JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY SEPTEMBER 1994 VOL. 9 919 Vaporization of Acids and Their Effect on Analyte Signal in Electrothermal Vaporization Inductively Coupled Plasma Mass Spectrometry* D. Conrad Gregoire Geological Survey of Canada Department of Natural Resources Ottawa Ontario Canada KIA OE8 Douglas M. Goltz Marc M. Lamoureux and Chuni L. Chakrabarti Otta wa-Carleton Chemistry Institute Department of Chemistry Carleton University Ottawa Ontario Canada KIS 566 The vaporization properties of HCI and HN03 under various furnace heating conditions were investigated. Drying-step temperatures of 140 "C (50 s) and pyrolysis-step temperatures of 400 "C (1 0 s) were effective in volatilizing most of the chloride from 10 1.11 of 1% v/v HCI however a small amount (40 ng) of acid was retained on the graphite even after pyrolysis at 400 "C. Under the same experimental conditions HNO was completely volatilized from the graphite tube.The effect of a range of concentrations of HCI HNO H2S04 and H3P0 on analyte signals was studied for Co Cu Ag Cs Pb Bi and U. Analyte signals were enhanced by as much as a factor of two in the presence of 1% v/v HN03 and H2S04. Phosphoric acid suppressed analyte signals for Ag and Bi and the use of HCI resulted in relatively small changes in analyte sensitivity. The use of a pyrolysis step in the heating programme reduced the effects associated with acid matrices but at the expense of signal intensity. A mixed modifier-carrier reduced the matrix effects associated with H2S04 and H3P04 and essentially eliminated them for HNO and HCI.Keywords Inductively coupled plasma mass spectrometry; electrothermal vaporization; interferences; acid; chemical modifier Electrothermal vaporization (ETV) as a means of introducing a sample into a plasma has great potential as an analytical tool for ultra-trace analysis. The ETV device was developed to overcome some of the inherent difficulties of solution nebulization in inductively coupled plasma mass spectrometry (ICP-MS). For example the sample transport efficiency using ETV was reported to be about 80% or greater compared with 10% or less for solution nebulization.' The ETV device can also be used to remove the bulk of the water and possibly matrix components through the use of a pyrolysis step in the temperature programme.This makes possible ICP-MS measurements under favourable conditions which include a dry plasma free from solvent 'loading' effects associated with solution nebulization sample introduction.2 have used mass spectrometry to study processes in the graphite furnace under vacuum con- ditions. Findings from these studies were then applied to problems in electrothermal atomic absorption spectrometry (ETAAS). GrCgoire et aL7 have noted that fundamental infor- mation obtained by ETAAS can be readily applied to studies in ETV-ICP-MS. Similarly it has been shown that ETV- ICP-MS can be used for studying numerous processes in the graphite furnace such as interference^,^,^ atomization and vaporization mechanisms.8 Although some studies on matrix effects associated with ETV have been reported only a few have investigated the chemical or physical behaviour of mineral acids in an ETV device and its effects on ICP-MS analyte signals.The effect of acids on 63Cu+ signals from an ICP-MS instrument using a rhenium strip electrothermal vaporizer has been reported by Park et d9 They showed that 1% HC1 HNO HF and H,SO suppressed Cu signals compared with non-acidified Cu solutions. For the acids studied they observed peak broadening with 3% acid which was attributed to unexplained chemical effects. These workers also found that 3% HNO or HF did not reduce the integrated signal whereas 3% HC1 or H,S04 A number of * Presented at the 1994 Winter Conference on Plasma Spectro- GSC contribution 40993 chemistry San Diego CA USA January 10-15 1994.did. Some characteristics of a tungsten furnace have been investigated by Tsukahara and Kubota," who showed that increased HCl concentrations resulted in an increase in the amount of tungsten vaporized. Judicious selection of acids used in ICP-MS aids in avoiding many troublesome spectroscopic interferences. Nitric acid is frequently the acid of choice for solution nebulization ICP-MS determinations because the N+ 0' and H+ species or any polyatomic combination of these are already present in the plasma and originate from water argon and atmospheric gases entrained into the argon plasma. The background spectra of HN03 is relatively simple resulting in fewer spectroscopic interferences compared with HCl which can produce a number of chloride interferences such as the interference 35C1160 + on V (m/z 51).Sulfuric and phosphoric acids are generally avoided since a large number of S and P containing polyatomic species can occur." The success of an ETV-ICP-MS determination is dependent on the complete vaporization of analyte from the graphite tube and the efficient transport of analyte to the argon plasma. The selection of acid as well as the concentration used can affect either analyte vaporization transport or both. The retention of water or acid within the graphite vaporizer even following extended high temperature ( 100-300 "C) evapor- ation is possible and could also have an impact on analyte ETV-ICP-MS signals. This paper reports on the effect of the vaporizer temperature programme on the retention of water HCl and HNO,.Analyte signal pulses for a number of elements are reported for vaporization under compromise multi-element conditions using HCl HNO H,S04 and H3P04 acids as diluents. The use of a mixed chemical modifier to control acid effects is assessed. Experimental A Perkin-Elmer SCIEX Elan Model 5000 ICP-MS instrument equipped with an HGA-600MS electrothermal vaporizer was used. The ETV system was fitted with a Perkin-Elmer AS-60 autosampler. The experimental conditions for both the Elan920 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY SEPTEMBER 1994 VOL. 9 5000 and the HGA-600MS are given in Table 1. Standard pyrolytic graphite coated graphite tubes were used for all experiments. Solution nebulization sample introduction was used to optimize the plasma and ion-lens settings of the mass spec- trometer.The ETV device was interfaced to the plasma with an 80 cm length of PTFE tubing (6 mm id.). The flow of argon from the HGA-600MS power supply was regulated with a pneumatic valve and operation of the HGA-600MS was com- pletely computer controlled. Under normal experimental con- ditions during the drying and pyrolysis steps of the temperature programme opposing argon gas flows (300 ml min-') from the ends of the graphite tube remove water and other vapour phase decomposition products through the dosing hole. In this way the matrix components are not introduced into the plasma. During the measurement of analyte signal pulses typically the high-temperature vaporization step a graphite probe seals the dosing hole and the flow of argon is directed from the far end of the graphite tube directly towards the plasma.For studying the temporal vaporization behaviour of HCl and HNO however the dosing hole was sealed during the entire temperature programme in order to observe the amount of ,'Cl I4Nl6O or "N produced from the vaporization of acid during each step in the temperature programme. While this method allowed observation of the temporal behaviour of these ions a quantitative determination of the acid retention after the drying or pyrolysis steps using this method could be misinterpreted. This was due to the larger than expected signals for 35Cl+ '5N'60+ and "N' which arose from the vaporiz- ation of acid that had condensed onto the cooler surfaces such as the transfer line during the drying and pyrolysis heating steps when the dosing hole was sealed. The temporal study of vaporization of HC1 and HNO was accomplished by monitoring the 35Cl+ 14N160+ and "N+ ions and the integrated ion counts were calculated using an 'in-house' Turbo Pascal program using the trapezoidal rule of numerical integration.For studying the effect of acids on analyte signals the integrated signal was reported using the Elan software provided. For all experiments 5-20 pl of sample were pipetted onto the walls of the graphite furnace. For each experiment a minimum of five runs were performed. All measurements were blank subtracted using water as a reagent blank for the acid studies a dry tube for the water studies and NASS-3 sea-water standard as the chemical modifier for experi- ments involving the use of a mixed carrier.The ICP-MS measurements were made using the high-resolution mode (0.7 u Table 1 Instrumental operating conditions and data acquisition parameters at 10% peak height) because of the possible peak overlap of 36Ar+ on the 35Cl+ peak. All solutions were prepared with ultra-pure water obtained from a Milli-Q water purification system (Millipore Mississauga Ontario Canada). The HCl HNO and H2S04 were Ultrex 11 ultrapure-reagent grade (Baker Analyzed J. T. Baker Canada Toronto Ontario). Phosphoric acid (Baker Analyzed) was laboratory-reagent grade. The "N-labelled HNO was purchased from CIL (Cambridge Isotope Laboratories Woburn MA USA) with a concentration of 40% v/v and the isotopic purity was 99% "N. Stock standard solutions were supplied by Spex Industries Edison NJ.The NASS-3 Open Ocean Reference Material for Trace Metals was obtained from the Institute for Environmental Chemistry of the National Research Council of Canada. For the mixed carrier work NASS-3 sea-water was subjected to column chromatography'* to remove any trace elements. A further reduction in trace contaminants was also achieved by diluting the purified NASS-3 sea-water 500-fold with de-ionized water. The carrier gas for all of these experiments was high-purity argon (99.9 %). Results and Discussion Vaporization Properties of Water It is known that water adsorbed on a graphite surface can react during the pyrolysis step to form hydrogen. Interestingly it has also been observed that water is actually difficult to remove from graphite even at elevated temperatures.' Frech and co-workers demonstrated' and Welz reported14 that in an uncoated graphite tube water was actually retained in sufficient amounts even after 15 min at 1200 OC.13*14 These workers also estimated the amount of residual water in a graphite tube following the drying step to be 1 pmol when 1 pl of water was deposited on a graphite tube and dried at 80°C.16 The ICP-MS background spectra have been docu- mented for a dry argon and the intensity of 36ArH+ signal at rn/z=37 has been shown to be indicative of the amount of water present.The temporal behaviour of the vaporization of water from the graphite furnace as monitored by the argon hydride ion is shown in Fig. 1. Corrections for any spectral overlap from the ,'Cl+ ion was accomplished by monitoring the 35Cl -+ ion and applying the appropriate correc- tion.The HGA-600MS was equipped with a pneumatically controlled graphite probe which effectively sealed the dosing hole. The positive pressure of argon within the graphite tube made it unlikely that significant amounts of entrained air or water vapour entered the graphite tube through the dosing hole. inductively coupled plasma mass spectrometer - R.f. power 1000 w Outer argon flow rate Intermediate argon flow rate Aerosol carrier argon flow rate Sampler/skimmer cones platinum 15.0 1 min - 850 ml min - 900 ml min-' Electrothermal vaporizer H GA-600MS - Sample volume Internal argon flow rate Dry step Dry ramp Pyrolysis step Pyrolysis ramp Vaporization step Heating rate 5-10 ~1 300 ml min- 80-140°C (30 S ) 2-20 s 400-1400 "C (10 S) 1-2 s 2400 "C (10 s) 2000 "C s - 2oo I 180 ," 160 2 140 I c 0 m 120 2 3 100 4- .- v) 6 80 .- 60 & 40 v 20 w - m .- Data acquisition - Dwell time 20 ms Scan mode Peak hop transient Points per spectral peak 1 Signal measurement Integrated signal pulse 0 10 20 30 40 50 60 70 80 Time/s Fig.1 Temporal behaviour of water in the electrothermal vaporizer monitored using the 36ArH+ polyatomic ion drying temperature = 140 "C (0-50 s); pyrolysis temperature = 400 "C (50-60 s); and vaporiz- ation temperature = 2400 "C (60-70 s)JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY SEPTEMBER 1994 VOL. 9 The temporal behaviour of 36ArH+ in the ETV device confirmed that the bulk of the water from a lop1 sample is removed during the drying step of the temperature programme.A pyrolysis step of 400°C was added however no significant signal for 36ArH + was observed until the high-temperature vaporization step at 60-70 s. The temporal behaviour of 36ArH+ as shown in Fig. 1 provides only a qualitative view of water retained on the graphite surface of the electrothermal vaporizer because some of the 36ArH + signal obtained during the vaporization step originated from water that had condensed on the graphite cones of the ETV device as well as on the transfer line. Retention of water in the ETV device is of some importance in ETV-ICP-MS because of its role in the production of hydride and oxide species and was included as a pretext for the vaporization properties of acids.These hydride and oxide polyatomic ions can interfere with other analytes in a multi- element analytical scheme. For example the production of could interfere with the determination of trace amounts of Fe at m/z=56. The presence of water could also lead to the formation of refractory oxides for some analytes such as the rare earth elements. 4 0 ~ ~ 1 6 0 + Vaporization Properties of Acids Drying and pyrolysis steps in the temperature programme of the ETV investigations can result in the selective volatilization and removal of the bulk of the sample matrix material from the graphite furnace. During the drying step most of the solvent matrix can be driven from the surface of the graphite furnace with the bulk of the remaining portion being removed at higher temperatures during the pyrolysis or vaporization steps.In the case of oxyacids it is believed that these can be retained within the surface of the graphite itself even at temperatures of up to 1000 OC." Pyrolytic graphite coated graphite tubes are less porous and therefore less prone to solvent entrapment than uncoated tubes and thus the retention of acids is greatly reduced. By studying the temporal behaviour of these acids a better understanding can be obtained of how they can affect analyte sensitivity in ICP-MS either by signal suppression (matrix effect or oxide formation) or by signal enhancement (improved analyte volatility and/or trans- port efficiency). Hydrochloric acid (i) Eflect of drying-step temperature. To determine the importance of the temperature of the drying step on the removal of chloride from the graphite furnace 1% v/v HCl was deposited on the walls of a pyrolytic graphite coated graphite tube.The temporal behaviour of HCl during the heating cycle was monitored using the ,'Cl+ ion. Hydrochloric acid was chosen because it is a common sample diluent it is a fairly volatile acid and thus its behaviour is similar to other volatile acids such as HF or possibly HNO,. The 35Cl+ ion was monitored and the counts generated from this ion were proportional to the amount of HC1 present. To investigate the role of the temperature of the drying step on acid retention this temperature was varied from 90 to 160°C and the HCl remaining on the surface during the high-temperature vaporiz- ation step was monitored. A pyrolysis step was not used.The intention of these experiments was to assess the retention of Cl in the ETV device after the drying step qualitatively. The results of these experiments shown in Fig. 2 demonstrate the limits of the drying step in removing chloride from the furnace. Only a slight increase in the integrated ,'Cl+ ion count was obtained as the drying-step temperature was increased from 90 to 160°C with the ion counts reaching a plateau after 120°C. Similarly the number of counts measured during a subsequent vaporization step did not change significantly above 120 "C. The fact that increasing the drying step beyond v) +- 5 150 0 m 125 Q c 100 .- 2 75 -0 Y- - m C .- 50 0 -0 + - 25 e I e B o C - rn Drying step Vaporization step 90 100 110 120 140 160 Drying temperaturePC 92 1 v) 3 + 30 8 rn 0 v -..25 c C 0 20 '4= 2 .- L 0 15 L Y- 0 10 a v) .- 5 b In -0 0 2 I m c C - Fig. 2 Effect of drying temperature on the integrated signal of 35Clf during the drying and vaporization steps using a 1 YO v/v HCI solution drying time = 50 s; and vaporization temperature = 2400 "C ( 10 s) 140°C had no effect on the amount of 35Cl+ removed from the ETV device could also indicate that the acid was actually condensed on cooler regions of the vaporizer. Therefore even with an increased drying temperature or drying time the retained acid was not removed from the ETV device until the high-temperature vaporization step. To determine whether the 35Cl + signal actually originating from chloride condensed or adsorbed on the transfer line rather than from the graphite tube 1% v/v HCl was dried in the graphite furnace in a manner similar to previous experi- ments.Prior to the high-temperature vaporization step the transfer line was disconnected and the ETV device was cleaned by three separate 2400 "C heating cycles for 10 s each. The transfer line was then re-connected and a 10 pl sample of water was deposited on the walls of the ETV device. The vaporizer was then subjected to a drying step and heated to 2400°C during which time the 35Clf intensity was monitored. In this manner any 35Clf residing in the transfer line would be vaporized and detected by the ICP-MS instrument. This experiment showed that approximately 65% of the retained C1 observed was actually originating from the transfer line. Therefore for a 5 pl sample of 1% v/v HC1 which yielded 18 000 counts during the vaporization step approximately 12000 counts would originate from the transfer line.This result is important to those using an ETV design' encorporat- ing a waste vapour venting system located some distance downstream from the vaporizer surface. These devices allow for condensation of water and acids which can easily be volatilized and carried with the analyte to the argon plasma during the high-temperature vaporization step. (ii) Eflect of pyrolysis-step temperature. For these experi- ments a drying-step temperature of 140°C was used. This temperature was reached by constantly increasing the tempera- ture of the ETV device or ramping for 10s until the desired temperature was reached. The drying-step temperature was then held constant for 40 s.Following the drying step the HCl retained in the graphite tube during the pyrolysis step was again monitored using the 35Cl+ ion. Residual chloride remain- ing on the graphite was measured during the high-temperature vaporization step. Three temperatures were chosen for these experiments such that a constant heating rate was maintained under normal ramping conditions. The results of these experi- ments are summarized in Fig. 3 which shows that a greater portion of chloride can be removed when a very high pyrolysis temperature is used. A significant portion of chloride however remained on the graphite even after heating to 1400°C as shown by the number of 35Cl+ ion counts remaining during922 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY SEPTEMBER 1994 VOL.9 400 640 1400 Pyrolysis tempera t u re/"C Fig.3 Effect of pyrolysis temperature on the integrated signal of 35Cl+ during the pyrolysis and vaporization steps sample solution 1 % v/v HC1; drying temperature = 140 "C (50 s); pyrolysis time = 12 s; and vaporization temperature = 2400 "C (10 s) vaporization. It should be noted that a pyrolysis temperature of 1400°C is excessively high and would result in significant analyte losses for many volatile elements such as Cd Zn Ag and Pb for which a pyrolysis temperature of 500°C should never be exceeded unless a chemical modifier is used.lg The effect of the temperature ramp was also investigated for its possible role in the vaporization of chloride from the graphite surface.The temperature ramp time in the drying step was varied from 2 to 20s and from 1 to 10s for the pyrolysis step. These ramp times were chosen as they are typical of those used in ETAAS and ETV-ICP-MS. The temperature ramp of both the drying step and pyrolysis step had no significant effect on the amount of acid retained. The temporal behaviour of the vaporization of HCl during an entire ETV temperature programme (dry to clean-up) was monitored using the 35Cl+ ion and is shown in Fig. 4. A 140 "C drying step was used followed by a pyrolysis step of 1400°C. For this experiment water was also monitored using the 36ArH + ion during the entire temperature programme. Though not shown in Fig. 4 very little chloride from HCl was vaporized until most of the water was removed from the surface of the ETV device which occurred at about 20s into the heating cycle.Following the vaporization step at 2400"C a clean-up step of 2650°C was included during which a chloride signal was still observed. The high-temperature vaporization step and clean-up step have been magnified in Fig. 4 (b) to reveal some details of the analyte pulses in these steps. The chloride pulse in the high-temperature vaporization step is a sharp peak lasting approximately 3-4 s. The more broadly shaped chloride pulse in the clean-up step could be indicative of chloride which is deeply intercalated in the graphite or chloride that has condensed on the transfer line or contact cones. The integrated area of the acid pulse in the clean-up step was close to half the integrated area for acid volatilized during the vaporization step.As pointed out under Experimental this temporal picture of chloride vaporization in the ETV device was obtained by sealing the dosing hole during the entire temperature pro- gramme. During the drying and pyrolysis steps the flow of argon originates from one end of the ETV device to the other instead of both ends and through the dosing hole giving the chloride a possible means of condensing on the transfer line. As a result the chloride peak during the high-temperature vaporization step in Fig. 4 is larger than expected and was caused by chloride that originated from chloride condensed onto the transfer line as well as from the ETV device. Under normal conditions with the dosing hole open for the drying and pyrolysis steps the chloride signal in the high-temperature vaporization step is much lower.Semi-quantitative analysis 7 00 600 500 400 300 200 r 'v) 100 z o 2 cn C .I- m -. t. > 0 20 40 60 80 100 120 80 90 100 Time/s Fig. 4 Temporal behaviour for the vaporization of 1% (v/v) HCI in a pyrolytic graphite coated graphite tube monitored using 35Cl+ drying temperature = 140 "C (0-60 s) pyrolysis temperature = 1400 "C (60-82 s); vaporization temperature = 2400 "C (82-92 s); and clean-up temperature = 2650 "C (92-102 s) for the amount of chloride retained under normal conditions from 20 pl of 1% (v/v) HCl following a 140 "C drying tempera- ture and a 400 "C pyrolysis step revealed that approximately 0.05% of the original amount of acid as chloride was retained under these conditions.This amount of chloride retained corresponds to approximately 1.1 x lop3 pmol of chloride or 40000 pg which is in large excess compared with the amount of analyte vaporized in the ETV device typically from 1 to 50pg. Nitric acid To study the retention of HNO on graphite a suitable ion either mono- or polyatomic must be chosen that is solely derived from HNO,. Mass scans in the spectral region of m/z values of from 14 to 60 were obtained for 1% v/v acid and compared with the background spectrum obtained with de-ionized water. These experiments were similar to those reported in the literature for solution nebulization,20 except in this case the plasma was dry. Choosing a suitable ion was difficult because of the similarity between the mass spectra of the HNO and the mass spectra of the blank which was water.A number of ions were evaluated including 14N160+ I2Ci4N+ 40Ar14N+ 14N'602t and I4N+ but 14N160+ was the only species that gave a satisfactory response for monitoring the vaporization of HNO,. NitricJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY SEPTEMBER 1994 VOL. 9 923 acid is a major source of 14N160+ however another important source within the ETV device is the formation of 14N160+ resulting from the vaporization of water which reacts with 14N+ in the plasma. In practice however the 14N160+ ion was found not to be suitable for quantifying the amount of HNO retained on the ETV device because the experiments were difficult to reproduce compared with HCl and were susceptible to high background ion sources such as residual water in the ETV device or water condensed in the transfer line.For this reason "N-labelled HNO was substituted and 15N+ was used as an alternative ion to 14N160+ for monitoring the temporal behaviour of HNO in the ETV device. In this way a direct correlation could be made between the integrated signal of an ion from the acid and the amount of acid remaining in the furnace. At m/z = 15 relatively low background counts were observed which were of the order of approximately 2000 counts s-'. As with previous experiments using HNO a 1% v/v solution was prepared and the temporal behaviour of the HNO was monitored using the 15Nf ion. As shown in Fig. 5 the 15N+ ion appeared to behave in a manner similar to the 35Cl+ ion. During the drying step of the temperature pro- gramme the bulk of the 15Nf was not removed until most of the water had evaporated from the graphite surface. The vaporization of water was observed by monitoring the intensity of the argon hydride (m/z=37) polyatomic ion during the heating programme.As with the chloride ion most of the 15N+ was removed during the drying step however 10-15% of the original "N+ appeared to be retained until the high- temperature vaporization step. The signal for "N' during the vaporization step appeared to be unusually large especially when compared with the chloride signal from 1% HCl. A pyrolysis step was added to remove the residual 15N+ from the ETV device. Neither a 400 nor a 900 "C pyrolysis step had any effect on reducing the intensity of the 15N+ signal during the high-temperature vaporization step indicating that the 15N+ was not originating within the graphite tube.It was indicated under Experimental and explained for the temporal behaviour of chloride from HCl that the temporal behaviour of "N+ was studied with the dosing hole sealed during the entire temperature programme. As with the chloride from HCl when the dosing hole was sealed and the argon flow was directed from one end of the ETV device to the other the vapours of HNO had an opportunity to condense on the transfer line or possibly on the graphite contact cones of the furnace housing resulting in a larger than expected "N+ signal. When the dosing hole was left open during the drying 1200 ~ 1000 1 *0° t I \ 0 10 20 30 40 50 60 Time/s Fig.5 Temporal behaviour for the vaporization of 1 % v/v HNO in a pyrolytic graphite coated graphite tube using a "N-labelled nitric acid solution drying temperature = 120 "C (0-40 s); pyrolysis tempera- ture = 400 "C (40-50 s); and vaporization temperature = 2400 "C (50-60 S) and pyrolysis steps (140 and 400 "C respectively) as would be the normal procedure with the ETV device the flow of argon originates from the ends of the ETV and exits via the dosing hole removing any vaporized HNO from the ETV device. Under these conditions almost all (> 99.97%) of the HNO could be removed from the ETV device during the drying step of the temperature programme. Experiments were performed on both old tubes (>200 firings) as well as new tubes and it was found that tube age was not a significant factor in the retention of HNO in the ETV device.Effect of Acid on Analyte Signals For these experiments the analytes were vaporized at 2400 "C following a 140°C drying step. A pyrolysis step was not used because the objective of these studies was to use experimental conditions that are similar to those reported in the literature and to explore the actual need if any of this heating step. For example under compromise conditions relatively low pyrolysis temperatures such as 200"C have been used to prevent pre- vaporization losses of volatile elements such as Ag or Pb during these heating steps." The implications of using any acid as diluent for ETV-ICP-MS determinations can be import- ant particularly for multi-elemental analysis where compro- mise conditions are required.The effects of 0.05% v/v and 1.0% v/v HNO HC1 H,SO and H,P04 were determined for seven elements Co Cu Ag Cs Pb Bi and U. These elements were chosen both to cover a large atomic mass range of the Periodic Table as well as a range of analyte volatilities. The results of these experiments are summarized in Table 2. For all elements vaporized in the presence of HNO a 1% v/v HNO matrix gave higher integrated signals than the 0.05% HNO matrix. For some elements such as Ag or Pb the effects were fairly pronounced with close to two-fold increases in the integrated signal with increased acid concen- tration. Initially it would appear as though higher concen- trations of HNO would simply mobilize more analyte from the graphite surface however this is probably not the primary cause for the higher analyte signals in higher concentrations of acid.It is possible that a carrier effect with higher acid concentrations is occurring whereby HNO is acting as the physical carrier. Another possibility is the increased degred- ation of the graphite surface in the ETV device with higher acid concentrations which could result in the formation and liberation of soot particles during the high-temperature vapor- ization step. These fine carbon particles could then act as physical carriers themselves or possibly as sources of nucleation for the analyte. For HCl increasing the acid concentration in the sample matrix from 0.05 to 1.0% v/v did not have a significant effect on any of the elements. The integrated signal remained unchanged within experimental error with increased HCl concentration.The integrated signal for most of the elements was fairly similar in either 1% v/v HNO or 1% v/v HCl except for U which was noticeably lower in the HCl matrix. The HC1 matrix could act in two ways that are beneficial to analysis by ETV-ICP-MS. Firstly the HC1 could act as a physical carrier and thereby improve the mass transport of the analyte to the plasma. Alternatively HCl could react with the analytes forming halides which are more volatile than the corresponding oxides making this a potentially desirable acid to use for many analytes. When H,SO was used the integrated signal for most of the elements studied increased significantly with the 1 YO acid compared with HNO and HC1.The effects of increasing the acid concentration of H,S04 were also significant indicating the potential for severe matrix effects. As with HNO when the H2S04 concentration was increased from 0.05 to 1.0% v/v the integrated signal also increased. This was especially true for elements such as Cs and Bi. Phosphoric acid gave data which were unlike those €or the924 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY SEPTEMBER 1994 VOL. 9 Table 2 Effect of acid concentration on integrated analyte signal pulses (50 pg) Integrated signal/counts c u 7144 6.4 11 417 8.1 10 530 3.6 10 155 4.7 15 720 6.5 21 455 5.6 Ag 12 507 3.8 22 294 9.9 20 060 1 .oo 19 863 3.1 13 497 6.3 18 065 5.0 23 254 9.9 13 225 2.5 c s 163 790 4.5 196 151 3.6 187 167 0.9 187 006 3.6 122 575 5.1 268 796 8.8 188 548 12.6 371 608 9.4 Pb 122 529 7.6 196731 7.7 250 100 0.70 248 335 6.3 129 097 5.7 239 324 7.0 Bi 174 329 8.5 263 292 5.6 212 667 6.1 199 736 4.3 136 053 4.2 368 462 7.4 206 83 1 5.6 192 756 4.0 U 136 369 7.7 225 123 5.4 106 140 6.3 99 589 12.0 116 543 5.3 217413 8.4 335 407 10.1 239 306 6.4 c o 16 275 2.5 19 659 5.6 19 833 1.1 19 616 4.8 25211 6.5 34 887 5.1 45 040 9.7 27 139 10.6 Parameter 0.05% HNO 1.0% HNO RSD* (Yo) (n=5) RSD (Yo) (n=5) 0.05% HCl RSD (Yo) (n=5) 1.0% HCl RSD (Yo) (n=5) 0.05% H,SO4 RSD (Yo) (n=5) 1.0% H2S04 RSD (%) (n=5) 0.05% H,P04 RSD (Yo) (n=5) RSD (%) (n=5) 1.0% H,P04 * RSD relative standard deviation.Table 3 Effect of pyrolysis temperatures on integrated analyte signal pulses (50 pg) Integrated signal/counts Parameter l%HN03 None* RSD (O/o)(n=5) 400 "C RSD (%) (n=5) 900 "C RSD (Yo) (n=5) 1% HCl None RSD (%) (n=5) 400 "C cu c s Pb Bi U 14 790 8.5 13 653 2.6 14 655 5.3 24 201 11.1 12 612 9.1 5502 4.08 268 020 6.1 263 190 2.7 268 030 5.86 235 590 5.1 183 270 8.1 175 660 7.04 163 100 4.3 109 710 6.5 81 765 3.33 171 655 4.8 207415 2.5 217 520 3.26 18 486 12.1 16 674 3.9 16 092 3.4 25 473 5.2 24 227 13 514 8.9 14016 20 336 4.5 19 314 158 210 5.0 156 355 182 330 8.7 125 300 170610 8.4 134 785 120 130 12.2 87 800 RSD (Yo) (n=5) 900 "C RSD (Yo)(n=5) 4.2 26 121 2.0 5.7 15 041 4.1 3.6 15 796 1.9 5.8 127 195 4.7 8.9 44 870 2.9 9.9 25 770 2.9 12.5 82 290 6.3 * No pyrolysis step used.other three acids because as the acid concentration was increased the integrated signal did not increase for all of the elements.Rather Co Ag and U decreased with increased H3PO4 concentration Bi remained unchanged and Cs increased significantly. In comparison with the other acid matrices the integrated signals for most analytes were greater in H,P04 than either HNO or HC1 and similar to the integrated signals obtained in the H,S04 matrix. Both H2S04 and H3P04 have very high boiling-points (> 200 "C) and will remain in the graphite tube especially when a pyrolysis step is not used. Although H,S04 and H3P04 are not commonly used in trace analysis in ETAAS they were included in this study for their potential as matrix interferences in ETV- ICP-MS. Real samples especially biological samples contain both S and P which in acidic media produce H2S04 and H3P04 during the drying and pyrolysis steps.It is possible that the matrix effects observed with these acids were entirely due to the presence of acids in the graphite tube and not to other processes such as those occurring in the plasma. The interaction between these acids and analytes in the ETV device is complex and requires further study. A pyrolysis step was used to study its effect on 1% v/v H N 0 3 and HCl since these acids are the most common ones used as diluents in ETV-ICP-MS. When a pyrolysis step was used a different analytical response was obtained. Two pyrol- ysis temperatures were used 400 and 9OO0C using 1% v/v acids as the matrix. The lower pyrolysis temperature at 400 "C was chosen because it is well below the standard pyrolysis conditions for volatile elements such as Ag.19 Therefore any effect seen would not be associated with analyte losses in the pyrolysis step at 400°C.The higher pyrolysis temperature at 900°C was chosen since temperatures in this range are used for a number of less volatile elements such as Cu or Co. Initially it was postulated that a high-temperature pyrolysis step would remove more acid from the surface of the ETV device and therefore result in lower integrated analyte signals as was illustrated in Table 2. The results of these experiments are summarized in Table 3. Maximum integrated signal inten- sities were achieved when a pyrolysis step was not used with 1% v/v HNO in the matrix. In a 1% v/v HNO matrix increased pyrolysis temperatures had a detrimental effect on the volatile elements such as Ag Bi and Pb.For the other elements there was no change in signal intensity within experimental error with the use of a pyrolysis step except for U which showed only a slight increase in integrated signal intensity with a pyrolysis step. The use of a pyrolysis temperature step for analytes in a 1% HC1 matrix did not result in an increase in the signal intensity for most of the elements studied. For most elements the integrated signals obtained without a pyrolysis step wereJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY SEPTEMBER 1994 VOL. 9 925 Table 4 Effect of acid concentration on integrated analyte signal pulses (50 pg) using NASS-3 as a physical carrier Integrated signal/counts Parameter 0.05% HNO 1.0% HN03 RSD ("/)(n=5) RSD (%)(n=5) 0.05% HC1 RSD (%) (n=5) 1.0% HCI RSD (Yo) (n=5) 0.05% H2S04 RSD (%) (n= 5 ) 1.0% H2S04 RSD (YO) (n= 5) 0.05% H3PO4 RSD (Yo) (n=5) 1.0% H3PO4 RSD (YO) (n= 5 ) co 30 382 9.4 28 464 7.9 23 085 2.3 21 648 2.9 40 437 3.42 42 471 5.34 46 581 9.1 27 581 5.6 ~~ c u 10 148 8.1 11 165 3.3 9297 10.4 9409 5.6 29 286 4.1 30 117 4.3 - - - - Ag 29 399 3.3 27 839 4.1 24 848 5.5 19 402 1.3 37 877 7.8 27 204 7.0 31 027 4.6 13 626 2.0 c s 125 492 8.1 115206 3.3 149 903 7.9 144 490 5.4 322 094 3.5 394 135 3.7 181 483 5.9 419 118 4.4 Pb 220 425 5.4 225 277 7.3 256 772 4.7 256 704 6.9 284 412 6.0 381 171 1.6 - - - - Bi 373 365 2.4 366 476 2.8 223 700 4.2 181 660 5.1 300 002 3.1 480 804 6.7 209 078 17.8 389 964 5.0 U 53 410 6.3 48 990 5.4 30 720 8.3 48 030 8.4 58 469 5.3 114 642 6.8 469 163 9.5 188 118 13.6 higher than when a pyrolysis step of 400°C was used.When the pyrolysis temperature was increased to 9OO0C the inte- grated signal decreased further for most elements in 1% HCl and in the case of elements that form volatile halides such as Ag or Pb the effect of using a pyrolysis step was even more dramatic. These results are consistent with the experiments completed using different acid concentrations (Table 2) when a higher concentration of acid resulted in an increase in analyte signal intensity. It is therefore reasonable that the introduction of a pyrolysis step that removes any traces of acid from the ETV device should result in lower integrated signals. These observations indicated that when a pyrolysis step was used the beneficial carrier effects of retained HCl are also removed and hence a lower signal is obtained.Effect of Chemical Modifier-Physical Carrier The addition of a chemical modifier<arrier was evaluated for the role in reducing the matrix effects associated with different acid concentrations. The physical carrier increases the inte- grated signal in ETV-ICP-MS by improving the mass transport of the analyte of interest.2'v22 The physical carrier chosen for this work was NASS-3 sea-water. Purified NASS-3 (see under Experimental) sea-water acts as a mixed modifier in the ETV device since sea-water is essentially a mixture of major ions which include Na K Mg Ca Sr C1 and Br. The NASS-3 sea-water is a readily available high-purity mixed modifier free of trace contaminant elements typical of other modifiers such as Pd.23 The results of the experiments with NASS-3 used as a physical carrier are summarized in Table 4.As before the analytes were vaporized immediately after the drying step and a pyrolysis step was not used. The integrated signals for all of the elements in the presence of NASS-3 were significantly enhanced indicating an improvement in the mass transport of the analytes to the plasma. The NASS-3 also reduced the effects of increased acid concentration for all of the acid matrices. The change in integrated signal when the acid concentration was increased was not nearly as significant when NASS-3 was present in solution. This was observed for all of the acids though less so for H3P04. The reproducibility (relative standard deviation RSD) of the integrated signals was also improved when NASS-3 sea-water chemical modifier- carrier was used.The use of NASS-3 as a chemical modifier in effect introduces into the sample matrix pg amounts of chloride. This excess of chloride (compared with pg amounts of analyte) serves to convert most of the analytes into their chloride form and can interfere or prevent the formation of analyte oxides which are more refractory. These oxides tend to vaporize or be reduced at much higher temperatures compared with the vaporization of chlorides and for some elements such as the rare earth elements oxides are vaporized in the graphite tube at a time when little carrier is present to ensure efficient transport to the argon plasma. Conclusions This work has shown that the use of aqueous external standards for calibration could lead to significant analytical error depending on the composition of the sample matrix.The use of a chemical modifier-carrier in both standard and sample solutions was shown to be essential to the successful application of ETV-ICP-MS to the quantitative analysis of complex mate- rials. Of the acids studied HC1 was least influential in altering analyte sensitivity especially in the presence of NASS-3 as the chemical modifier-carrier. Residual chloride remaining after the drying and pyrolysis steps served as an effective carrier for transporting analyte from the graphite tube to the argon plasma while at the same time prevented the formation of more refractory analyte oxides. The relatively lower sensitivity of analytes except for U in the presence of high concentrations of HNO compared with HC1 can be explained in part by the chemical nature of these acids.In the case of HNO the analytes formed oxides that were reduced to metal atoms during the high-temperature vaporization step making them less volatile compared with their corresponding halides. The behaviour of the less volatile H,PO and H,S04 acids and their effects on analyte signal pulses is complex and could involve both chemical and physical effects. Sample matrices containing large amounts of these acids will probably require detailed study for each analyte determined and could require alternative calibration schemes such as the method of standard additions or isotope dilution. References Carey J. M. and Caruso J. A. Crit. Rev. Anal. Chem. 1992 23 397. Gray A. L. and Date A. R. Analyst 1983 108 1033. Styris D. L. and Kaye J. H. Spectrochim. Acta Part B 1981 36 41. Sturgeon R. E. Michell D. F. and Berman S . S. Anal. Chem. 1983 55 1059. Bass D. A. and Holcombe J. A. Anal. Chem. 1987 59 974. Ham N. S. and McAllister T. Spectrochim. Acta Part B 1988 43 789.7 8 9 10 11 12 13 14 15 16 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY SEPTEMBER 1994 VOL. 9 Gregoire D. C. Lamoureux M. Chakrabarti C. L. Al-Maawali S. and Byrne J. P. J. Anal. At. Spectrom. 1992 7 579. Byrne J. P. Chakrabarti C. L. Gregoire D. C. Lamoureux M. and Ly T. J. Anal. At. Spectrom. 1992 7 371. Park C. J. Van Loon J. C. Arrowsmith P. and French J. B. Anal. Chem. 1987 59 2191. Tsukahara R. and Kubota M. Spectrochim. Acta Part B 1990 45 779. Evans E. H. and Giglio J. L. J. Anal. At. Spectrom. 1993 8 1. Sturgeon R. E. Berman S. S. Willie S. N. and Desaulniers J. A. H. Anal. Chem. 1981 53 2331. Frech W. Persson J. A. and Cedergren A. Prog. Anal. At. Spectrosc. 1980 3 279. Frech W. and Cedergren A. Anal. Chim. Acta 1976 82 93. Welz B. in Atomic Absorption Spectrometry VCH Weinheim 2nd. rev. edn. 1986 p. 203. Frech W. and Cedergren A. Anal. Chim. Acta 1976 82 83. 17 Karanassios V. and Horlick G. Spectrochim. Acta Part B 1989 44 1361. 18 Gregoire D. C. Prog. Anal. At. Spectrosc. 1989 12 433. 19 Slavin W. Graphite Furnace AAS. A Source Book Perkin-Elmer Ridgefield CT 1984 p. 18. 20 Tan S. H. and Horlick G. Appl. Spectrosc. 1986 40 445. 21 Ediger R. A. and Beres S. A. Spectrochim. Acta Part B 1992 47 907. 22 Grkgoire D. C. Al-Maawali S. and Chakrabarti C. L. Spectrochim. Acta Part B 1992 47 1123. 23 Grkgoire D. C. and Sturgeon R. E. Spectrochim. Acta Part B 1993 48 1347. Paper 31071 76F Received December 6 1993 Accepted March 10 1994