both the sensitivity and accuracy of the determination of non-metals by molecular absorption.However the extraction and back- extraction procedures are tedious and time-consuming. Laser-excited molecular fluorescence spectrometry (LEMOFS) of diatomic molecules can be used for the determi- nation of non-metals in a similar fashion to MAS of diatomic molecules. Inherently LEMOFS is more sensitive than MAS. Therefore LEMOFS provides an opportunity to remove matrix effects caused during the vaporization process by use of appropriate dilution which means that tedious extraction and back-extraction procedures can be avoided.The LEMOFS technique has the potential to be relatively free of spectral interferences compared with MAS because of the classical advantage of the additional selectivity of the detection of fluorescence at a different wavelength from excitation. Indeed inspection of the background signals in the region of the detection wavelength at all sample dilutions indicated no spectral interferences during the determinations of chlorine reported here. The high sensitivity available with LEMOFS Table 1 Detection limits for chlorine by graphite furnace MAS Detection Molecule Wavelength/nm limit/ng Reference AlCl 261.4 0.1 14 AlCl 261.4 2 15 GaCl 248.2 9 15 InCl 267.2 3 15 lnCl 267.2 9 16 MgCl 376.2 50 17 and its ability to detect total chlorine may be critical to some environmental samples such as chlorine-containing pesticides in soil water etc.It has been shown in the literat~re,'~ that among the many chlorine-containing diatomic molecules studied MAS of alu- minium monochloride has the lowest LOD for the determi- nation of chlorine (see Table 1 ). Naturally this molecule should be used for the study of LEMOFS. Unfortunately aluminium monochloride has no conventional non-resonance line for fluorescence detection which is required for the more modern fluorescence detection configuration in a graphite furnace; the so-called front surface-illumination.20 Indium monochloride offers a slightly worse LOD of chlor- ine than aluminium monochloride by MAS,13 but it has a relatively rich spectrum which allows for several fluorescence detection schemes.A literature search revealed that only two research groups have investigated LEMOFS of indium mon- ochloride for the determination of chlorine with either reson- ance fluorescence detection,2' or non-resonance fluorescence detection.22 For resonance fluorescence detection with trans- verse illumination Dittrich and Stark21 obtained an LOD of 20 pg for chlorine in a modified graphite tube atomizer. Both excitation and detection wavelengths were at 267.2 nm. Compared with the best LOD of 3.0ng by MAS of indium mon~chloride,'~ this was an improvement by a factor of 150. No determination of chlorine in real samples by LEMOFS was reported in ref. 21. For non-resonance fluorescence detection with the front-surface illumination approach.Anwar et a1.22 reported an LOD of 20 pg for chlorine. The excitation and detection wavelengths were 267.2 and 359.5 nm respect- ively with a laser spectral bandwidth of 0.01 nm that limited the resolution of the excitation spectrum. The effects of poten- tial interfering ions were studied and a method for the determi- nation of chlorine was examined by use of National Institute of Standards and Technology (NIST) standard reference mate- rials (SRMs). The results obtained for the determination of chloride in NIST SRM Orchard Leaves using nitric acid digestion compared well with the values reported by NIST. In an attempt to improve the sensitivity for the determination of chlorine the use of fluorescence detection within the same electronic transition as the excitation wavelength but with a different vibrational transition was evaluated. To our knowl- edge this concept has not been used before for analytical measurements by LEMOFS.High-resolution excitation spec- tra and fluorescence spectra were obtained and the vibrational transitions were resolved. The LODs for various excitation- detection schemes were obtained and compared under optimized conditions. The best scheme was applied to the determination of chlorine in NIST biological SRMs. Experimental Instrumentation The instrumentation used in this work was as described in a previous paper,2o except that a different dye laser was used. A laboratory-constructed grazing incidence dye laser23 was employed here. The maximum laser power at the indium chloride excitation wavelength of 267.21 nm was 15-20 pJ per pulse after the frequency doubler which resulted in 3-5 pJ per pulse inside the graphite furnace.This laser pulse energy was not able to saturate the fluorescence signal of the diatomic molecule according to a test in this work and a previous study.20 The laser was operated at 50 Hz for all experiments. A small portion of the dye laser beam was sampled with an optical fibre and was employed to trigger a boxcar averager that was used to process the fluorescence signal from a photo- multiplier tube (PMT). Coumarin 540A { lH 4H-2,3,5,6-tetra- hydro-8-( trifluoromethyl )quinolizino [ 9,9a 1 -gh] coumarin} was dissolved in methanol at a concentration of 16 mmol 1-' and was used as the laser dye for the oscillator.For theJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY APRIL 1994 VOL. 9 503 amplifier this dye solution was diluted with methanol by a factor of three. The rest of the instrumentation has been described in detail in ref. 20 including the optical system and data processing. Unlike the previous work," no colour filter was employed prior to the detection monochromator. 3 C'II Standard Solutions The standard solutions of chlorine (from NaCl) indium [from In(NO,),] as well as other cation or anion solutions used for interference studies were prepared daily on a class 100 clean air bench by serial dilution of a 1 mg ml-' stock solution. Standard solutions with concentrations of less than 1 mg ml-' were stored in polyethylene bottles. The stock solutions were made from high-purity salts (Spex Industries Metuchen NJ USA).All stock solutions and working solutions contained 5 and 0.2% nitric acid respectively. Preparation of Biological Samples Two biological standard reference materials SRM 1568a Rice Flour and SRM 1572 Citrus Leaves were obtained from NIST (Gaithersburg MD USA). The samples were dried at 105°C for about 2 h prior to use. Portions (0.1-0.25 g) of the SRMs were weighed out accurately into a beaker to which 5ml of nitric acid (Ultrex grade J. T. Baker Phillipsburg NJ USA) and 5 ml of de-ionized water were added. Rice Flour samples were then transferred onto a hot-plate and were heated to boiling for 1 h. The digested samples were diluted to 100 ml with sub-boiled distilled water before analysis.Citrus Leaves samples were heated in a beaker to near dryness. Nitric acid (1 ml) and hydrogen peroxide ( 1 ml) (J. T. Baker) were added to the samples followed by heating for 10 min and dilution to 100 ml with sub-boiled distilled water before analysis. Experimental Procedure A series of mixtures that contained both chlorine and indium were prepared daily as follows the indium solution the chlorine solution and sub-boiled water if necessary were mixed to make up desired concentrations of chlorine with In C1 mass ratios of between 100 and 1OOO. Aliquots of 20 pl of the solutions were introduced onto the L'vov platform inside the graphite tube furnace. For interference studies small vol- umes of solutions that contained other cations or anions were introduced onto the platform separately.After sample introduc- tion the furnace heating programme was started. In the vaporization step indium and chlorine were believed to evapor- ate together and to form indium monochloride molecules in the gas phase. These indium monochloride molecules were excited by laser radiation and the resultant fluorescence was measured. The data collection procedure was completely com- puter controlled. The furnace programmer signaled the com- puter 2 s prior to the vaporization step and the computer turned on the pulsed laser system immediately. Data collection started after approximately 1 s. Data processing and display followed automatically. The data points collected during the first second were used for baseline correction. Results and Discussion Spectroscopic Properties of Indium Monochloride Indium monochloride has been characterized by three major band systems (Fig.1):24 A3110++X1Z+; B3111eX'C' and C'll=X'C+ where X'C' is the ground state. Each band system consists of a series of vibrational transitions. According to ref. 24 all three of these band systems have the highest intensity at (0,O) vibrational transitions with wavelengths at 359.92 349.90 and 267.21 nm respectively. It has been found2' that the absorption coefficient in the C-system is appreciably higher than in the A- and B-systems which have been com- 267.2 nm X'Z * Fig. 1 Schematic partial energy diagram for indium monochloride. The solid and dotted arrows represent radiative and collisional transitions respectively monly used for indium monochloride molecular emission in flames.The higher absorption probability of the C-system can be attributed to the singlet nature of both the ground and excited states. Fig. 1 shows a partial schematic energy diagram of indium monochloride that illustrates the possible fluorescence detec- tion schemes. The transition that can be used as the excitation wavelength is 267.21 nm of the C-system from C'IIeX'C' (Fig. l) while the transitions that can be used as conventional non-resonance fluorescence detection wavelengths can be either 359.9nm of the A-system or 349.9nm of the B-system. In addition detection of non-resonance fluorescence within the same electronic transition as the excitation wavelength C'lleX'Z' is possible with high spectral purity laser radi- ation and a high-resolution monochromator for fluorescence detection.With the excitation wavelength at the 267.21 nm (0,O) vibrational transition non-resonance fluorescence detec- tion is possible at the 269.45 nm (0,l) vibrational transition and the 271.75 nm (0,2) vibrational transition etc. In previous reports of LEMOFS the excitation wavelength and detection wavelength were from different electronic transitions while for the detection scheme reported here both the excitation and the detection wavelengths were from the same electronic transition but from different vibrational transitions. For LEMOFS Dittrich and Starkz1 used a transverse illumi- nation approach where both excitation and detection wave- lengths were the same. Resonance fluorescence was detected by use of one of the three excitation wavelengths 267.21 269.45 and 271.75 nm that corresponded to the vibrational transitions (O,O) (0,l) and (0,2) respectively of the C-band system.Highest fluorescence was obtained with the excitation wavelength at the 267.21 nm (0,O) vibrational transition. They observed no fluorescence at the wavelengths of either the A- (359.9 nm) or the B-system (349.9 nm) with excitation wave- length at any of the wavelengths of 267.21 269.45 and 271.75 nm. This observation was attributed to the small laser energy at the three excitation wavelengths used with only 3.5 4.5 and 6 pJ respectively. Anwar et used non-resonance fluorescence detection at 359 nm with front-surface illumi- nation and obtained the highest fluorescence signal with the excitation wavelength at the 267 nm (0,O) vibrational transition.Measurements of fluorescence were made at the non-resonance line of 359 nm because severe scattering problems occurred for resonance fluorescence detection. Other non-resonance lines were not studied therefore a comparison between the non-resonance detection wavelengths was not possible. In the present work with the excitation wavelength at the 267.21 nm (0,O) vibrational transition high resolution non- resonance fluorescence spectra for all three band systems i.e. A- B- and C-systems were obtained. A direct comparison504 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY APRIL 1994 VOL. 9 of the LODs for chlorine was made for all the detection schemes studied. Chemical Optimization An excess amount of indium is needed to shift the equilibrium towards the formation of indium monochloride.Fig. 2 presents the effect of 1n:Cl mass ratio on the fluorescence signal produced by 200 ng of chloride. The various 1n:Cl mass ratios were obtained by variation in the amount of indium while the amount of chloride was kept constant at 200 ng of chloride. The excitation wavelength was 267.21 nm and the detection wavelength was 349 nm. The bandpass of the monochromator was 0.8 nm. These spectrometric conditions were not optimal as can be seen later but this would not have affected the optimization of the chemical conditions. Fig. 2 shows that the maximum indium monochloride signal was achieved at an 1n:Cl mass ratio of 80 above which a plateau of fluorescence signal was reached.Higher 1n:Cl mass ratios up to 1000 were studied for a lower amount of chloride at 20 ng and a plateau of fluorescence signal was obtained for 1n:Cl mass ratios between 80 and 1O00 the highest ratio studied. This result was in good agreement with that reported in the literature,2'-22 where an optimal fluorescence signal was obtained for 1n:Cl mass ratios above 100 up to 100000.22 In the present work all the experiments employed an 1n:Cl mass ratio of between 100 and 1000. Optimization of Furnace Conditions The vaporization and char temperatures were optimized to obtain the maximum indium monochloride fluorescence signal. Dittrich and Stark2' obtained the maximum fluorescence signal by use of transverse illumination with a modified graphite tube furnace at a vaporization temperature of 2800 "C and a char temperature of 900°C.Anwar et ~ 1 . ~ ~ reported a slow increase in fluorescence intensity as the vaporization temperature was increased from 1200 to 1600°C. At higher temperatures they observed a significant depression in the signal and suggested that this depression was probably a result of the thermal decomposition of indium monochloride. Fig. 3 shows the results obtained in this work for the optimization of vaporization temperature and char tempera- ture. From Fig. 3(u) it can be seen that the fluorescence signal increased from 900 to 1100 "C then reached a plateau between 1100 and 1600"C and finally dropped off at temperatures higher than 1600 "C. These results are similar to those reported by Anwar et a1.,22 but very different from the results in ref.21 where the optimal vaporization temperature was 2800 "C. This discrepancy may be explained by differences in geometry and 2.0 8 1.5 C 8 ! 1.0 - .4- 0 .- c - $ 0.5 0 40 80 120 160 200 In CI mass ratio Fig. 2 Optimization of 1n:Cl ratio. Experimental conditions 200 ng CI excitation/detection. 267.2/349 nm; monochromator bandpass 0.8 nm; char temperature 300 "C; and atomization temperature 1600 "C Vaporization temperature/"C 0 200 400 600 800 1000 1200 Char temper at u re/" C Fig. 3 Optimization of furnace conditions (a) vaporization tempera- ture; char temperature 300 "C; and (h) char temperature; atomization temperature 1600 "C. Experimental conditions as in Fig. 2 thermal characteristics between the modified graphite furnace in ref.21 and the unmodified commercial graphite furnace used in the present work and in ref. 22. A vaporization temperature of 1200°C was chosen in this work unless otherwise noted to minimize the blackbody emission from the graphite furnace. For optimization of the char temperature [Fig. 3(b)] it can be seen that the fluorescence signal remained constant from 20 to 400"C then decreased gradually from 400 to 800"C and rapidly from 800 to 1200 "C. A char temperature of 300 "C was employed in this work. Indium Monochloride Excitation and Fluorescence Spectra Dittrich and Stark2I did not report any excitation or fluor- escence spectra of indium monochloride. Anwar et obtained one excitation spectrum and one fluorescence spec- trum of indium monochloride.The excitation and detection wavelengths were 267.18 and 359.54 nm respectively. The excitation spectrum for the C-band system (0,O) vibrational transition at a peak wavelength of 267.21 nm was obtained and so was the fluorescence spectrum for the A-band system (0,O) vibrational transition at a peak wavelength of 359.54 nm. No investigation of non-resonance fluorescence spectra was performed in the B- and the C-system. In addition the excitation and fluorescence spectra had relatively low reso- lution; they used a laser beam with a relatively wide linewidth with a full width at half maximum of about 0.01 nm and the spectral bandpass of the monochromator was 1.6 nm. Slightly higher resolution excitation and fluorescence spectra were obtained in our work because the laser linewidth was much narrower (about 0.002-0.003 nm),23 and the monochromator used for fluorescence detection had slightly higher resolution.The bandpass of the monochromator was 0.8 nm at a slit- width of 0.5 mm which was sufficient to resolve all major fluorescence bands as can be seen later. Fig. 4 shows the excitation spectra obtained with different fluorescence detection wavelengths. The excitation spectra areJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY APRIL 1994 VOL. 9 505 100 ( a ) 267.21 nm I I 0 l2 I ( b ) 267.21 nm I 500 1 (c) I I 267.21 nm I 400 300 200 100 I I -0.07 0 0.07 0.14 0.21 0.28 0.35 0.42 Relative wavelengthhm Fig. 4 Excitation spectra obtained at various detection wavelengths. (a) Detection wavelength = 349 nm monochromator bandpass = 2.4 nm; 0 10 ppm C1 in 1 mg m1-l.In as In(NO,),; @ 1 mg ml-' In as In(NO,),. (b) Detection wavelength = 359 nm monochromator bandpass=2.4 nm; 0 10 ppm C1 in 1 mg ml-' In as III(NO,)~. (c) Detection wavelength = 269.5 nm monochromator bandpass = 0.8 nm; 0 1.25 ppm C1 in 1 mg ml-I In as In(NO,) plots of excitation wavelength obtained by tuning the dye laser wavelength versus relative fluorescence signal. As can be seen from Fig. 4 the excitation profiles show the same pattern at different fluorescence detection wavelengths. This obser- vation matched the C-band (0,O) vibrational transition in ref. 24. The excitation profile of the A-band system in Fig. 4(b) displayed more noise than the other spectra probably because the fluorescence intensity of the A-band system was signifi- cantly smaller than the other two bands.Fig. 4(u) contrasts the fluorescence signal of indium mono- chloride and the background scatter signal from the blank solution that contained indium nitrate. The solution that contained both indium ion and chloride displayed a structured excitation band in which the fluorescence signal varied with the excitation wavelength. In contrast the blank solution that contained indium only or chloride only (not shown here) only gave a flat or non-structured background throughout the C-band. The background signal was noticeable because the concentration of indium nitrate was high at 1 mg ml-'. When a solution at such a high concentration is introduced into a graphite furnace the laser radiation is scattered off the vapor- ized material in the gas phase during the vaporization step to produce a scatter signal that is independent of laser wavelength.Fig. 5(a) shows the fluorescence spectrum for the non- resonance detection scheme which included both the A- and B-systems. It should be pointed out that the bands in Fig. 5(u) were not completely resolved. For example the peak observed at 349.4 (+0.8)nm was probably a convolution of (0,O) (349.90 nm) and (1,l) (349.65 nm) transitions of the B-system with the (0,O) transition as the major component. However for simplicity the bands in Fig. 5 ( 4 are labelled with the major component of the peaks. It is clear from Fig. 5(a) that the most intense fluorescence was emitted at 349 nm which corresponds to the B-system (0,O) vibrational transition. Interestingly the fluorescence wavelength used by Anwar et al.22 359 nm which corresponds to the A-system (0,O) vibrational transition was a very weak band as demonstrated in Fig.5(u). Fig. 5(b) displays the fluorescence spectrum for different vibrational transitions in the C-system. The fluorescence spec- trum in Fig. 5(b) shows that intense fluorescence was emitted at the 269.45 nm (0,l) vibrational transition. The indium nitrate solution was used to monitor the background while the fluorescence profile of indium monochloride was obtained. As can be seen from Fig. 5(b) both indium nitrate solution and indium monochloride solution generated equivalent signals at the (0,O) transition which indicated that the signal was mainly from the scattered laser radiation off the vaporized materials in the furnace and the resonance fluorescence signal was masked by the scatter signal.The vibrational transitions of (O,l) and (0,2) etc. had very large fluorescence signals with negligible backgrounds. Detailed investigation showed that the small signals generated from the indium nitrate solution alone a C a v) 2 30 I I ( a ) I C f 340 350 360 370 380 300 200 100 0 267.0 269.0 271.0 273.0 275.0 277.0 Detection wavelengthhm Fig. 5 Non-resonance fluorescence spectra at an excitation wave- length of 267.2 nm. Monochromator bandpass = 0.8 nm; 0 5 ppm C1 in 1 mg ml-' In; m 1 mg ml-l In. (a) Conventional non-resonance fluorescence spectrum A B(2,O); B B(1,O); C B(0,O); D B(0,l); E B(0,2); F A(0,O); and G A(0,l); and (b) unconventional fluorescence spectrum A C(O.0) 267.21 nm; B C(O,1) 269.45nm; C C(0,2) 271.75 nm; D C(0,3) 274.07 nm; and E C(0,4) 276.40 nm506 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY APRIL 1994 VOL.9 at all vibrational transitions other than the (0,O) transition were fluorescence from contamination. It was possible to detect this contamination level due to the high sensitivity. By use of the same indium nitrate solution no contamination was observed in the excitation spectra with detection wavelengths at 349nm of the B-system [Fig.4(a)] and 359nm of the A-system due to poorer sensitivity. Detection Limits and Calibration Curve The LOD was defined as the analyte mass that produced a signal equal to three times the standard deviation (3s) of 16 measurements of the blank noise.25 The detection limits were obtained after subtraction of the blank signal by extrapolation of the calibration curves to a signal level equal to 3s.The measurement of the blank noise was performed either with the laser tuned to the analytical wavelength (the on-line measure- ment) or with the laser tuned away from the analytical wavelength (the off-line measurement). In the present work 0.2% nitric acid solution was used as the blank solution. No significant difference in the LODs was found between the on-line and off-line measurements. Table 2 summarizes the LODs for various detection wave- lengths. The excitation wavelength was the same for all detec- tion schemes at 267.2 nm. It is clear from Table 2 that the best LOD lOpg was obtained with the detection wavelength at 269 nm which corresponded to the C-system (0,l) vibrational transition.The detection wavelength 359 nm used by Anwar et gave the worst LOD among all detection schemes listed in Table 1 a factor of about 50 worse than the best LOD. Also listed in Table 2 are the LODs obtained by the other two research groups.21*22 Dittrich and Stark21 obtained an LOD of 20 pg with the resonance detection scheme. Anwar et ~ 1 . ~ ~ reported an LOD of 20 pg by use of the worst detection scheme i.e 359 nm. If they had used the best detection scheme i.e. 269 nm an LOD of 0.4 pg (400 fg) would have been obtained by assuming the 50-fold improvement in detection limit for the best detection scheme over the worst scheme. We cannot provide a reasonable explanation for the dramatic difference in LOD between our value and theirs for the same detection wavelength.Since no other detection scheme was investigated in ref. 22 a direct comparison of their results between different schemes was not possible. A calibration curve was obtained by use of the best detection scheme with signals obtained from 20 pl aliquots of standard solutions. Attenuation of the LEMOFS signal was necessary to ensure a linear response of the PMT during the construction of calibration graphs. The attenuation was carried out by insertion of calibrated neutral density filters before the mono- chromator. The linear dynamic range of the calibration curve was about four orders of magnitude. Precautions were required during preparation of standard solutions of chloride at low concentrations.In order to minimize contamination sub- boiled water was used to prepare all indium nitrate solutions as well as sodium chloride solutions below 200 ng m1-l. The precision obtained for all experiments was typically between 4 and 9% which was dependent upon the concentration of chlorine and the stability of the laser pulse energy. Table 2 of indium monochloride Detection limits for chlorine by graphite furnace LEMOFS Excitation/detection/ nm 267.21359 267.21349 267.21272 267.2 1269 267.2/26 7.2 267.21359 Detection limitlpg Reference 500 This work 200 This work 20 This work 10 This work 20 Dittrich et al.” 20 Anwar et a1.22 Effect of Metal Ions and Anions Barium and strontium have been reported to have significant enhancement effects on the fluorescence signal of diatomic molecule^.'^*'^*^^ Butcher et aL2’ reported that the addition of barium nitrate caused a significant enhancement in the mag- nesium monfluoride signal with a maximum enhancement of 100-fold after introduction of 1-2 pg of barium.Strontium nitrate was also shown to enhance the magnesium monofluor- ide signal to give a maximum 10-fold increase in signal size with the addition of 10 pg of strontium. The addition of very large amounts of barium or strontium caused suppression of the magnesium monofluoride signal. The effects of barium strontium and potassium on the indium monochloride signal are shown in Fig. 6. Unlike the results for magnesium monofluoride both barium and stron- tium had no significant effects on the indium monochloride signal up to about 1Opg.Potassium showed suppression of the indium monochloride fluorescence signal at amounts higher than 300 ng for 25 ng of chloride which was probably due to the competition between indium and potassium for chlorine in the gas phase. Other metal ions studied include copper magnesium nickel palladium and aluminum. No significant effects were found at amounts up to 10 pg. Fig.7 shows the effects of phosphate and bromide on the indium monochloride fluorescence signal. Phosphate was chosen because the SRMs used in this work contained a significant amount of phosphorus. Bromide co-exists with chloride. Fluoride has been shown previously2’ not to interfere with indium monochloride so it was not studied. As can be seen from Fig. 7(a) phosphate (as ammonium dihydrogen 4.0 r I 2.0 ’ 0 ‘ 1 I I I I 1 x 1 0 - ~ 1 x 0.1 1 .o 10.0 100 Amount of Ba [as Ba(NO,),I/pg A r 1 .- ; I CT I I I I J 0 2 4 6 8 Amount of SrIas Sr(NO,),I/pg *.O t 0 3.0 30 300 3000 Amount of K (as KNO,)/ng Fig.6 Effects of cations on InCl fluorescence signal.Excitation/detection 267.21349 nm for (a) barium (h) strontium and (c) potassium with 1.0 ppm of C1 for (a) and (6) and 1.25 ppm for (c)JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY APRIL 1994 VOL. 9 507 '= (0 6.0 I (a) 1 (b) 4.0 - 2.0 - I I I I I 0 1 .o 10 100 1000 Amount of Br (as NaBr)/ng Fig.7 Effects of anions on InCl fluorescence signal. Excitation/detection 267.2/349 nm for (a) phosphorus and (b) bromide with 1.25 ppm C1; monochromator bandpass 0.8 nm phosphate) did not affect the indium monochloride signal up to 1OOOng of phosphorus above which a dramatic decrease in signal size occurred.In the presence of about 5000ng of phosphorus the indium monochloride signal of 25 ng of chlor- ide was suppressed completely. Therefore it was important to keep the amount of phosphorus below 1000 ng for 25 ng of chloride which gives a P:Cl mass ratio of 40 1. Fig. 7(b) shows the effect of bromide on the indium monochloride fluorescence signal. As can be seen from Fig. 7(b) bromide had no effect on the indium monochloride signal up to about 500ng of bromide above which a gradual decrease in indium monochlo- ride signal occurred. No studies on mixtures of ions were performed in this work. In real samples there would be a variety of cations and anions present simultaneously therefore it is difficult to predict the convoluted effects of these ions on the indium monochloride fluorescence signal.Real Sample Analyses Two biological NIST SRMs were studied SRM 1568a Rice Flour and SRM 1572 Citrus Leaves. The dissolution method for these two samples is described under Experimental. The best fluorescence detection scheme was used to perform the analyses. Five parallel dissolutions of each sample were employed. The results obtained for the determination of chlor- ine in the two samples were in good agreement with the non- certified reference values (Table 3). Matrix effects caused during the vaporization process were observed and dilution was needed to remove them. Initially the analyses were carried out by dilution of the original sample dissolutions with 1 mg ml-I of indium nitrate solution in a 1:l v/v ratio.Incorrect results were obtained for both samples with this treatment and the signal profiles from both samples showed some changes compared with those of standard solu- Table3 Determination of chlorine in Rice Flour and Citrus Leaves by LEMOFS of indium monochloride Non-certified value/ This work/ Sample Pg g-' P8 g-' SRM 1568a 300 309 & 22 (n = 5)* SRM 1572 414 417+39 (n=5)* Rice Flour Citrus Leaves * n denotes the number of sample dissolutions that were analysed and used to calculate the standard deviation. 0 2.0 I 4.0 6.0 Time/s Fig. 8 Temporal signal profiles of standard solution and NIST SRMs. (a) 1.0 ppm C1 standard solution; (b) Rice Flour and (c) Citrus Leaves 1:l dilution of the sample solutions; (d) and (e) Rice Flour Citrus Leaves 1 7 dilution of the sample solutions; see text for details tions.Fig. 8 reveals that the signals from the two samples with 1 1 v/v dilution ratio appeared temporally earlier than that of the standard solution which indicated the presence of vaporiz- ation matrix effects. No experiments were performed to identify the exact causes. The phosphorus content in the sample solution was too low to cause any significant effect according to the interference study described earlier. When these sample solutions were further diluted by use of a 1:7 v/v ratio the appearance time of the signal for both samples was about the same as that for the standard solution (Fig. 8). Under this condition the correct results for the chlorine contents in both samples were obtained (Table 3) which indicated the removal of vaporization matrix effects by dilution.Conclusion The use of fluorescence from a vibrational transition within the same electronic transition as the excitation should be applicable to other diatomic molecules. This non-resonance scheme could be valuable for molecules with no conventional508 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY APRIL 1994 VOL. 9 non-resonance detection wavelength at a different electronic transition. It can be concluded that LEMOFS of indium monochloride is extremely sensitive which allows dilution to remove matrix effects caused during the vaporization process. The removal of vaporization matrix effects by dilution of the sample in LEMOFS is advantageous for real sample analyses compared with MAS where tedious extraction and back-extraction procedures have been needed to remove vaporization matrix effects.Compared with other methods discussed in the introduction LEMOFS measures total chlorine with high sensitivity and has micro-sample capability. Inductively coupled plasma mass spectrometry (ICP-MS) is a competitive technique with LEMOFS. However ICP-MS has a poor LOD for chlorine 40 ng s-l (ref. 27) which is equivalent to an absolute LOD of 400 ng for a counting period of 10 s. This is more than four orders of magnitude higher than the LOD of lOpg obtained in the present work by LEMOFS. The poor LOD for chlorine by ICP-MS is due to the high ionization energy of chlorine (13.02 eV) compared with that of argon (15.76 eV).It has been calculated that only 0.9% of chlorine is ionized in argon ICP-MS while most elements other than non-metals with high ionization energies such as halogens have a more than 90% degree of elemental ionization.27 The determination of halogens including chlorine as negative ions by use of an argon microwave-induced plasma (Ar MIP) has been investi- gated and found to be less sensitive2’ Detection of halogens as positive ions by use of He MIP as an ion source for MS has been investigated. The preliminary LOD for chlorine was 21 pg s-l which is equivalent to an absolute LOD of 210 pg for a counting period of 10 s . ~ ’ This He MIP-MS detection limit is higher than the LOD of 1Opg obtained in the present LEMOFS work.The determination of chlorine at trace levels by plasma source MS is still under investigation while little or no real sample analyses at trace levels have been reported. This work was supported in part by the National Institute of Health grant number GM32002. R.G.M. was supported by a Research Career Development Award from the National Institute of Environmental Health Sciences under Grant No. ESOOl30. E.G.S. was supported by an American Chemical Society Division of Analytical Chemistry Fellowship spon- sored by Glaxo. 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 References Singley J. E. in Kirk-Othmer Encyclopedia of Chemical Technology 3rd ed.; Wiley-Interscience New York 3rd edn. Tietz N W. Fundamentals of Clinical Chemistry Saunders Philadelphia 2nd ed.1976 p. 879. Chapman B. R. and Goldsmith I. R. Analyst 1982 107 1014. Subramaniam G. Chandra N. and Rao G. P. Talanta 1984 31 79. Kudermann G. and BlaufuB K.-H. Mikrochim. Acta ( Wien) 1990 11 273. Mehra H. C. and Frankenberger W. T. jr. Microchem. J. 1990 41 93. Galban J. Urarte M. and Aznarez J. Microchem. J. 1990,41,84. Hon P. and Townshend A. Anal. Chim. Acta 1980 115 395. Aoki T. and Munemori M. Anal. Chem. 1983 55 209. Zenki M. Komatsubara H. and Toei K. Anal. Chim. Acta 1988,208 317. Midgley D. and Gatford C. Microchem. J. 1990 42 225. Dittrich K. Spivakov B. Ya. Shkinev V. M. and Vorob’eva G. A. Talanta 1984 31 39. Dittrich K. CRC Crit. Rev. Anal. Chem. 1986 16 223. Tsunoda K. Fujiwara K. and Fuwa K. Anal. Chem. 1978 50 861. Dittrich K. and Meister P. Anal. Chim. Acta. 1980 121 205. Yoshimura E. Tanaka Y. Tsunoda K. Toda S. and Fuwa K. Bunseki Kagaku 1977 26 648. Dittrich K. and Vorberg B. Anal. Chim. Acta 1982 140 2337. Dittrich K. Spivakov B. Ya. Shkinev V. M. and Vorob’eva G. A. Talanta 1984 31 341. Dittrich K. Vorberg B. Funk J. and Beyer V. Spectrochim. Acta Part B 1984 39 349. Butcher D. J. Irwin R. L. Takahashi J. and Michel R. G. J. Anal. At. Spectrom. 1991 6 9. Dittrich K. and Stark H.-J. Anal. Chim. Acta 1987 200 581. Anwar J. Anzano J. M. Petrucci G. and Winefordner J. D. Analyst 1991 116 1025. Su E. G. Irwin R. L. Liang Z. and Michel R. G. Anal. Chem. 1992 64 1710. Donn6es Spectroscopiques Relatives aux Moldcules Diatorniques ed. Rosen B. Pergamon Oxford 1970. Long G. L. and Winefordner J. D. Anal. Chem. 1983 55 7139. Dittrich K. Hanisch B. and Stark H. J. Fresenius’ 2. Anal. Chem. 1986,324,497. Chong N. S. and Houk R. S. Appl. Spectrosc. 1987 41 66. Douglas D. J. and French J. B. Anal. Chem. 1981 53 37. V O ~ . 3 1980 pp. 938-958. Paper 31046508 Received August 3 1993 Accepted November 25 1993

ISSN:0267-9477    

DOI:10.1039/JA9940900501

出版商:RSC    

年代:1994

数据来源: RSC

摘要:

JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY APRIL 1994 VOL. 9 509 Reduction of Matrix Effects and Mass Discrimination in lnductively Coupled Plasma Mass Spectrometry With Optimized Argon-Nitrogen Plasmas Grace Xiao and Diane Beauchemin* Queen's University Department of Chemistry Kingston Ontario Canada K 7L 3N6 The effect of power (1.2-1.4 kW) and the addition of N (2-10°/0) to the outer gas of an Ar plasma was studied in an attempt to improve the analytical capability of inductively coupled plasma mass spectrometry (ICP-MS) for multi-element analysis. The experiments were all conducted at an optimum sampling position i.e. by adjusting the aerosol carrier gas flow rate for maximum sensitivity after each change in power and/or addition of N,. For each set of conditions the analytical capabilities of Ar-N plasmas in terms of sensitivity detection limits mass discrimination and susceptibility to the effect of 0.01 and 0.1 mol dmP3 Na on various analytes (Al V Cr Mn 56Fe 57Fe Co Ni Zn Cu As 77Se 78Se Mo Cd Sb and Pb) were evaluated.Some improvement in sensitivities and detection limits were observed upon addition of N to the outer gas in comparison with an Ar plasma at the same power. Both mass discrimination and the effect of either 0.01 or 0.1 mol dm-3 Na were reduced with the addition of NP. Compared with conventional operating conditions (Le. 1.2 kW no N,) a mixed-gas plasma with 10% N at 1.3 kW led to only slightly degraded sensitivities and detection limits (in fact improvements were observed for a few elements); but the effect of 0.01 mol dm -3 Na was eliminated across the mass range and that of 0.1 mol dmP3 Na was reduced to a uniform level across the mass range allowing the use of a single internal standard.Keywords lnductively coupled plasma mass spectrometry; mixed-gas plasma; effect of concomitant elements Although inductively coupled plasma mass spectrometry (ICP-MS) is more powerful than its optical emission spec- trometry (OES) equivalent in many respects (detection limits relative freedom from spectroscopic interferences capability of isotopic analysis e t ~ . ) ' - ~ its routine application is not as widespread as that of ICP-OES mainly because of its greater susceptibility to sample matrix effects. This arises from the fact that the sampling of ions in ICP-MS is not a passive process as is the measurement of light in ICP-OES.The matrix can therefore lead to solid deposits on the interface which gradually clog the sampler orifice inducing a downward drift in sensi- t i ~ i t y . ~ It can also have unpredictable effects on count rates from the analytes. Although signal suppression is in general ~bserved,~-'~ presumably owing to space charge enhancements have also been r e p ~ r t e d . ~ . ~ * ' ~ * ' ~ Instrumental conditions can be adjusted to decrease the magnitude of these effects or even eliminate them but this is usually achieved to the detriment of sen~itivity.~~'~~''*'~ Internal standardization can also to some extent compensate for these effect^.^.^." However because of the additional problem of mass discrimination if the analytes are spread over a fairly large mass range more than one internal standard may be required for adequate c~mpensation.~*~*~ The multi-element analysis of samples with complex matrices can therefore be fairly complicated requiring frequent recalibration and drift correction procedures," the method of standard addition^,^ or the isotope dilution technique for example ref.20 (which can only be used if there are two isotopes free of spectroscopic interferences for each analyte). The addition of N to the Ar outer gas has recently been reported to eliminate the enhancing effect of 0.01 mol dmP3 K2' but with a concurrent decrease in sensitivity. The latter was a result of the shift in the initial radiation zone2 (IRZ) which occurred upon the addition of N to the outer plasma gas,,' because the nebulization efficiency (i.e.aerosol carrier gas flow rate and sample uptake rate) was kept constant throughout the studies. Yet Ar-N mixed-gas plasmas have been shown to improve sensitivity when the sampling position was re-adjusted after each change in any operating param- * To whom correspondence should be addressed. eter.23-26 This adjustment could be achieved by increasing the nebulizer flow rate or physically moving the plasma closer to the interface either of which moved the IRZ closer to the sampler.22-26 The present work was therefore undertaken to investigate the effect of the percentage of N2 and the power on the analytical characteristics of the plasma for a large number of analytes covering the entire mass range. In this case however the results were obtained at an optimum sampling position near the tip of the IRZ by re-adjusting the aerosol carrier gas flow rate so as to maintain the best sensitivity after any change in power and/or percentage of N,.The ultimate goal of this work was to find optimum operating conditions in terms of sensitivity and detection limit with reduced mass discrimi- nation and hopefully no effect of concomitant elements (specifically Na) since the elimination of the last would allow the analysis of samples with a variety of matrices using a simple calibration with aqueous standard solutions. Experiment a1 Instrumental Conditions A Perkin-Elmer SCIEX Elan Model 500 inductively coupled plasma mass spectrometer (Thornhill Ontario Canada) was used throughout this study.Both the nebulizer (Meinhard C-3) and the double-pass spray chamber (Scott type) were standard components but a number of modifications were made to the instrument. For instance the torch was replaced by a low-flow PlasmaTherm torch. An x-y-z translation stage was installed under the torch box to allow precise positioning of the plasma in three dimensions with respect to the interface. A mass-flow controller (Model 1259B MKS Instruments Andover MA USA) was used to regulate the flow rate of the aerosol carrier gas the solution uptake rate to the nebulizer being maintained constant at 2 cm3 min-' using a peristaltic pump (Minipuls 11 Gilson Medical Electronics Middleton WI USA). A gas proportioner (Model 7371 Matheson Whitby Ontario Canada) was employed to add various percentages of N2 to the outer gas.The proportioner was connected to the input of the outer gas flow meter so as to510 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY APRIL 1994 VOL. 9 keep the total flow rate of the outer gas constant at 12 dm3 min-'. The operating conditions are summarized in Table 1. The distance between the sampler and the load coil was kept constant at about 1.8 cm. The resolution was set at 1.1 m/z (peak width at 10% of the peak height). Data acquisition was accomplished with the Elan 500 multi-element data acquisition software Version 2 in the maximum intensity mode. The species monitored were 27Al+ 'lV+ ',Cr+ "Mn+ 56Fe+ 57Fe+ 59C0t 60Ni+ 64Zn+ 6 5 C ~ f 75A~+ 76Se+ 78Se+ 9 8 M ~ + '14Cd+ 121Sb+ and ,08Pb+. (The 65Cu+ isotope was selected instead of 63Cu+ because of the expected 40Ar23Na+ interference).Reagents Three pairs of multi-element standard solutions each con- sisting of 0 and 100 pg dm-3 were prepared in respectively 1% v/v HNO 0.01 moldm-3 Na-1% HN03 and 0.1 mol dmP3 Na-lo/o HNO using lo00 mg dm-3 mono- element solutions (Spex Industries Edison NJ USA) distilled de-ionized water ( Milli-Q Plus Millipore Mississauga Ontario Canada) high-purity concentrated HNO (Seastar Sidney British Columbia Canada) and NaNO (AnalaR BDH Poole Dorset UK 99.5% purity). Procedure Initial optimization An all-Ar plasma was ignited at a power of 1.2 kW and allowed to warm up for 10min. The three-dimensional alignment of the plasma with the interface and the aerosol carrier gas flow rate were adjusted to maximize Rh' while aspirating 100 pg dm- of Li+ Rh+ and Pb'.Then the voltages of the ion lenses were adjusted to make the Li+ and Pb+ signals as equal as possible to one another without sacrificing the Rh+ signal. Finally the aerosol carrier gas flow rate was re-checked for maximum Rh+ sensitivity. Table 1 ICP-MS operating conditions Plasma conditions- Torch R.f. power/kW Reflected power/W N,-Ar outer flow rate/dm3 min - ' Proportion of N in outer gas (YO) Aerosol carrier gas flow rate/dm3 min Intermediate gas flow rate/dm3 min-' Sample delivery rate/cm3 min-' Bessel box stop/V Bessel box barrel/V Einzel lenses 1 and 3/V Bessel box end lenses/V Ni sampler orifice diameter/mm Ni skimmer orifice diameter/mm Interface pressure/Torr* Mass spectrometer pressure/Torr Measurement purameters- Replicate time/ms Dwell time/ms Scanning mode Sweeps per reading Number of replicates Points per spectral peak Muss spectrometer settings- Low flow 1.2- 1.4 t 8 12.0 2-10 0.82- 1.023 2.0 2.0 1 -4.6 to -8.0 1.3-2.3 -15.0 to -19.9 -4.8 to -7.8 1.14 0.89 (1.3-2.5) x 0.8-0.9 300 10 Peak hop 30 6 3 t * 1 Tom= 133.322 Pa.t One measurement at the central mass and the two others at +O.l m/z from the assumed peak centre; the average of these three points was computed. Formation of mixed-gas plasmas Nitrogen was introduced slowly into the outer gas stream of an already ignited Ar plasma at a selected power until the desired percentage of N was reached. The initial addition of N had to be carried out very carefully to prevent the plasma from extinguishing as a result of a sudden shrinkage of the plasma. The reflected power was kept below 8 W by manual fine tuning.Optimization of the sampling position After each modification of the plasma ( i e . an increase in incident power and/or the percentage of N2) a solution containing 100 pg dm-3 of Li+ Rh+ and Pb' was aspirated. The count rate for Rh+ was then maximized and those of Li+ and Pb+ made equal to one another by first adjusting the alignment of the plasma with respect to the interface then the aerosol gas flow rate and finally the voltages of the ion lenses as described above. Although the lens settings were found to be independent of the amount of N2 and only needed to be changed slightly when the power was increased from 1.2 to 1.3 kW the aerosol carrier gas flow rate had to be re-adjusted whether the power was increased or N was added.The various flow rates corresponding to the different sets of conditions which were used are summarized in Table 2. Study ofthe eflect of sodium The three pairs of solutions each consisting of a blank and a 100 pg dm- multi-element standard solution in a different matrix (i.e. Na free 0.01 mol dm-3 Na and 0.1 mol dm- Na) were aspirated in order of increasing Na concentration into the plasma under a given set of operating condition. Between each set the sample introduction system was rinsed with distilled de-ionized water for 5 min. Sodium was chosen for this study as it is a common matrix element (present in biological samples saline waters etc.) Data processing For each pair of solutions the count rate of the blank (which was always aspirated first) was subtracted from the count rate of the corresponding 100 pg dm- standard multi-element solution to yield a net count rate.The corresponding signal- to-background (S/B) and signal-to-noise (S/N) ratios were obtained by dividing the net count rate by respectively its corresponding blank count rate and the standard deviation of the blank count rate. The wide linear dynamic range of ICP-MS allows the use of the net count rate of one standard as a good estimate of the sensitivity. Similarly the detection limit which is defined as three times the standard deviation of Table 2 Aerosol carrier gas flow rates (dm3 min-') used to obtain an optimum IRZ position under various conditions Power/kW 1.2 1.2 1.2 1.2 1.2 1.3 1.3 1.3 1.3 1.3 1.4 1.4 1.4 1.4 1.4 N (%I 0 2 5 8 10 0 2 5 8 10 0 2 5 8 10 Flow rate 0.8 20 0.822 0.884 0.961 0.957 0.847 0.875 0.931 1.024 0.997 0.875 0.903 0.928 1.007 0.992JOURNAL O F ANALYTICAL ATOMIC SPECTROMETRY APRIL 1994 VOL.9 51 1 the blank count rate divided by the sensitivity is equal to three times the inverse of S/N. An increase in S/N therefore corre- sponds to an improvement in detection limit. For the assessment of mass discrimination the sensitivity in count rate per pg dm-3 was converted into count rate per mol dm-3 by multiplying it by the relative molecular mass of the element. The resulting molar sensitivity was divided by the natural abundance of the particular analyte isotope monitored so as to express it as if it were due to a single isotope with 100% natural abundance.Determination of the nebulization eficiency Dry silica gel was used to collect the aerosol exiting the spray chamber. A plastic tip from a 1000 pl micropipette filled with dried silica gel was used as a moisture absorber. The diameter of the tip was enlarged to 5 mm so as to match the diameter of the injector tube of the plasma torch and that of the outlet from the spray chamber. The collector was tightly connected to the outlet from the spray chamber. Care was taken to ensure that the normal back-pressure was present at the drain by placing the drain in a large beaker filled with water. The fine aerosol exiting the spray chamber was then absorbed by the silica gel. Its mass was measured after pumping an accu- rately known mass of water for 2 min (at 2 cm3 min-').The nebulization efficiency i.e. the ratio of the mass of water absorbed to that aspirated was determined at each of the aerosol carrier gas flow rates used during this work. Results and Discussion Nebulization Efficiency In a previous study,,' the IRZ moved whenever N was introduced because the nebulization efficiency was kept con- stant. In this study the IRZ was kept at an optimum position in terms of sensitivity by adjusting the aerosol carrier gas flow rate. Under these conditions however the count rate could change as a result of a varying nebulization efficiency. The transport efficiencies which were measured with the present sample introduction system (consisting of a Meinhard C-3 nebulizer and a Scott-type double-pass spray chamber) at the different aerosol carrier gas flow rates used during this work (see Table 2) for instance are shown in Fig.1. Clearly the efficiency increased fairly linearly with an increase in aerosol carrier gas flow rate (at a constant uptake rate of 2 cm3 min-'). However sensitivity reflects the production and transport of the aerosol as well as the ion yield in the plasma. Although a higher sensitivity could be expected with a higher nebuliz- ation efficiency the latter can also affect the plasma conditions since a variable amount of water will require a different amount 0.80 0.84 0.88 0.92 0.96 1.00 1.04 Aerosol carrier gas flow ratdl min-' Fig. 1 aerosol carrier gas flow rate Change in the aerosol transport efficiency as a function of the of energy (and hence residence time) from the plasma.In other words if the droplet size distribution is wider or narrower then the fraction of the droplets which have undergone desolv- ation vaporization atomization and ionization between the time of their introduction into the plasma and sampling by the interface will be different which could result in a different analyte signal. Thus one cannot strictly separate the nebulizer efficiency from the plasma conditions. Nonetheless in an attempt to make the results more compar- able i.e. assess only the effect of an addition of N2 on the analytical characteristics the blank-subtracted signal obtained at a given aerosol carrier gas flow rate was divided by the measured nebulization efficiency (from Fig.1 ) and multiplied by that corresponding to the optimum nebulizer flow rate under the same conditions but without N,. This method compensates for changes in nebulization efficiency only not for variations in droplet size distribution. Effect of N on Count Rates S/N and S/B at Various Powers As shown in Table 2 the aerosol carrier gas flow rate had to be increased to maintain the IRZ at an optimum position with respect to the sampler whenever power was increased and/or N2 was added to the plasma. This was because the distance between the sampler and the top of the load coil usually called sampling depth was held constant at 1.8 cm which was as close as the present system allowed. It is also shown in Table 2 that as the percentage of N was increased an increase in the aerosol carrier gas flow rate was necessary to maintain the peak response.The S/N S/B and net count rate observed in presence of N at 1.2-1.4 kW are compared with those obtained in an Ar plasma at the same power in Tables 3-5. At 1.2 kW the sensitivity was degraded compared with an Ar plasma except for V As Mo and Sb for which an enhance- ment occurred. In addition S/N and S/B were degraded for all elements except Fe and V. Increasing the power to 1.3 kW did not significantly improve anything except for the elements which had shown enhancements at 1.2 kW and that displayed even greater improvements. At 1.4 kW however S/N was similar or improved compared with that observed in an Ar plasma at the same power for all elements except A1 and Cd for which only degradations occurred at all percentages of N2 investigated.Nonetheless the improvement in S/N was observed while the corresponding sensitivity was degraded with the exception of V As Mo and Sb which were enhanced. It is shown in Tables 3-5 that for a given percentage of N the extent of suppression was very similar at different powers consistent with the relatively fixed IRZ position that was maintained. However most of the analytes except V As Mo and Sb were suppressed to an extent which increased with the percentage of N,. Therefore the best results in terms of sensitivity were in general observed with 2% N at a given power. The decrease in count rates with an increase in N may indicate the imperfection of the correction for nebulization efficiency but more likely shows that the effect of N is more complex than a simple shift in position of the IRZ.The central channel of the plasma has indeed been observed to become narrower in a mixed-gas plasma.,' These results are at first glance different from those of Lam and H o r l i ~ k ~ ~ who reported a modest enhancement of sensi- tivity over the mass range at higher power and higher nebulizer flow rate with the addition of N to the outer gas. Furthermore the enhancement was similar for 5-20% N,. The discrepancies could result from instrumental differences; for instance a different type of torch was used by Lam and Horlick and the plasma was physically moved closer to the sampler (the sampling depth was 1 cm i.e. nearly half that used in the present study). However the fact that Lam and H o r l i ~ k ~ did not correct their results for changes in nebulization efficiency can certainly account for at least part of the differences.The improvement in sensitivity observed for V As Mo and512 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY APRIL 1994 VOL. 9 Table 3 and a power of 1.2 kW Effect of the percentage of N2 in the Ar outer gas on analyte signals (S) SIN and SIB for 100 pg dm-3 with an optimum IRZ position Analyte A1 v Cr Mn Fe Fe c o Ni Zn c u AS Se Se Mo Cd Sb Pb 27 51 52 55 56 57 59 60 64 65 75 77 78 98 114 121 208 2% NJO% N2 5% N,/O% N2 8% NJO% N2 10% NJO% N2 S/countss-' SIN SIB 0.52 1.46 0.87 0.69 0.64 0.68 0.54 0.52 0.61 0.44 1.72 0.52 0.54 1.92 0.60 1.41 0.53 0.49 0.60 1.13 7.23 0.34 0.30 0.17 0.09 0.41 1.15 1.74 2.31 0.08 0.28 0.31 0.39 0.37 1.07 0.07 0.31 0.18 0.13 0.68 0.58 0.17 0.45 0.32 0.33 0.25 0.20 0.22 0.17 0.97 0.25 ~~ ~ Slcounts s - l SIN 0.57 0.4 1 1.49 4.16 0.88 0.29 0.67 0.10 0.6 1 0.85 0.64 1.12 0.52 0.1 1 0.50 0.3 1 0.55 0.40 0.43 0.23 1.46 0.32 0.41 0.38 0.4 1 0.18 1.66 0.36 0.47 0.43 1.1 1 0.57 0.42 0.85 SIB 0.26 4.42 0.12 0.06 0.79 1.3 1 0.18 0.22 0.74 0.20 0.15 0.27 0.37 0.16 0.1 1 0.10 0.22 Slcounts s-' SIN SIB 0.43 1.13 0.70 0.54 0.47 0.49 0.37 0.35 0.35 0.30 0.82 0.25 0.25 1.42 0.36 0.87 0.42 0.41 0.12 1.11 2.43 0.36 0.15 0.12 0.06 1.39 1.36 1.81 1.85 0.09 0.14 0.13 0.14 0.26 0.57 0.21 0.14 0.18 0.09 0.31 0.14 0.16 0.31 0.16 0.06 0.14 0.06 0.16 0.05 0.32 0.17 0.38 1.10 0.66 0.52 0.48 0.49 0.38 0.37 0.4 1 0.32 1.03 0.30 0.33 1.52 0.43 1.08 0.50 0.20 0.12 1.06 2.74 0.79 0.43 0.47 0.08 3.12 3.19 1.89 3.24 0.07 0.14 0.29 0.16 0.15 0.55 0.16 0.20 0.39 0.12 0.17 0.14 0.30 0.30 0.60 0.08 0.21 0.07 0.21 0.07 0.39 0.21 Table 4 Effect of the percentage of N2 in the Ar outer gas on analyte signals (S) SIN and SIB for 100 pg dm-3 with an optimum IRZ position and a power of 1.3 kW 2% NJOYo N2 Analyte A1 v Cr Mn Fe Fe c o Ni Zn c u AS Se Se Mo Cd Sb Pb mlz 27 51 52 55 56 57 59 60 64 65 75 77 78 98 114 121 208 Slcounts s - SIN 0.45 0.96 1.53 2.89 0.74 0.93 0.57 0.34 0.54 6.36 0.56 0.61 0.43 0.43 0.39 0.4 1 0.43 0.64 0.32 0.73 1.34 0.35 0.34 0.3 1 0.35 0.28 2.23 1.00 0.47 0.72 1.20 0.08 0.50 0.52 SIB 0.46 9.46 1.01 0.19 2.26 4.37 0.61 0.61 1.35 0.64 0.32 0.36 0.33 0.97 0.44 0.36 0.62 8% NJOYo N2 10% N2/O% N2 Slcounts s-' SIN 0.36 0.38 1.26 3.10 0.54 0.22 0.45 0.20 0.43 13.2 0.45 0.4 1 0.34 0.55 0.33 0.19 0.36 1.18 0.27 0.73 1.20 0.60 0.29 0.26 0.30 0.39 1.73 3.14 0.36 0.35 0.96 0.18 0.32 0.28 SIB 0.22 6.60 0.64 0.1 1 2.18 4.40 0.27 0.30 0.78 0.39 0.20 0.18 0.22 0.68 0.17 0.27 0.26 Slcounts s - l SIN 0.23 0.18 0.82 1.59 0.40 0.54 0.30 0.22 0.27 2.90 0.28 0.21 0.20 0.25 0.18 0.15 0.18 0.99 0.16 0.84 0.53 0.12 0.14 0.08 0.14 0.17 1.18 0.70 0.22 0.19 0.59 0.02 0.29 0.27 SIB 0.1 1 3.50 0.44 0.08 2.16 3.52 0.14 0.17 0.50 0.27 0.10 0.09 0.12 0.29 0.13 0.1 1 0.29 Slcounts sC1 SIN 0.25 0.40 0.97 2.2 1 0.44 0.38 0.33 0.18 0.33 5.97 0.34 0.32 0.27 0.25 0.26 0.10 0.30 0.70 0.2 1 2.15 1 .oo 0.22 0.25 0.14 0.25 0.27 1.40 0.88 0.29 0.22 0.82 0.3 1 0.27 0.35 SIB 0.12 4.4 1 0.43 0.07 2.37 4.06 0.13 0.18 0.66 0.32 0.13 0.12 0.17 0.37 0.1 1 0.19 0.26 Table 5 Effect of the percentage of N2 in the Ar outer gas on analyte signals (S) SIN and SIB for 100 pg dm-3 with an optimum IRZ position and a power of 1.4 kW 2% NJOYo N2 5% NJO% N2 8% N,/O% N2 10% N,/O% N2 Analyte m/z Slcounts s-' SIN SIB S/countss-' SIN SIB S/countss-' SIN SIB Slcountss-' SIN SIB A1 27 v 51 Cr 52 Mn 55 Fe 56 Fe 57 c o 59 Ni 60 Zn 64 c u 65 AS 75 Se 77 Se 78 Mo 98 Cd 114 Sb 121 Pb 208 0.45 1.66 0.77 0.60 0.60 0.58 0.46 0.42 0.46 0.35 1.46 0.36 0.39 2.55 0.5 1 1.28 0.53 0.20 0.30 1.60 6.25 2.70 1.20 1.78 0.19 6.31 2.62 3.58 4.46 2.10 0.40 1.57 0.68 1.85 1.05 1.36 0.88 0.53 0.84 0.28 0.53 0.98 0.38 0.87 0.93 0.36 0.46 0.36 0.58 3.22 1.07 0.29 1.16 0.52 0.4 1 0.42 0.41 0.34 0.32 0.41 0.28 1.51 0.37 0.38 1.86 0.42 1.14 0.36 0.12 0.19 2.07 6.04 2.48 0.92 0.76 0.10 7.92 2.35 2.25 3.83 1.08 0.22 2.54 0.32 0.78 0.75 5.40 0.63 0.43 0.52 0.30 0.24 0.90 0.29 3.88 0.87 0.32 0.24 1.24 0.51 2.07 0.70 0.2 1 0.8 1 0.34 0.29 0.30 0.29 0.23 0.22 0.30 0.19 0.94 0.22 0.24 1.55 0.34 0.99 0.38 0.03 0.30 1.84 3.19 2.53 0.32 0.60 0.02 1.64 2.57 2.38 4.30 1.15 0.19 1.01 0.16 0.81 2.69 1.91 0.58 0.21 0.12 0.12 0.16 0.24 0.07 3.69 1.01 0.58 0.11 1.15 0.37 5.06 0.98 0.60 0.62 0.6 1 0.62 0.65 0.64 0.71 0.74 0.84 0.74 1.07 0.20 1.08 0.53 0.68 0.69 0.44 0.06 0.09 2.08 3.33 1.72 0.47 0.30 0.04 12.5 1.52 1.69 3.16 0.70 0.09 1.04 0.22 1.22 1.00 2.75 0.53 0.78 0.39 0.28 0.21 1.02 0.25 3.12 0.66 0.37 0.19 0.59 0.27 0.57 0.50JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY APRIL 1994 VOL.9 513 Sb with 2-10%0 N at 1.2-1.4 kW is noteworthy.In particular V and Mo two analytes that are considered to be almost 100% ionized in an Ar plasma3 were enhanced to a greater extent than As and Sb which have a lower degree of ionization. However V and Mo form the strongest oxides of all the analytes of this study and the addition of N has been shown to be particularly efficient at reducing polyatomic oxide specie^.,^-,^ This certainly explains at least part of the enhance- men t observed. The notable increase in S/B observed for V from 1.2 to 1.4 kW with 2-10%0 N is surprising. It indicates that the background at m/z 51 was efficiently reduced by adding N to the Ar outer gas. However the only background to be expected with the chlorine-free solutions used for this study should come from 38Ar13C+ 36Ar15N+ and 36Ar'4NH+ and an increase in the last two components would have been expected upon the introduction of N,.An increase in background at m/z 51 has indeed been reported by Louie and S O O ~ who however added N2 to the outer gas as well as the intermediate and aerosol carrier gases. The lower background observed during this study suggests changes in the ionization mechan- isms and/or ion-molecule reactions leading to the formation of ions in the plasma when N is introduced into the outer gas and certainly warrants further investigation Nonetheless the Fe signals were suppressed while both S/N and S/B improved especially at higher power. This indicates as did previous s t ~ d i e s ~ ' * ~ ~ - ~ ~ that the addition of N decreases the interferences caused by Ar-containing polyatomic species (40Ar'60+ on 56Fe+ and 40ArOH+ on 57Fe+).In the case of 52Cr+ which is interfered with by 40Ar12C+ and 36Ar160+ the only improvement in S/B and especially S/N was observed with 2% N at 1.4 kW which is to some extent in agreement with a previous study at fixed nebulization efficiency,,' although improvements were seen with 2-10% N at 1.2 kW. Since the optimum IRZ position is determined mainly using Rh this might indicate that with 2% N the optimum sampling position is not the same for Cr and Rh a higher power being required to make their individual 'IRZ' coincide. For all the other elements the increase in S/N observed at 1.4 kW (relative to an Ar plasma at the same power) did not correspond to a reduction of the background and is therefore owing to an improvement in the stability of the plasma.In contrast it is shown in Table 6 that without N a higher power resulted in enhanced sensitivity for all analytes with the corresponding changes in S/N varying from suppressions to enhancements depending on the analyte and the power. Effect of N on Mass Discrimination The molar sensitivities (corrected to 100% natural abundance but not for changes in the degree of ionization) reported in Table 7 clearly show that for a given amount of N mass discrimination became more severe as power increased. On the other hand at a given power mass discrimination first decreased and then increased with an increase in the percentage of N,. The best compensation for m/z 27-65 was obtained with 5% N at 1.2kW whereas 2% N at the same power provided a better improvement for higher masses.An addition of N therefore improved mass discrimination as had been observed in a previous study performed however at fixed nebulization efficiency.21 Mass discrimination has been partly attributed to space charge effect^'^,'^ whereby lighter ions are repelled by heavier ones. The heavier the bulk of the ions extracted (i.e. the matrix) the worse will be the defocusing effect. The introduc- tion of a fairly small amount of N was nonetheless sufficient to bring about a reduction in mass discrimination. This can probably be explained as follows. When sampling an Ar plasma most of the ion defocusing is owing to the excessive Ar+ current. Upon addition of N to the outer gas which not only reduces the overall size of the plasma but also the diameter of the central channel,,' the Ar' current might be reduced while that of N + (a lighter ion) increases.This would in turn reduce the defocusing of analyte ions and therefore decrease mass discrimination. Further investigations are required to verify this hypothesis (e.g. measurements of the Ar' and N+ currents). Thus the simplicity of the present approach is another advantage of using mixed-gas plasmas. Effect of N on Non-spectroscopic Interference from Na Non-spectroscopic interference f rom Na in an Ar plasma The effect of Na on analyte count rates in an Ar plasma was seen as shown in Fig. 2 as a suppression for all analytes at the different powers with either 0.01 or 0.1 mol dm-3 Na.In general a greater suppression was observed with a higher Na concentration and/or power. However the shape of the signal curve as a function of m/z remained fairly constant whatever the power used which is consistent with a relatively constant IRZ position and indicates no change in mass discrimination as a function of power. Table 6 Effect of an increase in power on analyte signals (S) SIN and SIB for 100 pg drn-j at the optimum IRZ position Ratio of power/kW Slcounts s - ' SIN Element A1 v Cr Mn Fe Fe c o Ni Zn c u As Se Se Mo Cd Sb Pb m/z 27 51 52 55 56 57 59 60 64 65 75 77 78 98 114 121 208 1.311.2 1.53 1.19 1.52 1.57 1.51 1.50 1 S O 1.53 1.52 1.60 1.21 1.40 1.43 1.13 1.52 1.37 1.65 1.411.3 1.13 1.09 1.10 1.07 1.04 1.12 1.10 1.1 1 1.06 1.10 1.05 1.02 1.03 1.06 1.08 1.09 1.15 1.311.2 0.96 0.99 1.51 0.87 1.90 1.15 0.78 0.96 0.70 0.78 1.44 1.47 2.16 0.3 1 0.79 0.66 0.98 1.411.3 1.08 1.49 0.74 0.68 0.75 1.16 0.68 0.86 0.79 0.5 1 0.51 0.83 0.82 0.70 0.8 1 0.82 0.42 1.311.2 0.9 1 0.90 2.28 0.99 1.06 8.48 0.32 1.38 0.55 0.22 1.31 2.26 0.99 0.36 0.85 1.99 1.57 1.411.3 3.90 2.28 0.20 0.26 0.86 0.15 0.24 0.17 0.80 0.29 0.83 0.66 0.87 0.53 1.28 0.3 1 0.15514 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY APRIL 1994 VOL.9 0.80 Table 7 Molar analyte signals corrected to 100% natural abundance and expressed (in %) relative to that of 208Pb under various conditions while maintaining the IRZ in the optimum position. (These values were also corrected for changes in nebulization efficiency) ( C) - N in Ar outer gas (YO) 0 Power/kW - Element m/z 1.2 1.3 1.4 A1 27 26.1 24.1 23.7 v 51 31.5 22.8 21.6 Cr 52 58.1 53.6 51.6 Mn 55 77.7 74.1 69.2 Fe 56 71.2 65.1 58.9 Fe 57 70.2 63.8 61.8 Co 59 70.2 63.8 60.7 Ni 60 61.9 57.3 55.1 Zn 64 24.6 22.7 20.8 Cu 65 67.9 65.7 62.7 As 75 3.7 2.7 2.5 Se 77 7.1 6.0 5.4 Se 78 7.1 6.2 5.5 Mo 98 49.1 33.5 31.0 Cd 114 72.0 66.3 61.9 Sb 121 25.2 20.9 19.7 Pb 208 100 100 100 2 Power/kW 5 Power/kW 1.2 25.9 87.2 95.3 86.2 89.9 72.1 61.2 28.7 57.2 11.9 7.0 7.3 81.5 67.4 101 178 100 1.3 20.7 66.0 75.5 79.3 67.0 67.3 51.8 42.4 18.3 40.4 6.8 3.9 4.2 59.0 47.4 141 100 1.4 18.9 63.3 70.5 73.1 61.3 63.0 49.4 40.9 16.8 39.1 6.3 3.6 3.9 55.4 44.7 139 100 1.2 31.8 99.9 109 111 92.5 95.3 76.8 65.9 28.9 61.3 11.4 6.1 6.1 71.3 59.0 173 100 1.3 23.4 77.2 85.3 88.5 75.1 77.5 59.1 50.3 22.1 47.5 8.6 4.7 5.0 62.4 53.4 155 100 1.4 17.1 62.0 66.9 70.3 61.1 74.3 51.2 44.2 21.0 43.3 9.2 5.0 5.3 65.2 55.7 144 100 8 10 Power/kW Power/k W 1.2 1.3 1.4 1.2 1.3 1.4 19.9 62.7 70.8 99.9 54.0 60.2 45.2 38.0 15.3 41.6 5.3 3.2 3.2 45.8 38.5 122 100 13.8 43.7 48.9 50.4 40.0 40.5 29.2 24.1 9.3 23.3 3.2 2.0 2.0 90.6 32.4 28.1 100 9.8 14.9 16.1 12.8 32.3 51.5 57.9 47.0 34.3 57.2 62.3 49.3 36.7 60.0 64.8 53.7 32.2 46.6 56.1 49.2 33.0 51.6 56.7 50.1 25.6 39.6 44.3 52.9 22.3 34.0 38.6 38.8 11.5 15.1 17.7 22.8 21.9 32.8 36.6 38.4 4.3 5.6 7.1 10.8 2.1 3.2 3.9 6.1 2.5 3.5 4.1 6.4 89.1 111 122 119 39.5 46.4 51.0 63.6 36.0 40.5 45.0 58.7 100 100 100 100 1 .oo 0.80 0.60 0.40 0.20 0 Y -K 1 1 1 1 l l l 1 .oo (b) - 0.80 A > .- 0.40 0.20 1 I I ,;I 0 20 40 60 80 100 120 140 160 180 200 220 m/z Fig.2 Change in count rates observed in the presence of A 0.01; and B 0.1 mol dm-3 Na in an argon plasma. A relative signal of 1 indicates no effect from Na. Power (a) 1.2; (b) 1.3; and (c) 1.4 kW Non-spectroscopic interference from Na in an Ar-N plasma A reduction of suppression is clearly shown in Figs. 3-5 compared with that observed with an Ar plasma upon addition of the N under all sets of conditions. Furthermore at all powers an increase in the percentage of N2 further reduced the signal suppression from either 0.01 or 0.1 mol dm-3 Na. The effect of 0.01 mol dm-3 Na was essentially eliminated across the mass range at 1.3 kW with 10% N,. Another difference between these figures and Fig. 2 is that the shape of the signal curve as a function of m/z is much flatter which indicates that the effect of Na is more uniform across the mass range in a mixed-gas plasma than in Ar.The most striking case was with 10% N2 at 1.4 kW where the same degree of reduced suppression was observed on all analytes. The reason for this mass independence of the suppression in a mixed-gas plasma is unknown but it was also observed in the presence of 0.01 mol dm-3 Na only under certain sets of conditions (an example of which is shown in Fig. 6) during a separate unpublished study performed at fixed nebulization efficiency. Interestingly no mass-independent suppression could be obtained in the presence of 0.1 mol dm-3 Na with the addition of N at any power when the nebulization efficiency was kept constant.Nonetheless the uniform suppres- sion observed at the optimum sampling position implies that only one internal standard should be sufficient to compensate for the effect for the whole mass range which is a definite advantage. Choice of Optimum Operating Conditions With a Mixed-gas Plasma The conditions which gave the best overall signal or S/N were most efficient at reducing the suppressive effect of Na or at decreasing mass discrimination are summarized in Table 8. These conditions are compared with the best ones that were found at a fixed nebulization efficiency. In either case the selection of the conditions was made by comparison with the results obtained in an Ar plasma at 1.2 kW. A range of powers and/or percentages of N indicated that the best conditions depended on the analyte.Nonetheless the results are remark- ably similar despite the completely different approaches used for the two studies which were carried out several months apart. Adding N to the outer gas clearly improved the analytical characteristics of ICP-MS. However the best fea- tures were not all observed using one set of conditions. A small sacrifice in sensitivity and/or detection limits would be acceptable if it meant a relative freedom from effects from concomitant elements.JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY APRIL 1994 VOL. 9 1.00 515 ( a ) - 1.20 ( a ) I 1.00 t I t u - B K t K 1 1 1 1 1 1 1 1 1 20 40 60 80 100 120 140 160 180 200 220 20 40 60 80 100 120 140 160 180 200 220 Fig. 3 Change in count rates observed in the presence of A 0.01; and B 0.1 mol dmP3 Na in a mixed-gas plasma containing (a) 2; (b 5; (c) 8; and ( d ) 10% nitrogen at 1.2 kW. A relative signal of 1 indicates no effect from Na 0.20 "::r"- 0 20 40 60 80 100 120 140 160 180 200 220 L I 1 1 1 1 1 # I l l 20 40' 60 80 100 120 140 160 180 200 220 Fig.4 Change in count rates observed in the presence of A 0.01; and B 0.1 rnol dm-3 Na in a mixed-gas plasma containing (a) 2; (b) 5; (c) 8; and ( d ) 10% nitrogen at 1.3 kW. A relative signal of 1 indicates no effect from Na This is illustrated in Table 9 where the absolute sensitivities (ie. no correction was made for changes in nebulization efficiency) detection limits S/B and degrees of suppression by 0.01 and 0.1 mol dm-3 Na are compared observed in a conven- tional Ar plasma (at 1.2 kW) and mixed-gas conditions that eliminated the effect of 0.01 mol dm-3 Na.The latter conditions resulted in sensitivities within a factor of two of those obtained in a conventional plasma an improvement being observed for V As Mo and Sb. Lower detection limits only resulted for V Cr and Fe (by almost an order of magnitude in the case of s6Fe) although those of several elements (Al Zn Cu Se Sb and Pb) were within a factor of two of the corresponding ones in an Ar plasma. Furthermore the average degree of suppres- sion over the mass range caused by 0.01 rnoldmw3 Na was only 4.9 +_ 3.4% (average f standard deviation) as opposed to 36f10% in the Ar plasma. Finally the suppression by516 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY APRIL 1994 VOL.9 0.40 - 0.20 0) v) .- .- $ 0 4 1.20 4- CT 1 .oo 0.80 0.60 0.40 0.20 1 0' ' I I I I I I I I ' 20 40 60 80 100 120 140 160 180 200 220 l 1 1 1 1 1 1 1 1 1 1 ( d ) A I.--- -= t t 20 40 60 80 100 120 140 160 180 200 220 m/z Fig. 5 Change in count rates observed in the presence of A 0.01; and B 0.1 mol dm-3 Na in a mixed-gas plasma containing (a) 2; (b) 5; (c) 8; and ( d ) 10% nitrogen at 1.4 kW. A relative signal of 1 indicates no effect from Na 0.6 0.4 0.2 m o .g 1.6 1.4 1.2 1 .o 0.8 0.6 0.4 C 0) Q) .- - Q) I I I I I I I I I rl\ A 1.2 c 1.0 - - 0.6 0.4 0.2 1.2 1 o.2 0 t---J 20 40 60 80 100 120 140 160 180 200 220 0.2 1 01 ' I I I I I ' I 1 20 40 60 80 100 120 140 160 180 200 220 Fig. 6 Change in count rates observed in the presence of A 0.01; and B 0.1 mol dm-3 Na in a mixed-gas plasma containing (a) 2; (b) 5; (c) 8; and ( d ) 10% nitrogen at 1.3 kW with a fixed aerosol carrier gas flow rate of 0.84 dm3 min-'. A relative signal of 1 indicates no effect from Na 0.1 mol dmP3 Na was reduced to 46+6% instead of 87+5% in Ar.For comparison the results obtained with the best mixed- gas plasma conditions (in terms of suppression) found at fixed nebulization efficiency are reported in Table 10. They show a sIight over-compensation for the effect of 0.01 mol dm-3 Na with an average degree of suppression of -3.5 f3.1% over the mass range. Changes in sensitivities and detection limits were different on an analyte to analyte basis but nonetheless similar to those reported in Table9 although there was no significant improvement in detection limit such as that observed for Fe and V.Finally although these conditions reduced the effect of 0.1 mol dmP3 Na to a similar extent (with an average degree of suppression of 44k l6%) the extent ofJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY APRIL 1994 VOL. 9 517 Table8 Comparison of the best operating conditions found during this work at optimum IRZ position and an unpublished study carried out at fixed nebulization efficiency Parameter Best sensitivity Best detection limit Lowest mass discrimination Smallest effect from 0.01 mol dm-3 Na Fixed nebulization efficiency 1.3 kW 2% N 1.2-1.3 kW 0-2% N2 1.4 kW 2% N 1.3 kW 5% N Optimum IRZ position 1.4 kW 0-2% N2 1.2-1.3 kW 0% N2 1.2 kW 2 or 5% N 1.3 kW 10% N2 Table9 Comparison of the signal (S) S j N SIB and degree of suppression with 0.01 and 0.1 mol dm-3 Na observed under conventional conditions (1.2 kW O0/o N,) and at 1.3 kW 10% N at an optimum sampling position 0% N,,1.2 kW 10% N 1.3 kW Analytc Sjcounts s-' SIN SIB Dl(%)* 02(%)t S SIN SIB Dl*(%) 02t(%) '-A1 5'v ',Cr "Mn "Fe 57Fe "Ni 64Zn T U 75As 77se 78Se "Mo '"Cd '"Sb TO zO8pb 98000 61900 93200 140500 144600 3860 118300 26900 19700 32200 4880 720 2180 11900 18100 11800 25000 1520 2010 4450 17300 134 51 46700 4050 590 3150 2740 139 454 3734 5670 2950 2540 57 110 580 2530 5 3 3940 549 20 219 712 39 47 1210 2130 914 5 50 25 19 38 49 38 34 41 44 35 43 25 28 28 26 42 40 58 79 79 85 91 87 85 90 91 89 91 85 87 79 84 92 90 93 58800 100700 83400 105200 84700 2060 66900 15100 13000 15700 8400 380 1140 26500 11500 19000 15900 860 5630 5410 4310 1007 147 5210 832 335 2120 1110 65 173 1660 1520 2610 1990 10 676 553 225 27 16 5 50 138 13 79 196 11 26 20 1 259 167 198 8.9 5.6 9.4 7.3 8.2 6.0 5.2 2.5 4.3 1.5 - 2.9 1.8 3.1 3.3 2.6 5.8 10 38 39 45 50 49 48 50 49 50 49 52 44 38 40 52 51 36 * YO Suppression in 0.01 mol dm-3 Na.t YO Suppression in 0.1 mol dm-3 Na. Table 10 Comparison of the signal (S) SIN SIB and degree of suppression with 0.01 and 0.1 mol dm-3 Na observed under conventional conditions (1.2 kW 0% N2) and at 1.3 kW 5% N at a fixed aerosol carrier gas flow rate of 0.84 dm3 min-' ~ ~~ 0% N2 Analyte 2 7 ~ 1 51v 52Cr 55Mn "Fe 57Fe ("Ni "Zn "5cu "As 7hSe 7uSe '8Mo ' "Cd 12'Sb '08Pb 5 9 c 0 Sjcounts s I5 1000 143000 129000 17 1000 158000 3780 149000 30100 16700 30200 7710 744 2400 33800 23500 22100 49700 1 SIN 5370 33200 3 240 19600 43 78 36100 4590 519 2990 2360 132 63 10700 9120 8590 3310 SIB 208 11900 325 2330 2 15500 340 43 188 443 25 7 5060 1950 3230 174 3 Dl(%)* 1 .o 0.4 13.6 27.7 25.6 19.2 19.9 24.1 20.1 28.9 8.9 4.4 14.2 15.0 32.7 35.8 61.2 0 2 ( Yo ) t 78.2 74.4 79.6 85.4 90.2 87.5 85.1 87.3 88.4 89.0 82.2 83.7 85.7 80.8 90.6 89.5 94.4 S 65000 110000 92200 114OOO 96900 2290 79100 18000 18800 11800 19600 895 2720 26000 14700 25800 24900 SIN 2440 4280 2520 3710 40 20 4190 2070 789 4150 1820 77 103 2170 2740 6750 3220 SIB 82 975 166 866 1 1 1840 524 36 344 369 19 5 497 1170 868 47 2 D 1 *(Yo) - 1.7 -2.5 - 0.7 -0.4 - 3.4 - 2.9 - 1.9 -3.1 -8.1 0 - 7.6 - 1.4 - 9.9 - 0.7 -7.1 -6.5 - 1.9 D2t( Yo) 6.0 30.0 33.6 36.2 37.2 37.8 38.1 38.7 37.1 41.1 48.3 52.8 53.3 58.0 62.9 62.4 74.5 * YO Suppression in 0.01 mol dm-3 Na.t YO Suppression in 0.1 mol dm-3 Na. the reduction was not as uniform over the mass range (this is also evident from a comparison of Fig. 6 with Fig. 4). Conclusion Argon-nitrogen mixed-gas plasmas lead to only a small sacri- fice in sensitivity and detection limits for some elements compared with a conventional Ar plasma which however suffers from extensive suppression by Na matrices. These mixed-gas plasmas not only dramatically reduce or eliminate the suppression but by sampling the plasma at an optimum IRZ position the residual suppression is more uniform across the mass range and should therefore be easily compensated for by using a single internal standard. This is in contrast to an Ar plasma where the effect of concomitant elements is typically fairly important at the optimum sampling position requiring the use of less than optimum operating conditions (thereby sacrificing sensitivity) in order to reduce the effect.' More fundamental studies are underway to find the reason for these differences.These will include studies with other matrix elements (including heavier ones) and also comparing easily ionized with non-easily ionized matrix elements.518 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY APRIL 1994 VOL. 9 D.B. gratefully acknowledges the financial support of the Natural Sciences and Engineering Research Council of Canada (grant No. OGP0039487) and of the Advisory Research Committee of Queen’s University. 1 2 3 4 5 6 7 8 9 10 11 12 13 14 References Jarvis K.E. Gray A. L. and Houk R. S. Handbook of Inductively Coupled Plasma Mass Spectrometry Blackie Glasgow 1992. Olesik J. W. Anal. Chem. 1991 63 12A. Houk R. S. Anal. Chem. 1986 58 97A. Douglas D. J. and Kerr L. A. J . Anal. At. Spectrom. 1988,3 749. Olivares J. A. and Houk R. S. Anal. Chem. 1986 58 20. Beauchemin D. McLaren J. W. and Berman S. S. Spectrochim. Acta Part B 1987 42 467. Thompson J. J. and Houk R. S. Appl. Spectrosc. 1987 41 801. Tan S. H. and Horlick G. J . Anal. At. Spectrom. 1987 2 745. Gregoire D. C. Spectrochim. Acta Part B 1987 42 895. Crain J. S. Houk R. S. and Smith F. G. Spectrochim. Acta Part B 1988 43 1355. Vandecasteele C. Nagels M. Vanhoe H. and Dams R. Anal. Chim. Acta 1988 211 91. Kim. Y.-S. Kawaguchi H. Tanaka T. and Mizuike A. Spectrochim. Acta Part B 1990 45 333. Evans E. H. and Giglio J. J. J. Anal. At. Spectrom. 1993 8 1. Marshall J. and Franks J. J. Anal. At. Spectrom. 1991 6 591. 15 16 17 18 19 20 21 22 23 24 25 26 27 Gillson G. R. Douglas D. J. Fulford J. E. Halligan K. W. and Tanner S. D. Anal. Chem. 1988 60 1472. Tanner S. D. Spectrochim. Acta Part B 1992 47 809. Gregoire D. C. Appl. Spectrosc. 1987 41 897. Evans E. H. and Caruso J. A. Spectrochim. Acta Part B 1992 47 1001. Cheatham M. M. Sangrey W. F. and White W. M. Spectrochim. Actu Part B 1993 48 E487. McLaren J. W. Beauchemin D. and Berman S. S. Anal. Chem. 1987 59 610. Craig J. M. and Beauchemin D. J. Anal. At. Spectrom. 1992 Koirtyohann S. R. Jones J. S. Jester C. P. and Yates D. A. Spectrochim. Acta Part B 1981 36 49. Lam J. W. H. and Horlick G. Spectrochim. Acta Part B 1990 45 1313. Lam J. W. H. and McLaren J. W. J. Anal. At. Spectrom. 1990 5 419. Hill S. J. Ford M. J. and Ebdon L. J . Anal. At. Spectrom. 1992 7 719. Louie H. and Soo S. Y.-P. J . Anal. At. Spectrom. 1992 7 557. Montaser A. Fassel V. A. and Zalewski J. Appl. Spectrosc. 1981 35 292. 7 937. Paper 3/04898E Received August 12 1993 Accepted December 20 1993

ISSN:0267-9477    

DOI:10.1039/JA9940900509

出版商:RSC    

年代:1994

数据来源: RSC

摘要:

JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY APRIL 1994 VOL. 9 519 Inductively Coupled Plasma Mass Spectrometric Determination of Low-level Rare Earth Elements in Rocks Using Potassium-based Fluxes for Sample Decomposition Alessandro Rivoldini lstituto di Giacimenti Minerari Facolta di lngegneria Universita di Cagliari Piazza D'Armi 09 123 Cagliari ltaly Sandro Fadda Centro Studi Geominerari e Mineralurgici del Consiglio Nazionale delle Ricerche Piazza D 'Arm; 09 123 Cagliari ltaly The 14 naturally occurring rare earth elements Sc and Y were determined in fusion solutions by inductively coupled plasma mass spectrometry with limits of quantitation similar to an acid digestion procedure. Two potassium-based fluxes (K,C03 and K,B,07) were studied as fusion media. To maintain the level of total dissolved solids within the range permitted by instrumental criteria K and B were removed from the solution thus avoiding the necessity for considerable dilution.The validity of the procedures has been verified by precision and accuracy tests on selected Standard Reference Materials. Keywords Inductively coupled plasma mass spectrometry; scandium yttrium and rare earth elements; potassium-based flux; total dissolved solids; standard reference materials The lanthanides have become a very important suite of trace elements in geological samples. The precise and accurate determination of this elemental group in a wide range of rock types is an essential requirement for many geochemical and petrogenetic investigations. Inductively coupled plasma mass spectrometry (ICP-MS) has detection limits for the rare earth elements (REE) which are considerably better than other instrumental methods used in geoanalysis (X-ray fluorescence ICP atomic emission spec- trometry (ICP-AES) and neutron activation analysis).' For routine geological application the ICP-MS technique most commonly employs solution nebulization.In a solution-based analysis the sample dissolution procedure poses important limitations on the range of analytes that can be accurately quantified; incomplete digestion of some materials resistant to a conventional mixed-acid attack will preclude complete cover- age of some analytes preventing their accurate determination. This potential for partial dissolution is a serious problem in REE analysis because a large fraction of these elements often reside in accessory refractory mineral^,^*^*^*^ which could be very difficult to decompose totally depending upon their relative abundance (Table 1).The strongest acid treatments could be repeated or carried out in pressure digestion bombs heated in a conventional oven or in closed poly(tetrafluoro- ethylene) (PTFE) vessels for microwave-heated procedures. Even then for some unattacked material particularly that which is acid resistant a significant fraction of the analytes could still remain trapped. Table 1 Some REE and REE-bearing acid insoluble minerals Name Xenotime Florencite Britholite Fergusonite Formanite Microlite Pyrochlore Euxenite Pol ycrase Samarskite Aeschynite Zircon Formula YPO (La,Ce,Nd)Al,( PO,),(OH 16 (Y,Ce,Ca) (S~O,,PO,) (OH,F) (Y,Ce,La,Nd) (Nb,Ti)O Y,Ta04 (Na,K,Ca,REE,U),Ta,O (O,OH,F) (Na,K,Ca,REE,U),Nb,O,(O,OH,F) (Y,Ca,Ce,U,Th) (Nb,Ta,Ti),O (Y ,Ca,Ce,U,Th) (Ti,Nb,Ta),O (Y,Er,Ce,U,Fe) (Nb,Ta,Ti),O 16 (Y,Ce,Nd,Ca,Fe,Th) (Ti,Nb),(O,OH) ZrSiO An alternative decomposition procedure for geological samples is fusion with an alkaline flux which is an effective method of decomposing most rocks and refractories. However in view of the high level of dissolved solids originating from the flux prior to ICP-MS measurements it is necessary to increase the dilution factor of the sample solutions so resulting in a loss of analytical sensitivity by one order of magnitude with respect to the acid digest.6 Some trace elements could then be below quantitation limits.The level of total dissolved solids (TDS) in the nebulized solution is critical in ICP-MS analysis.Sample aerosols containing > 2000 pg ml-I TDS impair the efficiency of the ion-sampling system by reducing the orifice of the sampling cones causing both memory effects and significant drift in the analytical signal. The amount of TDS is controlled by the mass of sample taken for analysis and by the method of sample preparation. When using an acid digest final solutions for ICP-MS analysis are made up in dilute HNO,. This procedure has the advantage that salts are not added to the analytical solution unlike the fusion method. A sample mass of 0.2-0.4g can be tolerated and is consistent with acceptable detection limits in the analysis of geological samples. Previous work has involved making appropriate changes to the fusion procedures to achieve better detection limits in the solid samples [limits of quantitation (LOQs)].The removal of the large amount of salt added with the commonly used Li- or Na-containing fluxes is difficult and time consuming. Westeland and Kantipuly7 studied the efficiency with which Occurrence Acidic and alkaline igneous rocks pegmatites gneisses and schists Carbonatites pegmatites weathering crust deposits Granitoids pegmatites alkaline rocks carbonatites Granitic and alkaline pegmatites Pegmatites carbonatites nepheline-syenite alkaline rocks Granitoids pegmatites Alkaline and granitic pegmatites Alkaline rocks granites pegmatites Granitoids alkaline rocks volcanites520 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY APRIL 1994 VOL.9 K derived from the flux could be precipitated with HClO in a fusion procedure. They determined four selected cations (Ca2+ Eu3+ Ag' and Th4+) at trace levels in a programme of rock analysis for alkaline earth rare earth and other elements. Kantipuly et a[.* developed a procedure for the ICP-MS determination of Th and U in tourmaline using a sample fusion procedure. They found K2B407 to be an effective flux as after dissolution K could be easily removed by precipitation as KClO and B as the volatile methylborate. The aim of the present work was to determine low levels of REE in rock samples containing accessory refractory REE- bearing phases. A fusion procedure has been developed which takes advantage of the ease with which the flux cation can be removed by precipitation.Two K-based compounds were considered K2C03 and KzB407 each employed individually. A set of international silicate reference materials representing a range of rock types (Table 2) were analysed for REE to verify the precision and accuracy of the procedure. Experimental Chemicals and Reagents The fluxing agents used were K2C03 and KzB407 (Johnson Matthey analytical-reagent grade products). AnalaR grade BDH acids HClO HN03 HF and high- purity distilled de-ionized water were used throughout the work. Ethanol was from Carlo-Erba of analytical-reagent grade. Two synthetic multi-element REE calibration standards 10 and 200 pg l-l were prepared by dilution of Johnson Matthey 100 pg ml-1 multi-element solutions to cover the expected range of concentrations.All standards samples and blank solutions were made up to suitable volume maintaining an overall acidity of 2% HN03 and stored in polyethylene bottles. Solutions of Rh and Re (Johnson Matthey 100 yg ml-l) were added as internal standards at a concentration level of The standard solutions required for performance tests and instrumental optimization were from SPEX Industries. Five additional high-purity single-element solutions of Ba La Ce Pr and Nd required for spectral oxide interferences studies were prepared from Johnson Matthey standard solutions. 100 yg 1-I. Instrument a tion The ICP mass spectrometer used was a Perkin-Elmer SCIEX Elan Model 5000 equipped with a standard cross-flow nebulizer. The ICP-AES measurements of residual B levels after its Table 2 Silicate reference materials selected for accuracy and precision tests removal were performed by selecting the optimized instrumen- tal conditions for this element as described by Rivoldini and Cara.g The residual amount of K in the fusion solutions was measured by atomic absorption spectrometry (AAS).ICP-MS Operating Conditions The instrument was optimized following standard performance tests used for routine multi-element analysis across the whole mass range. The maximum Io3Rh response and uniform count rates for 24Mg and ,07Pb were obtained sensitivi- ties were respectively > 35000> 10000> 9000 counts s-' for the 10 pg 1-1 concentrations. Background count rates of <20 counts s-' were observed for blank solutions at m/z values where the electronic noise and photons are the only signals.Polyatomic interferences from the plasma and air gases (Ar O N and CO) or generated by the matrix acid were absent in the m/z range under study," the only exception being an interference on 45Sc presumed to be 12C16021H. A count rate of 1200counts s-l was observed for blank solutions at m/z 45 while for the 10 pg 1-1 solution of Sc the count rate was 29 000 counts s-l. The level of oxides and doubly charged ions was kept to a minimum (<2%) by monitoring CeO+ and Ba2+ signals in the test solution over an 8 h period of running the samples. With the system suitably optimized REE-O+ ions were found to be in the range 0.1-1.2%. However the severity of spectral interferences is dependent on relative elemental compositions and corrections could be necessary when light REE enriched rock types or extreme distributions are analysed.Correction equations were included in the software to take account of 137Ba and 142Ce oxide interferences on ions at m/z values of above namely 153E~ and 158Gd. The same type of correction can be programmed as a precautionary measure for 141Pr 142Nd 143Nd and lsoNd oxide interferences on respectively 157Gd 158Gd 15'Tb and 166Er because Nd and Pr can reach a relatively high concen- tration in common rocks. Instrumental operating conditions are shown in Table 3. It is appropriate to use such software corrections only when the respective interference is significant as unnecessary application of the correction will increase accompanying uncertainties. The ions chosen for the determinations are shown in Table 4.For isobaric overlaps the ELAN 5000 software provides default corrections automatically. Linear working curves obtained from two-point (10 and 200 pg 1-l) multi-element external standards were used for quantitation. Blank and standard solutions were prepared by following all the steps in the sample preparation procedure. Two separate internal standards Rh (for Sc Y La Ce Pr and Nd) and Re (for Sm Eu Gd Tb Dy Ho Er Tm Yb and Lu) were selected." Sample GA UB-N FK-N NIM-D NIM-L NIM-N NIM-S GSD-8 GSR-1 Source CRPG (Centre de Recherche Petrographique et Geologique France) ANRT (Association Nationale de la Recherche Technique France) NIM (National Institute of Metallurgy South Africa) IGGE (Institute of Geophysical and Geochemical Exploration People's Republic of China) Rock type Granite Table 3 Instrumental parameters for ICP-MS determination of Sc Y and REE Serpentine Potash feldspar Dunite Lujavrite Norite Syenite Stream sediment Granite Parameter Plasma r.f.power Outer gas flow Intermediate gas flow Aerosol carrier gas flow Sample uptake flow Sample delay Washing time Replicate time Dwell time Number of replicates Points across peak Value 1.050 kW 15 1 min-' 0.9 1 min-l 0.84 1 min-I 1.0 ml min-' 60 s 30 s 300 ms 100 ms 10 1JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY APRIL 1994 VOL. 9 521 Table 4 Isotopes chosen for the ICP-MS determinations and spectral oxide interferences Procedure After homogenization the finely powdered samples (200-300 mesh) were dried in an oven overnight at 110°C before weighing.Dehydrated fluxes were stored in a desiccator. Sub- samples of 0.2 g were thoroughly mixed with 2 g of dehydrated flux in platinum crucibles. Fusion with K,CO The crucibles were covered with a lid and gently heated for 15min over a low flame to allow CO to escape without spattering. The temperature was then gradually increased and fusions were made at 900-1000"C igniting over the full flame of a Bunsen burner for an additional 15 min. The crucibles were allowed to cool and then transferred with lids into 250ml glass beakers covered with sufficient water and the fusion cakes dissolved carefully on a warm hot-plate by reaction with 8 ml of HClO,. Once the perchlorate precipitation of K was complete,12 the crucibles and lids were rinsed and removed; solutions were then evaporated to white fumes.After cooling 40 ml of C2H50H-2% HClO were added and the dense and well formed precipitate of KC104 was separated by centrifu- gation and washed with the ethanol-2% HClO mixture. The centrifuging and decanting steps were repeated twice. The supernatant was transferred into glass beakers and evaporated to dryness on a hot-plate (150°C). Then 2 ml of HN03 were added the walls of the container rinsed and the contents heated to dryness again. The resulting residues were finally dissolved in 2 ml of HN03 plus water and made up to 100 ml with de-ionized water after adding 100 pg 1-1 of Rh and Re as internal standards. Fusion with KzB40-1 The crucibles containing sample and flux were placed in a muffle furnace at 1200 "C for 30 min.The crystal-clear melt was poured into 80 ml of water 5 ml of HC104 were added and mixed continuously on a magnetic stirrer until all the cake had dissolved. The K was precipitated by a similar procedure to that for carbonate fusion. After centrifugation the solution was transferred into a Poly( tetrafluoroethylene) (PTFE) beaker B and Si expelled by evaporating to dryness with 5 ml of HF. The sample treatment then followed the carbonate procedure. Results Removal of K and B The stoichiometric amount of K added to the analysis with 2 g of KzC03 is 1.13 g in a final volume of 100 ml. After precipitation the amount of K remaining in the carbonate solution was measured in 20 samples and was found to be in the range 5-15 pgml-' a level which did not interfere with the determination of the REE.Similar satisfactory results were obtained for removal of K from KzB407 fusion solutions the HF treatment of which resulted in complete loss of B 1-5 pgml-I compared with 0.28g in 100ml before volatilization. Detection and Quantitation Limits Detection limits (DL) were calculated as three times the standard deviation of the blank measured for 11 replicate determinations divided by the sensitivity. The L0Qsl3 were obtained as ten times the standard deviation of the blank divided by the sensitivity and multiplied by dilution factor (500). For each element the agreement between the LOQs obtained by the two fusion techniques is excellent; their mean values are given in Table 5 along with detection limits and chondrite levels.',.Precision and Accuracy The precision of the analysis was evaluated by decomposing 11 replicate samples of the reference material GSR-1 using both fusion procedures. For each set of solutions analysed the average (x) standard deviation (s) and relative standard devi- ation (RSD) were calculated. To test instrumental fluctuations one solution from both of the fusion procedures was run 11 times. The precision of the sample decomposition procedure was then calculated according to the following formula:15 where ssp =precision of the sample preparation; sTot = precision of the analysis; and sins =instrumental precision. The results obtained are listed in Table 6. Analytical accuracy was assessed by comparison of the results obtained (Table 7) with the compilation of Govin- daraju16 for nine reference materials chosen to match the range of REE concentrations under study.Some of these standards (NIM-D NIM-L and UB-N) have been reported to leave residues after repeated acid attack.17 When the complete pattern of REEs is not certified16 or the characterization is inadequate the comparison is made with determinations reported from other schemes of analysis using different tech- niques. In the case of NIM-D only a very small data set is as yet available for comparison and the values reported remain poorly characterized. Table 5 ICP-MS detection limits (DL) for Sc Y and REE with limits of quantitation (LOQ) and chondrite values Analyte DL/pg 1-' sc 0.057 Y 0.004 La 0.025 Ce 0.019 Pr 0.008 Nd 0.015 Sm 0.006 Eu 0.010 Gd 0.012 Tb 0.006 DY 0.013 Ho 0.004 Er 0.012 Tm 0.007 Yb 0.007 Lu 0.012 LOQ/pg 8-l 0.095 0.006 0.042 0.032 0.013 0.025 0.010 0.017 0.020 0.010 0.022 0.006 0.020 0.012 0.012 0.020 Chondritelpg 8-l - 0.2446 0.6379 0.09637 0.4738 0.1540 0.05802 0.2043 0.03745 0.2541 0.05670 0.1660 0.02561 0.1651 0.02539522 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY APRIL 1994 VOL.9 Table 6 ICP-MS results of precision tests for SRM GSR-1. X = average; sTot = standard deviation of the analysis; ssp = standard deviation of sample preparation procedure; and sins = standard deviation of instrumental measurements K2C03 fusion Analyte sc Y La Ce Pr Nd Sm Eu Gd Tb DY Ho Er Tm Yb Lu XIPS g-' 6.50 64.78 59.12 109.23 13.25 43.75 8.70 0.69 9.48 1.33 9.30 1.76 6.01 0.73 6.73 0.92 sTot/pg g-' RSD (%) 0.21 3.2 1.39 2.1 1.83 3.1 1.53 1.4 0.18 1.4 0.58 1.3 0.16 1.8 0.02 2.6 0.29 3.1 0.03 2.3 0.14 1.5 0.04 2.0 0.08 1.3 0.02 2.3 0.13 1.9 0.02 2.3 SspIPg g - 0.196 1.19 1.76 1.11 0.14 0.49 0.15 0.015 0.27 0.026 0.10 0.033 0.057 0.015 0.109 0.019 SinJpg g-' 0.076 0.71 0.50 1.05 0.10 0.30 0.04 0.013 0.10 0.014 0.09 0.01 0.05 0.013 0.07 0.007 Z/PLg g-' 5.93 61.86 55.16 103.90 12.38 41.02 8.50 0.67 9.05 1.30 8.84 1.71 5.75 0.68 6.34 0.87 sTot/pg 8-l 0.17 2.7 1 0.96 3.49 0.22 0.52 0.37 0.03 0.23 0.11 0.28 0.09 0.20 0.03 0.3 1 0.05 RSD (Yo) 2.9 4.4 1.7 3.4 1.8 1.3 4.4 4.5 2.5 8.5 3.2 5.3 3.5 4.4 4.9 5.7 %p/% g- 0.15 2.69 0.9 3.37 0.21 0.46 0.36 0.25 0.22 0.10 0.27 0.089 0.18 0.027 0.30 0.047 Sins/pg g - l 0.08 0.27 0.32 0.90 0.05 0.25 0.08 0.016 0.07 0.04 0.06 0.015 0.063 0.013 0.06 0.015 Table 7 Values (in pg g-') obtained by ICP-MS for Sc Y and the REE in reference materials NIM-D NIM-S NIM-N GA NIM-L FK-N UB-N GSD-8 and GSR-1 NIM-D NIM-S Analyte sc Y La Ce Pr Nd Sm Eu Gd Tb DY Ho Er Tm Yb Lu KzB407 5.81 0.32 0.94 0.53 0.088 0.22 0.032 0.066 0.062 0.024 0.088 0.03 1 0.035 0.01 1 0.15 0.019 K2C03 6.12 0.74 0.78 0.71 0.08 0.25 0.05 1 0.020 0.05 0.019 0.096 0.045 0.08 1 0.019 0.24 0.03 This work Ref.16 Ref. 17 - 7 0.2 0.87 - - - - - - - 0.928 - 0.116 - 0.02 - - - - - 0.198 - 0.067 - - - - - 0.339 - - This work K2B407 4.42 1.50 6.15 12.99 1.89 7.13 1.24 0.4 1 1.32 0.17 0.42 0.053 0.12 0.017 0.087 0.0 19 K2C03 4.33 1.68 5.92 12.76 1 .'80 6.82 1.18 0.37 1.04 0.10 0.39 0.056 0.14 0.017 0.10 0.027 Ref. 16 4 20 5 11.9 6 1 0.3 0.1 0.4 - - - - - 0.07 - Ref.18 - 1.14 4.47 10.07 1.39 6.16 1.14 0.27 0.83 0.33 0.05 0.12 0.08 0.01 - - Ref. 19 - 0.89 4.30 8.54 1.18 4.91 1.09 0.268 0.678 0.075 0.260 0.036 0.072 0.010 0.061 0.009 Ref.20 3.6 1.1 4.90 1.19 6.3 1.18 0.24 0.72 0.30 11 - - - - 0.06 - Ref.2 1 - - 5.39 9.89 6.34 1.07 0.55 0.41 0.29 0.18 - - - - - - Ref.6 - - 1.42 1.04 5.45 5.52 1.74 1.56 7.38 6.69 1.35 1.13 0.47 0.26 1.31 0.87 0.11 - 0.36 - 0.06 - 0.12 - 0.02 - 0.13 - 0.03 - 13.9 12.7 Ref.22 - 1.9 5.3 1.2 5.8 1.3 0.3 0.55 0.08 0.4 0.08 0.11 0.07 0.1 0.015 11 Analyte sc Y La Ce Pr Nd Sm Eu Gd Tb DY Ho Er Tm Yb Lu NIM-N GA This work This work K2B407 K2C0 Ref.16 Ref.18 Ref.20 Ref.21 Ref.6 Ref.23 Ref.24 K2B407 K2C03 Ref.16 Ref.18 - 37.4 - - - - - - 7.59 8.26 7 - 6.92 6.91 7 5.67 5.5 - 5.5 5.29 - - 5.25 21.34 20.18 21 17.3 4.28 3.48 3 3.09 2.8 3.14 2.65 2.80 2.67 3.09 2.85 43.05 42.34 40 42.1 6.84 6.84 6 6.32 5.5 6.24 5.40 5.06 5.69 6.32 6.31 76.07 78.19 76 82.8 0.93 0.9 - 0.77 0.67 - 0.68 0.67 0.75 0.77 0.66 8.92 9.23 7.3 7.63 3.52 3.51 3 3.32 3.9 3.4 3.07 2.80 3.22 3.32 2.72 28.46 29.18 27 27.1 0.82 0.74 0.8 0.78 0.82 0.91 0.67 0.82 0.85 0.78 0.742 4.47 4.71 5 4.81 0.70 0.62 - 0.57 0.58 0.67 0.55 0.47 0.60 0.57 0.523 1.29 1.33 1.08 1.15 1.12 0.97 - 0.91 0.96 1.07 0.92 0.65 0.93 0.91 0.794 4.17 4.08 3.8 3.83 1.03 1.02 - 1.00 1.06 1.14 0.98 0.99 1.10 1.0 0.967 3.35 3.27 3.3 3.15 0.198 0.70 0.69 - 0.63 0.23 0.23 - 0.63 0.66 - 0.67 0.67 0.73 0.63 0.64 0.64 0.67 0.617 1.93 1.88 1.9 1.87 0.096 0.33 0.31 - 0.10 0.10 - 0.61 0.63 0.7 0.65 0.66 0.71 0.61 0.53 0.68 0.65 0.603 1.94 1.85 2 1.86 0.12 0.11 0.2 0.11 0.1 - 0.10 0.05 - - 0.097 0.37 0.33 0.3 0.30 42.74 40.53 38 0.15 0.09 0.15 - 0.144 0.70 0.68 0.6 - - - - 0.21 0.17 - 0.22 0.22 - 0.21 0.19 - - - - - - 0.09 0.03 - -JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY APRIL 1994 VOL.9 523 Table 7 (continued) Analyte sc Y La Ce Pr Nd Sm Eu Gd Tb DY Ho Er Tm Yb Lu NIM-L FK-N This work This work K2B407 K2C03 Ref.16 Ref.18 Ref.20 Ref.21 Ref.6 1.00 22.75 225.58 244.68 23.82 52.64 4.28 1.44 6.0 0.7 1 3.15 0.70 2.3 1 0.4 1 2.64 0.51 0.75 22.39 248.62 235.23 23.89 52.59 3.96 1.17 5.5 0.64 2.89 0.68 2.29 0.39 2.38 0.51 0.3 22 250 240 48 5 1.2 0.7 - - - - - - 3 0.4 - 17 201 262 16.4 43.1 3.53 1.05 3.59 3.10 0.91 2.58 2.45 0.38 - - 0.5 18.8 215 270 19 45.5 3.8 0.95 2.35 2.75 0.55 1.55 2.60 0.70 - - - - 236 309 - 57.7 5.27 0.91 5.43 4.09 3.17 3.53 - - - - - 19.7 238 312 19.9 49.7 3.96 1.06 4.86 0.49 2.80 0.63 1.99 0.37 2.69 0.43 - 20.7 234 300 21.7 52.5 4.41 1.21 5.17 0.58 2.95 0.74 2.07 0.45 3.05 0.66 K2B407 0.08 0.40 1.12 0.98 0.093 0.36 0.06 0.4 1 0.06 0.008 0.053 0.015 0.039 0.007 0.05 0.010 K2C03 0.10 0.58 1.44 1.19 0.11 0.32 0.057 0.46 0.075 0.007 0.075 0.017 0.048 0.009 0.063 0.0013 Ref.16 0.05 0.3 1 1 0.3 0.06 0.42 0.05 0.01 0.06 0.04 0.04 0.01 - - - Ref. 18 - - 0.90 1.04 0.18 0.40 0.07 0.06 0.02 0.05 0.05 0.01 - - - - Ref.19 - 0.31 1 0.691 0.072 0.231 0.058 0.302 0.047 0.009 0.046 0.01 1 0.032 0.007 0.034 0.006 UB-N GSD-8 GSR- 1 This work Ref.25 Analyte sc Y La Ce Pr Nd Sm Eu Gd Tb DY Ho Er Tm Yb Lu K2B407 13.70 2.40 0.87 1.20 0.25 0.68 0.20 0.085 0.29 0.06 0.42 0.090 0.23 0.05 0.31 0.04 K2C03 14.58 2.86 1.68 1.19 0.19 0.77 0.20 0.096 0.32 0.06 0.41 0.097 0.28 0.05 0.28 0.05 Ref. 16 Ref.18 13 - 2.5 1.97 0.5 0.53 1 1.21 - 0.16 0.6 0.66 0.2 0.21 0.08 0.08 0.3 0.30 0.38 0.39 0.09 0.28 0.28 0.25 0.29 0.04 0.05 0.06 - - - - Ref. 19 - 2.19 0.75 1.14 0.135 0.644 0.219 0.08 0.283 0.06 0.377 0.086 0.255 0.042 0.249 0.038 10.5 10.2 2.27 2.30 0.16 0.42 0.56 0.93 0.10 0.12 0.50 0.54 0.25 0.15 0.09 0.10 0.30 0.27 0.06 0.06 0.044 0.34 0.09 0.1 0.26 0.21 0.023 0.04 0.36 0.22 0.05 0.06 This work This work K2B407 5.83 18.13 28.65 53.27 5.96 19.27 3.18 0.61 3.54 0.57 2.79 0.61 1.80 0.31 1.89 0.37 K2C03 6.03 17.48 28.29 54.64 6.04 19.57 3.24 0.60 3.62 0.54 2.78 0.61 1.80 0.29 1.92 0.36 Ref.16 5.7 18 30 54 21 5.7 3.8 0.56 3.5 0.54 2.6 0.96 1.8 0.36 2.1 0.36 K2B407 5.93 61.86 55.16 12.38 44.02 8.50 0.67 9.05 1.30 8.84 1.71 5.75 0.68 6.34 0.87 103.9 K2CO3 6.50 64.78 56.12 109.23 13.25 43.75 8.70 0.69 9.48 1.33 9.30 1.76 6.01 0.73 6.73 0.92 Ref. 16 6.1 62 54 108 12.7 47 9.7 0.85 9.3 1.65 10.2 2.05 6.5 1.06 7.4 1.15 Discussion and Conclusions Using the procedure described in this paper for both K2C03 and K,B407 the extra dilution factor (5000) that could otherwise be required for samples prepared by fusion was avoided as the cation deriving from the flux was quantitatively separated while completely recovering the analytes. Clear acid solutions were analysed by ICP-MS at a final dilution factor of 500 which is equivalent to an acid digestion procedure in which 0.2 g of sample is taken up into 100 ml of solution with a <2O00 pg ml-' nominal level of TDS.Determinations from the pgg-' level down to ngg-' in the solid were possible without separation or preconcentration producing complete data for all the REEs with a limit of detection of < 30 ng g-' in the solid i.e. below the chondritic values. The precision of the analysis was adequate for most geo- chemical and petrological applications. The carbonate fusion was found to be more precise (1.3-3.2% RSD) than fusion with tetraborate (1.3-5.7% RSD) as shown in Table 6. Precision of the sample preparation procedure was comparable to that of instrumental measurements. Significant discrepancies between the data obtained using the two fusion procedures do not exist.In comparing analyses of selected reference materials with published values (Table 7) some differences were observed but in general good agreement was obtained. Both the fluxes investigated are commensurate with the analytical aim of the work. Bearing in mind the desirability to simplify the analytical procedures an effort has been made to speed up the fusion which remains more difficult to streamline than sample preparation by acid digestion. Fusion with carbonate over a Bunsen burner is a laborious operation necessitating supervision. By contrast many cru- cibles can be accommodated in each batch heated in a muffle furnace so that fusion with tetraborate can be undertaken with minimal operator attendance and considerable savings in time.Boron removal is easy and quantitative but involves the additional use of HF. On the other hand the carbonate melt is readily decomposed by HC104 while the corresponding step in the tetraborate procedure is lengthy and requires continuous stirring. On balance fusion with K2B,07 is the most convenient method for large batches of samples. However the choice of fusion procedure also depends on the nature composition and mineralogy of the samples a prior knowledge of which would be helpful. In the analysis of rock samples which are known or suspected to contain appreciable amounts of chemically resistant REE-bearing phases mixtures of two or more fluxing agents can offer some advantage over the use of the single salts alone better decomposing properties lower melting points enhanced oxidizing power and higher reaction activity. In view of the above KNO or KC103 KOH524 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY APRIL 1994 VOL.9 and fluorides e.g. KF and KHF2 might then be more appropriate. In addition to the 14 REEs Sc and Y a set of other trace elements of interest in many petrological and geochemical studies were also investigated. Good preliminary results were obtained for Li Be Cs Co Zr Hf W Th and U while Ga Ge Nb Mo Sn Sb and Ta gave unsatisfactory data. It is concluded therefore that alkaline fusion with K salts offers a great advantage for ICP-MS low-level REE analysis by overcoming uncertainties in the effectiveness of the dissolu- tion step without nullifying such benefits as low limits of quantitation. The ICP-MS facilities were provided by ECOSYSTEMS Laboratorio di Analisi e Ricerca Scientifica Porto Torres (Sassari) Italy.The contribution of the technical support staff and the assistance of Ignazio Cau to operation of the instru- ment are gratefully acknowledged. References 1 2 3 4 5 Riddle C. Vander Voet A. and Doherty W. Geostand. Newsl. 1988 12 203. Dana J. D. and Dana E. S. The System of Mineralogy Wiley New York 1962. Lange N. A. Manuale di Chimica USES Firenze 1970. Fischesser R. Donndes des Principales Espbces Mine'rales SociCte de YIndustrie MinCrale Saint-Etienne 1970. Fleischer M. and Mandarino J. A. Glossary of Mineral Species 1991 The Mineralogical Record Tucson 1991. 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 Jarvis K. E. Chem. Geol. 1990 83 89. Westland A. D. and Kantipuly C. J. Anal. Chim. Acta 1983 154 355. Kantipuly C. J. Longerich H. P. and Strong D. F. Chem. Geol. 1988 69 171. Rivoldini A. and Cara S. Chem. Geol. 1992 98 317. Tan S. H. and Horlick G. Appl. Spectrosc. 1986 40 445. Doherty W. Spectrochim. Acta Part B 1989 44 263. Scott W. W. Standard Methods of Chemical Analysis ed. Furman N. H. Van Nostrand Princeton NJ 5th edn. 1955 vol. 1. Beauchemin D. and McLaren J. ICP In$ Newsl. 1985 11 441. Evensen N. M. Hamilton P. J. and ONions R. K. Geochim. Cosmochim. Acta 1978 42 1199. Meloum M. Militky J. and Forina M. Chemometrics for Analytical Chemistry Ellis Horwood Chichester 1992. Govindaraju K. Geostand. Newsl. 1989 13 1. Smith A. D. Gillis K. M. and Ludden J. N. Chem. Geol. 1990 81 17. Jarvis K. E. and Jarvis I. Geostand. Newsl. 1988 12 1. Jarvis K. E. Chem. Geol. 1988 68 31. Watkins P. J. and Nolan J. Geostand. Newsl. 1990 14 11. Bottazi P. Ottolini L. and Vannucci R. Geostand. Newsl. 1991 15 51. Zachmann D. W. Anal. Chem. 1988 60 420. Le Roex A. P. and Watkins R. T. Chem. Geol. 1990 88 151. Jochum K. P. Seufert H. M. and Thirlwall M. F. Geostand. Newsl. 1990 14 469. Totland M. Jarvis I. and Jarvis K. E. Chem. Geol. 1992,95 35. Paper 310605 7 H Received October 11 1993 Accepted December 12 1993

ISSN:0267-9477    

DOI:10.1039/JA9940900519

出版商:RSC    

年代:1994

数据来源: RSC

摘要:

JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY APRIL 1994 VOL. 9 525 Atomic Absorption Another Way for Phase Transition Characterization* Serge Walter and Andre Hatterer Ecole Nationale Superieure de Chimie de Mulhouse Laboratoire C. M.A. 3 rue Alfred Werner F68093 Mulhouse Cedex France Phase transitions ocurring in compounds present at ppm levels in powders can be qualitatively and quantitat- ively observed by atomic absorption spectrometry. Thermodynamic and kinetic aspects of atomic vapour formation during a phase transition are considered. A first generation apparatus for phase transition charac- terization by atomic absorption spectrometry (PTCAAS) measurements by a fluidized bed technique is presented. As a typical application of this new technique the determination of ppm amounts of a mixture of mercury([) and mercury(i1) salts in sodium iodide is discussed.The PTCAAS method allows thermal analysis of minor components present in powdered mixtures or as surface layers on solid substrates. By these means one can often obtain data concerning physical state oxidization state and behaviour during reactions or near equilibrium conditions of low concentration species for which usual thermal analysis X ray diffraction or electron microprobe techniques become ineffective. Thus it could be shown that drying a molar solution of sodium iodide containing less than 10 ppm of sodium iodomercurate results in the formation of solid mercury([) iodide and correlated destruction of the mercurate complex. Keywords Thermal analysis; trace analysis; atomic absorption; mercury iodides; phase transition Generally atomic absorption is used for elemental determi- nation regardless of the chemical nature of the compound being analysed.The aim of this paper is to show a new application allowing characterization of solids or adsorbed liquid layers on fine powders. During phase transitions some atoms of the compounds studied become free and thus can be used in order to get an atomic absorption signal. A knowledge of the transition temperatures results in information about parameters such as physical state oxidation state crystal type surface interactions etc which are generally investigated by other techniques such as thermal analysis. The high sensitivity of atomic absorption allows qualitative as well as quantitative determinations of minor species by integration of the absorb- ance signal.Moreover phase-transition characterization by atomic absorption spectrometry (PTCAAS) because of its high sensitivity can give information concerning equilibria especially when minor phases just begin to appear or when they are about to be completely transformed since conven- tional thermal analysis is generally ineffective under such conditions. Theoretical Aspects Two aspects have to be considered (i) the generation of free atoms in the fundamental state during a phase transition; and (ii) the stabilization of these free atoms in order to allow their transfer to an atomic absorption cell where quantitative photo- metric determination can be carried out. Generation of Free Atoms In order to show how free atoms are generated during a phase transition consider a chemical compound MeX where Me is an element that can be determined by AAS.During the phase transition the transformation MeXI-+MeXII is observed. At the equilibrium temperature the chemical potentials of both initial and final forms pi and pf are of course the same. Since the transformation is assumed to be completed under equilib- rium conditions the corresponding free enthalpy Gi equals Gf. Fig. 1 shows schematically the evolution of the free enthalp- ies over time during the phase transition. The bulk contri- * Presented at the XXVIII Colloquium Spectroscopicum Internationale (CSI) York UK June 29-July 4 1993. butions are shown by the lines 1’ (contribution of the initial form) and 1 (contribution of the appearing final form).Regardless of the speed of transformation which depends on a number of parameters such as activation nucleation growth etc. the discussion of which is not the aim of this paper the Time - Pure MeX Pure MeX Time - Fig. 1 Atomic vapour formation rate (a) as a function of time and (b) in correlation with free enthalpy variations during a phase trans- ition. A Free energy of activation during decomposition; D free energy of decomposition; G free energy of the sample in the initial state (before transition); G free energy in the final state (after transition); 1’ and 1 bulk contribution of the initial final forms to the free energy; 2‘ and 2 activation contribution of the initial final forms to the free energy; 3’ and 3 surface contribution of the initial final form to the free energy; 2“ and 3“ sum of 2’ and 2 3’ and 3 respectively; and 4 total free energy during a phase transition526 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY APRIL 1994 VOL.9 bulk free energies are assumed to show the variations given in Fig. 1. It is obvious that for equilibrium conditions the follow- ing relations always exist for an amount n of initial matter and consequently (3) If this were the only contribution to free energy there would be no free atom formation at all during a phase transition. But any change in physical state requires the movement of particles and thus becomes possible only when the moving particles cross over a transient activated state which is always of higher energy than both the initial and final states.The corresponding enhancement of chemical potential gives rise to activation contributions G to the free energy as shown by the dotted line 2‘ of Fig. 1. At each moment this contri- bution can be evaluated by the product of two functions. (i) A probability function P ( t ) rising from zero at the initial time ti of the phase transition to its final value (depending on the kind of the transformation and how close to the equilibrium the final conditions are) at the final time tf. This function describes the probability for a particle (atom or molecule) to leave its initial position in order to reach its final state. (ii) A function ni(t) depending on the number of particles able to undergo the transition. This number decreases from its initial value n to zero when the transition has reached completion.G ( t ) = Adn/dt = P ( t ) ~ i ( t) (4) Thus t h s contribution equals zero both at the initial and at the final time of the transition. Between these points it will show a maximum. The final species have a similar symmetrical behaviour. However the contributions of initial and final species can be quite different in magnitude since the forward and reverse directions of a phase transition may show major differences from a kinetic point of view. For simplicity Fig. 1 shows only an ideal situation. The dotted line 2” corresponds to the sum of the activation contributions from the initial (2’) and final (2) species. The surface contribution must also be considered. During the phase transition the size of the initial crystallites or droplets decreases. Simultaneously small crystals or droplets of the final form are created and grow.As long as these particles (crystals or droplets) are mechanically stable (ie. they do not form spontaneous cracks) their surface energy is higher than their bulk energy. Since the surface to volume ratio for approximately spherical particles is near 3/r (r being the radius of the particle) a decrease of the mean size of the particles results in an enhance- ment of their surface contribution. Thus the surface contribution ,us to chemical potential will increase while the particles are decreasing in size. Each particle containing an amount of matter dn will contribute to the total free surface energy G as follows dG = psdn ( 5 ) The total contribution depends on two factors (i) the surface to volume ratio of the particles (crystals or droplets); and (ii) the number of particles N .Thus since p s is a function of time as well as of particle size it is not a constant even in an homogeneous sample and the total free surface energy at time t is given by and where and where Ni and Nf are the numbers of particles in the initial and final forms respectively. Fig. 1 shows the surface contributions of both initial (broken line 3’) and final forms (broken line 3). Their sum is given by the broken line 3”. The total contribution to the free energy is given by line 4. GMex** being the free activation energy leading to the free fundamental atomic state D the activation energy gap AE, shows a minimum during the phase transition.Thus the ratio of the activated species N** to the fundamental species during the phase transition following Boltzmann’s law is shown in Fig. l(a). As the number of free fundamental atoms is pro- portional to this ratio it will result in a maximum while the phase transition is in progress. The possibility of formation of free atoms has therefore been shown. However since just after their formation free species are very close together they generally recombine very quickly and cannot be determined in this way. Therefore it is essential to avoid this recombination by a rapid dilution. Stabilization of Free Atoms Fig. 2 shows the chemical potential steps occurring while the phase transition proceeds. The activation step p* allows the transition by migration of the rearranging species.Vaporization of free molecules which involves higher energies can occur at pv. The highest activation state p** leads to decomposition and formation of atomic vapours at as discussed previously. Since chemical potential depends on partial pressure accord- ing to p=p,+RTlog PIP (9) a dilution of the atoms just formed results in a lowering of their partial pressure PIP and consequently reduces their final chemical potential ,up Thus provided the free atoms cannot react with the diluting gas they can be stabilized as required by sufficient dilution. This thermodynamic stabiliz- ation is enhanced by a kinetic stabilization generally thermal decomposition depends on low order reactions and is not strongly dependent on concentration factors.On the other hand recombination requires the meeting of at least two species; thus recombination reactions are at least of second order and their speeds V are at least proportional to the square of the partial pressure according to V = K(P/P0)’ where K is a kinetic constant. This double stabilization thermodynamic and kinetic allows the free atoms formed to be carried away to an absorption cell. Experimental Apparatus The experimental device used is shown by Fig. 3. A finely powdered sample (1-20 g) is heated (ramp rate 5-20 “C min-’) in a fluidized bed. Heat losses are reduced by warming up the double coating of the reaction cell the strong hot air flow in the heat gun blows around the fluidized bed region allowing the side walls and the sample to be approximatively at theJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY APRIL 1994.VOL. 9 527 Pf - Events Me + X + A r a * * i Activation Species MeX*' Atomic Dilution I Vaporization\ \ . t I 1 I Initial conditions near the sample surface -r- - -- - - - - - _ _ _ _ --- - _ - - _ _ - _ _ _ _ _ Final conditions in the inert gas flow I Fig.2 Chemical potential steps and variations during a phase transition pco chemical potential of both initial and final state; p* chemical potential of migrating molecules; pv chemical potential of free molecules (in the vapour phase); pd chemical potential of free atoms; p** chemical potential of molecules undergoing decomposition; and pr chemical potential of free atoms after dilution Fig.3 PTCAAS vapour generator 1 Fine glass filter; 2 gas cooler; 3 cold water flow; 4 coarse glass filter; 5 0.25 mm thermocoax chromel-alumel (N-type) thermocouple; 6 dou ble-channel recorder; 7 gas outlet towards the absorption cell; 8 hot-air outlet; 9 thin- walled thermocouple glass coating; 10 fluidized bed; 11 absorbance signal from spectrometer; 12 heat gun; 13 power supply (mains 220 V a.c.); 14 inert gas heater; 15 temperature regulator; and 16 inert gas (argon hydrogen etc.) inlet same temperature.The carrier gas has to be non-reactive also with both the sample and the atoms that become free during the phase transitions. Temperature is controlled by a 0.25 mm thermocoax N-type thermocouple in order to have fast thermal response. A double dust filtering prevents condensed particles being carried into the absorption cell.A cooling system (although not essential) lowers the gas temperature after filtering. The absorption cell is quite similar to those used for cold vapour generation techniques; the optical path is 200 mm long. A Varian AA-6 spectrometer is coupled to a Sefram double channel recorder for the measurement of both absorbance and temperature signals. The four-lamp turret single beam spectrophotometer is fitted with a mercury hollow cathode and a deuterium lamp. The mercury hollow cathode is operated at an intensity of 0.5 mA and the deuterium lamp intensity is adjusted so that it leads to a zero absorbance signal when the deuterium lamp replaces the mercury hollow cathode. Alternative measurements with either the mercury or deuterium lamp allows background correction and baseline control the latter allowing discrimination between molecular absorption (disappearing once the species are swept away) and spurious signals due to the presence of condensed particles (remaining partially on the cell windows after the experiment) in the analytical light beam.The determinations were carried out at 253.7 nm with a spectral bandpass of 0.2 nm (slit-width 60 pm). Samples Sodium iodide (Prolabo ref. 27 91 5 23 1) was used as a matrix. Mercury(1) and mercury(I1) salts were added to the sodium iodide as follows a known amount of mercury salt is dissolved in a molar solution of sodium iodide in order to obtain a solution containing about 1000 mg 1 - of mercury. This solu- tion is then used to adjust the mercury content of another molar sodium iodide solution to the required level (0.1-100 ppm).This solution is then slowly dried at 90 "C in order to avoid any loss of mercury. The solid crystalline sample is finely powdered by grinding in an agate mortar. The mercury content of the sample is controlled by direct carbon rod atomization as described by Hatterer et al.' A precisely weighed amount of powdered sodium iodide containing mercury salts is introduced into the device shown in Fig. 3 and heated after fluidization has started. Preparation of Mercury Salts About 10mollK' nitric acid reacting with a large excess of metallic mercury (Prolabo ref. 25 402 155) yelds a solution containing mercury(I1) nitrate (see refs. 2 and 3). For the preparation of mercury(r) an excess of elemental iodine (dis- solved in an aqueous solution of sodium iodide) is added to the mercury(I1) solution to carry out thourough oxidation as described in ref.2 Mercury(1) and mercury(11) iodides are prepared by the addition of an hypostochiometric amount of sodium iodide to the corresponding nitrate^.^ The precipitates are then filtered off and washed with water. An excess of sodium iodide solution is used for the formation of the soluble mercury complexes. Results and Discussion Reversibility of the Mercury(i1) Iodide Transitions Fig. 4 shows the absorbance signal (solid line) and the tempera- ture (broken line) uersus time obtained with a 10 g sodium iodide sample containing 100ppm of mercury. An atomic signal appears as soon as the sample temperature rises over the solid-solid transition temperat~re.~ The fine peak obtained by increasing the temperature is quite different from the noisy signal corresponding to the transition back to the initial form.This is in good agreement with the experimental behaviour of528 I JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY APRIL 1994 VOL. 9 Boiling-point (354 "c) 0.010 0.005 '.. /- \ / Phase transition ' - / / / / / -1 r 0 300 600 Time/s -I (1 29 "c) 4 0 900 Fig. 4 PTCAAS signal for 10 g of sodium iodide containing 1 mg of mercury ( 100 ppm) absorbance (solid line) and temperature (broken line) versus time mercury(11) iodide when the temperature is increased the transformation from the red low temperature form to the yellow high temperature form is very easy. Thus all of the mercury@) iodide contained in the sample undergoes transition very close to the theoretical transition temperature.However the transformation back to the initial form is much more difficult. The probability of nucleation is very low and thus a noticeable delay may occur before some crystals undergo their transition although their temperature is far beyond the trans- ition temperature. Therefore the corresponding PTCAAS signal becomes very noisy each transforming crystal giving rise to its own peak which can be observed when resolution time is sufficient. However the areas of the peaks observed for both an increase and a decrease in temperature are the same (differences less than 5%). The experiment was carried out with the same sample up to 30 times without any noticeable loss in peak area (standard deviation 3%) This shows that only a few atoms became free during the solid-solid transition so that the corresponding loss of matter is negligible in comparison with the initial amount present in the sample.Melting- and Boiling-points A second experiment (Fig. 5 ) with a sample containing ten times less mercury(1) iodide than in the first shows that the linearity of the signal corresponding to the solid-solid trans- ition is fairly good (the solid-solid transition peak decreases by a factor 10 5%) and can be used for quantitative determi- nations. As the melting-point is reached a second sharp signal appears about ten times greater than the previous one. Near the boiling-point evaporation becomes rapid and the corre- sponding liquid-gas transition gives rise to a third broad peak.This shows clearly that each transition (solid-solid solid-liquid and liquid-gas) gives rise to its own signal. Obviously since PTCAAS requires lowering of the partial pressures of the gas- phase components only condensed-phase transitions are well defined. Moreover since the fluidizing gas always lowers the partial pressure of the vapours below the boiling pressure the evaporation is completed before the boiling-point is reached and there are no more observable signals corresponding to phase formations below the boiling-point when the lowering temperature crosses over the melting and the solid-solid transition temperature. Both mercury(1) iodide and sodium iodomercurate can be used as a mercury source without any observable change in the PTCAAS signal.0.010 0.005 I 1 0 600 1200 1800 Time/s Fig. 5 PTCAAS signal for 10 g of sodium iodide containing 0.1 mg of mercury (10 ppm). Absorbance (solid line) and temperature (broken line) versus time Oxidation States and Equilibria When both mercury(1) and mercury@) iodide are present in the sample (Fig. 6) four peaks appear in the range 20-300 "C. Good resolution is obtained by lowering the temperature ramp rate to 8"Cmin-l. As foreseen with regard to the data in refs. 3 4 and 5 these four peaks respectively appear at the solid-solid transition temperature of mercury(I1) and mer- cury(1) iodide then at the melting temperatures of mercury(1) and mercury(11) iodide the latter having the highest melting- point. This experiment shows how atomic absorption can be used to determine quantitatively the oxidation states of mer- cury in a sample.On the other hand it must be taken into account that the initial solution contained no mercury iodides since mercury forms soluble complexes with sodium i ~ d i d e . ~ Evaporating a solution containing nothing but iodomercurate yelds a colour- less crystalline solid complex. In contrast with this pouring an iodomercurate solution onto solid sodium iodide results in the formation of a red precipitate showing decomposition of the complex into mercury and sodium iodide by an excess of sodium iodide. The question of whether or not this decompo- sition occurs at a very low mercury to sodium ratio is clearly I 0.010 n P I - W rn I_/ 0 0 C m ft a 2 0.005 0 600 1200 1800 Timeis Fig.6 PTCAAS signal for 10 g of sodium iodide containing 0.1 mg* mercury@) iodide (Hg,I,) and 0.1 mg of mercury(1) iodide (HgI,) absorbance (solid line) and temperature (broken line) versus timeJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY APRIL 1994 VOL. 9 529 answered by PTCAAS since the only peaks observed are those of mercury(1) and mercury(I1) iodides excluding the broad peak observed when pure iodomercurate decomposes. Thus it can be concluded that the decomposition of iodomercurate on solid sodium iodide is complete before heating of the sample. Conclusion It has been shown that AAS can be a very sensitive means for the determination of minor species by characterization of their transition temperatures. It should be noted that since PTCAAS is a cold vapour generation technique the signal-to-noise ratio essentially depends on the noise of the source (generally hollow cathode lamps) or of the detector (photomultiplier).New laser techniques (Niemax and co-w~rkers~’~) allowing significant lowering of the signal-to-noise ratio will certainly be essential for the expected development of PTCAAS techniques by providing a drastic enhancement in sensitivity. In order to be more easily applied the first generation device described here must be adapted to smaller samples and as far as possible be capable of being used with a simpler method of vapour-phase dilution rather than fluidization. However it has to be pointed out that the fluidized bed technique presents two major advantages firstly the sample temperature is very homo- geneous and therefore high-temperature ramp rates resulting in sharp easily detectable signals can be used; secondly since the fluidizing gas is quite dry (water content below 5 ppm in the cell) the sample is completely dry even before the tempera- ture begins to rise.Comparative measurements by PTCAAS and differential scanning calorimetry (DSC) (carried out by use of a Setaram DSC 111 calorimeter) showed that DSC could not work when there was less than 25 ppm of mercury in the sample. Since the sample is heated in a nearly closed crucible the traces of water present in the sample (even after drying by P,O,) which become free by heating lead to a ternary system Nal-H,O-HgI the behaviour of which is different from the binary system Nal-HgI dealt with here.On the other hand since the size of the sample is about 30 times higher in the prepared PTCAAS device (10 g in the described as 1 2 3 4 5 6 7 experiment 50g being the maximum capacity of the device) than in the DSC crucible (about 300mg) the sensitivity by PTCAAS for the determination of mercury iodides by their melting-points is at least 100 times better than by DSC. Compounds other than mercury iodides should be studied in order to determine how PTCAAS can be applied to other elements. Mercury compounds convert rather easily to atomic vapours and therefore it could be expected that this technique would not have a general applicability to low temperature transitions of compounds of other elements. However heating the samples using a carbon-rod like device would allow very high temperatures to be reached.Under these conditions a lot of high temperature phase transitions the determination of which would become very difficult by conventional methods such as thermal analysis could be studied since these high temperatures would result in reasonably good atomization conditions. Work is being done in order to develop a small high-temperature device to carry out PTCAAS measurements easily as electrothermal AAS measurements. References Hatterer A. Mougenel V. and Walter S. Analusis 1987 15 486. Gmelins Handbuch der Anorganischen Chemie ed. Kotowski A. Verlag Chemie GmbH Weinheim 1965 part B Lfg. 1 34 pp. 97 98 118 and references cited therein. Comprehensive inorganic Chemistry eds. Ballar J. C. Emeleus H. J. Nyholm R. and Trotman-Dickenson A. F. Pergamon Oxford 1973 3 p. 393 and references cited therein. Gmelins Handbuch der Anorganischen Chemie ed. Kotowski A. Verlag Chemie GmbH Weinheim 1967 part B Lfg. 2 34 pp. 820-825 832 833 and 844 and references cited therein. Pascal P. Nouveau Traitb de Chimie Minirule Masson et Cie Paris 1962 5 Zinc Cadmium Mercure pp. 710 and 711. Niemax K. Naturwissenschaften 1991 78 250. Schnurer-Patschan C. Zybin A. Groll H. and Niemax K. J. Anal. At. Spectrom. 1993 8 1103. Paper 3/05391 A Received September 8 1993 Accepted December 21 1993

ISSN:0267-9477    

DOI:10.1039/JA9940900525

出版商:RSC    

年代:1994

数据来源: RSC

摘要:

JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY APRIL 1994 VOL. 9 531 Spectral Interference on the Lead 283.3 nm Line in Zeeman-effect Atomic Absorption Spectrometry* U. Kurfurst University of Fulda (Fachhochschule) Marquardstr. 35 0-36039 Fulda Germany J. Pauwels Commission of the European Communities Joint Research Centre 6-2440 Geel Belgium In Zeeman-effect atomic absorption spectrometry a spectral interference for lead on the 283.3 nm line was observed when solid samples containing high sulfur contents are analysed. From experiments and published data it is evident that this interference is due to molecular absorption by S2 molecules. In real samples selective vaporization of the sulfur compounds by thermal pre-treatment is not possible with the graphite furnace-platform equipment used.However with appropriate atomization conditions the interfering signal can be sufficiently separated. Keywords Atomic absorption spectrometry; Zeeman-effect background correction; molecular interference; lead determination; solid sample analysis At present the most powerful technique for background correc- tion in atomic absorption spectrometry (AAS) is the one based on the Zeeman effect. Since its discovery many analytical problems caused by a strong background have been solved. However even using Zeeman-effect atomic absorption spec- trometry (ZAAS) spectral interferences are possible when the background is structured by atomic or molecular absorption in between the Zeeman split wavelength pattern of the analytical line. In direct ZAAS ( i e . the spectral source is in the magnetic field resulting in splitting of the emission line) spectral inter- ference can be observed by an interfering absorption line because the background measurement takes place ‘beside the analytical line’ by the split (0) line components.An example of such an interference by an atomic line was shown by Stephens and Murphy’ at an early stage of the development of the Zeeman technique whereas an example of molecular line interference was first shown by Massmann2 with flame atomization and Kurfiirst3 with electrothermal atomization. In inverse ZAAS (ie. the analyte atom volume is in the magnetic field resulting in splitting of the absorption line) the background is measured ‘under the analytical line’. Massmann has also shown spectroscopically that in this case atomic and molecular absorption can cause an interference when the interferent line shows a Zeeman splitting.2 Over several years further examples of spectral interferences have been published and these are presented in Table 1.Evidence for the identification of the molecular compound that is responsible for the observed interference effect varied. Apart from Massmann2 only Ohlsson and Frech” have carried out real high-resolution spectroscopic measurements in order to identify several interferences in ZAAS owing to small molecules. While the identification of all interfering atoms are based on reliable tabulated spectroscopic data data on the identifi- cation of interfering molecules are lacking. In only a few cases have spectroscopic considerations or calculations on overlap- ping lines from molecular band spectra been p e r f ~ r m e d .~ ? ~ ~ . ~ ~ . ~ ~ In some studies considerations of vapour-phase chemistry have been used to obtain sufficient plausibile evidence for the molecule that is supposed to be responsible for an observed interference e f f e ~ t . ~ * ~ * l ~ In some papers only the analytical signals of a spectroscopic interference effect were documentated because no information * Presented at the XXVIII Colloquium Spectroscopicum Inter- nationale (CSI) Post-Symposium on Graphite Atomizer Techniques in Analytical Spectroscopy Durham UK July 4-7 1993. about spectral overlaps or the appearance of a interfering molecule was a ~ a i l a b l e . ~ . ~ ~ The present paper shows an interference effect on the analytical lead line at 283.3 nm frequently used in electrother- mal AAS measurements which was observed during the analy- sis of solid samples using direct ZAAS.Additional experiments Table 1 Observed atomic and molecular interferences in ZAAS Analytical line/ Element nm Atomic inter$erence- Ag Au Bi Fe Ga Hg Pb Pd Pt Zn 328.1 267.6 223.1 271.9 287.4 253.9 261.4 247.6 265.9 213.9 Molecular integerence- Ag 328.1 Au 267.6 Bi 306.8 Ca 422.7 Cd 326.1 c o 3 84.61388.2 Cr 357.9 Cr 425.4 Fe 246.31248.3 250.11252.3 Fe 358.11386.0 Ga 403.3 Hg 253.7 In 325.8 K 404.4/404.7 Mg 383.8 Mn 279.5 Ni 378.4 0 s 378.2 Pb 217.0 Pb 283.3 Pd 244.31247.6 Sb 206.8 Se 196.0/204.0 Sn 286.3 T1 276.8 Zn 213.8 Interfering component Rh c o Fe Pt Fe c o c o Pb Eu Fe PO BaO OH CuH PO CN CN CuH PO CN CuH PO PO CuH CN AlBr CN CN A10 s2 PO PO NO PO InBr cs PO NO Ref.4 5 6 1 7 8 1 7 9 9 10 9 8 11 12 4 2 12 7 12 13 13 12 12 12 13 12 12 14 12 13 4 13 13 4 This work 12 14 12 15 16 4 3 12 16532 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY APRIL 1994 VOL. 9 and published spectroscopic and vapour-phase chemistry data are evidence for the presumption that the S 2 molecule is responsible for this interference. The aim of this study was to document this effect and to give systematic proposals for avoiding serious analytical errors in solid sampling AAS. Experimental Direct ZAAS measurements were carried out using an SM-30 spectrometer (Griin-MeDtechnik Ehringshausen Germany) in combination with an audiofrequency electrode discharge larnpI7 for lead positioned in the gap of a permanent magnet (0.9 T).The direct ZAAS configuration has been described in ref. 3. The instrument is equipped with a Massmann-type graphite furnace (tube length 50 mm with an internal tube diameter of 8 mm). The samples are introduced into the furnace using a graphite platform boat.18 Interference of the vapour from pure sulfur was measured using a 10mm long cylindrical quartz cell with flat windows containing some micrograms of elemental sulfur. It was evacu- ated to a primary vacuum refilled with argon up to 1 x lo4 Pa and sealed off (Fig. 1). The cell was placed in a tube furnace aligned in the light beam of the SM-30. The furnace was slowly heated from room temperature up to 800°C. Temperature light transmission and the ZAAS signal were measured and recorded. Results and Discussion Analytical Effects Non-compensated background interference in measurements of lead by ZAAS at the 283.3 nm line can be observed during the analyses of solid samples containing sulfur; the ZAAS signal of 2 mg of a powdered grass sample is shown in Fig.2. Light beam (A = 283.3 nm) ..... - \ [FA?) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . '.' \ / Tube furnace \ Quartz cell (1.5 x 1 cm 0.d.) Fig. 1 of light from Pb 283.3 nm by sulfur vapour Experimental set-up for the demonstration of the absorption 0.03 I 1 100 0.02 0 C (D e a 0.01 0 0 1 2 3 4 5 6 7 8 Time/s Fig. 2 by direct ZAAS Analytical signal for Pb at 283.3 nm in a 2 mg grass sample The negative peak indicates a stronger absorption for (one of) the o-components than that of the n-component.To avoid inaccurate absorption measurements three tech- niques were investigated selective separation of the interfering compound by thermal pre-treatment of the sub-samples; in situ digestion; and separation of the background from the analyte signal by an appropriate temperature programme during atomization. Three thermal pre-treatment (prolysis) curves are shown in Fig. 3. Although the negative signal that appears for a pure sulfur sample vanishes completely at a pre-treatment tempera- ture of 700 "C (line C) the negative background signal from the grass material is constant up to 950°C (line B) and can still be observed up to 1300°C.The thermal stability of biologically bound sulfur is well known from a technique for the determination of sulfur where temperatures of approxi- mately 1000°C and catalytic surfaces must be applied to free the bound sulfur. The pre-treatment curve for lead in the grass sample is shown in Fig. 3 line A. It is known that losses of lead take place at temperatures above 500 "C. Hence selective vaporiz- ation of the interfering compound from biologically bound sulfur is definitely not possible with the instrumental set- up used. Oxidative destruction of the matrix was also investigated by adding a sufficient amount of nitric acid to the test sample directly in the sample boat. As expected the negative signal did not appear because the sulfur was oxidized leaving the tube as SO2.However complete chemical transformation could only be achieved if the acid was pipetted very carefully onto the sample pulver so that the sample was soaked totally moistening all of the sample particles. Because this in situ reaction is not easy to control with the direct introduction of solids (and is what leads to the existence of negative signals) and moreover the advantage of the solid sampling technique is partly lost by the additional steps required and the use of reagents (giving blank values) this method is not recommended for routine measurements. Considering the different appearance temperatures and times of the background and analyte signals solid samples which show this interference effect can however still be analysed for lead if the two signals are resolved further by choosing a smoother temperature ramp during the atomization step.The peak height value is then not influenced by the interference. Choosing an integration window (e.g. an integration period I A 1.0 0.5 0 -0.5 - 1 500 1000 1500 TemperaturePC Fig. 3 Thermal pre-treatment curves for Pb at an atomization tem- perature of 2200°C. A Peak height of the analyte in a grass sample plotted upwards (normalized to a sample mass of 1 mg). The uncom- pensated background signals are plotted downwards (B) for a grass sample. C Negative signal (peak height) of 1 mg samples from a mixture of 9 + 1 graphite-elemental sulfurJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY APRIL 1994 VOL. 9 533 from 1.2 to 2.6s see Fig.2) the influence of the interfering signal on the peak area value can be neglected.This has been shown by the analysis of certified reference materials where the described interference has been seen." Particular attention should however be paid to the fact that the observed effect can also appear with samples with lower sulfur contents if the vaporized compounds are allowed to condense on the cooler parts of the graphite tube. As shown in Fig. 4 a slight but increasing negative peak can appear in successive runs. This effect can be avoided by burning out the furnace after each analysis. Identification of the Interferent A negative signal appears if the sample contains a large amount of sulfur (e.g. grass 0.3% and coal 2%). With the direct introduction of some micrograms of pure sulfur into the furnace and vaporization without thermal pre-treatment the range of the signal processing system is totally exceeded (in the negative direction). It is known that in the gas phase elemental sulfur forms several molecular compounds at room temperature and atmospheric pressure S8 molecules dominate but at increasing temperatures and decreasing pressure S S4 S2 molecules and atomic sulfur are formed by stepwise dissociation.20 While the large sulfur molecules show no line spectrum S2 has an absorption band spectrum in the region of 240-630 nm.20,21 The table of persistent band heads given by Pearse and Gaydon2' shows that for this molecule a line coincidence with the absorption lines in AAS can be expected.Herzberg22 gave a highly resolved absorption spectrum for S vapour.It showed the band head at 282.91 nm degrading to the red as was also found and has been described by Rosen and D e ~ i r a n t . ~ ~ Encouraged by these data from the spectroscopic literature the influence of sulfur vapour on the ZAAS signal of the lead 283.3 nm line was investigated by a model experiment. Using the set-up with the quartz cell (Fig. 1) in the light beam of the spectrometer it could be shown that a negative signal appears with an increase in temperature as is shown in Fig. 5. This signal reaches a maximum at only 250°C because of the reduced pressure and decreases again slightly with further heating. This course corresponds to the dissociation of the large sulfur molecules to a lower average number of atoms in each molecule as the number of diatomic and atomic sulfur species increases.20 Conclusion The S2 molecule could be identified with a high degree of certainty as being responsible for the observed spectral h r $ 1 I 100 50 0 t 0 2 4 6 8 10 12 14 16 Time/mi n Fig. 5 A Light transmission; B signal of direct ZAAS (283.3 nm); and C temperature of a quartz cell containing sulfur vapour interference on the lead 283.3 nm line without the expendi- ture necessary to obtain high-resolution spectroscopic measurements.With the proposed separation of the analyte and back- ground signals by appropriate temperature programme and signal processing serious measurement errors can be avoided. However if part of the true analyte signal is not integrated great attention must be payed to ensuring true analytical results are achieved (e.g.standard additions or use of a certified or standard reference material with an identical matrix). The use of an alternative analytical line should preferably be considered. Unfortunately the 217.0 nm line is often too sensitive for solid samples and has a narrow working range (low intensity) while the 261.4 nm line is two orders of magnitude less sensitive compared with the line under discussion. Because of the importance of the 283.3 nm line it will be of value in studies of the vapour phase chemistry of sulfur and its compounds under electrothermal atomization conditions. Hence better methods of chemical modification or sample treatment could possibly be found to overcome this problem (e.g. oxygen ashing).A negative peak can be also observed with inverse ZAAS (the furnace is in the magnetic field) on introducing pure sulfur or sulfur containing solid samples into the furnace e.g. the grass material used in Fig. 2. An analytical signal of a Time -t Fig. 4 Evolution of the ZAAS signal for Pb 283.3 nm from a grass sample for a succession of analyses534 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY APRIL 1994 VOL. 9 Time/s Fig. 6 Analytical signal for Pb at 283.3 nm in a 0.2 mg coal sample by inverse ZAAS pulverized coal sample achieved with the a.c. inverse Zeeman- effect AAS (using a Perkin-Elmer 2-3030) is shown in Fig. 6. No explanation can be given for the difference in the appear- ance times of the interference signal relative to the analyte signal with the different ZAAS techniques (see Figs.2 and 6). It could for example be caused by the differences in the tube-platform configuration (Grun) and the cup-in-tube arrangement ( Perkin-Elmer) respectively. A spectroscopic difference due to the application of a modulated magnetic field to the absorption volume p perkin-Elmer) could also be caused by the observed but not understood strengthening of the absorption band spectra of the S2 molecules by a magnetic field which is suspected of being influenced by the Zeeman effect on disturbances of the vibration term levels.20 It could be expected that this interference can also disturb AAS measurements of sample types other than solids and with instruments equipped with background compensation systems other than that using the Zeeman effect.The authors thank E. Freistedt for valuable help in producing the quartz cell equipment and L. De Angelis and J. Kress for their assistance with the measurements. References 1 Stephens R. and Murphy G. F. Talanta 1978 25 441. 2 Massmann H. Talanta 1982 29 1051. 3 Kurfurst U. Fresenius’ 2. Anal. Chem. 1983 315 304. 4 Wennrich R. Frech W. and Lundberg E. Spectrochim. Acta Part B 1989 44 239. 5 Wibetoe G. and Langmyhr F. J. Anal. Chim. Acta 1985 176 33. 6 Trostle D. Beals T. Kuczenski R. and Shaver M. At. Spectrosc. 1991 12 64. 7 Carnrick G. R. Barnett W. and Slavin W. Spectrochim. Acta Part B 1986 41 991. 8 Wibetoe G. and Langmyhr F. J. Anal. Chim. Acta 1984 165 87. 9 Wibetoe G. and Langmyhr F. J. Anal. Chim. Acta 1986,186,155. 10 Frigge C. and Jackwerth E. Spectrochim. Acta Part B 1992 47 787. 11 Manning D. C. and Slavin W. Spectrochim. Acta Part B 1987 42 755. 12 Ohlsson K. E. A. and Frech W. J. Anal. At. Spectrom. 1989 4 379. 13 Doidge P. S. Spectrochim. Acta Part B 1991 46 1779. 14 Wibetoe G. and Langmyhr F. J. Anal. Chim. Acta 1987 198 81. 15 Radziuk B. and Thomassen Y. J. Anal. At. Spectrom. 1992 7 397. 16 Le Bihan A. Cabon J. Y. and Elleouet C. Analusis 1992,20,601. 17 Kurfurst U. Fresenius’ 2. Anal. Chem. 1985 322 660. 18 Hadeishi T. and Le Vay T. Fresenius’ J. Anal. Chem. 1990 337 264. 19 Rosopulo A. Grobecker K.-H. and Kurfurst U. Fresenius’ Z . Anal. Chem. 1984 319 540. 20 Gmelins Handbuch der Anorganischen Chemie Band 8 Schwefel Teil A Verlag Chemie Weinheim 1953. 21 Pearse R. W. B. and Gaydon A. G. The Zdentijcation of Molecular Spectra Wiley New York 1950. 22 Herzberg G. Spectra of Diatomic Molecules Van Nostrand Reinhold New York 2nd edn. 1950. 23 Rosen B. and DCsirant M. Bull. Soc. Sci. LiGge 1935 233. Paper 3 f06347.J Received October 25 1993 Accepted January 4 1994

ISSN:0267-9477    

DOI:10.1039/JA9940900531

出版商:RSC    

年代:1994

数据来源: RSC

摘要:

JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY APRIL 1994 VOL. 9 535 Determination of Total Mercury in Scalp Hair of Humans by Gold Amalgamation Cold Vapour Atomic Absorption Spectrometry* Carlos G. Bruhn Aldo A. Rodriguez Carlos Barrios and Victor H. Jaramillo Departamento de Analisis Instrumental and Departamento de Farmacia Facultad de Farmacia Universidad de Concepcidn Casilla 237 Concepcidn Chile Jose Becerra Departamento de Botanica Facultad de Ciencias Naturales y Oceanograficas Universidad de Concepcidn Concepcidn Chile Urcesino Gonzalez Departamento de Estadistica Facultad de Ciencias Fisicas y Matematicas Universidad de Concepcidn Chile Nuri 1. Gras Laboratorio de Analisis per Activacidn Comisidn Chilena de Energia Nuclear Centro de Estudios Nucleares La Reina Santiago Chile Olga Reyes Environmental and Occupational Network of the World Health Organization P.0. Box 105 Concepcidn Chile Seremi Salud Secretaria Regional Ministerial de Salud Octava Region de Chile Concepcidn Chile Cold vapour atomic absorption spectrometry (CVAAS) with a laboratory-built system using a Au-Pt grid for mercury amalgamation was applied in the determination of total mercury (Hg-1) in human scalp hair. Two sample dissolution procedures by HNO digestion were tested and compared in a poly(tetrafluoroethy1ene) (PTFE) bomb for 1.5 h at 11 0 "C (procedure l) and in sealed Pyrex ampoules for 24 h at 50f 10 "C (procedure 2). After optimization of the aeration gas flow rate (1 00 ml min-') de-amalgamation temperature (700 "C) and releasing time (1 9 s) the analytical methodology was characterized and validated.The linear working range extended from 0.5 to 12.5 ng and the characteristic mass was 0.29 ng of Hg (mass of analyte giving an absorbance of 0.0044). The repeatibility of measurements (within-day variation) expressed as percent relative standard deviation (RSDYO) was 5.5% at 1.25 ng of Hg (n =8) and 3.7% at 12.5 ng (n=14) and the mean reproducibility (between-day variation) estimated from calibration curves on different days was 6.4 f 1 .I 940 (n =5). The absolute detection limit (3 x ob) was 0.13 ng of Hg and the limit of quantitation (10 x ob) was 0.43 ng of Hg ( ~ 0 . 1 1 mg kg -' in hair). Analytical precision (8.4 & 4.0% RSD) and accuracy (4.7 f 2.5% mean relative error) were satisfactory for ppm and sub-ppm levels of Hg-T in several biological and environmental certified and standard reference materials (CRMs and SRMs) including hair.Procedure 2 was selected as it is much simpler requires inexpensive reagents and is more amenable for routine application. Mean recovery of Hg-T in hair spiked with Hg standard solutions was 105% in the range 0.83-1.27 mg kg-' In addition two human head hair samples were assayed as prospective laboratory control materials. Thus sub-sample homogeneity analytical intra-laboratory variability and external quality control were assessed to confirm the methodology in use. Between-day variation in hair analysis for Hg-T conducted on pooled scalp hair over a period of 5 months was 4.8% RSD (1.1 0 f 0.053 mg kg-' for the 95% confidence interval; n = 16). Instrumental neutron activation analysis used as an independent method showed significant correlation in results with CVAAS both in several biological and environmental SRMs and CRMs and in human hair samples of pregnant and nursing women (Hg concentration range = 0.1 -6.9 mg kg -'; n = 21 ; ? = 0.880 p < 0.0001).Keywords Gold amalgamation cold vapour atomic absorption spectrometry; mercury; hair; sample preparation ; analytical characterization and validation ; quality assurance Increased concern about the health effects of Hg and its compounds in humans exposed to very low environmental concentrations of Hg is due to sub-clinical effects' whose symptoms are difficult to detect and measure.' Clinical and epidemiological evidence indicates that pre-natal life is more sensitive to the toxic effects of Hg than adult life in particular when it is present as methylmercury ( Me-Hg).3 Recent studies have shown effects on the early childhood development and mental ability of children whose mothers were exposed during pregnancy to 3-4 times the tolerable recommended weekly intake set by the World Health Organization (WHO) and the Food and Agriculture Organization ( FAO).4 Until better * Presented at the Second Rio Symposium on Atomic Absorption Spectrometry Rio de Janeiro Brazil June 21-28 1992.methods are developed for determining the difference in con- duct and mental activities of people with low and slightly elevated levels of Hg in their bodies it is helpful to monitor prospective health-risk groups in problem sites.Since the 1960s the concentration of Hg in hair has been used to assess the level of impregnation of the organism in order to show both acute and chronic exposure. People accumulate Hg in their bodies mainly by intake and absorption of Hg from their diets' and from the air in their local environment.6 The major source of human exposure to Me-Hg is through the diet mainly from consumption of fish and fish product^.^ In this case Me-Hg is the main organomercury species incorporated into the human hair structure.' Usually the percentage of Hg present as Me-Hg in human hair varies between 80 and 100%; and total mercury (Hg-T) levels in536 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY APRIL 1994 VOL. 9 newly formed hair of humans with increased dietary exposure reflect very accurately the concentration in Thus hair analysis provides a convenient tool for estimating body burdens of Hg-T and Me-Hg in exposure to Hg through the diet.Relative to other body organs and fluids scalp hair contains elevated concentrations of trace elements (about 300 times more Hg than in blood) which makes analysis Hair sampling and conservation is easier than for other biological tissues and fluids; it grows about 1 cm per month is painlessly removed and does not require specialized equipment and refrigerated storage facilities. Concentrations of Hg in new hair growth indicate the body burden of Hg during the time of growth.'O However hair is susceptible to external contami- nation from sources such as atmosphere dirt dust sweat and by the use of cosmetics and pharmaceutical preparations.Thus a more realistic assessment of the problems involved makes it difficult to generalize about the usefulness of hair as an indicator of exposure. Nevertheless under certain conditions including consistent sampling washing and analysis and excluding exposures to low levels of Hg vapour in which case adventitious contamination can originate misleading results,' hair can be a good indicator of exposure to Hg through the diet for population survey purposes." Various techniques have been developed for the determi- nation of low levels of Hg in biological and environmental samples the most widely used being radiochemical neutron activation analysis,I2 atomic fluorescence ~pectrometry,'~ and cold vapour atomic absorption spectrometry (CVAAS).14 When there is a need to selectively measure Me-Hg or other organomercury species usually gas chromatography with elec- tron capture detection (GC-ECD) is used whereas CVAAS is applied for Hg-T and inorganic Hg determination and by difference organic mercury.I5 Nevertheless Horvat et tested different isolation techniques (ion-exchange extraction volatilization and distillation) for the determination of Me-Hg with final measurement via CVAAS or GC and results were comparable for almost all biological and environmental samples including human hair.Recently a study comparing Hg results for various kinds of certified reference materials (CRMs) and concentration levels using different isolation techniques and final measurements by CVAAS (distillation and ion-exchange) and GC (extraction distillation and volatil- ization) showed good agreement indicating that currently available CRMs for Hg-T can also be used to establish recommended values for Me-Hg and organic Hg.17 Nowadays CVAAS is the most widespread method used to determine Hg in biological samples.Since its inception described first by Poluektov and c o - ~ o r k e r s ' ~ ~ ~ ~ several modifications have been introduced to the procedure aiming for improvements in sensitivity precision lower interferences and increased practi- cality of the methodology. Probably one of the most important improvements has been the introduction of a preconcentration step by amalgamation preferably on gold,2s25 of the Hg vapour generated in the reduction reaction.Hence the sensi- tivity is increased by more than an order of magnitude because the collected Hg can be released rapidly into the absorption cell by heating the gold trap.25 In this paper a simple laboratory-built cold vapour system including a preconcentration step by amalgamation of Hg in an Au-Pt grid and operated in batch mode is characterized and validated. Optimization of three parameters (i.e. gas flow rate de-amalgamation temperature of the Au-Pt grid and releasing time of the trapped Hg) and application of the system in the determination of ppm to sub-ppm levels of Hg-T in several CRMs and SRMs is presented and discussed. Two wet digestion procedures are tested and compared for the determination of Hg in human scalp hair. The preparation of hair samples was assessed for homogeneity analytical intra- laboratory variability and external quality control to confirm the reIiability of the methodology in use.The system was applied to the determination of Hg in human scalp hair of pregnant and nursing women using a simple and more amen- able wet digestion procedure in a Pyrex ampoule. The CVAAS methodology was compared with instrumental neutron acti- vation analysis (INAA) used as a reference method for external quality control of the results. . Experimental Apparatus A Perkin-Elmer Model 380 atomic absorption spectrometer equipped with a deuterium-arc background corrector a labora- tory-built cold vapour system operating in batch mode and a Beckman Lin-Log 10 in strip chart recorder were used throughout. A Hg hollow cathode lamp (Perkin-Elmer) was operated at 6 mA and the analytical wavelength was 253.7 nm selected using a spectral bandpass of 0.6 nm.Absorbance was measured in peak height mode. The laboratory-built cold vapour system is illustrated in Fig. 1 and consists of a cylindrical borosilicate glass reaction vessel that is conical shaped at the bottom (35 ml volume) (3) and is supplied with a 24/29 ground joint connector with two outlets. One outlet is a borosilicate glass T-tube which enters into the reaction vessel ending in a conically shaped tip near the bottom. One side of this T-tube is connected through silicone rubber tubing to a peristaltic pump (Minicor Digital Controller Lyon France) (2) for the addition of the reductant solution (1) into the vessel. On the other side a controlled flow rate (100 ml min-') of carrier gas (mercury-free air) (4) is introduced continuously into the solution and interrupted only upon the addition of the reductant.The carrier gas stream leaves the reaction vessel via a side arm connected by a short piece of Tygon tubing to a 90 x 3.3 mm (i.d.) glass tube filled with desiccant [dry Mg(C104),] (5). This tube is connected on the other side to a 70 x 3.3 mm (id.) quartz tube stocked with an Au-Pt gauze (90% Au-10% Pt) (Perkin-Elmer Part No. 11 1489) ( 6 ) for collection and preconcentration by amalga- mation of Hg vapour. This tube outlet is connected uia a short piece of Tygon tubing to the absorption cell. A T-shape open ended glass tube [ 170 mm (1) x 12 mm id. (S)] aligned in the optical path of the spectrometer is used at room temperature as is the absorption cell.For the transfer of the collected Hg to the gas flow and into the absorption cell the quartz tube is wound externally with a coil of Nichrome wire (22 gauge wound approximately 20 turns per 20 mm electrical resistance 2.8 a) (7) and is connected to a variable transformer (Variac 0-220V a.c.). The quartz tube is heated to 700°C until the Hg vapour is released. The temperature measurements were carried out with a Pt-Pt/Rh (10%) thermocouple (calibrated in the range 100-1500 "C) and a digital micro-multimeter (Keithley DMM-17) by placing the thermocouple within the quartz tube in the Au-Pt gauze. The applied voltage in the Variac was measured with a DA-8604 Lutron Multimeter.The efficiency for Hg trapping was examined under optimized 8 1253.7 nm ==i===l- A Fig. 1 Cold vapour atomization system for Hg determination by AAS 1,10% m/v SnC1 in 30% v/v HC1; 2 peristaltic pump; 3 reaction vessel with 24/29 ground joint connector; 4 carrier gas stream (Hg-free air); 5 dry Mg(ClO,),; 6 Au-Pt gauze for collection of Hg; 7 coil of Nichrome wire; and 8 absorption cellJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY conditions by placing a second Au-Pt gauze after the first. Although the trapping efficiency of the second gauze was not established experimentally when two subsequent measurement cycles were performed with 20 ng of Hg using both gauzes in tandem configuration the absorption signal was obtained from the first gauze whilst the second one did not trap Hg in detectable amounts.This observation was consistent with the results obtained in the recovery and accuracy studies con- firming the efficient Hg trapping with a single Au-Pt gauze. The quartz tube was cooled after each measurement cycle by use of an external flow of clean dry air for 3 min. After this cooling period the gauze temperature was < 100 "C and a new measurement cycle could be started. No variation in the electrical resistance of the Nichrome wire was noticed despite continuous oxidation of its surface by atmospheric oxygen. Reagents All reagents were of analytical-reagent grade ( Merck Darmstadt Germany). The acids used were HNO (65% m/m maximum Hg content 0.0000005%) and HCl [purified using a quartz sub-boiling still (H.Kurner Rosenheim Germany)]. The reductant (SnCl solution) was prepared by dissolution of 10 g of SnC1,*2H2O (maximum Hg content 0.000001%) in 30 ml of concentrated HC1 and dilution to 100 ml with ultrap- ure water (Milli-Q system Millipore Bedford MA USA). The solution was aerated for 30min with Hg-free air to minimize the Hg content. A standard solution of 1000mgl-' of Hg2+ was prepared from a Titrisol concentrate and a 50mg1-' standard stock solution of Hg2+ in 5% v/v HN03 was prepared by dilution of the former. Working solutions were prepared daily by appropriate dilution of the 50mg1-1 stock solution in 5% v/v HNO,. The carrier gas for the determination was compressed and filtered air pre-purified by passing it through a gas wash- bottle filled with copper strands to remove traces of Hg.No difference was found in baseline levels when pure nitrogen (99.98%) was purified in a similar manner and used as car- rier gas. Sampling and Pre-treatment Considering the morphological structure and chemical com- position of hair it is unlikely that trace elements will be evenly distributed in it and there are regions which incorporate them preferentially (ie. disulfide and isodipeptide bridges found in the cortex and medulla respectively of the hair follicle).26 Before a hair sample is taken it is essential to decide the aim of the analysis. When making a diagnosis or monitoring a trend of treatment only hair follicles in the anagen (i.e. growing) development phase should be used otherwise the results can be quite inc~nsistent.~~ According to B e n ~ z e ~ ~ the segments of hair follicles that are still biologically active are representative of different time periods from those which are in the catagen (Le.transition) and telogen (ie. dead) phase. However such selective sampling is unnecessary in epidemiol- ogical investigations where hair analysis is aimed either at monitoring the environment or place of work for specific elements,26 or in non-occupational exposure assessment of populations. Hair offers a good way of discerning long-term variations in trace element concentrations and this can be done by measuring the variation along the length of long hair equivalent to several months or by taking samples periodically.28 In this work a United Nations Environmental Programme (UNEP)-WHO-International Atomic Energy Agency (IAEA) procedure4 with some modifications was adopted for sampling and sample preparation.Approximately 2 g of human hair sample (strands no longer than 10 cm) were collected from the APRIL 1994 VOL. 9 537 occipital region of the head. Hair was cut with stainless-steel scissors as close as possible to the scalp placed with the proximal ends on the same side into a numbered polyethylene bag and the proximal ends were identified by a line drawn outside of the bag to enable later sectional analysis in case the results indicated high exposure. In particular the segment of hair between 0 and 5 mm from the scalp provides information concerning the last 2-3 weeks before corresponding to the most recent Hg exposure. In the case of pregnant women at the moment of delivery hair strands of 9 cm length from the proximal end can provide a record of Hg variations during pregnan~y,~ and indirect information of the exposure experi- enced by the foetus.Normally 85% of the hair is in the anagen phase in healthy hair;26 therefore chronological characteriz- ation is possible for specimens obtained as close as possible to the scalp discarding hairs longer than 10 cm since Hg concen- trations at the far end are influenced for example by repeated washing and application of cosmetic preparations. To remove surface contamination with minimal leaching of internal material different procedures for washing hair samples have been However it appears from studies of the removal of elements on washing that Hg is largerly retained in the hair and not lost.2,27*32 In the present study a fraction of the hair (about 70-80% of the sample) was cut into segments of 2-5 mm in length transferred into a Pyrex bottle and washed by the IAEA recommended pr0cedu1-e.~~ Each 10 min washing period was divided into 5 min of manual shaking and 5min of ultrasonic cleaning.The chopped hair was cleaned successively with 99.5% acetone (25 ml) three portions of ultrapure water (25 ml each) and again acetone with the wash liquid being decanted off after each step; the hair was dried at room temperature for 24 h in a dust- and Hg-free area within a clean hood protected from draughts and was then transferred into numbered clean polyethylene bags. No further sample homogenization was attempted between the first acetone and first water washings because no reliable homogenization technique was available in this laboratory.Sample Digestion Two sample digestion procedures were studied for Hg-T determination in hair. Acid digestion in a poly(tetrafluor0- ethylene) (PTFE) bomb27 (Procedure 1) was tested first with several biological CRMs in addition to hair and was applied to some molluscs and fish samples too. Acid digestion in a sealed Pyrex arnp0ule~~7~~ (Procedure 2) was tested with human hair samples samples spiked with Hg and CRMs. Procedure 1 Hair ( z 100 mg) was placed in a 23 ml PTFE digestion bomb (No. 4745 Parr Instruments Moline IL USA) and digested in 2.5 ml of HNO under pressure for 1.5 h at 110 "C. After a 2 h cooling period the bomb was carefully opened and the clear digest was transferred into a 25 ml calibrated flask and diluted to volume with ultrapure water.Procedure 2 Digestion under pressure in a sealed Pyrex ampoule35 (volume 14ml) was carried out by addition of 2ml of HN03 to the weighed hair sample ( ~ 1 0 0 m g ) placed in the Pyrex vessel. The vessel was sealed and after standing for 30min it was placed into a sand-bath at 50 & 10 "C and left to react for 24 h. After this digestion period the sample was completely dis- solved. The ampoule was removed from the sand-bath cooled to room temperature and washed with copious amounts of ultrapure water to remove external contamination. After a drying period at room temperature the ampoule was cooled with liquid nitrogen (for about 5 min) notched with a diamond538 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY APRIL 1994 VOL.9 knife and opened. The digest was quantitatively transferred into a 25 ml calibrated flask and diluted with ultrapure water. Extreme precautions must be taken with the materials and reagents used if reproducible results for Hg at low concen- trations are to be achieved. The Pyrex digestion vessels and all glassware used in this determination of Hg must be carefully cleaned by steaming for 1 h in HNO vapour (e.g. within a large beaker covered with a glasswatch and containing a special glass-made support to hold the glassware to be cleaned) followed by several washings with ultrapure water and drying within a clean hood. Analytical Techniques and Procedures The CVAAS technique with a reduction step using SnCl aeration and preconcentration of the Hg vapour by amalga- mation in the Au-Pt gauze was used in the determination of Hg-T in scalp hair The INAA method was used as an independent method for external analytical quality control of the results obtained by CVAAS.CVAAS An aliquot (approximately 0.5-1.0 ml) of Hg standard (in 5% v/v HNO,) or sample solution was introduced into the reaction vessel with a micropipette and ultrapure water was added to give a volume of x10ml. The vessel was connected to the system and the reductant [2 ml of 10% SnCl solution in 30% v/v HCl (Suprapur)] was added with the use of a peristaltic pump. The carrier gas caused rapid mixing of the sample with the reducing solution and the Hg vapour was transferred by the carrier gas to the amalgamation unit for collection.After 3 min of collection the Au-Pt gauze was heated rapidly to remove Hg and the carrier gas stream transferred the Hg vapour to the absorption cell for measurement. Quantification was performed by the standard additions approach because sample matrix effects were not fully matched by the composition of the standard working solutions. Samples from separate sub-samples were analysed in replicate. The Hg content was assessed in duplicate reagent blanks and was subtracted from the sample contents. The Hg-T content based on dry mass is reported as meankstandard deviation (SD). Two separate sub-samples were dried at 110 "C for 1 h or to constant mass and the mean percent loss of mass measured was used to calculate concentration on a dry mass basis. A blank and a control sample were digested and analysed with each sample batch as part of the internal analytical quality control.Samples giving signals beyond the linear working range were diluted in 5% v/v HNO before continuing the analytical procedure. INAA About 100mg of hair sample packed in a quartz vial were irradiated in a 5 MW nuclear reactor RECH-1 (Centro de Estudios Nucleares La Reina Chilean Nuclear Energy Commission Santiago Chile) for 24 h at a thermal neutron flux of approximately 0.3-1.0 x lo1 n cmP2 s-'. After a decay period of 3 weeks the irradiated samples were counted using gamma ray spectrometry with a Ge( Li) detector coupled to a 4096 multichannel analyser. The measurement was performed on the 203Hg photopeak at 279.1 keV and the interference produced by the 75Se photopeak in this gamma emission was corrected by calculating the relation between this emission and the emission at 265 keV.Standard solutions of Hg [in the form of Hg(N03)2] were prepared and used as comparators by dropping appropriate solutions on filter-paper impregnated with 10% m/v aqueous solution of thioacetamide. Results and Discussion Optimization Several factors affecting the determination of Hg by CVAAS were assessed and established previously in the laboratory- built system {i.e. cell geometry (35 ml) final solution volume (10 ml) trapping time (180 s) type of desiccant [dry Mg(ClO,),] and the efficiency (x 100%) of the Au-Pt gauze). In the present work three parameters were optimized the aeration gas flow rate the de-amalgamation temperature and the releasing time of Hg vapour.The air flow rate was studied between 50 and 300 ml min-' at a de-amalgamation temperature of 700°C and a releasing time of 19 s. As shown in Fig. 2 the effect of gas flow rate on the peak height absorbance of Hg indicate constant and relatively higher sensitivity between 50 and 100 ml min-'. A gradual decrease in sensitivity was observed between 100 and 300 ml min-l which is consistent with the fact that a low gas flow rate supports an increase in peak height absorbance due to relatively lower dispersion of Hg vapour in the air stream. A gas flow rate of 100 ml min-' was selected to keep efficiency and repeatability of the Au-Pt gauze relatively constant for approximately 100 heating cycles with pure Hg solutions.Nevertheless deactivation of the Au-Pt gauze was observed in particular with samples of some biological matrices (e.g. seafood). In such cases the Hg absorption signals became lower and erratic possibly due to purgable organic species adsorbing on the gold gauze,36 and thus affecting both the trapping efficiency for Hg and its release. No attempts were made to measure integrated absorbances because these signals were similar in shape to the Hg standards but inconsistent in peak height. These effects disappeared and the sensitivity was restored when the quartz tube with the Au-Pt gauze was first cleaned with hot 20% v/v HNO and washed with ultrapure water followed by a 30 min drying period at 110 "C in an oven; and second when the tube was heated twice to 700°C under optimum gas flow rate conditions.The temperature of the Au-Pt gauze measured with a Pt-Pt/Rh( 10%) thermocouple was varied as a function of the applied voltage (Variac) and was related both to peak height absorbance (at 10 and 20ng Hg) and time elapsed between setting up of the resistance heating and achievement of maxi- mum absorbance (releasing time) for Hg (20 ng). The results are given in Fig. 3. It is clear that a de-amalgamation tempera- ture between 600 and 800°C and a time between 28 and 15 s were required to achieve complete release of 20 ng of Hg from the trap. A temperature of 700°C was found to be suitable to maximize the absorption signal without damaging the efficiency of the Au-Pt gauze and a releasing time of 19 s was necessary to attain maximum peak height absorbance at this temperature.At this temperature using air as carrier gas the rate-limiting factor for the release of Hg from the gauze is probably the diffusion out of the amalgam.37 Higher temperatures should be avoided as they lead progressively to constriction of the Ov30 * $ 0.15 Q 0.10 0 50 100 150 200 250 300 350 Air flow rate/ml min-' Fig. 2 Effect of air flow rate on peak height absorbance of A 10 and By 20 ng of HgJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY APRIL 1994 VOL. 9 539 0.30 0.25 a 5 0.20 -e a 0.10 0 0.15 I] 0.05 in Hg for 100 mg of hair dissolved in 25 ml. Hence the limit of quantitation ( 10ab) was estimated as 0.43 ng (uiz. 0.11 mg kg-' in Hg). The accuracy and precision of the method was assessed with several biological and environmental National Institute of Standards and Technology (NIST) Standard Reference Materials (SRMs) and CRMs.As shown in Table 1 the results obtained by Procedure 1 agreed quite well with the certified values and in particular the results obtained in three human hair CRMs (between 2.16-12.30 mg kg-' in Hg) with both digestion procedures were comparable and consistent with the certified values. The mean relative error was 4.0+2.7% and the mean imprecision was 8.7&3.7% RSD. Moreover the results obtained by Procedure 2 in two 'blind' reference mate- rials (copepoda and fish homogenate) delivered in the first external quality control exercise of a Coordinated Research Programme (CRP)4 conducted by IAEA in this CRP were consistent with the target values (Table 1).The efficacy of the two digestion procedures was compared on National Institute of Environmental Studies (NIES) RM No. 5 and Community Bureau of Reference (BCR) CRM 397 and as shown in Table 1 these results were in good agreement with the certified value digestion in sealed Pyrex ampoules being slightly more accurate and precise. The latter is much simpler and requires inexpensive reactors instead of the classic and expensive PTFE bombs which are almost always available in limited amounts. Besides this procedure is more amenable for routine application since digestion can be carried out for several tubes at a time within a small sand-bath or heating block. Furthermore recovery of Hg in two undigested hair samples of known and low Hg content was investigated for Procedure 2.The undigested samples were spiked with known amounts of Hg added as an Hg(NO& solution to the digestion vessels subjected to digestion by Procedure 2 and analysed. As shown in Table 2 the results indicate a mean recovery of 105% for relatively low Hg content (0.83-1.27 mg kg-') which is con- - - - - - - 0.35 I - - - - 30 25 -$ E 20 'E 0 15 . ((J a K -10 5 - 5 U 0 1 I I I I I 1 I 10 550 600 650 700 750 800 850 900 950 TemperaturePC Fig.3 Effect of de-amalgamation temperature on peak height for A 10 and B 20ng of Hg. C Effect of de-amalgamation temperature on releasing time for 20 ng of Hg metal fibres in the gauze reducing on one side the surface area for amalgamation and its trapping ~apacity.~' Characterization and Validation of Analytical Methodology A typical calibration curve in CVAAS obtained by linear regression corresponded to A = 0.0153MH - 0.0012; r2 = 0.9998 (4 =peak height absorbance; M = mass of Hg in ng) and the linear working range extended between 0 and 12.5 ng of Hg in 5% v/v HN03.The reciprocal sensitivity obtained from the working curve was 0.29 ng ml-' (peak height absorbance 0.004). The repeatability of measurements (within-run vari- ation) expressed as RSD% varied between 3.0% (1.25 ng of Hg) and 7.5% (12.5ng of Hg) and the reproducibility (between-day variation) was 6.4 1.1 YO estimated from cali- bration curves obtained with a blank 2.5 5.0 10.0 and 12.5 ng of Hg in five consecutive days. The detection limit ( 3 4 obtained in 14 blank runs alternated with 0.625 and 1.25 ng of Hg standards respectively corresponded to 0.13 kO.01 ng (95% confidence level) and was equivalent to 0.033 mg kg-' Table 1 Results for Hg-T (mg kg-l; mean$95% confidence limits) in several biological and environmental SRMs obtained by CVAAS and INAA; values of n given in parentheses. Experimental detection limits in biological materials 0.011 mg kg-l (CVAAS) and 0.03 mg kg-' (INAA); in hair 0.033 mg kg-I (CVAAS); quoted variation is one standard deviation Result Sample CVAAS INAA Certified value RSD (Yo) Error (YO) by CVAAS by CVAAS NIST RM 50 (Albacore Tuna) NIST SRM 1566 (Oyster Tissue) NIST SRM 1646 (Estuarine Sediment) IAEA MA-A-1 (Copepod) DORM-17 (Dogfish Muscle) IAEA SAMPLE No.1 (Biological material) IAEA SAMPLE No.2 (Biological material) NIESS RM No. 5 (Human hair) GBW 0910111 (HH Human hair) BCR CRM 397 (T.E. in Human hair) 0.93 k 0.04 ( 5 ) 1.06k0.05 (2) 0.032 k 0.003 (3) ND* 0.28 k 0.01 (3) ND 0.264k0.019 (2) 0.477 & 0.01 1 (2) 3.97 k 0.56 (3) ND 13.20$-0.11 (2) 0.95k0.10 4.3 -2.1 0.060 f 0.008 (6) 0.057f0.015 13.3 5.3 0.068 f 0.004 (6) 0.063 k0.012 5.9 7.9 0.30 & 0.02 (3) 0.28 k 0.01 6.7 7.1 0.746f0.110 (4) 0.798 0.074 14.7 - 6.5 0.280 0.028 (6) 0.28 f 0.01 10 0 0.479 f 0.053 (8) 0.47 k 0.02 11.1 1.9 4.19f0.41 (5)§ 4.40 f 0.29 (5)7 2.12f0.10 (5R 4.4 f 0.4 4.4 f 0.4 2.16 0.21 9.8 6.6 4.7 - 4.8 0 - 1.9 11.59f 1.52 (2)§ 11.67k0.54 (3)TI 12.30 f 0.50 12.30&0.50 13.1 4.6 - 5.8 - 5.1 * ND =not determined. -f Marine Analytical Chemistry Standards Program (National Research Council Canada). $ NIES =National Institute for Environmental Studies Japan Environmental Agency. 5 Acid digestion in Parr Bomb.7 Wet digestion in sealed Pyrex ampoules. 11 Shanghai Institute of Nuclear Research Shanghai China.540 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY APRIL 1994 VOL. 9 Table2 pyrex ampoules and CVAAS; values of n given in parentheses Recovery of Hg-T in scalp hair by wet digestion in sealed Mean & SD Added Found* Recovery (YO) 0.83+0.11 (8) 0.54 0.5710.10 105.5 1.27 1 0.30 (5) 0.90 0.94 10.13 104.4 * Average of duplicates. sistent with the accuracy obtained previously at higher concen- trations and comparable with Hg recoveries from human hair digest samples at higher relative concentration^.^^ Between- day variation in hair analysis was assessed from 16 independent analyses conducted on a sample of pooled scalp hair over a period of 5 months.With a mean concentration of 1.10 mg kg-' and an SD of 0.098 mg kg-' the 95% confidence limits were 1.10f0.053 mg kg-' and the RSD was 4.8%. Sub-sample Homogeneity and Intra-laboratory Variability Despite the limitations in the present work for further homo- genization of the hair samples the homogeneity of hair sub- samples and the possibility of using a laboratory-prepared hair control sample to assess analytical intra-laboratory variability were investigated. The former was estimated by comparison of RSDs(%) obtained in independent determinations of Hg-T for several sub-samples from two laboratory-prepared human head hair samples available in large amounts with the RSD(%) found when independent sub-samples of human hair RM NIES No.5 were analysed in a similar manner. Intra-laboratory variability was estimated by batch-to-batch comparison of results of control hair sub-samples included within each sample batch with its mean value. Hence two human head hair samples obtained in reasonably large amounts from two normal unexposed volunteers were prepared as indicated above (i.e. not powdered but cut into small segments). In eight different days several sub-samples of both materials were dissolved by Procedure 2 and analysed repeatedly by CVAAS. As is shown in Table 3 samples A and B have practically the same Hg-T content by CVAAS and results obtained in tripli- cate by INAA were in close agreement with those achieved with CVAAS.The variability in the Hg-T content [RSD(%)] indicate a maximum imprecision of 14% in sample A and a better precision of 6.8% in sample B. The latter result is quite satisfactory as it is consistent with the imprecision obtained previously (6.6%) with NIES RM No. 5 the latter at an approximately 4-fold higher concentration. Despite similar Table 3 Assessment of intra-laboratory variability in the determi- nation of Hg-T by CVAAS Day No.* 1 2 3 4 5 6 7 8 Mean+95% confidence interval RSD (%) Sample A$ 0.99 0.74 0.75 0.73 0.78 0.8 1 0.99 1.09 0.86 & 0.12 13.95 Sample B§ 0.76 0.86 0.86 0.97 0.88 0.89 0.91 0.88 & 0.06 6.82 * Within a one month period. t Wet digestion in sealed Pyrex ampoules. $ Results for INAA 0.89k0.15 mg kg-l; n=3.3 Results for INAA 0.99 1 0.20 mg kg- '; n = 3. Hg-T content in both samples the relatively large difference in RSD(%) can be attributed in part to the difficulty of achieving good homogenization in sample A during prep- aration. Sample B was considered adequate for intra- laboratory variability assessment and is used successfully for this purpose. External Analytical Quality Control As shown in Table 1 comparable results were obtained by CVAAS and INAA for Hg-T in several biological and environ- mental SRMs (Oyster Tissue NIST SRM 1566 Albacore Tuna RM-50) and CRMs (Copepod MA-Al/TM) two human hair RMs (NIES No. 5 and BCR CRM 397) and two biological specimens sent as 'blind' samples by IAEA for analytical quality control exercise within the CRP.Upon application of the t-test for differences between two means there were no significant differences 0 < 0.05) in the results of both methods. By regression analysis a correlation coefficient of 0.993 was established. The mean detection limit for Hg-T determined in several biological RMs was estimated as 0.02 mg kg-' in Hg. Application to Human Hair In order to test the applicability of the proposed method to the analysis of real samples several scalp hair samples of healthy pregnant and nursing women aged 16-38 resident in fishing villages of the coastal zone of the Eighth Region of Chile were analysed in duplicate. A comparison was made between the results obtained by CVAAS and INAA as a reference method (Table 4). There was a significant correlation between hair Hg content obtained by CVAAS and INAA (mean sfr SD 2.813 sfr 2.050 and 2.635 f 1.910 mg kg-I in Hg respectively) [ Y(CVAAS) = 1.007 x (INAA) + 0.160 r2 = 0.880 n = 21 p < O.OOOl] and no significant differences were estab- lished by paired t-test (mean difference = 0.178 standard error = 0.155 t calculated = 1.152 p = 0.2628).Some of the larger differences between CVAAS and INAA results (e.g. samples 040-91 062-91 and 016-92) may be attributable to Hg losses which occur either during the rather long irradiation step (24 h) which is followed by a cool-down period of 3 weeks to let the shortlived radionuclides decay or during the opening of the quartz vial to transfer the irradiated sample to a new polyethylene vial for the counting step. Also the inter- Table4 Results of Hg-T in human scalp hair obtained by CVAAS and INAA n = 2 Hg-T/mg kg - Sample 001-91 002-9 1 004-9 1 009-9 1 012-91 025-91 039-91 040-9 1 042-9 1 058-91 060-9 1 062-9 1 099-9 1 101-91 165-91 179-9 1 252-91 256-91 004-92 006-92 016-92 CVAAS 3.53 3.78 4.65 4.44 2.55 2.87 2.50 2.50 2.13 1.95 2.84 2.49 1.11 1.21 0.43 0.42 3.63 3.99 4.57 4.31 7.03 6.79 6.39 6.85 4.22 4.51 4.57 4.68 0.73 0.69 0.14 0.15 0.66 0.92 0.32 0.34 4.30 4.50 1.57 1.29 0.76 0.78 INAA 3.37 4.12 3.93 4.06 2.67 2.24 3.26 3.05 2.26 2.15 3.10 2.36 0.98 1.08 0.23 0.28 3.35 3.76 4.02 3.99 6.23 6.35 3.14 4.49 4.68 4.81 4.57 4.67 0.41 0.40 0.08 0.07 0.47 0.70 0.28 0.26 5.64 5.21 1.34 1.52 0.48 0.53JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY APRIL 1994 VOL.9 541 ference of the 279.6 keV photopeak of 75Se on the 279.1 keV photopeak of '03Hg which is corrected for by the ratio between the 75Se emission at 279.6 keV and 265 keV could have been over- or under-estimated (e.g.sample 004-92) depending on the Hg level since the Se levels measured by INAA in these samples were relatively low. Last but not least the homogeneity of samples prepared by cutting hair in segments of 2-5 mm is limited and should not be disregarded as a source of variability in results. The present CVAAS methodology is being used to assess Hg levels in hair of pregnant and nursing women residents in coastal areas of the Eighth Region of Chile as part of a study to evaluate the exposure of selected human groups to Hg. Conclusions The determination of Hg-T in human hair by CVAAS was achieved with a laboratory-built cold vapour system when suitable sampling washing digestion and analytical procedures were followed.In particular complete sample dissolution in sealed Pyrex ampoules at 50k 10 "C in a sand-bath proved to be reliable amenable for large number of samples and simpler in comparison to the classic digestion in PTFE bombs. The relatively long sample digestion period (24 h) is compensated for by the unlimited number of samples that can be dissolved at once. According to the results obtained in homogeneity and intra-laboratory variability tests reasonably good precision (6.8-13.9% RSD) was attained by this procedure in hair samples with sub-ppm Hg-T content despite the use of hair cut into small segments (2-5 mm in length) rather than pow- dered hair. Good agreement in results was obtained between CVAAS and INAA both for several biological and environmen- tal SRMs and CRMs including human hair (0.06-12.3 mg kg-' in Hg) as well as for real human hair samples (0.1 and 6.9 mg kg-' in Hg).This study was supported by the International Atomic Energy Agency under research contract No. 6331/RB and by the Direcci6n de Investigacih Universidad de Concepcion Concepcih Chile (Project 4.91.71.001). The authors are grate- ful to Dr. M. Horvat (Josef Stefan Institute Ljubljana Slovenia) €or helpful suggestions and discussions and to Dr. K. Okamoto (National Institute for Environmental Studies 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 WHO Environmental Health Criteria 101 -Methylmercury World Health Organization Geneva 1990 pp.100-103. Horvat M. Prosenc A. Smrke J. and Liang L. Working paper presented at the Second Research Coordination Meeting (RCM) Coordinated Research Programme on Assessment of Environmental Exposure to Mercury in Selected Human Populations as Studied by Nuclear and Other Techniques Report on the Second RCM NAHRES-13 IAEA's Section of Nutritional and Health-Related Environmental Studies Department of Research and Isotopes Vienna Austria 1992 p. 65. Taylor A. Ann. Clin. Biochem. 1986 23 364. Airey D. Sci. Tot. Environ. 1983 31 157. Dermelj M. Horvat M. Byrne A. R. and Stegnar P. Chemosphere 1987 16 877. Zmijeska M. J. Radioanal. Chem. 1977 35 389. Shull M. and Winefordner J. D. Anal. Chem. 1971 43 799. Hatch W. R.and Ott W. L. Anal. Chem. 1968 40 2085. WHO Environmental Health Criteria 101-Methylmercury World Health Organization Geneva 1990 pp. 19-23. Horvat M. May K. Stoeppler M. and Byrne A. R. Appl. Organornet. Chem. 1988 2 515. Horvat M. Working paper presented at the First Research Coordination Meeting (RCM) Coordinated Research Programme on Assessment of Environmental Exposure to Mercury in Selected Human Populations as Studied by Nuclear and Other Techniques Report on the First RCM NAHRES-7 IAEA's Section of Nutritional and Health-Related Environmental Studies Department of Research and Isotopes Vienna Austria 1991 p. 93. Poluektov N. S. and Vitkun R. A. Zh. Anal. Khim. 1963 18 33 (Russian). Poluektov N. S. Vitkun R. A. and Zelyukova Y. V. Zh. Anal. Khim. 1964 19 873 (Russian).Kaiser G. Gotz D. Tolg G. Knapp G. Maichin B. and Spitzy H. Z . Anal. Chem. 1978 291 278. Kaiser G. Gotz D. Schoch P. and Tolg G. Talanta 1975 22 889. Ulfvarson U. Acta Chem. Scand. 1967 21 641. Matsunaga K. and Takahashi S. Anal. Chim. Acta 1976,87,487. Yamamota Y. Kumamaru T. and Shiraki A. Z . Anal. Chem. 1978 292 273. Welz B. Melcher M. Sinemus H. W. and Maier D. At. Spectrosc. 1984 5 37. Bencze K. Fresenius' J. Anal. Chem. 1990 337 867. Bencze K. Fresenius' J. Anal. Chem. 1990 338 58. Chatt A. and Katz S. A. Hair Analysis. Applications in the Biomedical and Environmental Sciences VCH Publishers New York 1988. Ibaraki Japan) and Dr. C. Chifang (Institute of High Energy Physics Academia Sinica Beijiing China) for kindly supplying NIES NO. 5 Human Hair and GBW 09101 Reference Materials respectively.125 131. 29 Assarian G. S. and Oberleas D. Clin. Chem. 1977 23 1771. 30 Chittleborough G. Sci. Tot. Environ. 1980,14 53. 31 Salmela S. Vuori E. and Kilpio J. O. Anal. Chim. Acta 1981 32 Dermelj M. Horvat M. Byrne A. R. and Stegnar P. References Friberg L. and Vostal J. in Mercury in the Environment eds. Friberg L. and Vostal J. Chemical Rubber Company Press Cleveland 1972 pp. 183-186. Airey D. Environ. Health Perspect. 1983 52 303. Clarkson T. W. in Trace Elements in Human Health and Disease eds. Prasad A. S. and Oberleas D. Academic Press New York Coordinated Research Programme on Assessment of Environmental Exposure to Mercury in Selected Human Populations as Studied by Nuclear and Other Techniques Report on the First Research Coordination Meeting NAHRES-7 International Atomic Energy Agency Section of Nutritional and Health-Related Environmental Studies Department of Research and Isotopes Vienna Austria 1991 pp. 1-14. Gorchev H. G. in The Food Laboratory Newsletter eds. Hofsten B. Blomberg-Johansson B. and Vaz R. Swedish National Food Administration Uppsala 1987 No. 10 pp. 31-36. Phelps R. W. Clarkson T. W. Kershaw T. G. and Weatly B. Arch. Environ. Health 1980 35 161. 1976 V O ~ . 11 pp. 464-470. Chemosphere 1987 16 877. Ryabukhin Yu. S. Activation analysis of Hair as an Indicator of Contamination of Man by Environmental Trace Element Pollutants. Report IAEA/RL/50. International Atomic Energy Agency Vienna 1978 p. 1. 34 Horvat M. Zvonaric T. and Stegnar P. Vestn. Slov. Kem. Drus. 1986 33 475. 35 Horvat M. Lupsina V. and Pihlar B. Anal. Chim. Acta 1991 243 71. 36 Schroeder W. H. Hamilton M. C. and Strobant S. R. Rev. Anal. Chem. (Tel-Aviv) 1985 8 179 and refs. cited therein. 37 Baeyens W. and Leermarkers M. J. Anal. At. Spectrom. 1989 4 635. 38 Kaiser G. Gotz D. Tolg G. Maichin B. and Spitzy H. Fresenius' Z . Anal. Chem. 1978 291 278. 39 Pineau A. Piron M. Boiteau H. L. Etourneau M. J. and Guillard O. J. Anal. Toxicol. 1990 14 235. 33 Paper 31048201 Received August 10 1993 Accepted November 23 1993

ISSN:0267-9477    

DOI:10.1039/JA9940900535

出版商:RSC    

年代:1994

数据来源: RSC

摘要:

JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY APRIL 1994 VOL. 9 543 Determination of Silver in Waters and Soil by Electrothermal Atomic Absorption Spectrometry After Complexation and Sorption on Carbon Akie K. Avila Departamento de Quimica Analitica da Universidade Federal do Rio de Janeiro llha do Fundiio Rio de Janeiro Brazil Adilson J. Curtius* Departamento de Quimica da Pontificia Universidade Catolica do Rio de Janeiro Rio de Janeiro Brazil A method for the preconcentration of silver is proposed. Silver is complexed by the ammonium salt of dithiophosphoric acid 0,O-diethyl ester and sorbed on to carbon. After desorption in a small volume of nitric acid solution silver is determined by electrothermal atomic absorption spectrometry. In de-ionized water the yield was about 70% indicating incomplete complexation and/or sorption on the carbon.In purified sea- water (sea-water submitted to solvent extraction to extract silver) the yield was less about 40% probably due to the effect of concomitants in the sea-water. The method was applied to sea and river waters and to a soil reference material. Spiked sea-water samples produced an average recovery of 102% indicating good accuracy if analytical solutions in purified sea-water are used for the calibration. The detection limit (k=3) for 200 ml of sea-water is 0.3 ng I-'. After acid dissolution of the soil sample ascorbic acid should be added prior to the complexing agent to avoid interference of iron(iii). Because of the small final volume high enrichment factors can be obtained. Keywords Silver determination; carbon; dithiophosphoric acid 0,O-diethyl ester; atomic absorption spectrometry; sea and river water In order to determine very low concentrations of silver in samples such as natural waters and alloys several preconcen- tration procedures applied before determination by atomic absorption spectrometry (AAS) have been proposed.Kinrade and Van Loon' used solvent extraction with isobutyl methyl ketone after complexation with a mixture of ammonium pyrrolidine- 1-yl-dithioformate and diethylcarbamic acid for several elements including silver. The best pH range for silver was 4.5-6.0. Re-extraction of the complexes with nitric acid solution increased the stability of the metals in solution according to Jan and Young.2 These workers found a limit of detection (LOD) of 0.02 pg 1-' for silver using electrothermal atomic absorption spectrometry (ETAAS) after re-extraction.Adsorptive preconcentration procedures were also proposed. Silver and other noble metal ions were separated by Kenawy et aL3 from aqueous solutions using cellulose-hyphan as an ion-exchanger. Preconcentration of metals from acidic solu- tions on a thin layer of activated carbon using dithizone as chelating agent was accomplished by Beinrohr et aL4 The metals were extracted from the carbon with 14 mol 1-' nitric acid and determined by ETAAS. The reported LOD for silver was 1.6 pg I-'. Yang and Jackwerths used a resin Amberlite XAD-4 instead of activated carbon. About 15 elements could be enriched using different complexing agents. Performing coprecipitation with cobalt pyrrolidin-1-yl- dithioformate and detection by ETAAS Bloom and Crecelius' found an LOD of 0.1 ng 1-' for silver in sea-water using 200 ml of sample.Reductive coprecipitation with sodium tetra- hydroborate has been used by several The LOD for silver in sea-water was reported to be 0.7 ng 1-' using 900 ml of sample with detection by ETAAS.' The sodium salt of dithiophosphoric acid 0,O-diethyl ester (NaDDTP) was used by Bode and Arnswald'o~'' to complex a number of elements including silver. The complexes were extracted with carbon tetrachloride and several elements were measured by photometry. The DDTP complexes are stable in * To whom correspondence should be addressed. Present address Departamento de Quimica da UFSC 88040-900 Florian6polis S.C.Brazil. strong mineral acid solution which is a real advantage in many analytical applications. Separation of several trace elements from high-purity gallium and aluminum12 and from high- purity iron chromium and rnangane~e'~ have been described. Complexes of trace amounts of the impurities were formed in the acid solutions of the samples and filtered through a small filter-paper covered with a layer of activated carbon. After treating the carbon with a nitric acid solution the trace impurities were determined by either discrete nebulization into the flame or by loop-flame AAS. A similar procedure was used by Monte and Curtius14 to preconcentrate molybdenum in solutions of biological and geological reference materials using either discrete nebulization into the flame AAS or ETAAS. In the present paper the determination of silver after com- plexation with NH4DDTP and sorption on to carbon is studied.High enrichment factors could be expected due to the small final separation volume. Another advantage of this complexing agent is that several components of natural samples such as aluminium barium calcium iron(n) mag- nesium manganese sodium lithium strontium etc. are not complexed.'O~l' Experimental Instrumentation A Perkin-Elmer Z 3030 atomic absorption spectrometer equipped with an HGA-600 furnace and an AS-60 Autosampler also from Perkin-Elmer was used. The back- ground was corrected using the Zeeman system. The conditions used were those recommended by the manufacturer unless otherwise stated. Argon 99.9% was used as purge gas.All measurements were recorded as integrated absorbance. Usually 10 p1 of the solution was pipetted into the graphite tube. No chemical modifier was used. The temperature programme is given in Table 1. A pyrolytic graphite coated graphite tube Perkin-Elmer No. B 010-9322 with a totally pyrolytic graphite platform Perkin- Elmer No. 010-9324 was used.544 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY APRIL 1994 VOL. 9 Table 1 by ETAAS Temperature programme for the determination of silver Argon flow Step Temperature/"C Ramp/s Holdls rate/ml min- ' 1 90 10 10 300 2 120 1 15 300 3 600 1 30 300 4* 1500 0 5 0 5 2650 1 4 300 6 20 10 10 300 * Read in this step. A Varian-Techtron hollow cathode lamp for silver was operated at 8 mA. The wavelength was set to 328.1 nm.Chemicals and Materials All chemicals used were of analytical-reagent grade unless specified otherwise. The water was de-ionized. The following aqueous stock solutions were used. Silver and silicon. A 1000mg1-' solution of each was prepared from the respective Titrisol solution (Merck). Dithiophosphoric acid 0,O-diethyl ester ammonium salt 1 YO m/v. Prepared from commercial reagent No. 17779-2 (Aldrich 95% purity). Ascorbic acid 20 g I-'. Prepared from the Merck reagent. Iron 10 g 1-l. Prepared by dissolving the iron metal powder (Merck 99.5% purity) in a 1 + 1 solution of nitric acid. Carbon. Graphite RW-B from Ringsdorff-Werke. A 3% m/v suspension in water was prepared. Medium hard filter-paper was obtained from Schleicher and Schull (No.589/2) and used as a support for the carbon. The membrane filter was from Millipore 0.45 pm. Whatman 41 filter-paper was used. Samples and Pre-treatment The surface sea-water samples were collected around the coast of Rio de Janeiro. The river samples were from the Carajas Hill region Para State Brazil (Parallel 5" 45 S between merid- ians 51" 0 0 and 50" 0 0 W-GR). The water samples were filtered through a 0.45 pm Millipore membrane filter and acidified with nitric acid to pH 1.0. They were stored in polyethylene bottles at low temperature. About 0.2000g of Soil 5 from the International Atomic Energy Agency (IAEA) was digested with 15 ml of concentrated hydrochloric acid and 5 ml of concentrated nitric acid first at room temperature and then heated to 95 "C. After the evolution of NO2 fumes had ceased the mixture was evaporated close to dryness and mixed with about 40 ml of a solution containing 1 YO v/v nitric acid and 1 % v/v hydrochloric acid.The resulting mixture was filtered through a Whatman 41 filter-paper. The small residue in the filter-paper was washed with 5 ml of warm concentrated hydrochloric acid and then with 20ml of warm de-ionized water. The solutions were collected 5 mg of ascorbic acid were added in order to reduce iron(II1) to iron@) which is not complexed and the volume was made up to 100 ml. The digestion procedure was as proposed by Kimbrough and Wakamura." Preconcentration A filter-paper of 2.5 cm in diameter was covered with 300 mg of carbon by filtering 1Oml of a 3% m/v carbon suspension through it using a water vaccum system which was activated only after placing the suspension on the filter.A 1 ml aliquot of a 1% m/v solution of NH,DDTP was added to 100 or 200 ml of the water sample previously adjusted to pH 1.0 with nitric acid or to 100ml of the soil solution containing 5.0 mg of ascorbic acid. After stirring for 10 min with a magnetic stirrer the solution was passed through a filter covered with carbon. After drying at 110 "C for 20 min the carbon was scraped into a 25 ml beaker 1 ml of concen- trated nitric acid was added and evaporated to dryness. The residue was treated with 2ml of 4.5% v/v nitric acid at room temperature and centrifuged. Silver was measured in the super- natant solution. This procedure was based on those described in the literature for other element~.'~-'~ For the river waters and for the soil solution analytical solutions in de-ionized water at pH 1.0 were submitted to the same preconcentration procedure for calibration purposes.For sea-water samples analytical solutions in purified sea-water were also submitted to the same procedure. For the purifi- cation 2 ml of a 1% m/v NH,DDTP solution were added to 700 ml of a sea-water sample adjusted to pH 1.0. After stirring for lOmin the silver complex was extracted with 100ml of carbon tetrachloride. The silver concentration in the purified sea-water was below the LOD of the proposed method. Blanks were run parallel to all determinations. Results and Discussion Preconcentration The optimized conditions for the preconcentration were estab- lished using either 100ml of de-ionized water or 1OOml of purified sea-water enriched with 20 ng of silver and submitting these solutions to the preconcentration procedure.The effect of pH was studied by adding nitric acid to the initial solution. The silver absorbance in the final solution decreased slowly as the pH increased up to pH 3. For higher pH values the decrease was stronger; pH 1.0 was used in this work. After adding the complexing agent the solution was stirred for some time and then filtered. The time of stirring is important. The silver signal increased almost linearly with stirring time up to 5 min and from then on remained approxi- mately constant. A stirring time of 10min was chosen. After filtration the carbon containing the silver complex was treated with concentrated nitric acid and dried.The residue was suspended in nitric acid solutions of different concentrations. For nitric acid concentrations in the range 2-9% v/v the silver absorbance was almost constant. For lower concentrations lower signals were obtained. A nitric acid concentration of 4.5% v/v was chosen in this work. As shown in Fig. 1 at least 300mg of carbon on the filter are needed to obtained the highest signal either in the de-ionized water or in the purified sea-water. Fig. 2 shows the effect of the mass of NH,DDTP on the silver signal. While for de-ionized water 2mg are sufficient to obtain the maximum signal for sea-water at least 5 mg are needed. A mass of 10 mg of complexing agent was chosen in this work. Both figures 0.200 1 1 A I -e 23 o.150 0.100 t 1 0 100 200 300 400 500 Mass of carbon/mg Fig.1 Integrated absorbance of silver after the preconcentration procedure versus carbon mass A 100 ml of de-ionized water containing 20ng of silver (initial solution); and B 100ml of purified sea-water containing 20 ng of silver (initial solution)JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY APRIL 1994 VOL. 9 545 v) 0.200 0 3.0 6.0 9.0 12.0 15.0 Mass of NH,DDTP/mg Fig. 2 Integrated absorbance of silver after the preconcentration procedure versus mass of NH,DDTP. A and B as in Fig. 1 also show that the yield in sea-water is about 60% of that in the de-ionized water. The lower yield in the sea-water is discussed later. Pyrolysis and Atomization Temperature Curves Pyrolysis and atomization temperature curves are shown in Fig.3 for silver in 4.5% HN03 and for preconcentrated silver from enriched de-ionized water and enriched purified sea- water. As expected the curves are very similar indicating that pyrolysis temperatures of up to 600°C can be used without significant loss of silver for the three situations presented. No chemical modifier was used since after the preconcentration the matrix should be very simple. Pyrolysis and atomization temperatures of 600 and 1500 "C respectively were chosen. Yield of the Preconcentration Procedure For a 100% yield of silver after the preconcentration procedure curves A and B should coincide with curve C in Fig. 3. As shown in this figure only about 70% of silver in de-ionized water was recovered. This yield could be explained by partial complexation or most probably by incomplete retention on the carbon.The carbon used was spectrographic graphite which should have a limited number of active sites in compari- son with activated carbon. Attempts to use activated carbon from several manufacturers were not successful because of the high silver blanks obtained. The yield was even lower for the purified sea-water samples around 40%. Concomitants in the purified sea-water could eventually compete for the active sites of the carbon or even change the activity of the surface. As shown in Fig. 1 increasing the mass of carbon above 300 mg does not increase the silver absorbance which makes these hypotheses less likely. Another 0.250 * 0.200 a S 0 0.150 : a (21 0.100 c ?! f 0.050 m - 0 C A 200 400 600 800 1000 1200 1400 1600 Temperatu rePC Fig.3 Pyrolysis and atomization temperature curves for silver (pyrol- ysis temperature of 600 "C for the atomization temperature curves and atomization temperature of 1500 "C for pyrolysis temperature curves). A and B as in Fig. 1; C 0.1 ng of silver in 4.5% v/v HN03 not preconcentrated possibility would be the effect of preconcentrated concomitants on the determination of silver in the graphite furnace. A spectrographic analysis of the carbon after retention of the complex formed in the purified sea-water showed that only silicon was specially enriched 40 ppm while the pure carbon had a much lower silicon concentration around 0.4ppm. By submitting 100 ml of de-ionized water containing 20 ng of silver and different concentrations of silicon to the preconcen- tration procedure it was verified that the silver signal decreases almost linearly with silicon concentration.For the average silicon concentration in sea-water 2.8 ppm16 the drop was 20%. The appearance time of silver taken from the absorption pulses for preconcentrated sea-waters purified or not is around 0.74s which is larger than for the de-ionized water 0.59 s also indicating a matrix effect on the silver signal. In conclusion the effect of the silicon explains at least partially the lower yield for the sea-water samples in relation to the de-ionized water. The literature is not informative on the complexation of silicon by NH,DDTP but the formation of H,Si03 a colloidal precipitate at low pH could be ~0nsidered.l~ Since the yields are very reproducible the preconcentration procedure can be used for analytical purposes.The only disadvantage of the incomplete recovery is that the analytical solutions must be submitted to the same procedure. By using an initial sample volume of 200 ml and a final volume of 2 ml enrichment factors of 70 and 40 can be obtained for de-ionized and sea-waters respectively. Analysis of Waters In order to estimate the accuracy of the procedure the recovery test was applied to a sea-water sample. Different masses of silver were added to 100 ml sample aliquots and the resulting solutions were submitted to the preconcentration procedure. The results obtained by subtracting the absorbance of an unspiked sample submitted to the same procedure are shown in Table 2.An analytical curve in purified sea-water also submitted to the preconcentration procedure was used. The recoveries around loo% showed good accuracy. Fig. 4 shows a typical analytical curve in purified sea-water after the precon- centration procedure. This figure also shows a curve obtained Table2 Recovery of silver in spiked sea-water after pre- concentration procedure n = 3 Mass of silver Mass Recovery added/ng found/ng w.) 3.0 2.9 97f5 10.0 10.4 104 f 3 20.0 20.8 104 f 4 30.0 31.6 10514 0.300 B 0 0.10 0.20 0.30 [AgI/ng ml-' Fig. 4 A Analytical curve for silver in purified sea-water after precon- centration and B standard additions curve for silver in a sea-water sample after preconcentration546 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY APRIL 1994 VOL.9 by the standard additions method applied to a sea-water sample with the solutions also submitted to the preconcen- tration procedure. Since both curves are parallel matrix effects are compensated for thereby demonstrating that the cali- bration can be performed using analytical curves in purified sea-water. The LOD ( k = 3) for 200 ml aliquots of sea-water and a final volume of 2 ml is 0.3 ng 1-I. The reference sea water NASS-2 Open Ocean Seawater from the National Research Council Canada was analysed using 200 ml aliquots. The silver concentration found was 37 & 8 ng 1-l. Unfortunately the silver concentration is not certified in this material. The analysis of the coastal surface sea-water samples around the city of Rio de Janeiro Brazil showed silver concentrations in the range 13 & 5-20 3 ng l-' which agree with the concentrations in sea-water samples from the Tacoma Marine determined by ETAAS after a coprecipitation procedure.6 River waters were also analysed using in this case analytical solutions in de-ionized water submitted to the preconcentration procedure for calibration.Reference water from the National Institute of Standards and Technology NIST SRM 1643b Trace Elements in Water was diluted 1:250 and 100m1 as the initial volume was analysed. A silver concentration of 9.70 k 0.27 ng ml-I was obtained which compares well with the certified value of 9.8 k0.8 ng ml-I. Three water samples from rivers in the Carajas Hill Para State Brazil where copper mines are being explored were analysed. The silver concen- trations ranged from lo+ 3 ng 1-I (Itacai Unas River) to 44+ 5 ng 1-' (Salobo River closer to a mine).Analysis of a Reference Soil A reference soil Soil 5 (IAEA) was submitted to the preconcen- tration procedure after the acid dissolution described under Experimental. A higher than expected silver concentration was obtained when analytical solutions in de-ionized water also submitted to the preconcentration procedure were used for calibration. By adding different masses of iron(II1) to 100ml of de-ionized water containing 20ng of silver it was found that the integrated absorbance for a volume of 20 pl injected into the graphite tube increased from 0.262s when no iron was present to about 0.360 s for an iron mass of 0.3 mg. For larger masses of iron at least up to 1.0 mg the signal remained approximately constant.By adding different amounts of ascor- bic acid to the initial silver solutions containing 0.5 mg of iron(m) a decrease in signal to about 0.250s was observed when 2mg of ascorbic acid were present and the signal remained constant for larger masses of ascorbic acid at least up to 5mg. According to Bode and Arnswald," iron(II1) is partially complexed and the compound precipitates and decomposes while iron@) is not complexed. In fact when the complexing agent is added the solution containing iron(1n) becomes opaque. The higher silver signal in the presence of iron(Ir1) must be due to its precipitation and retention of the precipitate on the carbon. This contributes to the theory already discussed that poor retention on the carbon or incom- plete complexation might be responsible for a yield below 100% for the preconcentration in de-ionized water.To avoid the interference iron(rI1) should be reduced to iron@) before the addition of the NH,DDTP. Ascorbic acid 5.0 mg was added to the soil solution and also to the analytical solutions. The concentration of silver found in Soil 5 was 1.89 & 0.09 pprn which compares well with the recommended value of 1.9 ppm. Conclusions The proposed method for the preconcentration and determi- nation of silver is adequate for the analysis of waters and soils. The small volume required by ETAAS coupled with the carbon preconcentration procedure leads to high enrichment factors. For real samples having a relatively high concentration of iron(In) such as soil samples a reducing agent should be added prior to the complexation.The analyte addition tech- nique is not necessary provided the analyte solutions are submitted to the same procedure as the sample. The authors are grateful to the Brazilian Science and Technology Ministry for providing financial support. A.K.A. had a scholarship from Brazilian Research Agency CNPq (Conselho Nacional de Pesquisa e Desenvolvimento Tecnologico). 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 References Kinrade J. D. and Van Loon J. C. Anal. Chem. 1974 46 1894. Jan T. K. and Young D. R. Anal. Chim. Acta 1978 50 1250. Kenawy I. M. Khalifa M. E. and El-Defrawy M. M. Analusis 1987 15 314. Beinrohr E. Rojcek J. and Garaj J. Analyst 1988 113 1831. Yang X. C. and Jackwerth E. Fresenius' 2. Anal. Chem. 1987 327 179. Bloom N. S. and Crecelius E. A. Anal. Chim. Acta 1984,156,139. Skogerboe R. K. Hanagan W. A. and Taylor H. E. Anal. Chem. 1985 57 2815. Danz H. J. and Jackwerth E. Fresenius' 2. Anal. Chem. 1987 326 57. Nakashima S. Sturgeon R. E. Willie S. N. and Berman S . S. Anal. Chim. Acta 1988 207 291. Bode H. and Arnswald W. Fresenius' 2. Anal. Chem. 1962 185 99. Bode H. and Arnswald W. Fresenius' Z . Anal. Chem. 1962 185 179. Berndt H. and Messerschmidt J. Fresenius' Z . Anal. Chem. 1981 308 104. Berndt H. Messerschmidt J. and Reiter E. Fresenius' 2. Anal. Chem. 1982 310 230. Monte V. L. A. and Curtius A. J. J. Anal. At. Spectrom. 1990 5 21. Kimbrough D. E. and Wakamura J. R. Envivon. Sci. Technol. 1989 23 898. Bruland K . W. Chemical Oceanography eds. Riley J. P. and Skirrow C. Academic Press 2nd ed 1983 p. 172. Kolthoff I. M. Sandell E. B. Meehan E. J. and Bruckenstein S. Quantitative Chemical Analysis 4th edn. Macmillan New York 1969 p. 652. Paper 3/05646E Received September 20 1993 Accepted November 16 1993

ISSN:0267-9477    

DOI:10.1039/JA9940900543

出版商:RSC    

年代:1994

数据来源: RSC

摘要:

JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY APRIL 1994 VOL. 9 547 Speciation and Preconcentration of Trace Elements With Immobilized Algae for Atomic Absorption Spectrophotometric Detection H. A. M. Elmahadi and Gillian M. Greenway School of Chemistry University of Hull Hull UK HU6 7RX The properties of two covalently immobilized algae Chlamydomonus reinhartii and Selenestrum capri- cornutum were compared for the preconcentration of Cu2+ Ag' C?' and Cp'. These reagents were found to have different properties with Chlamydomonus reinhartii having a lower capacity for Cu2+ and a higher capacity for Cr". With the exception of Cr6+ the maximum preconcentration of the metal ions investigated occurred at different pH values for the different algae suggesting that different binding sites were involved.The accuracy of the preconcentration method for the analysis of estuarine sediment (using both reagents) was illustrated with a reference sample obtained from the National Research Council Canada. A flow system for the speciation of C f + and C?' was then developed using both reagents. It was shown that no interference occurred between the two species. Further interference studies showed that metal ion interferences were minor even when the interfering ion was present in excess. The effects of sodium chloride sodium hydrogencarbonate and humic acid were also studied and these were found to interfere by competing for the metal ions to be complexed. Keywords Immobilized algae; preconcentration; speciation; trace metals; atomic absorption spectrometry Recently algae and other organisms such as yeast have been exploited as analytical reagents to preconcentrate trace metals such as copper cadmium lead and gold.' This work is usually carried out on material that has not been immobilized using * batch methods such methods are effective but time-consuming.Immobilization of the algae either by physical entrapment in silica gel polymeric materials,2 or by covalent attachment to water insoluble substrates such as controlled-pore glass3 allows rapid on-line analytical methods to be developed. Organisms such as algae have been shown to perform very well for this type of application because they have a small uniform cell size and a number of different metal binding sites on their cell walls4 These sites include amine and carboxyl groups from amino acids and polysaccharides sulfhydryl groups and unmethylated pectins.In comparison to chemical chelating reagents which normally would have only one binding site the wide range of sites in algae provide excellent potential for obtaining high selectivity for different elements and their different oxidation states by altering elution conditions. Previous work reported the covalent immobilization of the alga Selenestrum capricornutum on controlled-pore glass and its utilization for the preconcentration and determination of copper lead zinc cobalt mercury and cadmium by flow injection atomic absorption spe~trometry.~ In the present work a different alga Chlamydomonus reinhartii is immobilized and its performance is compared with the previously immobilized alga.Both algae are then investigated for their application in the determination of different oxidation species of chromium and for their use in the analysis of real samples. Experimental Reagents The cultivation and immobilization of both Selenestrum capri- cornutum and Chlamydomonus reinhartii were carried out as described previously for Selenestrurn c~pricornutum.~ The coval- ent immobilization procedure consisted of first silanizing the controlled-pore glass and then using the bifunctional properties of gluteraldehyde to cross-link the algae to the silanized glass through the amino groups on the surface of the cell wall of the algae. All other reagents with the exception of humic acid were of analytical-reagent grade and all were obtained from Merck (Poole Dorset UK).The water used throughout this work had a resistivity of 12 MQcm and was produced by reverse osmosis followed by ion exchange. A National Research Council Canada reference sediment (MESS-1) was obtained from the Laboratory of the Government Chemist (Teddington Middlesex UK) for the determination of copper and zinc. Reference Sample Preparation Duplicate 1 g samples of MESS-1 were accurately weighed into 25 ml platinum crucibles and digested to near evaporation with 5 ml of nitric acid (68%) and 5 ml of hydrochloric acid (36.5%). A 2.5 ml portion of hydrofluoric acid (49%) was then added after the addition of 3.0ml of water to remove any silicates. This was repeated until effervescence ceased. The samples were then treated with 0.05 mol 1-1 hydrochloric acid diluted to 50 ml with water and stored in polyethene bottles.Aliquots (25 ml) of solution were then taken and adjusted to pH 7.5 with NaOH prior to dilution to 50 ml with phosphate buffer (pH7.5) for the determination of Cu2+. The same procedure was carried out for the determination of Zn2+ but this time tris( hydroxymethyl methylamine) (TRIS) buffer at pH 7.5 was found to give the maximum preconcentration. Instrumentation Preconcentration and Speciation Procedures The instrumentation and preconcentration procedures were as described pre~iously.~ The atomic absorption spectrometer used for this work (Varian Model AA75) had been recently serviced but lacked sensitivity due to the age of its optics and the background correction was not operational.An air- acetylene flame was used for all elements as safe facilities were not available for a nitrous oxide-acetylene flame. These factors did not however prevent illustration of the preconcentration and speciation properties of the reagents. Conditions were optimized for each element and backgrounds were checked. Wavelengths of 324.8 213.9 242.8 and 357.9 nm were used for copper zinc silver and chromium respectively. The immobil- ized algae were packed into a glass tube ( 5 cm x 2.5 mm id.) and incorporated into a flow injection manifold consisting of a peristaltic pump injection valve and two three-way valves. For preconcentration volumes of up to 5ml of sample in buffer solution were passed through the minicolumn which was then washed with water.The accumulated ions were548 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY APRIL 1994 VOL. 9 released from the minicolumn by injecting an eluent reagent into the water carrier stream prior to the column. The plug of preconcentrated ions was then transported to the atomic absorption spectrometer in the flowing stream. A flow rate of 2 ml min-l as selected in previous work was used.5 For the determination of different oxidation species of chromium two flow injection manifolds were connected by a T-piece (Fig. 1). One manifold was used to preconcentrate Cr6+ and the other to preconcentrate Cr3+ using the conditions described under Results and Discussion. The two different forms were retained on their respective columns and washed with water prior to sequential elution.Results and Discussion Comparison of the Performance of Chlamydomonus reinhartii With Selenestrum capricornutum For this work four metal ions namely Cu2+ Ag' Cr3+ and Cr6+ were investigated for both algae in order to show a range of different pH dependences for binding. The Cu2+ and Cr3+ are hard metal ions showing cationic behaviour Ag+ however is a soft metal ion tending to exhibit covalent bonding with ligands containing nitrogen and sulfur. The Cr6+ forms CrOd2- and thus has anionic behaviour. The capacity of the immobilized Chlamydomonus reinhartii was determined as in the previous work (for 0.1 g of immobil- ized algae) where the capacity C was calculated from c=(ci-cf)l/lm where ci is the initial concentration and cf is the final concentration after equilibration (in moll-') m is the mass of controlled-pore glass (in g) and V is the volume of the solution (in ml).Immobilized Chlamydomonus reinhartii had exchange capacities of 5.45 mmolg-1 for Cu2+ and 0.927 mmol g-' for Cr6+ whilst immobilized Selenestrum capricornutum had exchange capacities of 9.70mmolg-l for Cu2+ and 0.828 mmol g-' for Cr". Therefore the immobilized Chla- mydomonus reinhartii was found to have a considerably lower capacity for Cu2+ than Selenestrum capricornutum but the reverse was found to be true for Cr6+. Effect of pH and Eluent Concentration The TRIS and phosphate buffers were used in the pH range 4-10 (0.1 moll-l) in an investigation of the effect of pH on preconcentration. Maximum chelation of different metal ions occurred at definite pH values depending on the buffer used.The optimum pH values for the different metal ions are compared for the different immobilized algae in Table 1. TRIS buffer was the most effective for Cr3-' and Ag+ and phosphate buffer was best for Cu2+ and Cr6+ when using immobilized Chlamydomonus reinhartii. With immobilized Selenestrum 2 ml min-' rn Metal ion solution To waste 4 u t I 5 2 HNO8 I I 4 U 2 Fig. 1 Manifold for speciation of Cr3+/Cr6+ 1 3-way valve; 2 peri- staltic pump; 3 injection valve; 4 column of immobilized reagents; and 5 atomic absorption spectrometer Table 1 Comparison of the pH values required for maximum sensi- tivity for the immobilized algae Chlamydomonas reinhartii Selenestrum capricornutum pH of phosphate pH of Metals buffer Tris buffer cu2 + 6.50" 6.50 Cr3 + 7.50 7.50* Ag+ 8.50 5.00" Cr6 + 4.00* 5.00 pH of phosphate pH of buffer Tris buffer 7.50* 7.50 8.50 8.50* 7.50" 5.50 4.00* 5.00 * Maximum sensitivity.capricornutum the best results were obtained with phosphate buffer for all metal ions except Cr3+. The effect of buffer is particularly noticeable for Ag' and is probably due to the effect of competing anions as is discussed later. Only Cr6+ has maximum preconcentration- at the same pH for both algae. This is in agreement with results obtained by Darnel1 et aL6 who found that it did not bind at high pH values where the overall charge of the algal surface is negative but that it did bind at low pH values where the surface is positively charged allowing electrostatic binding of the CrOd2- ion. Maximum preconcentration for the other metal ions occurred at different pH values for the two algal species which suggests that different binding sites were involved and implies that each of the algae has unique binding properties.As would be expected Cr3+ a hard acid and Cu2+ (borderline) were retained at pH values greater than 5.0 and eluted at pH values less than 2.0. The elution conditions required were the same for both algae in the cases of Cr3+ and Ag'. They both required 0.1 mol I-' thiourea in 0.1 moll-' nitric acid to allow complete elution from the column (40 pl for Ag+ and 60 1-11 for Cr3+). The Cr6' was eluted from both algal columns with 40 pl of 0.1 mol 1-1 NaOH. The Cu2+ was eluted with 40 p1 of 0.1 mol 1-1 nitric acid from the mini-column of immobilized Chlamydomonus reinhartii but a higher acid concentration of 0.5 moll-' was needed to elute it from the Selenestrum capricornutum mini-column.This together with the capacity data supports the claim that Cu2+ binds more strongly with Selenestrum capricornutum than with Chlamydomonus reinhartii. Calibration and Recovery Calibrations were carried out using the proposed procedure and the curves for Chlamydomonus reinhartii are shown in Fig. 2. The figures of merit comparing the results for the different algae are given in Table 2. The poor detection limits for the chromium ions are due to the instrumental limitations described earlier. In agreement with the previous discussion the enhancement factor for Cu2+ with Selenestrum capri- cornutum was 20 times greater than that with Chlamydomonus reinhartii.Chlamydomonus reinhartii had a higher enhancement factor for Cr3+ and a lower one for Cr6+ in comparison with Selenestrum capricornutum but the values were the same for both algae with Ag+. The improvement in sensitivity reflects the enhancement factors. The recoveries were determined by comparing the signal obtained by direct injection with that obtained after preconcen- tration and elution with the appropriate volume and concen- tration of acid. The recoveries for Chlamydomonus reinhartii were 92.0 92.0 83.2 and 78.0% for Cu2+ Ag+ Cr3+ and Cr6+ respectively. Those for Selenestrum capricornutum were 100.2 88.2 83.2 and 72.0% for Cu2+ Ag' Cr3+ and Cr6+ respectively. This means that the best recoveries were seen for Cu2+ and Ag'.The recovery for Cr6+ was lower for both algae.JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY APRIL 1994 VOL. 9 549 Table 2 Analytical performance of the on-line preconcentration flow injection atomic absorption spectrometric system for 5 ml of sample (sampling rate 20 h-l) Linear range/ Metal ion Reagent* ng ml-' cu2 + sc CR Cr3 + sc CR Cr4 + sc CR Ag + sc CR 5-45 30-150 200-2000 240-2400 600-5000 300-3000 55-500 50-400 Correlation coefficient 0.9992 0.9999 0.9991 0.9993 0.9997 0.9998 0.9998 0.9995 RSDT (Yo) 1.5 (40) 1.2 (60) 1.6 (1000) 1.5 (750) 1.5 (2000) 1.3 (2000) 1.5 (200) 1.2 (100) Detection limit $/ng ml - ' 0.05 1 .o 20 25 40 30 2.0 2.0 Enhancement factor5 4000 200 50 40 25 33 125 125 Improvement in sensitivitfl 118 56 61 58 31 55 30 30 * SC = Selenestrum capriconutum; CR = Chlamydomonus reinhartii.7 n = 5; values in parentheses are the concentration of the metal ion in ng ml-'. 5 20 - value. 5 Factor by which the limits of detection decrease using preconcentration as opposed to direct injection. 7 Comparison of gradients of the calibration curves before and after preconcentration. 160 140 prepared. These consisted of a 1.0 pg ml-I standard of Cr3+ made-up in pH 8.5 TRIS buffer as a reference and five solutions containing 1.0 pg ml-' of Cr3+ in different concentrations of Cr6+ (1.0 4.0 6.0 8.0 10.0 pg ml-l) made-up with the same buffer and at the same pH as the reference solution. Aliquots ( 5 ml) of these solutions were preconcentrated and eluted under the optimum conditions for Cr3 + mentioned earlier.The absorbance for the solutions containing the Cr6+ were then compared with the absorbance for the reference solution and no difference was seen. This meant that Cr3+ could be detected in the presence of Cr6+ in 10-fold excess and that no inter- ference was observed. The reverse experiment was then carried out using Cr6+ as the reference solution and Cr3+ as the interfering ion with the optimum preconcentration conditions for Cr6+. The results showed that Cr6+ could also be detected without interference by Cr3+ present at a 10-fold excess. 0 1 2 3 4 Similar results were obtained for both oxidation species when Several possibilities were investigated for the flow injection manifold and the configuration used here can be seen in Fig. 1. As it is normal for low levels of chromium to be detected the sample would usually need to be preconcentrated.However 120 E loo E E .- o) 80 r 2 \ a % 60 40 20 COnCentratiOn Of metal iOn/ngml-' ( X 10 for Cr3+and Cr") Fig.2 Calibration curve for a series of metal ion standards using immobiiized Ch~amy~omonus reinhartii (on the y-axis mm corre- sponds to an absorbance of 0.0012) Chlamydornonus reinhartii was used instead. Analysis of a Sediment Reference Material An estuarine sediment reference material was analysed for Cu2+ and Zn2+ using the standard procedure with preconcen- tration on both types of algal column. This analysis was carried out to investigate whether the preconcentration procedure was applicable to real samples. The results can be seen in Table 3 and show that accurate results can be obtained by this method for these elements at the mg kg-' level.Both algae performed well as preconcentration agents and excellent recoveries were obtained; it must be remembered however that the levels being measured were relatively high. The interference studies described later illustrate some of the problems that can occur at lower levels. Speciation of C?+ and Cr6+ To study the selectivity that can be achieved for the different oxidation species by using different buffer conditions with immobilized Selenestrum capricornutum a set of solutions were as the preconcentration procedure would only elute a particular oxidation state it is necessary to quantify both oxidation states rather than deducting the value for one oxidation state from the total.Varying degrees of automation for this procedure are possible but for this work the procedure was kept very simple. As discussed previously Chlarnydomonus reinhartii had a higher capacity for Cr6+ and Selenestrum capricornutum had a higher capacity for Cr3+ so as both columns were available they were both used in the manifold. For the analysis two aliquots were removed from the test solution and these were then buffered. A 5 ml aliquot of test solution containing 2pgml-I of each of Cr3+ and Cr6+ in phosphate buffer (0.1 mol 1-1 pH 4.0) were passed through the Chlamydomonus reinhartii column via one of the three-way valves and a 5 ml aliquot of test solution containing 2 pg ml-' of each of Cr3+ and Cr6+ in TRIS buffer (0.1 mol l-l pH 8.5) were passed through the Selenestrum capricornutum column using the other three-way valve.The two columns were washed with water and the species were then sequentially eluted Cr3+ with 60 pl of thiourea in 0.1 mol 1-1 HC104 and Cr6+ with 40 pl of Table 3 Analysis of Sediment Reference Material MESS-1; values given in mg kg-' Chlamydomonus reinhartii Selenestrum capricornutum Metal ion Found Certified Recovery (YO) Found Certified Recovery (YO) cu2 -k 25.5 k 2.4* 25.1k3.8 101.6 25.2 -I- 1.2* 25.1 k 3.8 100.40 Zn2 + 195.3 & 2.6" 191.0+ 17.0 102.2 193.0 & 2.3* 191.0+ 17.0 101.10 * Standard deviation of three determinations.550 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY APRIL 1994 VOL. 9 0.01 D c ' I I i i m i i Time - Fig. 3 Speciation of Cr3+/Cr6+ (2 pg ml-') A (Cr") and B (Cr3+) with Selenestrum capricornutum; C (Cr6+) and D (C?+) with Chlamydomonus reinhartii; E (Cr") and F (Cr3+) with both algal species 0.1 mol I-' NaOH.By using the calibrations for chromium the correct values were obtained for each oxidation state and the resultant traces can be seen in Fig. 3. These results illustrate that immobilized algae can be successfully used for the determi- nation of different oxidation states of chromium. Interferences Metal Ion Interferences As reported earlier only very minor interferences were observed for Selenestrum capricornutum even in large ex~ess.~ Similar if not better results were obtained for Chlamydomonus reinhartii as can be seen in Table 4. The common alkaline earth metals ions have no interfering effect. A slight negative interference was seen for the determination of Cu2+ in the presence of Zn2+? and Hg2+ was seen to enhance the signal for Ag'.This slight enhancement could be due to the release of ions that had not been completely recovered from previous determinations. Other Interferences The effects of other species that may be present in natural waters were investigated for immobilized Selenestrurn capri- cornutum. The species investigated included sodium chloride (Cl-) sodium hydrogen carbonate ( HC03 -) sodium dihydro- gen orthophosphate (phosphate buffer H,PO,-) and humic acid. The experiments were carried out for the determination of Cu2+ and Zn2+ at concentrations of 40 and 90ngml-' respectively. The metal ions were prepared in either different concentrations of NaCl which were adjusted to pH 7.5 with 0.1 mol I-' NaOH or solutions of NaHCO with a pH of 7.8.No buffers were used as the anion in the buffer might have interfered with the results. In Fig. 4 the results are compared with those obtained in 0.1 mol 1-' phosphate buffer at pH 7.5 (9.70 mg I-' H2P04-). For the sodium chloride solutions an increase in absorbance was seen for both Cu2+ and Zn2+ with increasing concen- tration up to a maximum after which the absorbance decreased. A 3.55 mgml-' Cl- (0.1 moll-') solution gave the highest absorbance for Cu2+ and 1.77 mg ml-' of C1- (0.05 moll-') the highest for Zn2+. As can be seen in Fig. 4 a similar trend was observed for sodium hydrogencarbonate. Comparing the results for sodium chloride solution and phosphate buffer a depression in the absorbance signal is 160 140 E IZ0 E '4 .P 100 c a c Y m 2 80 60 40 I I I I I 0 2 4 6 8 10 Concentration of anions/mg I-' Fig.4 Effect of anions on the absorbance Cu2+ and Zn2+ A effect of phosphate buffer on Zn2+ (90 ng ml-'); B effect of phosphate buffer on Cu2+ (40ngml-l); C effect of sodium chloride on Zn2+ (90 ng ml-I); D effect of sodium chloride on Cu2+ (40 ng ml-I); E effect of sodium hydrogen carbonate on Zn2+ (90ngml-I); and F effect of sodium hydrogen carbonate on Cu2+ (40 ng ml-I). Concentrations should be divided by 2 for the phosphate buffer; a peak height of 1 mm corresponds to an absorbance of 0.0012 Table 4 Interference effects of high concentrations of different interfering ions (20 pg rn1-l) on preconcentration process for particular metal ions using immobilized Chlamydomonous reinhartii Metal ion (ng nil-') Interfering ion Cr6 + Cr3 + cu2 + Ag+ Hg2 + Zn2 + Ca2 + Mg2+ Ag + (400) Cr3+ (3000) Cr6+ (5000) Cu2+ (100) Change in peak height (%) 0.00 0.00 - - 0.00 - 0.00 0.00 0.00 0.00 0.00 - - 0.00 0.00 0.00 + 12.0 0.00 0.00 0.00 0.00 0.00 0.00 - 4.0 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY APRIL 1994 VOL.9 55 1 Table 5 Effect of interfering anions in the presence of 0.5 pg ml-I of humic acid (HA) in the determination of Zn2+ and Cu2+ Zn2+ (90 ng ml-l) Cu2+ (40 ng ml-l) Concentration/ Concentration/ Interfering ion mg ml-l Absorbance Interfering ion mg ml-' Absorbance HC0,- 3.05 0.162 HC0,- 6.10 0.138 c1- 1.77 0.132 c1- 3.05 0.096 H2P04- 9.70 0.156 H2P04- 9.70 0.120 observed for sodium chloride solutions for both Zn2+ and Cu2+.This would suggest that C1- is competing for metal ions and is more successful than H,PO,-. On the other hand there is an increase in absorbance signal especially for Zn2+ in the presence of sodium hydrogencarbonate solution suggest- ing that HC03- is less effective at competing for metal ions than both H2P04- and C1- ions. In work with real samples buffer would also be present and therefore further experiments were carried out for the determi- nation of 90 ng ml-' of Zn2+ in 0.1 moll-' phosphate buffer (9.70 mg ml-I H,P04-) in the presence of either 3.05 mg ml-1 HC03- or 1.77 mg ml-I C1-. For the straight buffer solution an absorbance of 0.168 was observed however when the HCO - ions were present the absorbance increased to 0.42.An enhancement was also seen for C1- but to a lesser extent (absorbance of 0.21). It is well known that metal ions bind to humic acid and that humic acid present in natural waters will therefore interfere with analysis. To investigate this effect different concentrations ranging from 0.05 to 10 pg ml-' of humic acid were adjusted to pH 7.5. Standard Cu2+ and Zn2+ solutions were then added to the humic acid solutions to give concentrations of 40 and 90 ng ml-1 of Cu2+ and Zn2+ respectively. Aliquots ( 5 ml) of these solutions were then preconcentrated and eluted. The absorbance values obtained are shown in Fig. 5. This clearly illustrates that humic acid strongly inhibits the preconcen- tration of Cu2+ and Zn2+ with practically no binding at the 10 pg ml-' level.Table 5 shows the more complex effect of having both humic acid and either C1- HC03- or H2P04- present. The increase in absorbance values indicate that the addition of HC0,- and H2P04- to a solution containing either Zn2+ or Cu2+ releases the metal ions from the humic acid so that they are available to bind with the algae even though the signal stays below that obtained in the absence of humic acid. The addition of chloride ions to the metal ion and humic acid produces a similar absorbance to that for chloride alone. This work shows that although accurate results were obtained for the estuarine sediment sample the effect of the matrix can be considerable at the ngml-' level. Although humic acid caused a large depression in the signal this effect was counteracted by the presence of other anions in solution including those from the buffer.This means that the method would still be of use for measuring low levels of metal ions in natural waters as long as standard additions calibration tech- niques were used. Further work is being carried out on real samples using a more sensitive detection system. 0 2 4 6 8 10 [Humic acidl/pg ml-' Fig. 5 Effect of the presence of humic acid on absorbance of A Cu2+ (40ngml-l) and B Zn2+ (90ng rn1-I); a peak height of 1 mm corresponds to an absorbance of 0.0012 maximum preconcentration pH eluent conditions capacity and recovery shows that each alga has unique binding proper- ties probably due to the presence of different binding sites on the cell walls. The effects of interfering metal ions on the preconcentration process was shown to be negligible however the presence of different anions had considerable impact. Therefore for the analysis of real samples at ultra-trace levels it would be essential to use standard additions for calibration. References Mahan C. A. Majidi V. and Holcombe J. A. Anal. Chem. 1989 61 624. Darnell D. W. and Gabel A. US EPA Res. Dev. [Rep.] EPA EPA/600/9-89-072 Int. Conf. New Front. Hazard. Waste Manager. 3rd 1989 pp. 217-225. Elmahadi H. A. M. and Greenway G. M. J. Anal. At. Spectrom. 1991 6 643. Crist R. H. Oberholser K. Shanks N. and Nguyen M. Environ. Sci. Technol. 1981 15 1212. Elmahadi H. A. M. and Greenway G. M. J. Anal. At. Spectrom. 1993 8 1011. Darnell D. W. Greene B. Hosea M. McPherson R. A. Henzel M. and Alexander M. D. in Trace Metal Removal from Aqueous Solutions ed. Thompson R. Royal Society of Chemistry Cambridge 1986 Special publication 61 p. 1. Conclusion This work has confirmed that immobilized algae are useful reagents for preconcentration and speciation. The data for Paper 3105222B Received August 31 1993 Accepted November 1 1 1993

ISSN:0267-9477    

DOI:10.1039/JA9940900547

出版商:RSC    

年代:1994

数据来源: RSC

摘要:

JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY APRIL 1994 VOL. 9 553 Calibration in Flame Atomic Absorption Spectrometry Using a Single Standard and a Gradient Technique lgnacio Lopez Garcia Pilar Viiias and Manuel Hernandez Cordoba* Department of Analytical Chemistry Faculty of Chemistry University of Murcia E-30071 -Murcia Spain To obtain calibration graphs a standard solution of the analyte was pumped using an increasing flow rate while pure water was aspirated for compensation through a T-piece. In this way an on-line concentration gradient was obtained and a constant flow was supplied to the nebulizer. The absorbance was monitored via a computer which facilitated data acquisition and handling. In spite of the problems posed at low flow rates by the pulsations caused by the rollers of the peristaltic pump this calibration strategy gave graphs comparable to those obtained by means of the separate measurement of several standard solutions.A calibration graph could be obtained in only 50 s with errors of less than +3%. The mode of operation also permits the on-line dilution of moderately concentrated solutions which cannot be measured by direct aspiration and simplifies the standard additions method. Keywords Automation ; automated calibration; on-line dilution; flame atomic absorption specfromefry. The usual method of obtaining calibration graphs for flame atomic absorption spectrometry (FAAS) involves the prep- aration of a stock standard solution of the analyte. A set of appropriately diluted solutions covering the response range of the spectrometer is prepared and the solutions individually measured.Owing to instrumental drifts recalibrations are frequently required and although this whole process cannot be considered as being tedious it is time consuming. Thus if a different strategy that saved time and afforded the same practical results could be found it would be of interest. For this reason a number of different approaches particularly the use of flow injection (FI) methodology have been discussed in the literature in an attempt to simplify the calibration procedure. In addition to the extensive work of Tyson and co-workers,l-ll several other researchers have discussed this point using very different strategies.12-21 Several years ago Bysouth and TysonZ2 reported an interes- ting flow manifold based on a fixed-speed pump together with a computer-controlled stream switching valve and pump which permitted automated on-line dilution of standards.These workers stated in their conclusions that one of the major sources of error arises from the peristaltic pumps which produce pulsations in the flow and thus in the absorbance. Recently Starn and HieftjeZ3 have proposed an automated on-line gradient calibration procedure using a high- performance liquid chromatography (HPLC) pump module for mixing stock and diluent solutions in order to obtain the desired concentrations. However these workers indicated that a drawback lay in the lag in the response of the system when the concentration gradient is stepped which is probably due to the plumbing configuration of the HPLC pump used.With this in mind a different approach was attempted. Firstly it was demonstrated that the HPLC pumps available in this laboratory obtained from different suppliers to that of Starn and Hieftje's pump also led to a lag in the response of the system. This being so the use of low-cost easily available peristaltic pumps was again considered. If a stock standard solution is pumped at a linearly increasing flow rate from zero up to the nebulizer uptake rate and a simple T-piece with a tip immersed in pure water is included in the manifold immediately before the nebulizer it should be possible to obtain an on-line concentration gradient whilst the nebulizer is fed at a constant flow. Since the manifold is very simple and the spectrometer response is fast inertial effects when varying the concentration along the gradient should be minimal and the absorbance measured should be * To whom correspondence should be addressed. very close to that obtained when using the conventional aspiration mode. Hence in spite of the predictable problems owing to the pulsations of the pump it was decided to investigate whether such a strategy could provide reliable analytical results.This paper reports the results obtained and discusses other features (on-line dilution and the standard additions method) which as occurs with other related a p p r o a ~ h e s ~ ~ ~ ~ ~ are a consequence of this non-conventional mode of operation. Experimental Preliminary experiments were performed on a Pye Unicam SP1900 atomic absorption spectrometer coupled to a Hewlett- Packard 4742A integrator.For the remainder of the experi- ments a Pye Unicam 919 Solaar spectrometer equipped with a burner of 5cm in length was used. The analogical non- damped output of the spectrometer was connected to a per- sonal computer uia a PClab 818PG data acquisition card. Software written in the laboratory permitted the direct vis- ualization of the absorbance-time relationship. The data file for each experiment was saved in ASCII format and sub- sequently read by using a commercial software package which allowed the easy treatment and plotting of data. The measure- ments were made at 422.7 nm for calcium and 324.8 nm for copper using slit-widths of 0.5 nm. Air-acetylene flames were used exclusively. Stock solutions of copper and calcium (1000 pg ml-l) were obtained from Panreac Spain.The manifold used is shown schematically in Fig. 1. All the connecting lines were of 0.8 mm id. poly( tetrafluoroethylene) (PTFE) tubing. The T-piece was made of Perspex and was located immediately (about 10 mm) before the nebulizer. The length of the aspiration tube (distance between the nebulizer and the solution being aspirated through the T-piece) was about 10cm. When the peristaltic pump was stopped the nebulizer uptake rate was measured to be 8.4 ml min-'. The peristaltic pump used (Gilson Minipuls 3) is supplied by the manufacturer with an interface allowing remote control which made it possible to obtain programmed flow rates by varying the voltage applied. For a voltage range of 0-5 V the pump rate depends on the applied voltage (1) rpmrnax x V vmax rpm = where rpm is the pump rate in revmin-1 at time t when a voltage is being applied.In all cases the maximum pumping speed (rp%ax) was adjusted to match the nebulizer uptake rate. To obtain the value of rpmmax with the pump stopped a solution of analyte was aspirated through the T-piece and554 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY APRIL 1994 VOL. 9 water was pumped at an increasing rate until the absorbance fell to zero. This value was regularly checked since the prolonged use of the pump tubes can produce a drift in the flow rate owing to wearing. It was verified that the simple precaution of lubricating the pump tubes with a few drops of silicone oil considerably extended their useful life.To obtain a time-variable flow rate the applied voltage was varied linearly over the range 0-5V by using a polarographic function generator (Amel Model 566). The voltage ramps applied could be selected over a wide range those of 50 and 100 mV s-l being the most suitable. This produced accelerations in the flow rate of 1.4 and 2.8 pl sP2 respectively in the equipment used. General Procedure To obtain a calibration graph dip the aspiration tube of the pump into the most highly concentrated standard solution that provided a response within the linear response range and use a blank for compensation through the T-piece. Run the flow rate gradient and obtain the signal using a recorder or preferably a computer. The former provides a calibration graph directly and with the latter it is easy to perform a convenient fit.The abscissa axis of the plot is directly related to the analyte concentration. To quantify an unknown sample with the pump stopped aspirate the problem solution through the 1-piece. Following this general procedure and by varying the nature of the solutions being aspirated and pumped the manifold permits other possibilities such as on-line dilution standard additions and the easy use of releasing agents as is discussed below. Theory Consider a solution containing a concentration of analyte C being pumped through the manifold shown in Fig. 1 using a linear variation of the pumping rate with time ranging from 0 for t =O to UT for t = T where UT is the nebulizer uptake rate. The difference between UT and the pumping rate Up can be compensated for through the three-way connector by aspirat- ing another solution with a concentration C in such a way that where U is the aspirated time-dependent flow rate.In this way the concentration of the solution reaching the spec- trometer is time variable ranging from C to C and an on-line concentration gradient is obtained. As the solution is entering the nebulizer at a constant flow rate assuming that the proportionality constant of the detector is the same for the solutions being pumped and aspirated the absorbance meas- . . . . . . (T}. . . . . . . {P.). Fig. 1 Diagram of manifold P peristaltic pump; RG ramp analogic generator; PC personal computer; T T-piece; FAAS spectrometer; C concentration of pumped solution; C concentration of aspirated solution; Upt pumping rate; UT nebulizer uptake rate; Uat aspirated time-dependent flow rate ured at time t is given by (3) Combining eqns.(2) and (3) and taking into account that Up is directly proportional to the time elapsed from the starting of the pump t the absorbance can be expressed as a function of t the time necessary to reach the nebulizer uptake rate T and the concentrations of the solutions aspirated and pumped as (4) It must be noted that if the proportionality constant is not the same for both the pumped and aspirated solutions eqn. (4) is not valid because the constant of the mixed solution is time dependent. Even so eqn. (4) could be applied close to both t=O and t = T as the composition of the solution entering the nebulizer is very similar to that of the solutions being aspirated and pumped respectively. Unless otherwise stated the remain- der of the discussion assumes the invariability of the pro- portionality constant.Three cases can be discussed depending on the values of C and C,. A. Pure Water is Aspirated for Compensation In this case as C is zero eqn. (4) is expressed as t T A = K - C and a straight line is obtained if A is plotted against time. It should be noted that eqn. (5) can be re-written as A,=KC because the term tCpT- indicates the instantaneous concen- tration of the solution being pumped Cpl. So when the highest concentration which provides an absorbance within the linear response range of the instrument is pumped through the manifold at a variable rate and the absorbance is continuously monitored by a plotter or preferably a computer a calibration graph is obtained.An unknown solution is then easily quant- ified with the pump stopped by aspirating the problem through the 1-piece. This method has interesting advantages over the conventional one in which calibration graphs are obtained the calibration graph is obtained in a few seconds using a single solution and the number of data acquired by means of the computer is very high which means that their fitting to linear or polynomial functions is more reliable. B. Concentration of the Solution Being Pumped is Zero The solution being aspirated for compensation through the T-piece is diluted on-line which is useful for extending the calibration range and to obtain analytical signals from solutions that are too concentrated to be aspirated directly.In this case (C = 0) and eqn. (4) which is rewritten as A = KC,(,-+) is only valid for that portion of the absorbance-time plot in which the measurements are within the linear response range. For this reason a curve is obtained at the start of the experiment and only when the aspirated solution is sufficiently diluted is a straight line with a slope of -KC,TV1 obtained. The concentration of the unknown can be calculated from the absolute value of the slope of this linear portion and from the slope of a calibration graph obtained as indicated under (A) using a solution with concentration C, lslope of dilution1 C - slope of calibration Cp (7)JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY APRIL 1994 From a theoretical point of view this approach should be valid irrespective of the degree of dilution suffered by the concentrated solution.However the greater the degree of dilution required the shorter the time interval in which a linear absorbance-time plot is obtained and so the approach is restricted in practice to moderately concentrated solutions. C. Both C and Cp Differ from Zero This is in practice an equivalent of the standard additions method where the ‘unknown’ is the solution being aspirated which is merged with the pumped standard solution. The absorbance-time plot for signals within the linear response range must follow eqn. (4). The concentration of the ‘unknown’ C can be calculated either by comparison of the slope of the graph with that of a calibration graph following (8) Slope of standard additions Cp -~ - Slope of calibration C - C or by using the intercept on the time axis tad.In this case eqn. (4) is expressed as tad ’tad - T c,=c - (9) The above considerations are valid if and only if the essential condition of the analyte response factor is the same for both the solution being aspirated and the solution being pumped. If this is not so a different strategy is necessary. When the cause of the different behaviour is physical such as differences in density or viscosity a simple solution is to use for calibration purposes a sample of known concentration and similar physical properties to those of the unknown. If a chemical interference occurs a releasing agent can be used. It is not necessary to add this agent to both the standard (pumped) and the unknown (aspirated) solutions but only to the reference solution since owing to the increasing flow rate of the pumped solution the concentration of the releasing agent increases with time.When the chemical interference is slight it might even be possible to omit the releasing agent since any interference associated with the ‘unknown’ is diluted when this is merged with the solution being pumped. With increasing dilution the so-called ‘limit dilution method’2427 predicts that the interference can be circumvented. Results and Discussion Linear and Non-linear On-line Calibration A calibration graph obtained when a solution containing 5 pgml-1 of copper was pumped through the manifold using an acceleration of 1.4 pl s - ~ for the pump flow rate and using pure water as the compensating solution through the T-piece is shown in Fig.2(a). The plot corresponds to a direct recording obtained by a computer in which the software took 1000 pairs of absorbance-time data. To aid understanding the axis of the abscissa which is really a time axis has been converted into a concentration axis by using the equation Cp = C,tT-’. In the instrument used (Solaar Model) the calibration graph obtained using the conventional method was linear up to about 4pgml-l. The points drawn on the line indicate for comparison the absorbance values obtained when the pump was stopped and several solutions of copper were aspirated through the T-piece. The experimental data obtained by the direct recording were fitted to both a linear and a second- order equation.A plot is shown in Fig. 2(b) of the residual values C ( y - 9 ) for the second-order fit y being the experimen- tal value of the absorbance and j the absorbance value provided by the least-squares fitting. Similarly good fitting results were obtained up to 4 pg ml-1 in the case of the linear fitting [Fig. 2(c)]. A number of similar experiments were also performed using calcium as the analyte. In this case to obtain 0.5 0.4 a 2 0.3 e m z 0.2 0.1 0 40L. 9 (a) ii 20s I 1 I 1 I I Time - 0.02 lb’ -0.02 !- 2 I ’ C‘C’ a 0.02 555 -0.04 I I 1 I I 0 1 2 3 4 5 [Cul/pg ml-’ Fig. 2 (a) Time tracing showing the production of a calibration graph. Points indicate the absorbance values obtained in the conventional way when different concentrations of copper were aspirated; (b) and (c) residual values obtained from a second order (0-5 pg ml-l) and a linear fitting (0-4 pg ml-’) respectively a wide non-linear region a calibration graph up to 10 pg ml-1 was used.The main statistical parameters obtained are summa- rized in Table 1. The errors associated with measuring ‘unknown’ solutions using the calibration graphs obtained and fitted as described were lower than +3% over the whole concentration range studied. When the quality of the fits was judged by the ‘quality defined as where N is the number of data points not including the zero point high QC values of close to 30% were obtained. These poor QCs did not vary significantly when weighting factors were used in the fits and were nearly independent of the number of data points used. This is due to the initial portion of the calibration graph being severely affected by the pulsation of the peristaltic pump which considerably increases the average relative deviation i.e.the QC value. The QC values calculated for different portions of a calibration graph are shown in Fig. 3. It is clear that if the first 3% of data are discarded QC values decrease considerably i.e. the fit is better. However it appears to be unnecessary to excluded556 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY APRIL 1994 VOL. 9 Table 1 Parameters obtained in the fitting of calibration graphs for calcium and copper s standard deviation; R2 multiple correlation coefficients; r correlation coefficient; F F values; a 95% confidence level was used for the calculation Function Parameter A = a + bCp + CC,? a+s c + s R2 F w - 9 ) 2 b+Sb A = a+ bC a+s b+Sb r w - 9 ) 2 Analyte 0-5 pgml-l Cu - 0.00 176 f 0.00049 -0.00417 f0.000089 0.128 f 0.00046 0.9989 440475 0.026 0-4 pg ml-I Cu 0.001 f 0.0005 0.114 k 0.00022 0.9987 0.057 0-10 pg ml-’ Ca 0.0001 2 f 0.00022 0.0340 -t 0.0001 -0.OOO75 f 1 x 0.9998 1.6 x lo6 0.002 0-4 pg ml-’ Ca 0.002 & 0.0002 0.03 1 1 0.0001 0.9987 0.002 25 30b R rn // / 10 - 8 - 6 - 4 - 2 - s 0 - I I I Range of analyte concentration/pg MI-’ Fig.3 calibration graph of (a) Ca; and (b) Cu Quality coefficients obtained from different portions of a these data since as is indicated in Fig. 4 the fit coefficients to a second-degree polynomial did not significantly vary within the calibration graph.As has been pointed the accuracy of the fitted graphs is better judged from the analytical results obtained for the ‘unknowns’. Eflect of Instrumental Vuriables An important point t o be considered is the effect on the reliability of the calibration graph of both acceleration of the I I I 1 1 I I I ( b ) m Y f “t 1 1 I 1 I I 1 1 J ,Q ,Q 20 ,Q \Q ,Q 2‘ Qy’ o” Q9’ 2’ 0’ 0’ Range of Ca concentration/pg ml-’ Fig. 4 Variation of the parameters a b and c (graphs (a) (b) and (c) respectively) for different portions of a calibration graph for calcium fitted to a second order equation flow rate and the data collection rate. It is predictable that if the gradient rises too fast problems can arise owing to the inertia of the instrumentation. Calibration graphs obtained when different ramps were used for the programmed flow rate are shown in Fig.5. For all cases the graphs which are displaced on the x-axis for better visualization were obtained from 100 absorbance-time data pairs taken by computer. The straight lines plotted on the graphs correspond to the least-squares fitting of the data. Inertia effects are evident for ramps faster than 200 mV s-l (i.e. 5.6 pl sP2) which produce a decrease in the slope of the straight line and consequently surplus errors when quantifying an unknown. Naturally the optimum values for the ramp and the data collection rate are related. The results for a number of calibration graphs for calcium obtained using 10 pg ml-1 for C and different values for the ramp and data collection rate are summarized in Fig.6. The values of Z ( Y - ~ ) ~ at theJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY APRIL 1994 VOL. 9 557 I [Cul- Fig.5 Calibration graphs for Cu obtained using different ramps for the increasing pump rate. Curves A-E correspond to flow-rate acceler- ations of 1.4 2.8 5.6 14 and 28 pl s-' respectively Fig. 6 Effect of both the flow rate acceleration and data collection rate on the error associated with a second degree polynomial fitting of the signal obtained for 0-10 pg ml-' of Ca (see text for details) intermediate concentration of 5 pg ml-' are plotted on the z-axis. The points drawn on the graph correspond to combi- nations of both variables leading to values of Z(y-9)' below 0.00014 and so to errors below &2% at that concentration level.The experimental data indicate that it is possible to use a 200 mV s-' ramp if at least 20 measurements per second are taken. For a 100mVs-' ramp the data acquisition rate can be changed to 0.2 s-'. From a practical point of view when a computer is used for data collection and handling a data acquisition rate of 0.2 s-' with a ramp of 50 mV ~ ~ ' ( 1 . 4 p1 s - ~ ) appears to be adequate since a reliable calibration graph is obtained in only 100 s and the data file generated is not voluminous. When an x-y recorder is used instead of a computer a ramp of 100mVs-I appears to be a suitable choice since the most common data acquisition rate of these devices is 0.1 s-'. In this way the calibration graph is obtained in only 50s. In any case it should be noted that the inertia effects which are the principal factors which limit using high ramp values are dependent on the instrumentation used and must be experimentally verified. As already indicated because of the action of the rollers the peristaltic pump produces pulsations in the flow delivered and so undesirable fluctuations are obtained in the absorbance values especially at the beginning of the calibration graph when the pump is working at a low rate.The pulsation effect could not be sufficiently damped using the usual pulse sup- pressors as inertia effects increased considerably. However owing to the way in which the data are obtained this drawback is not in practice as severe as first appears. Several portions of a calibration graph for calcium are shown in Fig.7. The y- axis has been amplified for better visualization and the straight lines indicate the least-square fit of the data. The noise obtained for the same instrument when a solution of 3 pgml-' of calcium is measured using the conventional mode is also included for comparison. The absorbance-time plot obtained is periodical with a time-variable frequency and an amplitude which is also variable. The signal could be treated by means of the computer using transforms but a simple least-square fit appears to be sufficient for practical purposes since the number of data taken by the computer is sufficiently high to make the fit reliable. Thus for example when a calibration graph for calcium is obtained in the 0-10 pg ml-' range the point for 1 pg ml-' is obtained at a pump rate of about 3 rev min-' and the period of the signal is about 2 s which means that ten measurements are obtained by the computer in this period.As the pulsations produce fluctuations not higher than an absorbance of about 0.006 it appears reasonable from a practical point of view that ten measurements can provide a sufficiently reliable mean absorbance value within the period and this reliability further increases if the data collection rate is increased. On-line Dilution Performance Another feature of the manifold is that it permits the on-line dilution of moderately concentrated solutions which cannot be measured by direct aspiration. Direct recordings obtained when pure water was pumped are shown in Fig. 8(4 using a linearly programmed flow rate with a ramp of 50 mV s-' and several solutions of copper in the 5-80 pg ml-' range were aspirated for compensation through the T-piece.The solid lines on the graph indicate the least-square fitting for the absorbance values within the linear response range of the instrument. As shown in Fig. 8(b) and (c) with the exception of the solution containing 80 pg ml-' of copper there was a linear relationship between the concentration of the solution aspirated and both the slope and the intercept of the lines. A comparison of the slopes with that of a calibration graph led to the results shown in Table 2. The poor result obtained for the 80 pg ml-' copper solution is owing to the fact the sample needs a 20-fold dilution to fall within the linear response range.As a consequence the fitting is only made with the data acquired in the last 5 s of the experiment which reduces its reliability. Of course it is possible to use a slower ramp to extend the rectilinear portion of the graph but this produces a decrease in the sampling frequency. In practice the approach appears to be only suitable up to a 10-fold dilution. On-line Standard Additions Method The approach discussed here can present problems as occurs when FAAS is used in the conventional manner owing to differences between the composition and physico-chemical properties of the unknown solutions and those used as stan- dards. The use of 1inea.l~ programmed flow rates is a con- venient way to overccme such difficulties to add common releasing agents or to serform the standard additions method.For matrix interfereaGes a straightforward solution is to use a sample of known concentration for calibration purposes spiked if necessarj with analyte to cover the entire range of concentrations n*:eded. To check the Zkpplicability of the standard additions method a 4pgml-' solution of copper was pumped with a pro- grammed flow rate using different solutions of the analyte for compensatiot. The recordings obtained are shown in Fig. 9. As expected once the data were fitted to straight lines the558 +? t s a Q JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY APRIL 1994 VOL. 9 0.01 A ;I I - I ICal - Fig. 7 Amplified time tracing showing different portions of a calibration graph for 0-10 pg ml-' of Ca. A-D curves corresponding to 0-1 1-2 2-3 and 3-4 pg ml-' of Ca respectively.Straight lines indicate the linear fitting of the data within each range. Curve E shows the signal fluctuation with the pump stopped when 3 pg ml-' of Ca was aspirated in the same instrument Table 2 Results obtained using the on-line dilution procedure for several solutions of copper 1.5 0 0 20 40 60 80 100 Timefs 0 20 40 60 80 ICul/pg ml-' Fig. 8 (a) Signals obtained when distilled water was pumped with an increasing flow rate and solutions containing 5 10 20 40 and 80 pg ml-' (curves A-E respectively) of Cu were aspirated through the T-piece. Straight lines correspond to the linear fitting (A = a + bC) of the data with absorbance values below 0.45 (maximum value within the linear range); (b) and (c) variation of a and b fitting parameters respectively with the Cu concentration Analyte content/pg ml-' Found using Nominal eqn.(7)* 5 10 20 40 80 4.95 f 0.02 10.3 f 0.03 19.7 & 0.04 40.3 & 0.04 75.1 & 0.07 * Mean value &confidence interval for five measurements (95% confidence level). 0.4 0.3 0.1 ~ 0 10 20 30 40 50 Time/s Fig.9 Time tracings obtained when a 4pgml-' Cu solution was pumped at an increasing rate (2.8 pl sP2) and solutions containing 0 1 2 3 and 4 pg ml-' of the same analyte (curves A-E respectively) were used for compensationJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY APRIL 1994 VOL. 9 - - - - 559 0.8 0.6 - - % 0.4 z 0.2 - 0 0.16 0.14 0.12 0.10 Q) 0.08 0 m + v) 2 0.16 0.14 0.12 0.10 0.08 0 10 20 30 40 50 d 10 20 30 40 50 0 - 0 15 30 45 0 % 150 1 ‘d) 1 .o 1 I I I I 0 15 30 45 Ti rn e/s Fig.10 Effect of citric acid on the iron signal (a) a 5 pg ml-’ iron(m) solution was pumped at an increasing flow rate and solutions con-_thing 5 pg ml-’ of iron with different amounts of citric acid (0 0.005 0.01 0.02 0.03 and 0.04% m/v curves A-F respectively) were used for compensation; (b) variation of the citric acid concentration in (a) with time; (c) and ( d ) as for (a) and (b) but the iron solution being pumped also contained 1% m/v NaC1. Curves A-D correspond to 0.015 0.05 0.08 and 0.12% m/v citric acid respectively in the solutions used for compensation. Line E on (d) indicates the variation in the NaCl concentration with time slopes were found to be proportional to the concentration of the solution being aspirated.The slopes and intercepts were obtained for a number of ‘unknowns’ and the application of eqns. (8) and (9) permitted their concentrations to be calcu- lated. The results are shown in Table 3. For the case of a typical chemical interference the pro- portionality constant of the spectrometer is different for the sample and for the standard solution. This means that the response factor of the instrument must be time variable since the mixture entering the nebulizer is constantly changing its composition along the gradient. If the interference is not too severe it should disappear when the concentration gradient is carried out since when the flow delivered by the pump increases the interferent to analyte ratio decreases. To check this a number of experiments were made using iron(Ir1) and citric acid solutions.The recordings obtained when a 5 pg ml-’ Table 3 Results obtained in the determination of copper (pg ml-I) using the standard additions procedure Nominal Found using eqn. (8) Found using eqn. (9) 0.5 0.48 f 0.03 0.48 Ifr 0.03 1 .o 1.01 f0.02 0.98 & 0.02 1.5 1.51 f0.02 1.51 k0.02 2.0 2.00 & 0.02 2.01 & 0.03 2.5 2.49 & 0.03 2.52 f 0.03 3.0 3.01 k 0.03 2.99 f 0.03 * Mean value &confidence interval for five measurements; 95% confidence level. solution of iron was pumped at an increasing rate and solutions of the same concentration but containing different amounts of citric acid were used for compensation are shown in Fig. 10. As expected and as shown in Fig. lO(a) the lower the inter- ferent concentration the lower the degree of dilution needed to suppress the deleterious effect of citric acid.The addition of 1% m/v sodium chloride to the standard solution allowed suppression of the interference with a lower degree of dilution [Fig. lo@)]. If the chemical interference is severe a releasing agent must be used. The recordings obtained when a 10 pg ml-’ solution of calcium was pumped in the programmed mode and solutions containing different concentrations of phosphate were used for compensation are shown in Fig. ll(a). It is clear that the dilution suffered by the interferent along the gradient i.e. the increase in the analyte to interferent ratio was not sufficient to overcome the interference. The plots shown in Fig. ll(b) were obtained when the standard additions method was per- formed on ‘unknowns’ containing 4 pg ml-’ of calcium and different amounts of phosphate. In this case a 10 pg ml-’ calcium standard solution containing 2% m/v EDTA (ethylen- ediaminetetraacetic acid disodium salt) was pumped at an increasing rate.It can be seen from Fig. ll(b) that even for a phosphate to calcium ratio of 20 a rectilinear plot is obtained in the final portion of the gradient. Comparison of the slope of this rectilinear portion with that shown by the same portion of a calibration graph permitted the calcium concentration of ‘unknowns’ to be calculated by applying eqn. (8). The results are shown in Table4. The intercept of these straight lines560 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY APRIL 1994 VOL. 9 0'4 0.3 ~' 0.4 0.3 a 8 rn 0.2 $ 2 0.1 0 0 10 20 30 40 50 Time/s - 1 I I 0 15 30 45 Time/s 2.0 1.5 - 1.0 E a z 0.5 & 0 r I - E 0) 2 0 0 0 15 30 45 Time/s Fig.11 Effect of phosphate on the Ca signal (a) A 10 pg ml-' Ca solution was pumped at an increasing flow rate and solutions containing different amounts of phosphate (0 5 x 1 x lo-' and 2 x lop3 mol l-' curves A-D respectively) were used for compensation; (b) variation of phosphate concentration in (a) with time; (c) as for (a) but the solution being pumped also contained 2% m/v EDTA and the solutions aspirated contained in addition to phosphate 4 pg ml-' of Ca. The lines E and F in (d) indicate the variation in the Ca and EDTA concentrations respectively Table 4 containing phosphate Results obtained for the determination of calcium in samples Content*/pg ml-l [P0,3-]/mol 1-1 5 10-4 1 10-3 2 x 10-3 5 x 10-4 1 x 10-3 2 x 10-3 5 x 10-4 1 x 10-3 2 10-3 Nominal 2 2 2 3 3 3 4 4 4 Found using eqn.(8) 1.98 & 0.02 2.01 kO.01 1.97 k 0.02 3.02 f 0.02 2.99 k 0.01 2.98 k 0.02 4.02 k 0.03 3.97 f 0.02 4.03 & 0.03 Found using 1.99f0.01 1.98 & 0.02 2.01 kO.01 2.98 k 0.02 3.02 k 0.02 3.02 0.03 3.98 & 0.02 4.03 0.03 4.01 f 0.02 eqn. (9) * Mean value &confidence interval for five measurements; 95% confidence level. allowed the concentrations to be calculated by means of eqn. (9) and the results are also shown in Table 4. Conclusions The strategy studied here permits reliable calibration graphs to be obtained in a few seconds by using a single standard solution and on-line dilution of moderately concentrated solu- tions.Practical aspects of the standard additions method are also simplified. For this procedure when a chemical inter- ference is present it is not necessary to add a releasing agent to the sample but only to the standard solution used An important point for the success of the approach is the use of a computer for data acquisition and handling since in this way many absorbance-time data pairs can be obtained and the fits can circumvent the inertial effects of the instrument as well as the drawback owing to the pulsations caused by the peristaltic pump. As indicated by Bysouth and Tyson,22 these pulsations are confirmed as the main problem associated with these types of approaches. The manifold presents other features whose uses are now being investigated.Financial support from the Spanish Direccion General de Investigacion Cientifica y Tecnica (DGICYT) (Project PB 90-0302) is gratefully acknowledged. 1 2 3 4 5 6 7 8 9 10 11 12 13 References Tyson J. F. Anal. Proc. 1981 18 542. Tyson J. F. and Idris A. B. Analyst 1981 106 1125. Tyson J. F. Appleton J. M. H. and Idris A. B. Anal. Chim. Acta. 1983 145 159. Tyson J. F. Analyst 1984 109 319. Tyson J. F. and Appleton J. M. H. Talanta 1984 31 9. Tyson J. F. Adeeyinwo C. E. Appleton J. M. H. Bysouth S. R. Idris A. B. and Sarkissian L. L. Analyst 1985 110 487. Tyson J. F. Anal. Chim. Acta 1986 179 131. Tyson J. F. Mariara J. R. and Appleton J. M. H. J. Anal. At. Spectrom. 1986 1 273. Tyson J. F. and Bysouth S. R. J. Anal. At. Spectrom.1988,3,211. Tyson J. F. Fresenius' 2. Anal. Chem. 1988 329 663. Beinrohr E. CsCmi P. and Tyson J. F. J. Anal. At. Spectrom. 1991 6 307. Olsen S. Rfiii6ka J. and Hansen E. H. Anal. Chim. Acta 1981 136 101. Reis B. F. Jacintho A. O. Mortatti J. Krug F. J. Zagatto,JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY APRIL 1994 VOL. 9 561 E. A. G. Bergamin H. F. and Pessenda L. C. R. Anal. Chim. Acta 1981 123 221. 14 Araujo M. C. U. Pasquini C. Bruns R. E. and Zagatto E. A. G. Anal. Chim. Acta 1985 171 337. 15 Fang Z. Harris J. M. REiEka J. and Hansen E. H. Anal. Chem. 1985 57 1457. 16 Sherwood R. A. Rocks B. F. and Riley C. Analyst 1985 110 493. 17 Zagatto E. A. G. GinC M. F. Fernandes E. A. N. Reis B. F. and Krug F. J. Anal. Chim. Acta 1985 173 289. 18 Koicielniak P. J. Anal. At. Spectrom. 1987 2 329. 19 Whitman D. A. and Christian G. D. Talanta 1989 36 205. 20 Sperling M. Fang Z. and Welz B. Anal. Chem. 1991 63 151. 21 Fang Z. Welz B. and Sperling M. Anal. Chem. 1993 65 1682. 22 Bysouth S. R. and Tyson J. F. J. Anal. At. Spectrom. 1987,2,217. 23 Starn T. K. and Hieftje G. M. J. Anal. At. Spectrom. 1992,7,335. 24 Thompson M. and Ramsey M. H. J. Anal. At. Spectrom. 1990 5 701. 25 Bosch Reig F. Bosch Mossi F. Peris Martinez V. and Pastor Garcia A. J. Anal. At. Spectrom. 1992 7 47. 26 Gilbert P. T. Jr. Anal. Chem. 1959 31 110. 27 Shatkay A. Anal. Chem. 1968 40 2097. 28 Knegt J. and Stork G. Fresenius’ 2. Anal. Chem. 1974,270 97. 29 de Galan L. van Dalen H. P. J. and Kornblum G. R. Analyst 1985 110 323. 30 Hu Y. Smeyers-Verbeke J. and Massart D. L. J. Anal. At. Spectrom. 1989 4 605. Paper 3/06307K Received October 21 1993 Accepted January 4 1994

ISSN:0267-9477    

DOI:10.1039/JA9940900553

出版商:RSC    

年代:1994

数据来源: RSC