摘要:

Low-flow Interface for Liquid Chromatog rap hy-Inductively Coupled Plasma Mass Spectrometry Speciation Using an Oscillating Capillary Nebulizer LANQING WANG,* SHELDON W. MAY AND RICHARD F. BROWNER? School of Chemistry and Biochemistry Georgia Institute of Techndogy Atlanta GA 30332-0400 USA STANLEY H. POLLOCK School of Pharmacy Mercer University Atlanta GA 30341 USA The application of a novel nebulizer the oscillating capillary nebulizer (OCN) is described for use in speciation studies. The nebulizer has certain features which make it very suitable for this application without modification at both micro-flows (1 p1 min-') and macro-flows (1 ml min-I). Short- and long- term precision at typical operating flows are comparable fo a normal (1 ml min-l) concentric glass nebulizer. Column-to- nebulizer dead volume is approximately 1 pl.The narrow drop size distribution for the nebulizer at low flows leads to excellent sensitivity when coupled to a micro-LC column. Post- column peak broadening introduced by the interface is minor at flows 2 5 p1 min-' but widens the peaks noticeably at flows between 1 and 5 p1 min-'. The very high efficiency of the nebulizer at flows <50 pl min-' is exemplified by the faci that no drain is necessary at these flows in the open spray chamber as no visible liquid condenses on the chamber walls. The ICP-MS response for the OCN (counts per ng of Se injected) does not change when water is replaced by methanol as solvent whereas with a conventional nebulizer a solvent change of this type inevitably results in a significant change in response.The OCN was used for the reversed-phase LC separation of a mixture of five organic Se compounds of pharmacological significance at flows of 12 50 and 400 p1 min-'. With use of a 0.5 mm id column a flow rate of 12 pl min-' and a 60 nl injection good peak separation was found with an average efficiency of B 10 000 plates and a detection limit of around 30 pg. Keywords Nebulizer; speciation; inductively coupled plasrna mass spectrometry; oscillating capillary nebulizer; seleniui n analysis; drug analysis Inductively coupled plasma mass spectrometry provides a highly effective element-selective detection system for ipeci- ation studies. When used in conjunction with an appropriate separation technique ICP-MS offers excellent sensitivity and selectivity for multi-element identification and quantitatjon of organometallic compounds.' Nevertheless LC-ICP-MS still suffers from a number of drawbacks many of them a direct consequence of limitations inherent to the system interface. The most frequently used configuration for LC-ICP interfac- ing involves passing column effluent through a narrow bore tube to a standard pneumatic nebulizer. The nebulizer gener- ates a polydisperse aerosol,2 which is subsequently size- conditioned by passing it through a spray chamber before being injected as still-wet aerosol into the plasma.In the process of passing through the spray chamber droplets larger * Present Address Centers For Disease Control Clinical Biochemistry Branch 4770 Buford Highway Atlanta GA 30341.USA. t Author to whom correspondence should be addressed. Journal of Analytical Atomic Spectrometry than about 5 pm diameter are eliminated reducing the percent- age of analyte reaching the plasma to approximately 1-3% of what was injected onto the c01umn.~ While this type of analyte loss is undesirable in normal sample injection when it occurs following chromatographic separation it may render the eluting bands undetectable. In LC separations the total injected amount may only be a few hundred nanograms and after separation each eluting band may contain only a few tens of nanograms of material or less. An LC interface that typically discards 2 97% of each eluting peak before detection is clearly far short of ideal. Another problem encountered with LC-ICP-MS interfacing is the difficulty of introducing organic solvents directly into the plasma.For example if pure methanol acetonitrile or tetrahydrofuran are continuously introduced into an Ar ICP at a solvent flow of x 1-2 ml min-l they will cause plasma instability and coat the sampling cone with carbon. As the interest in speciation has grown over the last decade so researchers have been energetically seeking ways to over- come these problem~.l*~-~ The most effective approach to date appears to be the development of low-flow nebulizers coupled to low-flow separation techniques. Consequently several types of nebulizer have been developed specifically for low-flow ICP-MS and ICP-OES interfacing. These include the direct injection nebulizer (DIN),',' the microflow ultrasonic nebulizer (F-USN),~ the glass-frit nebulizer,iO~ii the Meinhard high efficiency nebulizer ( HEN),l2,I3 the micro-concentric nebulizer ( MCN)14 and the Thermospray n e b u l i ~ e r .l ~ - ~ ~ The DIN was the first microflow concentric nebulizer developed specifically for LC-ICP interfacing" and has since been used extensively by Houk et a1.'9920 for interfacing low-flow LC columns with both ICP-OES and ICP-MS. With the DIN the nebulizer tip is positioned inside the central channel of the plasma torch only a few millimetres upstream from the base of the plasma. Because of its construction the DIN is essentially 100% efficient in passing analyte to the plasma Also as the DIN operates at quite low liquid flow rates (30-120 pl min-I) the plasma remains stable even when the DIN is used with organic solvents." In addition the dead volume between the end of the LC column and the ICP is very low.However with no possibility for desolvation the entire solvent load of aerosol and vapour passes directly into the plasma causing local plasma coo1ing.21,22 On the basis of purely analyte transport considerations an approximately 30-fold improvement com- pared with a normal pneumatic nebulizer would be predicted at a nebulizer flow of 100 pl min-'. However only typically a 2.5-fold improvement in detection limits is observed,' primarily because of solvent-induced plasma cooling but also possibly because of incomplete analyte vaporization in the The work of Tarr et aL9 illustrates the capability of the Journal of Analytical Atomic Spectrometry December 1996 Vol.11 (1 137-1 146) 1 137micro-ultrasonic nebulizer (p-USN) to generate a stable aerosol at liquid flow rates as low as 5 p1 min-'. The micro liquid flow is fed through a 50 pm diameter silica capillary tube onto the surface of an ultrasonic nebulizer which generates an aerosol with an extremely low Sauter mean diameter (2 pm) and quite narrow size distribution. Transport studies with the p-USN have demonstratedg that it is capable of providing essentially 100% transport efficiency at 5 pI min-' flow rate. However from a practical perspective the p-USN requires a reasonably high level of skill to operate and optimize and like all ultrasonic nebulizers it is relatively expensive. Conventional pneumatic nebulizers are generally incapable of generating a stable aerosol at liquid flows much below 300 pl min-' as the relatively large diameter of the liquid tube precludes formation of a stable liquid jet under these con- ditions.However a new version of the Meinhard concentric glass pneumatic nebulizer the 'high efficiency nebulizer' (HEN) has a smaller centre liquid tube and smaller gas orifice ring than the higher flow version. Studies by Hobbs and O l e ~ i k ~ ~ have shown low detection limits and high transport efficiency (20%) for the HEN. However the liquid flow rate is a little high (30-50 pl min-') to interface capillary electrophor- esis (CE) and 0.5 mm diameter LC columns to an Ar ICP. The isolated droplet generator developed by French et and used by Olesik and H ~ b b s ~ ~ shows considerable promise as a device for introducing small amounts of material into a plasma as its transport efficiency should be close to 100%.The use of this device for chromatographic coupling has not yet been reported. Consequently the issues of how efficiently the relatively large droplets that it generates atomize in the plasma and also how well it would work with mixed aqueous/ organic solvents and solvent gradients has not yet been addressed. A totally different type of nebulizer the electrospray nebul- izer operates effectively at extremely low flow rates (< 1-30 pl min-') which makes it especially suitable for CE interfacing. With a pneumatically assisted electrospray (Ionspray) nebulizer it is possible to work with flow rates from = 1 to 200 pl min-' .26 In this technique the capillary nebulizer tip is held at a potential of 3-5 kV relative to ground.Liquid travelling through the capillary picks up charge forms a Taylor cone and disintegrates into small droplets due to electrostatic repulsion. Subsequent drop instability and breakup leads to the formation of an even finer highly charged aerosol. The aerosol itself can be used directly as an ion source for inorganic mass spectrometry as described by Agnes and H o r l i ~ k . ~ ~ ~ ~ At the moment the major limiting factors with elemental electrospray-MS are the complexity of the mass spectra formed and restrictions on solvent composition and liquid flow rates. In another approach to low flow separations Olesik et used the end of a capillary electrophoresis column with solvent flows of approximately 2 pl min-' as the central tube for a coaxial pneumatic nebulizer.The aerosol generated by this device was shown to have a Sauter mean diameter of 4.2 pm at a flow rate of 1 ml min-' which is similar to that expected from a normal pneumatic neb~lizer.~' No drop size data were given for low- and micro-flow oper- ation presumably because the aerosol number density was too low to allow meaningful measurement of drop size by laser scattering techniques. Consequently despite the notable improvements in low-flow nebulizers that have been made in recent years really effective interfacing of ICPs with micro-scale separation techniques is still hindered by the lack of a simple inexpensive and effective low- and micro-flow nebulizer. In this paper the ability of the oscillating capillary nebulizer (OCN) to work as a micro-flow low-flow and normal-flow nebulizer is examined using solvent flows as low as 1 pl min-'.The use of the OCN for interfacing 1 138 Journal of Analytical Atomic Spectrometry December ICP-MS with macrobore microbore and capillary LC columns is also described. EXPERIMENTAL Oscillating Capillary Nebulizer and Spray Chamber The OCN was fabricated in-house as described previo~sly.~' The nebulizer consists of two concentric fused silica capillary tubes mounted concentrically with compression fittings. The tube dimensions were as follows. Inner tube id 50pm od 142 pm; outer tube id 450 pm od 440 pm. Liquid samples were fed into the inner capillary with a Hewlett-Packard Model 1090 liquid chromatography pump which is capable of delivering continuous flows with 1 pl min-l resolution.At a liquid flow rate of 10 p1 min-l or less a short length of 20 pm id silica capillary tube was placed in-line to minimize pdmp pulsation. The OCN was operated with a single-pass cylindri- cal spray chamber. Aerosol was generated axially at one end of the chamber and passed to the plasma torch through a tube connected to the opposite end of the chamber. The length and internal diameter of this spray chamber were identical to those of the standard Scott spray chamber. At low flows (e.g. < 50 p1 min-I) no drain tube was necessary for the spray chamber as there was no deposition or collection of water on the sides or end of the spray chamber. At flows > 50 pl min-' a drain tube was necessary to remove solvent build-up from the spray chamber. A standard TR-30-C3 Meinhard nebulizer was used in some experiments for comparison purposes.A standard double-pass Scott type spray chamber was used with the Meinhard nebul- izer. A Matheson mass flow controller (Model 8270) was used to control nebulizer gas flow rates. Ar back-pressure was set at 120 psi for the OCN and 40 psi for the Meinhard nebulizer. LC-ICP-MS Systems The inductively coupled plasma mass spectrometer used was an ELAN 250 (Perkin-Elmer SCIEX). The instrumental con- ditions and operating parameters are listed in Table 1. For all ICP-MS studies the dimensions of the fused silica capillary tubes used in the construction of the OCN were 1 liquid Table 1 Instrument conditions and data acquisition parameters Instrument conditions ICP-MS Forward power Reflected power Argon flow rates/l min-' Plasma Intermediate Nebulizer Meinhard OCN Sample flow rate Sampling position Data acquisition parameters -Continuous-FI-ICP-MS Multielement monitoring mode Dwell time Scanning mode Points per spectral peak Resolution Measurement time Replicate Multielement monitoring in Dwell time Replicate LC-ICP-MS graphic mode Perkin-Elmer SCIEX ELAN 250 1.2 w <5 w 12 1.2 0.9 1.1 1-2000 pi min-' 17 mm above the load coil Elan 5000 program 7Li 24Mg lo3Rh '14Cd 208Pb 50 ms Peak hop 3 Normal 1 s 4000 Se (m/z = 77,78,82) 83Kr 150 ms 6000 (background) 1996 Vol.11(central) capillary 50 pm id 142 pm od; 2 gas (outer) capillary 250 pmid 440pmod.At liquid flow rates less than 50 p1 min-' samples were introduced with a Rheodyne 7010 injector fitted with a 2ml PEEK injection loop which pro- duced a steady-state signal lasting for 40min or more. For on-line column separation either a Rheodyne 7413 injector with a 1 p1 internal loop or a zero dead-volume Valco injector with a 60 nl internal loop (Valco Houston TX) were used. The same inlet systems were also used for flow injection studies. For high liquid flow rates (i.e. > 100 pl min-I) samples were introduced directly from the solvent reservoirs of the LC pump. Inlet filters were placed in both the solvent delivery line and the injection port of the injector. Connections from the sample injector to the column inlet and the column outlet to the liquid capillary tube of the OCN were made with 5 cm long 0.005 in id PEEK tubing.The total dead volume between the outlet of the LC column outlet and the nebulizer tip was estimated to be less than 1 pl. Optimum separation conditions for the system are given in Table 2. The analytical columns used for separation of the mixture of selenide compounds were 1 a 25 cm long 0.5 mm id Keystone Kappa C18 column (Keystone Scientific) packed with 5 pm ODs-Hypersil material; 2 a 25 cm long 1 mm id Keystone C column packed with 5 pm BDS-Hypersil mate- rial; and 3 a 25 cm long 3.2 mm id Alltech C18 column with 5 pm Absorbosphere Hs packing material. The optimum flow rates recommended for these columns are 12 pl min-' 50 pl min-' and 0.4 ml min-' respectively. Separation con- ditions were optimized by changing the composition of the organic modifier the pH value the concentration of ionic buffer and the elution gradients.After column separation the eluent was transferred directly through the nebulizer and spray chamber into the plasma source. Three isotopes of selenium were monitored continuously in the graphic mode at mass- charge ratios of 77 78 and 82 respectively. Before the separa- tion the ICP-MS device was optimized by introduction of 1000 ppb Se in 1% HN03. Reagents and Samples Stock solutions of Li Mg Sc Se Rh Cd La and Pb were prepared from 1000 ppm atomic standard solutions ( Fisher Scientific). For LC use mobile phases were prepared by dissolving solid LC grade ammonium acetate (0.1 mol I-') in LC grade water (Fisher Scientific) and adjusting the pH value to 6.0 with high purity 6 moll-' HC1.All mobile phases were filtered through a 0.45-pm pore filter before use. The columns were refreshed with methanol and water overnight. For FI studies filtered deionized water was used as the carrier stream. The organic selenide compound and its four derivatives were Table 2 Optimum chromatographic conditions for selenide speciation synthesized and purified in-house by the S.W. May research They included phenyl-2-aminoethyl selenide ( PAESe) 4-hydroxyphenyl-2-aminoethyl selenide ( HO-PAESe) 4-fluorophenyl-2-aminoethyl selenide (pF-PAESe) (RS)a- methylphenyl-2-aminoethyl selenide [(RS)-MePAESe] and phenyl-2-acetamidoethyl selenide (N-acetylPAESe). Stock solu- tions of each selenide compound were prepared by dissolving the selenide salt in de-ionized water at 1000ppm Se concen- tration for each compound.The mixture of 5 selenides was made by mixing an equal volume of each compound solution together (166.7 ppm Se for each compound). RESULTS AND DISCUSSION An ideal nebulizer for interfacing both LC-ICP-MS and CE-ICP-MS would possess the following characteristics. Firstly it would generate a stable aerosol over a wide liquid flow rate range. This includes flows at or below 1 pl min-' for CE use. Secondly it would generate a primary aerosol with a small Sauter mean diameter and a narrow size distribution. Thirdly aerosol properties would be very similar when using either organic or aqueous solvents. Fourthly the aerosol would be readily desolvated before reaching the plasma source.Fifthly the dead volume between the end of the LC column and the tip of the nebulizer would be small of the order of 1-2pl. This would ensure the absence of peak broadening occurring between the end of the column and the ICP detector. The primary purpose of the present study was to evaluate the applicability and utility of the OCN for interfacing both low-flow (100-10 pl min-') LC and micro-flow (10-1 pl min-l) LC to an ICP-MS. We believe that very low-flow and micro-flow chromatographic separation will become an important aspect of LC-ICP-MS interfacing for three main reasons firstly increased speed of separation; secondly small sample size needed (60 nl-1 pl); thirdly reduction in solvent use with a corresponding reduction in operating cost and need for solvent disposal; and fourthly reduction of solvent and matrix loading to the plasma.From the perspective of LC-ICP-MS interfacing the last factor probably has the greatest relevance. With reversed-phase LC separations an organic solvent such as MeOH or CH3CN is usually present often in high concentration sometimes approaching 100%. In LC separations of complex mixtures the major limiting factor in the process becomes the need to adjust solvent composition to obtain optimum column separation conditions not to obtain optimum plasma operating conditions desirable though this may be. Consequently using typical normal bore packed columns and LC solvent flow rates of 1-2 ml min-' there is a substantial organic solvent load delivered to the plasma. In the absence of an efficient desolvation device the plasma may Column Dimension Stationary phase Sample flow rate Injection volume Mobile phases Eluent gradients Compounds Narrow bore Alltech CI8 25 cm long x 3.2 mm id Adsorbosphere HS 5 pm particle 0.4 ml min-' 1 P1 A 0.1 mol 1-' CH3COONH4 pH=6.0; B 100% MeOH t=O 90% A 10% B t=20 70% A 30% B t = 50 50% A 50% B Phenyl-Zaminoethyl selenide (PAESe) Hydroxyphenyl- 2-aminoethyl selenide (HO-PAESe); 4-fluorophenyl-2-aminoethyl selenide ( F-PAESe); Phenyl-2-acetamidoethyl selenide (N-acetyl PAESe); (R,S)a-methylphenyl-2-aminoethyl selenide [(R,S)- MePAESe] Micro bore Keystone CI8 25 cm long x 1.0 mm id BDS-H ypersil 5 pm particle 50 pl min - ' 1 1.11 Same Same Same Capillary Keystone CIS 25 cm long x 0.5 mm id ODs-H ypersil 5 pm particle 12 pl min-' 1 pl or 60 nl Same Same Same Journal of Analytical Atomic Spectrometry December 1996 Vol.11 1 139therefore become quite unstable and in ICP-MS use unwanted carbon deposition will occur on the sampling cone. Additionally for ICP-OES use strong atomic carbon lines and intense molecular bands (e.g. CN CH and OH) may cause increased spectral background and possible spectral interference^.^^"' Potentially the most problematical issue to be faced with low- or micro-flow chromatography-ICP interfacing arises from the significant reduction in total analyte mass transported to the plasma per second (wot),36*37 which results from the lower rate of solvent delivery to the nebulizer under low- or micro-flow conditions. In general ICP-MS signals are pro- portional to yet and so a reduction in solvent flow rate with everything else remaining constant will produce a proportional reduction in signal at the detector.At higher flows (e.g. 2400 pl min-') the amount of analyte reaching the plasma per second is no longer a simple linear function of solvent flow rate38*39 because 1 analyte transport qot becomes less efficient at higher liquid flows; and 2 higher solvent loading cools the plasma,40 which may lead to a lower ion concen- tration in the plasma. However to counterbalance this effect low-flow nebulizers generally have higher analyte mass trans- port efficiencies than high flow nebulizers. Consequently the net reduction in analyte delivery rate to the nebulizer may be partly or totally offset by higher analyte transport efficiency.This will lead to a non-linear drop in signal with reducing LC flow rates. This phenomenon is discussed in more detail el~ewhere.~' Continuous Microflow OCN-ICP-MS It was of interest initially to determine the long-term stability of a low-flow sample introduction system attached to an ICP-MS compared with the more usual high-flow systems. Long-term stability The OCN was run at microflow rates and the ion counts were monitored continuously while aspirating solutions containing 100 ppb each of Li Rh Cd and Pb both in water and in methanol for 1 h. Flow rates of 1 2 5 10 20 and 50 p1 min-' were used. Fig. 1 shows ICP-MS traces for lithium-7 rhodium-103 and lead-208 at 1 pl min-' flow rate in 100% H,O and Fig.2 shows ICP-MS traces for rhodium-103 at flows of 1 2 and 10 p1 min-' in 100% MeOH.The nebulizer gas flow rate was optimized individually for each solvent giving an Ar flow for water of l.lOlmin-' and for methanol of 0.81 1 min-'. A 2 ml PEEK tubing sample loop was used for the long term stability test. After monitoring steady-state signals for approximately 1 h the injection valve was switched back to pure solvent. Precision data are listed in Table 3. The long-term stability of the OCN at microflow rates improved by about a factor of 2 from l-lOplmin-'. The higher drift rate and noise level observed at the low flows is at least partially a result of the limited flow stability of the LC pump used which was being operated close to or at the lower limit of its pumping range.The stability of analyte signals for 100% MeOH solutions improved from 1 to 10 p1 min-' but then became degraded when the liquid flow rate exceeded 20 pl min-'. At MeOH flows much above 50 pl min-' serious plasma instability was induced by the higher organic loading. This manifested itself as an intense green C atomic emission plasma flickering and noticeable graphite build-up on the sampling cone following a 3-h run. Signal response The relationship between analyte response and solvent flow rate is a key issue in the evaluation of low- and micro-flow LC nebulizer interfaces. The effect on the response of varying 1000 - 500 - cn c 2 1 w o / s \ I 0 lo00 2000 3000 4000 Time/s Fig. 1 ICP-MS traces for 100 ppb each of (a) Li; (b) Rh; and (c) Pb in 100% water (m/z 7 103 and 208 respectively) at 1 pl min-' flow rate; nebulizer gas flow rate= 1.1 1 min-' 600 - 3001 - 0,- 1600 1 mt1 I I 0 lo00 2000 3000 4000 Timels Fig.2 ICP-MS traces for 100 ppb Rh (m/z 103) in 100% MeOH at (a) 1; (b) 2; and (c) 10 pl min-' flow rates; nebulizer gas flow rate= 0.81 1 min-' liquid flow rate using the OCN with both water and methanol is shown in Fig. 3. When using water a nearly linear correlation was established (coefficients of variation = 0.989 0.981 and 0.935 for rhodium-103 cadmium- 1 14 and lead-208 respect- ively) as the liquid flow rate was raised from 1 t o 20 pl min-'. At solvent flows greater than 20plmin-' for water and between 10 and 20 pl min-' for MeOH the signal decreased with increasing liquid flow rate due very probably to the increasing interaction between combined solvent aerosol and solvent vapour with the plasma.The response for methanol peaked at a solvent flow of 20 pl min-' and decreased thereafter. 1 140 Journal of Analytical Atomic Spectrometry December 1996 Vol. 1 1Table3 Long-term precision (YO RSD) for OCN and Meinhard nebulizer in order to evaluate some critical analyte transport issues such as the effect of very low solvent flow rates on peak broadening. Flow rate/ Solvent pl min- ' Li m/z = 7 Rh m/z = 103 Pb m/z = 208 OCN* Water 1 12.2 12.0 20.2 2 9.78 8.66 15.1 5 8.29 6.71 14.7 10 6.06 6.70 10.8 20 11.1 8.01 11.8 50 5.59 3.81 5.80 Methanol 1 14.3 13.8 20.2 2 10.7 9.43 13.4 10 t 7.34 8.76 20 t 6.09 5.73 50 t 11.4 7.33 Meinhard nebulizert Water 2000 4.73 3.34 5.56 * Signal precision was calculated based on 1 h steady-state sig- nal using 100 ppb Li Rh and Pb in water and methanol; nebulizer gas flow was 1.10 1 min-' for water and 0.8 1 min-' for methanol observation height was 18 mm power was 1.8 kW.Isobaric interference with carbon. * Same conditions as the OCN except nebulizer gas flow rate was 0.90 1 min-'. l5I / 7 l o t f A - 0 - 10 20 30 40 50 60 Liquid flow rate/pl min-' Fig. 3 Response for continuous flow OCN-ICP-MS versus liquid flow rate for (a) water and (b) methanol. A 7Li; B lo3Rh; C lI4Cd; and D "'Pb A high analyte transport efficiency E in the range of 50-80% has been observed for the OCN when operated at microflow rates (1-10 pl min-'). Other groups have also found high E values for low-flow neb~lizers.''*~~ For example Olesik et aLZ9 observed a 20% E for a HEN operated at 50 pl min-' and Liu et al.measured similar values of approximately 55% for both a HEN and a thimble glass frit nebulizer operated at 11 pl min-'. Clearly E is strongly influenced by low nebulizer liquid flow. Low-flow Flow-injection OCN-ICP-MS As a preliminary step to evaluating the LC performance of the OCN-based interface a flow injection study was carried out Long term FI stability Long term stability for the OCN in the FI mode was evaluated by measuring the YO RSD for both peak area and peak height based on 12 repetitive 1 p1 injections of a 100ppb solution containing Li Mg Rh Cd and As as shown in Table 4. The 5 elements tested which cover quite a wide mass range gave better precision when peak area was measured rather than peak height.The typical YO RSD varied from 2 to 6%. The poor precision obtained with peak height measurements was presumably due to noisy instrument response under these conditions; for example the average peak height for the five elements obtained at 1 pl min-' flow rate was below 1000 counts s-'. Degradation of the signal peak area precision at 50 p1 min-' for both peak height and peak area was mainly due to decreased peak area and increased solvent throughput into the plasma at higher flow rates. As is illustrated in Fig. 4 the long term precision for OCN flow injection data at flow rates < 20 p1 min - l based on peak area measurements was better than that from the continuous flow system at similar flow rates. At 50 pl min-' the situation reversed and continuous flow had the better precision. Peak area was used rather than peak height because of the dispersion broadening of peaks at low flow rates.Peak broadening versus liquidflow rate Fig. 5 illustrates the effect of liquid flow rate on the flow injection profile for an injection volume of 60 nl. It is clear that at low values liquid flow rate has a strong influence on the profile of flow injection and hence also on chromatographic signals. At flow rates in the range 1-5 pl min-' response was low and peaks were broad. At liquid flow rates > 5 pl min-' the peak profile became narrower and relatively larger. However at liquid flow rates 2 2 0 pl min-' both peak height Liquid flow rate/pl min-' Fig. 4 Long term precision data for 100 ppb each of (a) Li; (b) RH; and (c) PB in aqueous solution (m/z 7 103 and 208 respectively) using FI-OCN-ICP-MS and continuous flow OCN-ICP-MS.Flow injection volume= 1 p1 min-' nebulizer gas flow rate= 1.10 1 min-' Journal of Analytical Atomic Spectrometry December 1996 Vol. 11 1 141Table4 Long term precision (YO RSD) for flow injection OCN-ICP-MS. Signal precision calculated from peak heights or peak areas of 12 consecutive 1 pl injections of a solution of 100 ppb Li Rh Cd and Pb in 1% HNO,; nebulizer gas flow rate= 1.10 1 min-' observation height= 18 mm power= 1.8 kW Peak height Peak area Flow rate/pl min-' 1 2 5 10 20 50 1 2 5 10 20 50 Li m/z=7 34.7 6.88 8.44 6.35 6.28 3.53 2.58 1.93 5.36 6.74 12.0 10.9 Rh m/z= 103 34.0 4.69 10.1 2.58 6.1 1 5.48 3.06 3.61 2.66 5.59 10.7 10.7 C d m/z= 114 35.9 20.4 13.9 6.16 7.18 12.8 24.5 11.6 4.56 5.73 4.59 12.2 P b m/z = 208 32.3 13.2 13.9 5.53 9.14 7.62 4.96 8.21 4.1 1 7.82 14.0 17.5 30 PI min" min-' I Q t c - 2000 lo00 0- 100 200 300 400 500 600 700 800 Tim& Fig.5 Influence of solvent flow rate on peak shape and size for (a) Li (m/z=7) and (b) Rh (m/z=103) l.Oppm with FI-OCN-ICP-MS; injection volume = 60 nl nebulizer gas flow rate= 1.10 1 min-' and peak area decreased because of the limited time available for signal accumulation.The optimum flow rate for a 60 nl injection using peak area measurements was found to occur at a flow rate between 10 and 15 pl min-l. Similar trends were observed for all of the elements tested. Linear dynamic range The absolute detection limit and linear dynamic range for a 60 nl injection volume at 10 pl min-' flow rate was determined. Fig.6 shows Rh ICP-MS traces with duplicate injections of amounts ranging from 0.6 to 600 pg. Below 0.6 pg no signal could be detected above the blank and baseline noise. In all situations where signal could be readily detected peak shape did not appear to be affected by the amount of Rh injected. Calibration curves using either peak area or peak height showed good linearity over at least a lo4 dynamic range for all elements tested giving correlation coefficients close to 1 .OOO. Analyte concentrations ranged from 100 ppb to 100 ppm. Memory effects For flow injection at low flow rates as for low-flow chromatog- raphy the time taken for the analyte signal to return to 2-t 1 - II It Time/min Fig.6 Signal versus analyte mass injected in FI-OCN-ICP-MS. Element 103 Rh 60 nl injected 10 pl min-' solvent flow gas flow rate= 1.10 1 min-' baseline is a critical issue. Fig. 7 shows the memory effect for a 60nl injection of a 100ppm Li Rh Cd and Pb solution at 10 pl min-l flow rate. As can be observed ion counts for Rh decreased from a maximum of 300000 to about 10000 in about 20 s. After 40 s the signal had dropped to <OS% of the peak maximum indicating that the system washout time was adequate for LC-ICP-MS use. Efect of solvent composition For a typical reversed-phase LC separation an organic solvent (e.g. methanol acetonitrile or tetrahydrofuran) is added to an aqueous mobile phase as a modifier in order to optimize separation conditions.Numerous attempts have been made to understand the effect of organic solvent loading on ICP-MS signals.33934,40-42 F rom a fundamental point of view the change of ICP-MS signals for organic solvents can be attributed to the interplay of two major factors the change of aerosol I I 0 40 80 120 Timels Fig. 7 Washout profile in FI-OCN-ICP-MS 100 ppm of Li Cd Rh and Pb in 60 nl injection at 10 pl min-' flow rate; nebulizer gas flow = 1.10 1 min-' 1 142 Journal of Analytical Atomic Spectrometry December 1996 Vol. 11characteristics and the interaction between solvent and plasma. Typically the interaction of solvent aerosol and solvent vapour with the plasma absorbs energy and therefore decreases the signal intensity. On the other hand the addition of an organic modifier into an aqueous solution reduces the surface tension and so favours the formation of an aerosol with a smaller mean drop size.This tends to increase analyte transport efficiency resulting in an increased Kot and hence an increase in ion count. For a typical pneumatic nebulizer the interaction between solvent and plasma dominate^;^' on the other hand for the DIN the signal has shown a dramatic increase with the addition of 20% methanol to the aqueous solution as reported by Shum et When using the OCN the ICP-MS was able to tolerare the use of 100% methanol at 50 p1 min-' for at least 2 h without the use of a solvent removal process. For acetonitrile although the plasma was stable at 50 p1 min-' flow rate it is preferable to use a liquid flow of 10plmin-' or less as the sampling orifice becomes blocked by carbon deposition after 1 h at the 8000 1lOO:O 7525 5050 25:75 0:lOO I 0 100 200 Ti mels Fig.8 Influence of H20-MeOH ratio on response for Cd Rh and Pb ( 1.0 ppm) using FI-OCN-ICP-MS; injection volume = 60 nl; liquid flow rate = 10 p1 min-' higher flow. Particle sizing studies3' indicate that a change in water methanol ratio has little influence on aerosol size distri- butions for the OCN. For continuous microflow OCN- ICP-MS signals the optimum gas flow rate varied considerably with solvent composition. For example optimum gas flow rates were 1.10 and 0.81 1 min-' for pure aqueous and methanol solvents respectively. The order of the optimum gas flow rate for water and methanol with the microflow OCN-ICP-MS system was similar to that with the PN at higher liquid In contrast to the higher flow condition the optimum gas flow rate for the microflow FI-OCN-ICP-MS system was found to be the same for both water and methanol.Peak profiles were generated for a 60 nl injection at a 10 pl min-' flow rate for a mixture of 1.0 ppm of Rh and 1.0 ppm of Pb both in pure methanol and in pure water and at different gas flow rates. It was observed that signals for both Rh and Pb reached a maximum with increasing gas flow rate up to 1.151min-' but that the peak shapes were basically unchanged. Unlike the situation observed with conventional high flow nebulizers the trends were very similar for both water and methanol. Fig. 8 shows ICP-MS traces for 60 nl injections of Rh Cd and Pb at different water-methanol solvent compositions ranging from 100% H,O to 100% MeOH.At a 10plmin-' flow rate signal responses and peak shapes were almost identical for the three elements in solvents of all compositions. This observation may be contrasted with the results of studies using high flow injection pneumatic nebulization (PN)- ICP-MS.31*35 This behaviour is probably the result of the constancy of the aerosol drop size distribution for the OCN with different solvents and the extremely small amount of injected sample which therefore exerts little matrix effect on the plasma. The uniformity of response for a wide range of solvent compositions is a very valuable OCN character- istic for LC-ICP-MS interfacing because this allows chromatography conditions to be optimized to consider possible response variation with composition.The use of gradient elution in a mode also becomes quite straightforward. without the need changing solvent fully quantitative phenyl-2-aminoethyl selenide (PAESe) SeCH*C+NHCS CH @ 4- h yd roxy- p hen yl-2-am i noet h yl sel enide phenyl-2-acetamidoethyl selenide (HO-PAESe) (N-acetyl PAESe) 4-fluoro-phenyl-2-aminoethyl selenide (RS)a-methyl-phenyl-2-aminoethyl selenide (pF-PAESe) (( RS)-MePAESe) Fig. 9 Structures of selenide species examined Jourrlal of Analytical Atomic Spectrometry December 1996 Vol. 1 1 I 143Selenide Speciation by LC-OCN-ICP-MS The phenol-2-aminoethyl selenide group of compounds devel- oped by May et al.,45 has attracted considerable interest for its highly effective antihypertensive activity.46 These compounds are substrates of the enzyme dopamine P-monooxygenase.Separations using macro and micro-LC columns Fig. 9 shows the structures of the molecules PAESe HO-PAESe F-PAESe N-acetyl PAESe and (RS)-MePAESe and Fig. 10 shows a chromatogram of a standard mixture of these species measured using an OCN-LC-ICP-MS system and monitoring the selenium-82 isotope. The optimum separa- tion conditions are given in Table 2. Peaks were identified by retention times established by separate injections of each species. Monitoring at the 77 and 78 isotopes of Se gave very similar peak shapes to selenium-82 despite a difference in background response. Retention times were almost identical for each Se compound at both 50 pl min-' and 0.4 ml min-' flow rates.Peaks for the five selenide compounds were well resolved with comparable resolution under both sets of conditions. A very striking and unexpected feature of these results is that the sensitivity of the OCN-ICP-MS system for each selenide compound was observed to be about 10 times higher for a solvent flow of 50plmin-I than for a solvent flow of 0.4 ml min-'. However the relative peak areas of compounds in both chromatograms remained similar for both sets of conditions. LC-ICP-MS does not behave as a concentration responsive detector. Therefore no signal enhancement is expected to arise in this system from concentration enrichment that might occur on a small diameter column. Consequently the higher signals observed in this study at low flow rates can only be a result of either 1 higher analyte transport efficiency to the plasma,47 giving rise to a greater species population in the plasma; or 2 lower solvent loading in the plasma leading to more efficient ionization of the existing species.48 In the absence of solvent effects the ICP-MS signal sensitivity should be approximately proportional to the total analyte mass transport rate to the plasma qot.The analyte concentration on the column is enriched by approximately a factor of 10 in going from a 3.2 mm diameter column to a 1 mm diameter column but the liquid flow rate is correspondingly decreased by a factor of 8 so the net improvement in total analyte mass transport to the plasma caused by the column should only be a factor of 1.3. Clearly the one order of magnitude signal enhancement which occurs on changing from a conventional 4.6mm id column (a) (PAESe) (N-acetylPAESe) t II 24 28 32 36 40 44 Time/min Fig. 10 Speciation of mixture of 5 selenides 166 ng of each species in 1 pl injection Se m/z=82.(a) OCN at 0.4ml min-' on 25 cmx 3.2mm id C18 column; and (b) OCN at 50 plmin-' on 25 cm x 1.0 mm id C column running at 0.4 ml min-' to a 1 mm id column running at 50 p1 min-' cannot be accounted for simply by concentration effects occurring on the narrow-bore column. The net signal increase can probably best be explained by considering a combination of higher analyte transport efficiency and better plasma ionization conditions. If it is assumed that the conven- tional concentric nebulizer running at 0.4 ml min-' is 1.5% efficient and that the OCN running at 50 pl min-' is approxi- mately 50% efficient (which are reasonable assumptions based on recent data),49 the ratio of (qot at 400 pl min-I) (Pi& at 50 pl min- ') = 4.2.Consequently a significant part of the increased ion count may be attributed to the more efficient analyte delivery which occurs at low flows with the OCN compared with the lower efficiency which occurs at high flows with a conventional nebulizer. The remaining factors are probably accounted for by a combination of column concen- tration ( 1.3-fold see above) and enhanced plasma ionization conditions ( 1.9-fold). Separation with capillary LC column A Se-specific chromatogram of the same selenide mixture as that shown in Fig. 10 was repeated using a 0.5 mm id capillary C18 column.As is shown in Fig. 11 the 5 selenides were adequately resolved using the capillary column. One notable difference observed between these and earlier results (Fig. 10) is that for separations carried out at 12 pl min-' relative signal ratios for eluting bands decrease significantly with increasing chromatographic retention time. This stands in marked contrast to experiments carried out at 50 pl min-' and 0.4 ml min-' where the relative peak sizes hardly change with eluent flow. With the microbore column run at 50 pl min-' separation of the selenide mixture took about 40min. Using the capillary column at 12 pl min-' the same separation took roughly twice as long. Consequently the drop-off in response for the later eluting peaks as seen in Fig.11 may be largely a result of increased band dispersion resulting from extended retention times. Absolute limits of detection for Se-speciation were estimated using peak height based on a 3a criterion. Each selenide had a different response factor the order of response being MePAESe > N-acetyl-PAESe x pF-PAESe > PAESe > OH- PAESe. Four different systems were compared including the Meinhard nebulizer at 0.4 ml min-' with the narrow-bore column and the OCN at 0.4 ml min-' 50 p1 min-' and 12 pl min-' with the narrow-bore microbore and capillary columns respectively. Results are shown in Table 5. A 50 pg absolute limit of detection under these circumstances corre- sponds to a concentration detection limit of approximately 0.8 ppm. For all the selenide compounds tested the following general 60000 ~ 4 m [ ( " ' 30 40 50 60 70 80 90 Time/min Fig.11 Speciation of 5 selenides at Se m/z 82 12 p1 min-' flow on 25 cm x 0.5 mm id C1 column. (a) 16.7 ng of each Se species in 1 pl injection; and (b) 1 ng of Se in 60 nl injection 1 144 Journal of Analytical Atomic Spectrometry December 1996 Vol. 11Table 5 Absolute limits of detection for 6 selenide compounds with OCN at different solvent flow rates Absolute limit of detection of selenide compound/pg ~~~~~~ OH-PAESe PAESe pF-PAESe N-acetyl-PAESe MePAESe Nebulizer OCN at 12 pl min-'* 50 00 150 155 215 OCN at 50 pl min-'t 30 50 70 90 100 OCN at 0.4 ml min-'* 150 200 250 3 50 400 PN at 0.4 ml min* 450 ?OO 1200 1250 2000 * 0.5 mm id C18 column. 1.0 mm id C18 column.* 3.2 mm id C1 column. Table 6 Plate numbers for 5 selenides on c18 column Theoretical plate number for selenide compound* Nebulizer OH-PAESe PAESe pF-PAESe N-acetyl-PAESe MePAESe OCN at 12 pl min-lt 8000 3CI000 5000 7000 9000 OCN at 50 pl min-'* 10 200 1 (I 400 40 500 50 700 20 800 OCN at 0.4 ml min-'§ 10 450 3c3 200 30 000 40 100 30 900 PN at 0.4 ml min- 10 700 2c1 000 20 700 50 000 20 500 * Calculated from N = 5.54 (tR/wl,2)2 where t = retention t h e w1,2 =peak width at half height. 0.5 mm id C column. * 1.0 mm id c18 column. 3.2 mm id c18 column. conclusions can be drawn. First the absolute LODs obtained by the OCN-ICP-MS combined with any size column were superior to the values obtained using the conventional PN-ICP-MS system. Secondly the best powers of detection were observed for the LC-OCN-ICP-MS in the microflow rate region.Clearly in any situation where analyte solution volumes and concentrations are severely limited the micro- scale LC-OCN-ICP-MS offers an attractive means for a;hiev- ing high detection power as well as good separation efficiency. Furthermore the system consumes much less valuable LC grade solvent compared with conventional flow separations. For reference purposes the estimated chromatographic efficiencies were determined for the four systems. Theo:.etical plate numbers for each of the five selenides were calculated for each system based on the peak width at half peak height. Results are shown in Table 6. Typical values were in the range of 10000 to 30000 theoretical plates. CONCLUSIONS This work demonstrates that the oscillating capillary nebulizer is a very versatile aerosol generator for both micro- and macro-LC-ICP-MS interfacing.It operates stably over an exceptionally wide range of liquid flow rates from 1 pl inin-' to 2 ml min-'. The continuous flow and flow injection experiments in the microflow rate range indicate good stability and precision for both OCN-ICP-MS and OCN-LC-ICP-MS systems. The ability to nebulize a wide variety of solvents with high analyte transport efficiency provides excellent flexibility for quantitat- ive multielement analysis. Because of its low dead volume the OCN is useful for interfacing different sized LC columns for both ICP-AES and ICP-MS detection especially when using micro-scale separation techniques.The uniform response obtained at micro-flow rates when using either organic solvent mixtures of different composition or aqueous solutions is another attractive feature of the OCN. A study of aerosol transport and performance characteristics of the both OCN and other low-flow nebulizers has been completed and will be reported el~ewhere.~' This work is based on research carried out with financial support from NIH (HL 28167) and NSF (CHE88-01813). REFERENCES 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 Vela N. P. Olson L. K. and Caruso J. A. Anal. Chem. 1993 65 585A. Canals A. Hernandis V. and Browner R. F. J. Anal. At. Spectrom. 1990 5 61. Smith D. D. and Browner R. F. Anal. Chem. 1982 54 1411. Wiederin D. R. Smith F. G. and Houk R. S. Anal. Chem. 1991 63 218.Liu H. Montaser A. Dolan S. P. and Schwartz R. S. J. Anal. At. Spectrorn. 1996 11 307. Brotherton T. J. Shen W. L. and Caruso J. A. J. Anal. At. Spectrorn. 1989 4 39. Lawrence K. E. Rice G. W. and Fassel V. A. Anal. Chem. 1984 56 289. Wiederin D. R. Smith F. G. and Houk R. S. Anal. Chem. 1991 63 219. Tarr M. A. Zhu G. and Browner R. F. Anal. Chem. 1993 65 1689. Lin L. Olson L. and Caruso J. A. 21st FACSS Meeting St. Louis MI USA 1994 Paper 336. Liu H. Clifford R. H. Dolan S. P. and Montaser A. Spectrochim. Acta Part B 1996 51 27. Olesik J. W. Kinzer J. A. and Harkleroad B. Anal. Chem. 1994 66 2022. Liu H. and Montaser A. Anal. Chem. 1994 66 3233. Wiederin D. H. Pittsburgh Conference on Analytical Chemistry and Applied Spectroscopy Chicago IL USA 1996.Koropchak J. A. and Veber M. Crit. Rev. Anal. Chem. 1992 23 113. Labonda F. De Loos-Vollebregt M. T. C. and De Galan L. Spectrochim. Acta Part B 1991 46 1089. Vanhoe H. Moens L. and Dams R. J. Anal. At. Spectrom. 1994 9 815. LaFrenier K. E. Fassel V. A. and Eckels D. E. Anal. Chem. 1987 59 879. Shum S. C . K. Pang H. and Houk R. S. Anal. Chem. 1992 64 2444. Shum S. C. K. and Houk R. S. Anal. Chem. 1993 65 2972. Alder J. F. Bombelka and Kirkbright G. F. Spectrochim. Acta Part B 1980 35 163. Fister J. C. and Olesik J. W. Spectrochim. Acta Part B 1991 46 869. Journal of Analytical Atomic Spectrometry December 1996 Vol. 1 1 1 14523 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 Hobbs S. E. and Olesik J. W. Anal. Chem. 1994 66 2022. French J. B. Etkin B. and Young R.Anal. Chem. 1994,66 685. Olesik J. W. and Hobbs S . E. Anal. Chem. 1994 66 3371. Huggins T. G. and Henion J. D. Electrophoresis 1993 14 531. Agnes G. R. and Horlick G. Appl. Spectrosc. 1994 48 649. Agnes G. R. and Horlick G. Appl. Spectrosc. 1994 48 655. Olesik J. W. Kinzer J. and Olesik S . V. Anal. Chem. 1995,67 1. Browner R. F. Canals A. and Hernandis V. Spectrochim. Acta Part B 1992 47 659. Wang L. Tucker A. May S. W. and Browner R. F. Anal. Chem. 1996 in the press. May S. W. Herman H. H. Roberts S. F. and Ciccarello M. C. Biochemistry 1987 26 1626. Boorn A. W. and Browner R. F. Anal. Chem. 1982,54 1402. Kreuning G. and Maessen F. A. M. J. Spectrochim. Acta Part B 1989 44 367. Weir D. G. and Blades M. W. J. Anal. At. Spectrom. 1996,11,43. Browner R. F. Boorn A. W. and Smith D. D. Anal. Chem. 1982 54 533. Browner R. F. ‘Fundamental Aspects of Aerosol Generation’ in Analysis by Inductively Coupled Plasma Atomic Emission Spectrometry ed. Boumans P. W. J. M. John Wiley New York USA Part 11 1987. Farino J. Miller J. Smith D. D. and Browner R. F. Anal. Chem. 1987 59 2303. Bates L. C. and Olesik J. W. J. Anal. At. Spectrom. 1990 5 239. 40 41 42 43 44 45 46 47 48 49 Long S. E. and Browner R. F. Spectrochim. Acta Part By 1988 43 1461. Pan C. Zhu G. and Browner R. F. J . Anal. At. Spectrom. 1990 5 537. Pan C. Characterization of Solvent-Plasma Interaction for Inductively Coupled Plasma Atomic Emission Spectrometry and Inductively Coupled Plasma Mass Spectrometry Ph.D. Thesis Georgia Institute of Technology Atlanta GA USA 1991. Shum S. C. K. Johnson S. K. Pang H. and Houk R. S. Appl. Spectrosc. 1993 47 575. Canals A. Hernandis V. and Browner R. F. Spectrochim. Acta Part B 1990 44 8. May S. W. Wimalasena K. Herman H. H. Fowler L. C. Ciccarello M. C. and Pollock S. H. J. Med. Chem. 1988,31 1066. Pollock S. H. Herman H. H. Fowler L. C. Edwards A. S. Evans C. and May S . W. J. Pharmacol. Exp. Ther. 1988,246,227. Tarr M. A. Zhu G. and Browner R. F. J. Anal. At. Spectrom. 1992 7 813. Bates L. C. and Olesik J. W. J. Anal. At. Spectrom. 1990 5 239. Jarrett J. and Browner R. F. Spectrochim. Acta Part B 1996 submitted for publication. Paper 61061 OOA Received September 4 1996 Accepted October 14 1996 1 146 Journal of Analytical Atomic Spectrometry December 1996 Vol. 11

ISSN:0267-9477    

DOI:10.1039/JA9961101137

出版商:RSC    

年代:1996

数据来源: RSC

摘要:

Effect of Different Spray Chambers on the Determination of Organotin Compounds by High-performance Liquid Chromatography-Inductively Coupled Plasma Mass Spectrometry Journal of Analytical Atomic Spectrometry CRISTINA RIVAS LES EBDON AND STEVE J HILL* Analytical Chemistry Research Unit Department of Environmental Sciences University of Plymouth Drake Circus Plymouth Devon UK PL4 8AA One of the potential problems to overcome when coupling liquid chromatography with atomic spectrometry is the low sample transport efficiency. Often this is a consequence of the interface design and in particular the nebulizer and spray chamber configuration. The present study reports on a comparison of several spray chambers (single-pass doublepass and cyclone types) with respect to both design and performance characteristics.In order to evaluate performance for speciation studies different configurations have been evaluated utilizing HPLC-ICP-MS. Various organotin compounds (monobutyltin dibutyltin tributyltin and triphenyltin chlorides) were coinjected and the effects on resolution sensitivity and SNR assessed. Of the spray chambers evaluated a cyclone spray chamber (internal volume 22 ml) with cooling jacket was found to offer the best performance and gave a transport efficiency of 7.5% without loss of chromatographic resolution and sensitivity. Keywords Spray chamber design; high-performance liquid chromatography-inductively coupled plasma interface; organotin speciation One of the major weaknesses of ICP-MS is the low efficiency of the transportation of the sample to the plasma.The aerosol droplets produced by a nebulizer should be of a diameter of less than 1Opm so that the desolvation volatilization and atomization are as rapid and efficient as possible. The com- monly used pneumatic nebulizers produce a wide distribution of droplet size usually up to m in diameter.' The main use of a spray chamber is thus to separate the larger droplets from the smaller ones allowing the latter to reach the plasma. Unfortunately 98-99% of the sample is lost in this process when employing commonly used spray chamber design^.^.^ Separation of the droplets occurs as a result of a variety of processes that eliminate the larger drops according to the different trajectories inside the spray chamber and collisions with the walls or with a bead placed inside the chamber The larger drops go to waste and the smaller droplets are carried into the torch and subsequently to the plasma.Several attempts have been made to improve the efficiency of nebulization Most common designs of spray chambers employ flow reversal cyclones or impact beads. The two first types cause changes in the flow direction as well as impaction on the walls of the chamber. Those employing impact beads operate by placing a device which intercepts the flow of the aerosol and provides an in-line impaction site. The performance of a nebulizer-spray chamber system can be evaluated through its analytical performance provided that the operating parameters are fully optimized for each of the * To whom correspondence should be addressed.systems. In the present study seven different types of spray chambers were tested following this procedure. Two spray chambers were Scott-type (double and single pass) one was a miniaturized laboratory-made double pass and the remaining four were of the cyclone type. The presence of organotin species in the environment has been of great concern in recent years owing to their high toxicity particularly to marine organisms. Levels as low as 1 pgl-' of tributyltin (TBT) as tin in water can affect the population and mariculture of marine organisms especially shellfish." Considerable effort has been made to improve the sensitivity and selectivity of the analytical methodology used to detect such species and in particular to develop techniques capable of determining quantitatively the chemical form of the analyte.One of the more popular approaches used to separate the different species has been HPLC since it provides a simple and rapid method without the requirement for derivatization of the analytes prior to the chromatographic separati~n.''-'~ Thus the nature of the coupling of the HPLC and the ICP-MS instruments is critical. Any chromatographic detector should not markedly increase the dead volume of the system otherwise chromatographic resolution will be lost. Several attempts have been made to couple chromatography with ICP-MS whilst ensuring that there is minimal increase in dead volume particularly by placing the nebulizer and spray chamber close to the end of the column. A reduced volume spray chamber of the cyclone type has been reported by Wu and Hieftje4 for ICP-AES.The internal volume of this spray chamber was 40ml and advan- tages in transport efficiency were reported. It therefore seemed timely to investigate a number of different spray chambers for directly coupled HPLC-ICP-MS including a reduced volume cyclone chamber based on the Wu and Hieftje design but modified for ICP-MS. Four different organotin compounds were separated and the performance of the nebulizer-spray chamber system evaluated considering both the sensitivity for each analyte and the chromatographic resolution obtained for the organotin species. Various parameters for each spray chamber were evaluated including transport efficiency shape and internal volume. The present paper reports the findings of this study.EXPERIMENTAL Chemicals The tributyltin chloride (96%) dibutyltin (DBT) chloride (96%) monobutyltin (MBT) chloride (95%) and triphenyltin (TPhT) chloride (95%) were obtained from Aldrich Gillingham Dorset UK. Stock solutions of each organotin ( 1000 pg g- ') were prepared in HPLC-grade methanol (Rathburn Chemicals Journal of Analytical Atomic Spectrometry December 1996 Vol. 11 (1 147-1 150) 1 147Walkerburn Peeblesshire UK) and stored in the dark at 4°C. HPLC-grade methanol was also used to prepare the mobile phases for the separation studies. Appropriate dilutions of the stock solutions were made when necessary with the mobile phase employed for the chromatographic separation. The triammonium citrate and citric acid used as buffers were obtained from Fisons Loughborough Leicestershire UK as analytical-reagent grade.Lipase and protease enzymes employed in the enzymatic extraction of fish tissue were obtained from Aldrich Chemicals. Sodium dihydrogenphosphate buffer was BDH AristaR grade (BDH Poole Dorset UK). Dichloromethane was HiPerSolv grade obtained from BDH. Milli-Q de-ionized water ( Millipore Bedford MA USA) or equivalent was used throughout the study. Instrumentation The HPLC was carried out using an inert gradient pump (Varian Model 9010 Warrington Cheshire UK). A 200 pl injection loop was fitted to a chemically inert injection valve (Cheminert Model C1 valve Valco Instruments Houston TX USA). The two analytical columns (250 x 4.6 mm) were packed in-house with 10 pm Partisil SCX-10 (Thames Chroma- tography Maidenhead Berkshire UK). The mobile phase used throughout the study at a flow rate of l.Omlmin-' consisted of a mixture of methanol water and citrate buffer.The chromatographic conditions used are shown in Table 1. The ICP-MS instrument used for the study was a PlasmaQuad 2 + (Fisons Instruments Elemental Winsford Cheshire UK). The operating conditions used are shown in Table 2. The nickel sampler and skimmer cones (Fisons Instruments Elemental) had orifices of 1.0 and 0.7 mm respect- ively. Using a gas blender (Signal Instruments Camberley Surrey UK) an addition of oxygen (1.4%) was made to the argon aerosol carrier gas to avoid deposition of carbon on the cones. The spray chamber was cooled with a recirculating chiller (Endocal RTE- 100 Neslab Instruments Newingdon NH USA).The ion lens settings were optimized every day to give the best performance. Seven spray chambers (designated as A-G) were investigated. They were conventional Scott double-pass (A) and single-pass (B) spray chambers a miniaturized laboratory-made double- pass spray chamber'* (C) and four different cyclone spray Table 1 Chromatographic parameters Column (and guard column) Partisil SCX-10 10 pm 250 x 4.6 mm (guard column 25 x 4.6 mm) Step gradient methanol-water (70% methanol 1.5 min; 85% methanol 9.5 min) ammonium citrate (pH 5.8 4.0 min; pH 3.4 7 min) Mobile phase (gradient elution) Buffer (gradient elution) Flow rate 1 ml min-' Injection volume 200 pl Step gradient citric acid- Table 2 Operating conditions for the ICP-MS system Outer gas flow/l min-' Intermediate gas flow/l min - ' Aerosol carrier gas flow/l min- Oxygen bleed (YO) Forward power/W Nebulizer Spray chamber Signal monitored (m/z) 1 15 0.75 0.85 1.4 1500 Meinhard Various 120 chambers.The first of the cyclone designs had no liquid cooling jacket (D) and the last three (E F and G) were cooled via an integral jacket These latter spray chambers differed only in the internal volume and the shape of the indentation or 'dimple'. The internal volumes of the various spray chambers are given in Table 3 and the designs and shapes of the spray chambers in Fig. 1. In each case a Meinhard nebulizer was used (Type A Fisons Instruments Elemental). RESULTS AND DISCUSSION Transport Efficiency The silica gel trap method3," was employed in all cases to obtain the transport efficiency.An argon flow rate of 1 1 min-' was used with a sample uptake rate of 1 ml min-' (achieved using a peristaltic pump). Four U-tubes filled with silica gel were connected to the exit of the spray chamber. Each U-tube was weighed before and after the passage of 50ml of Milli-Q water (also weighed). The spray chamber was cooled (except for types C and D) to -4°C. The experiment was conducted in triplicate. The results obtained are shown in Table 4. As expected the cyclone spray chambers gave better trans- port efficiency possibly as a result of allowing larger drops to pass through the system. The lowest transport efficiency was found for the miniaturized double-pass spray chamber in this case probably because of the large ratio of impaction surface to internal volume and the more contorted gas flow necessary to exit this small spray chamber.It is also interesting to note that although the uncooled cyclone spray chamber D had the highest transport efficiency the SNR obtained from the chromatogram was poor. Cooling of the spray chamber (e.g. type G) although reducing the transport efficiency much improved the SNR. The extent of this improvement was influenced by the overall design (Table 4). Effect of Internal Volume on Wash-out Time and Resolution One of the main considerations in the coupling of HPLC with ICP-MS is the dead volume of the interface. This volume affects the resolution of the chromatographic system poten- tially losing the separation previously achieved by the column. Clearly the internal volume of the spray chamber can increase the dead volume of the system.The different internal volumes of the spray chambers used in the present study are shown in Table 3. The wash-out time of a sample introduction system is the time required to clean the system and is most commonly defined as the time required for the signal to return to 1% of the original maximum. In the present study this time was calculated by monitoring the decrease in the signal of a solution of 100 ng ml-1 of "'In in 70% methanol after being replaced by a 70% methanol solution without In. The wash-out curves obtained for both the Scott-type double pass (A) and the new cyclone spray chamber (G) can be seen in Fig. 2. Surprisingly the wash-out time in the cyclone spray chamber is slightly longer (24 s) than in the much larger Table 3 Internal volumes of the spray chambers Spray chamber (A) Scott-type double pass (B) Scott-type single pass (C) Miniaturized double pass (D) Cyclone (no jacket) (E) Cyclone (jacketed) (F) Cyclone (jacketed) (G) Cyclone (jacketed) Internal volume/ml 88 40 13 20 40 27 22 1 148 Journal of Analytical Atomic Spectrometry December 1996 Vol.1 13.4 9.5cm 14.2 cm 15.0 cm ~ 4 7 -f+Ti.4 cm .9 cm Drain (A u + 1-0.6 cm Plasma Jacket Drain (s) L T . 0 cm I "Jacket Plasma ,6 cm 4 1-0.6 cm Drain Fig.1 Types of spray chamber used A Scott-type double pass; B Scott-type single pass; C miniaturized double pass; D cyclone w~ thout jacket; E F and G cyclones with jackets Table4 Transport efficiency and SNR values for TBT with each spray chamber Spray chamber Transport efficiency & s (YO) SNR (A) Scott-type double pass 2.45 f 0.07 39 (B) Scott-type single pass 2.92 & 0.20 59 (C) Miniaturized double pass 1.74 & 0.05 41 (D) Cyclone (no jacket) 8.05 k 0.12 50 (E) Cyclone (jacketed) 3.95 & 0.07 56 (F) Cyclone (jacketed) 5.85 & 0.10 56 (G) Cyclone (jacketed) 7.53 & 0.04 108 internal volume double-pass spray chamber (15 s).This would indicate that there are stagnant pockets within the cyclone spray chamber which are not being efficiently removed. Optimizing the size of the dimple is intended to minimize this problem. 100 - - _ 80 17 I I 8 1 60 70 80 90 100 Time/s Fig. 2 Wash-out times for 100 ng ml-' of "'In in 70% methanol for two spray chambers.Solid line new cyclone spray chamber G; and broken line Scott-type double-pass spray chamber A Using the present chromatographic system TBT and TPhT are not base-line resolved.I2 Any loss of resolution in the system will adversely affect the separation of species. The resolution (R,) between these two species was calculated for each of the spray chambers using the equation:" t2 - t l R,= 1.18 - w1+ w2 where tl and tz are the retention times of the peaks for TBT and TPhT respectively and w1 and w2 are the peak width at half-height for TBT and TPhT respectively. The ICP-MS experimental parameters (ie. flow rates torch position and lens settings) were sequentially optimized for each system in order to obtain the best signal (counts s-l) using a lOOngml-' solution of "'In made up in the mobile phase (70% methanol 30% water).The results obtained are shown in Table 5. As can be seen all of the spray chambers gave similar resolution although the spray chambers with the smallest internal volumes tended to give slightly better results. Figures of Merit The system giving the best performance using the above parameters was selected for further study. The cyclone spray chamber with cooling jacket G was one of the three spray chambers giving the best resolution between TBT and TPhT and the transport efficiency was also second highest of those tested. Using the cyclone spray chamber D although better transport efficiency was achieved the resolution was not so good as with spray chamber G. The miniaturized double-pass spray chamber C gave good resolution but the sensitivity obtained was inferior to that for spray chamber G.This was attributed to the lower transport efficiency. An additional disadvantage of spray chamber C was the noisy signal obtained as a result of the small internal volume. Following these considerations the cyclone spray chamber G was chosen as the best for use in the present application. Examples of the chromatograms obtained with both the Table 5 Resolution between TBT and TPhT for each spray chamber Spray chamber (A) Scott-type double pass (B) Scott-type single pass (C) Miniaturized double pass (D) Cyclone (no jacket) (E) Cyclone (jacketed) (F) Cyclone (jacketed) (G) Cyclone (jacketed) Resolution 0.81 0.83 0.84 0.74 0.76 0.75 0.83 Journal of Analytical Atomic Spectrometry December 1996 Vol.11 1 149300 450 600 750 900 1050 Time/s Fig. 3 Chromatograms obtained with Scott-type double pass A and new cyclone spray chamber G. Solid line new cyclone spray chamber G; and broken line Scott-type double-pass spray chamber A Table 6 LOD for TBT TPhT DBT and MBT with Scott-type double pass spray chamber A and the new cyclone spray chamber G LOD/ng g-' Spray chamber TBT TPhT DBT MBT (A) Scott-type double pass 1 1 2 1 (G) Cyclone 0.44 0.26 1.40 0.23 Table 7 Repeatability (RSD) for cyclone spray chamber G using three chromatographic runs at each of the concentrations RSD (Yo) Nominal concentration investigated/ng g- Anal yte 10 100 500 TBT 8.3 4.2 2.7 TPhT 8.0 4.5 4.0 DBT 6.5 3.6 2.0 MBT 7.3 10.5 7.6 Scott-type double pass spray chamber A and spray chamber G can be seen in Fig.3. The detection limits (3s of the baseline noise) obtained for the four analytes with this new spray chamber are presented in Table 6. The repeatability for spray chamber G was also tested by use of triplicate chromatographic injections of the four analytes using peak height at three concentrations (approximately 10 100 and 500 ng g-' as tin). The results obtained expressed in terms of the RSD are shown in Table 7. Analysis of Fish Tissue CRM NIES-11 A certified reference material (fish tissue NIES-11) from the National Institute of Environmental Studies of Japan (NIES) was analysed. This material is certified for TBT (1.3 f 0.1 pg g-' as TBT) and has an indicative value for TPhT (6.3 pg g-' as TPhT). The extraction procedure employed based on a recently published method by Ceulemans et was as follows.Approximately 1 g of fish tissue was accurately weighed. Lipase and protease (0.05 g of each) and 40ml of citric-phosphate buffer (prepared by dissolving 21.0 g of citric acid 14g of dihydrogen sodium phosphate and 64 ml of ethanol in 11 of water followed by adjustment of the pH with ammonia solution to pH 7.5) were added. The mixture was left overnight in a shaker at 37°C. The following day the aqueous phase was extracted into 10 ml of dichloromethane three times. The organic phase was then rotary evaporated until dryness. The dry extracts were redissolved into a mixture of 66% methanol and 33% mobile phase prior to the analysis. The separation of the analytes was performed using the chromatographic conditions described previously utilizing the new cyclone spray chamber type G.The method proved to be adequate for the analysis of this material. The result obtained for TBT (n=6) following this procedure (1.29k0.05 pg g-' as TBT) is in good agreement with the certified value (1.3k0.1 pg g-'). CONCLUSIONS A modified spray chamber has been designed and characterized for use with HPLC-ICP-MS. The main parameters for its construction and its analytical characteristics have been evalu- ated. This spray chamber gave improved detection limits for all organotin compounds used in the evaluation. These improvements can be attributed to a higher transport efficiency (7.5%) without loss of resolution and proved to be superior to the conventional Scott-type double pass spray chamber and other designs used in this study.The authors thank the European Community (Standards Measurements and Testing Programme) for supporting this work and for the provision of a student bursary (to C . R.). REFERENCES 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 Ebdon L. Foulkes M. and Hill S. J. Microchem. J. 1989,40 30. Sharp B. L. J. Anal. At. Spectrom. 1988 3 939. Ebdon L. and Cave M. R. Analyst 1982 107 172. Wu M. and Hieftje G. M. Appl. Sprectrosc. 1992 46 1912. Veres S. A. Briickner P. H. and Denoyer E. R. At. Spectrosc. 1994 96. Wu M. Madrid Y. Auxier J. A. and Hieftje G. M. Anal. Chim. Acta 1994 286 155. Nam S.-H. Lim J.-S. and Montaser A. J. Anal. At. Spectrom. 1994 9 1357. Wiederin D. R. Houk R. S. Winge R.K. and D'Silva A. P. Anal. Chem. 1990 62 1155. Wiederin D. R. Smith F. G. and Houk R. S. Anal. Chem. 1991 63 219. Alzieu C. Thibaud Y. HCral M. and Boutier B. Rev. Trav. Inst. P2ches Marit. 1980 44 301. Ebdon L. Hill S. J. and Jones P. Analyst 1985 110 515. Rivas C. Ebdon L. and Hill S. J. Quim. Anal. 1995 14 142. Rivas C. Ebdon L. Evans E. H. and Hill S. J. Appl. Organomet. Chem. 1996 10,61. Kadokami K. Uehiro T. Morita M. and Fuwa K. J. Anal. At. Spectrom. 1988 3 187. Dauchy X. Astruc A. Borsier M. and Astruc M. Analusis 1992 20 41. Astruc A. Dauchy X. Pannier F. Potin-Gautier M. and Astruc M. Analusis 1994 22 257. Dauchy X. Cottier R. Batel A. Astruc A. and Astruc M. Environ. Technol. 1994 15 569. Ambrose A. J. Ebdon L. Foulkes M. E. and Jones P. J. Anal. At. Spectrom. 1989 4 219. Browner R. F. in Inductively Coupled Plasma Emission Spectroscopy Part I I Applications and Fundamentals ed. Boumans P. W. J. M. Wiley New York 1987 pp. 244-288. Snyder L. R. Glajch J. L. and Kirkland J. J. Practical HPLC Method Development Wiley New York 1988. Ceulemans M. Witte C. Lobinski R. and Adams F. C. Appl. Organomet. Chem. 1994 8 451. Paper 6/03482 I Received May 20 1996 Accepted August I 1996 1 150 Journal of Analytical Atomic Spectrometry December 1996 Vol. 11

ISSN:0267-9477    

DOI:10.1039/JA9961101147

出版商:RSC    

年代:1996

数据来源: RSC

摘要:

Feasibility Study of Low Pressure Inductively Coupled Plasma Mass Spectrometry for Qualitative and Quantitative Speciation GAVIN O’CONNOR LES EBDON AND E. HYWEL EVANS* Department of Environmental Sciences University of Plymouth Drake Circus Plymouth Devon UK PL4 8AA HONG DING LISA K. OLSON AND JOSEPH A. CARUSO Department of Chemistry University of Cincinnati Cincinnati OIi 45221 -01 72. USA Low-pressure ICPs (LP-ICPs) formed with helium have been utilized as ion sources for MS. In the first part of the paper a description is given of a plasma operated at 90 W forward power used to ionize organotin and organolead compounds introduced by GC with detection limits of 11 2 and 14pg obtained for tetraethyltin (TEtSn) tetrabutyltin (TBuSn) and tetraethyllead ( TEtPb) respectively. The same plasma operated at 45 W forward power yielded fragment ions of TEtSn and TBnSn.In the second part a customized mass spectrometer for use with an LP-ICP is described. The optimum power for the generation of fragment ions of perfluorotributylamine (PFTBA) was found to lie between 5 and 8 W. Two optima for skimming distance were observed at 6 and 8 mm downstream of the sampler orifice. Mean ion kinetic energies for fragment ions of PFTBA at 69,219 and 502 m/z were 1.2 1.7 and 2.1 eV respectively at 6 W forward power but increased when the power was increased to 8 W. Molecular and other fragment ions were observed for chloro- iodo- and dibromobenzenes introduced using GC. Keywords Low-pressure inductively coupled plasma; helium plasma; mass spec trorne t r y ; element-selec tiue detection; molecular ion The requirement for information about the chemical composi- tion and structure of organic compounds has led to the development of numerous instrumental techniques.For example the elemental composition of organometals can be determined by ICP-MS while structures can be elucidated by techniques such as electron impact MS (EI-MS) and nuclear magnetic resonance (NMR) spectrometry. Additionally an extra degree of specificity can be afforded by coupling separa- tion techniques such as LC or GC to the above spectrometric instruments. Mass spectrometer ion sources essentially fall into two categories i.e. those which create molecular species and frag- ment ions such as conventional electron impact (EI) chemical ionization (CI) and fast atom bombardment (FAB) and those which create atomic species such as ICPs and MIPs.Coupled with suitable chromatographic inlets these have led to the development of two of the major instrument classes found in virtually all modern analytical laboratories. However no exist- ing mass spectrometer ion source allows routine monitoring of both atomic and molecular species particularly in a tune- able mode. For trace metal speciation element-selective detectors coupled with chromatography of one form or another have been used successfully to quantify organometallic compounds * To whom correspondence should be addressed. Journal of Analytical Atomic Spectrometry at sub-pg levels.’ However qualitative identification of the compounds has relied on comparison of chromatographic retention times and it has proved impossible to identify unknown peaks without the use of a complementary technique.This effectively limits the analysis to the determination of known compounds. A major advantage of element-selective detection is the ability to calibrate for an unknown compound with a known compound containing at least one element in common. This is possible because the compounds are completely broken down to their constituent atoms and then ionized and/or excited to facilitate either optical or mass spectrometric detec- tion of monatomic ions. Hence the signal intensity does not depend on the nature of the compound but only on the nature of the element chosen for detection. If it were possible to operate a single instrument as an element-selective detector and a qualitative analyser this would allow compounds to be identified qualitatively in one mode of operation and deter- mined quantitatively in another and would provide a powerful technique for the determination of a wide range of compounds both known and unknown.MIPs operated under reduced pressure have previously been investigated as possible sources for fragment ions of organic corn pound^.^-^ However these experiments were gen- erally performed by introducing the pure compound or head- space vapour so were not representative of analysis at trace levels. Atmospheric pressure ionization (API) sources have also been developed utilizing an MIPS or GD.6 The latter workers noted that the predominant ions produced using the API source were the molecular (M’) and (M + 1)’ ions while the same GD source used under reduced pressure produced mol- ecular ions accompanied by fragmentation.However these workers used continuous sample introduction and no work was performed with transient signals. Recently Evans and co-workers have developed a low- pressure ICP (LP-ICP) as an ion source for MS,’-’ which is capable of providing information about both the elemental composition and molecular structure of the analyte. In particu- lar Evans et aL9 have operated a low-pressure helium ICP at powers of between 4 and 40 W and 1 mbar (1 bar= lo5 Pa) pressure to produce molecular ions and fragmentation spectra similar to an EI source for organometallic and halogenated species introduced by GC. On increasing the power and pressure the degree of fragmentation was increased until com- plete fragmentation and hence element-selective detection was achieved at a power of 150 W and pressure of lOmbar with detection limits in the 10-1OOpg range for a variety of organohalogen and organometallic compounds.Clearly this demonstrates the potential of the technique as a dual mode detector. Journal of Analytical Atomic Spectrometry December 1996 Vol. 11 (1 151 -1 161) 1 151UTILIZATION OF A COMMERCIAL ICP-MS INSTRUMENT Experimental Instrumentation The ICP-MS instrument used was a VG PlasmaQuad I (VG Elemental Winsford Cheshire UK). The original torch-box was designed for an atmospheric pressure argon ICP so it did not have a sufficient tuning range to impedance match low pressure plasmas.Hence the tuning range was extended by increasing the vacuum capacitance approximately four-fold thereby reducing the high reflected power to zero for helium LP-ICP operation at 100 W forward power. The low-pressure torch was a quartz tube 145 mm long with 6mm 0.d. and 4mm i d . One end of the torch was connected to a GC interface with an Ultra-Torr fitting and the other end was connected to a modified aluminium sampler (Fig. 1). The sampler-skimmer spacing was reduced by approximately 2 mm using spacers behind the skimmer plate. Additional pumping at the expansion stage was provided by a 1500 1 min-' E1M-80 rotary vacuum pump (Edwards High Vacuum Crawley Sussex UK). The pressure achieved with this interface in the absence of any helium flow was 7.0 x mbar and 4.4 x lo-' mbar with a helium flow of 562 cm3 min-'.The gas chromatograph was a Hewlett- Packard Model 5700A. The capillary column (40 m x 0.32 mm i.d.) had an SE 54 non-polar stationary phase (J&W Scientific Austin TX USA). The transfer line from the GC outlet to the interface T was a 1/8 in 0.d. copper tube through which a 0.32 mm i.d. deactivated fused silica capillary column was inserted to transfer the GC eluent. The transfer line was heated with a heating tape powered with a variable transformer (Fisher Scientific Pittsburgh PA USA) and the helium plasma gas line was heated to the same temperature as the GC transfer line using the same type of heating device. The heated transfer and gas lines were necessary for complete and reproducible transfer of GC analytes and to minimize any temperature disturbance at the mixing T.A four-way valve (Valco Anspec Ann Arbor MI USA) was employed for venting in some of the experiments. Two mass-flow controllers capable of flows of up to 3000 and 25 cm3 min-' (Tylan General Torrance CA USA) respectively were utilized to regulate the plasma gas and GC make-up gas flows. EI-MS spectra of the organotin compounds were provided by Jim Carlson (Department of Chemistry University of Cincinnati) from a computer data base or obtained experimentally using a GC-MS instrument (6890 Series Hewlett-Packard Palo Alto CA USA). Ion lenses Quadrupole I Skimmer diffusion diffusion Pump Pump - 1 o oven Plasma gas Fig. 1 with an LP-ICP source Modifications to an atmospheric ICP-MS instrument for use Reagents Tetraethyltin (TEtSn) tetrabutyltin (TBuSn) and tetraethyllead (TEtPb) compounds (Aldrich Milwaukee WI USA) were used to prepare standard solutions and a Lead in Fuel SRM was used for analysis (SRM 2715 NIST Gaithersburg MD USA).Methanol and hexane (Fisher Scientific Fair Lawn NJ USA) were used for solution preparation for the tin and lead compounds respectively. Stock solutions (1000 pg g-') were prepared from the alkylmetal compounds as m/m concen- tration of the metal moiety. Fresh analytical solutions were prepared daily from these stock solutions. Results and Discussion Optimization of Operating Conditions Operating conditions for the GC instrument were similar to those used in previous studies. Briefly the initial oven tempera- ture was held at 80°C for 2 min then increased to 290°C at a rate of 32 "C min-'.However the temperature of the GC injection port proved to be important for the quantitative recovery of TBuSn. As shown in Fig. 2 the signal-to-noise ratio (S/N) increased substantially with increasing injection port temperature before starting to plateau at 400 "C. The position of the capillary termination just behind the load coil of the ICP had to be adjusted carefully to obtain good S/Ns. As shown in Fig. 3 there was an optimum distance between the capillary outlet and the load coil. Shortening the length of the torch by 10 mm resulted in a 62-180% increase in S/N [Fig. 3(d)]; however shortening the quartz tubing any further [Fig. 3(e)] resulted in unstable signals possibly owing 90 A I ' - ' - - l 80 70 h 5 6 0 8 1 30 20 10 cn 200 250 300 350 400 Temperature/"C Fig.2 Effect of GC injection port temperature on S/Ns for TEtSn (a) and TBT (El) Position of the capillary S/N TESn S/N TBuSn (a) -1 Erratic signals (b) 52 78 000 *** (c) 74 168 -15.5 cm - (d 120 485 0.0 -14.5 cm- -11 cm- Unstable signals Fig.3 Effect of the position of the GC capillary and length of the LP-ICP torch on S/Ns for TBuSn and TEtSn 1 152 Journal of Analytical Atomic Spectrometry December 1996 Vol.1 11 120 I 1 500 s 400 .?! E 300 z 2j 200 100 360 444 562 604 715 He flow/ml min-' Fig. 4 Effect of He outer plasma gas flow on S/Ns for TEtSn (H) and TBuSn (8) c PowerMl E - 50 40 30 20 10 v) 415 371 329 300 232 TemperaturePC Fig.5 Effect of (a) forward power and (b) transfer line temperature on S/Ns for TEtSn (B) and TBuSn (a) to interference between the metal sheath of the capillary and the load coil.For total metal determination the forward power helium plasma gas flow and GC transfer line temperature were optim- ized to yield the best S/Ns for m/z 120. The helium plasma gas flow did not affect the S/N greatly over a range of 285- 652 cm' min-l (Fig. 4) but the influence of forward power and transfer line temperature were more significant (Fig. 5). The optimal ranges of forward power and transfer line tempera- ture required for good S/Ns were 75-90 W and 30O-39O0C respectively. Optimal operating conditions for the low-pressure plasma are listed in Table 1. Good S/Ns were observed for 1OOpg injections of TEtSn and TBuSn (Fig.6) although the TEtSn peak was broader than the TBuSn peak because it was affected by the solvent front. A calibration was successfully performed with at least three orders of magnitude linear dynamic range less than 5% RSD for both compounds and detection limits of 19 and 11 pg for TEtSn and TBuSn respectively. Figures of merit are summarized in Table2. Peaks were also observed at m/z 135 Table 1 Operating conditions for the low-pressure inductively coupled plasma Forward power/W Reflected power/W He outer plasma gas flow/cm3 min-' Expansion stage pressure/mbar* Transfer line temperature/"C 90 2 562 4.4 x lo-' 325 * 1 bar= lop5 Pa. TBuSn I 0 1 2 3 4 5 Time/min Fig.6 Separation of 1OOpg each of TEtSn and TBuSn by GC-LP-ICP-MS with element-selective detection at m/z 120 and without solvent venting and 150 for both TEtSn and TBuSn (Fig.7). These were thought to be due to Sn-CH at m/z 135 and Sn-(CH,) at m/z 150 although peak intensities were low and high concen- trations of the analyte compounds (20 and 50 ng respectively) had to be used. The molecular ion for TEtSn at m/z 236 was not observed and that for TBuSn at m/z 348 was beyond the scanning range of the mass spectrometer. Solvent venting was performed with a Valco four-way valve in order to improve the quality of GC separation by removing the solvent. The gas chromatograph was operated under atmospheric pressure for the first minute then the column outlet was connected to the low-pressure plasma by switching the valve. A helium GC make-up gas at a flow of 5 cm3 min-' was used to compensate for the difference in the helium flow in the plasma when the chromatograph was operated at atmospheric pressure.Solvent venting improved the detection limits from 19 and 11 to 7 and 2 pg for TEtSn and TBuSn respectively (Table 2). The same operating conditions were used for the determination of TEtPb for which figures of merit are also given in Table 2. Analysis of SRMs The chromatogram obtained for the determination of TEtPb in the NIST SRM 2715 Lead in Fuel is shown in Fig. 8. The SRM was diluted two-fold with hexane before analysis. The lead concentration was found to be 838 +2 pg g-' of Pb in the form of TEtPb compared with the certified value which is 784+4 jig g-' of Pb in the form of TEtPb. The value found was outside the confidence limits of the certified value exhibit- ing positive bias.The reason for this is not known although 800 700 600 500 400 v) 300 3 200 b 8 1 70 i7j 60 50 40 30 20 10 0 1 2 3 4 5 6 Time/min Fig. 7 m/z 150 50 ng injection for fragment ions of TEtSn and TBuSn Single-ion monitoring at (a) mjz 135 20 ng injection and (b) Journal of Analytical Atomic Spectrometry December 1996 Vol. 11 1 153Table 2 Figures of merit of alkyltin compounds using GC-LP-ICP-MS 6 TEtSn -(a) TEtSn I TBuSn TEtPb Parameter Linear dynamic range rz* Slope of log-log plot RSD (Yo) Detection limit/pg Without Venting venting 0.9999 0.9997 1.013 0.9986 4.7 1.2 103 103 19 11 Without Venting venting 0.9962 0.9999 0.9321 0.9983 4.0 3.1 7 2 103 103 Without Venting venting 0.9997 0.9976 0.9842 0.9001 5 7.1 6 14 102 lo2 * rz = regression coefficients.0.5 1 1.5 2 2.5 3 Timdmin Fig. 8 Chromatogram for the analysis of NIST SRM 2715 Lead in Fuel by GC-LP-ICP-MS using element-selective detection at m/z 208 the TEtPb standard could have degraded somewhat given the lability of this organometallic species. Fragmentation studies Mass spectra of TEtSn and TBuSn were also obtained at a forward power of 45 W instead of the usual 90 W. The signal intensity at m/z 135 was greatest at a forward power of 75 W for both TEtSn and TBuSn (Fig. 9). However higher mass fragments such as those at m/z 150 and 235 were favoured at lower forward powers (5~45 W) as shown in Fig. 10 (a) and (b). The most predominant fragments were those of lower mass such as elemental tin and monomethyltin.The spectra of TEtSn and TBuSn were similar as is shown in Fig. 11. This was consistent with the observation of Heppner,2 suggesting that the compounds were fully ionized in the plasma and the fragments are recombinants of the ions rather than fragments from the original molecules. Hence fragmentation did not occur in an analogous manner to an EI or CI source under these operating conditions. At higher analyte concentrations more high mass fragments 1200 i O 100 95 75 60 45 PowerMl Fig. 9 Effect of forward power on signal intensity for fragment ions at mlz 135 200 I i - 150 5 F g loo $ 50 0 30 45 50 65 25 I I O ' ,''Zd /- 50 70 ' PowerNv Fig. 10 Effect of forward power on signal intensity for fragment ions at (a) m/z 150 and (b) m/z 235 4 "t I 110 130 150 170 190 210 230 250 0 .. - - 110 130 150 170 . . 1 9 0 . 210 230 Fig. 11 Mass spectra obtained by GC-LP-ICP-MS for 20 ng each of TEtSn and TBuSn showing the most probable assignments of the fragment ions (i.e. m/z 170 and 200) were observed (Fig. 12). Greater amounts of analyte may require more energy for ionization hence fragmentation was incomplete under these conditions and these ions could reflect the structure of the original compound more accurately. EI spectra are shown in Figs. 13 and 14 and compared with the spectra obtained with the LP-ICP it is obvious that the plasma conditions were still not 'soft' enough to produce similar fragmentation patterns to EI spectra. In short the forward power and partial pressure of analyte influences the degree of fragmentation and even lower power and pressure could favour the formation of molecular and fragment ions rather than recombinants.Recently an LP-ICP 1 154 Journal of Analytical Atomic Spectrometry December 1996 Vol. 118 1 .- m $ 70 w c t .- 3 60 .p 50 w 40 30 20 10 0 . . 110 130 150 170 190 210 230 . 250 Mass Fig. 12 Mass spectra obtained by GC-LP-ICP-MS for 500 tig each of TEtSn and TBuSn showing the most probable assignments of the fragment ions sustained at 5 W using only the GC helium flow as the plasma has been de~eloped,~ and the results suggest that lower power and pressure do indeed favour the formation of molecular ions. The instrumentation has now been developed further and results are reported in the next part of this paper. Conclusions Elemental quantitation and some limited fragmentation has been achieved with a single instrumental configuration.Optimization of the ion sampling and skimming process and the power between 1 and 15 W and the utilization of a mass spectrometer with a greater mass range is required. These aspects have been addressed in the next part of this paper which describes the development of a customized LP-ICP-MS instrument. UTILIZATION OF A CUSTOMIZED LP-ICP-MS INSTRUMENT To date only commercially available ICP-MS instruments have been modified and used as low-pressure plasma mass spectrometers. However a number of disadvantages are inherent in this and have been addressed as detailed below. Generation of an LP-ICP Commercial ICP-MS instruments were originally designed to form argon plasmas at gas flow rates of between 15 and 17 1 min-' and forward powers of between 1300 and 1500 W.Most commercial instrument manufacturers use a 1.5 kW rf generator with a matching network designed to couple rf power into an atmospheric argon plasma. With such a gener- ator it is impossible to form a stable low-pressure helium plasma at low power and the matching network is inappropri- ate for low-pressure plasmas in almost all gases. Hence a 1.5 kW 27.12 MHz rf generator was modified to give a stable output between 1 and 300 W and a new rf matching network was purchased (RF Applications Eastbourne East Sussex UK) to enable coupling of rf power from the modified generator into the low-pressure plasma source. The LP-ICP torch consisted of a 140 mm long quartz tube of 3 in o.d.with a t in 0.d. side-arm through which the plasma gas could be introduced. The low-pressure sampling cone was machined from aluminium (Machine Shop University of Plymouth) and had a 2mm orifice and an Ultra-Torr fitting for a %in pipe so it was possible to form a vacuum seal between the low-pressure torch and sampler. The low-pressure torch was interfaced at the rear end with a gas chromatograph (PU 4550 Pye Unicam Cambridge UK) fitted with an on-column injector by way of a heated transfer line held at a temperature of 250 "C. The GC capillary extended through the transfer line and into the torch the vacuum seal being made using a combination of Ultra-Torr and Swagelock fittings with graphite ferrules. This configuration has been described pre- viously in more detail.g Customization of the Mass Spectrometer Another disadvantage associated with using a commercial ICP-MS instrument is that it is designed for elemental analysis so the quadrupole mass range only scans up to 255 m/z.This would be a great disadvantage for the determination of organometallic or other high relative molecular mass com- pounds because the molecular ion peak would appear above m/z 255. In order to overcome these problems a dedicated instrument was constructed by conversion of a Hewlett- Packard Mass Selective Detector (HP-MSD) for use with an LP-ICP ion source. The HP-MSD is ideal for such an application because it is a compact bench-top instrument of proven capability with a mass range of between 10 and 800 m/z. However it was necessary to make several major modifi- cations to the instrument in order to couple the LP-ICP and extract and focus ions from this source.These modifications are described below. Ion sampling interface The theory underlying the design of ion-sampling interfaces for plasma MS has been developed and discussed in length by other The Hewlett-Packard 5970a MSD is normally operated as a GC-EI-MS so it required modification to enable it to accept ions produced by an external ion source. Hence a new ion- sampling interface was constructed using established theory as a guide. The position of the first Mach disc down stream from the sampler orifice is expressed as'' so it will not be repeated here. where X is the position of the Mach disc downstream of the sampler orifice P1 is the pressure in the expansion stage Po is the pressure in the source and Do is the diameter of the sampler orifice.In order for a representative sample to be obtained from the plasma the skimmer orifice must be placed upstream of the first Mach disc. Initial studiesg suggested that the Mach disc would occur at about 3.6 mm behind the sampling orifice when argon was used as the plasma gas at a flow of 1000 cm3 min-'. Hoglund and Ro~engren'~ stated that the flow conditions appropriate to eqn. (1) prevail when the pressure ratio across the sampler orifice is greater than two or analytically for an ideal gas when where y is the ratio of heat capacities at constant pressure and Journal of Analytical Atomic Spectrometry December 1996 Vol. 1 1 1 155n 6oOo.5Ooo 4ooo. 3ooo. l-n-n-r 2030 loo0 n i 40 50 60 70 80 90 121 14 1 179 T & 230 249 m/r Fig. 13 EI spectra of TEtSn obtained (a) experimentally by GC-MS; and (b) from a computer library volume. However studies by Gray'' and Luan et indicated that subsidiary barrel shocks and Mach discs form downstream of the primary Mach disc when Po/Pl < 100 for an argon ICP. Because the pressure in the torch and the expansion region can change fairly considerably when different plasma gases are used (e.g. in the present study a plasma formed with helium at a flow rate of 3 cm3 min-' was investigated) the skimmer which was a conventional nickel skimmer for a VG PlasmaQuad I1 with a 0.7mm orifice was mounted on a copper plate that could be moved back and forth by adding or removing copper spacing washers.Hence the sampler- skimmer spacing could be optimized empirically. The vacuum in the expansion stage was maintained using two rotary pumps (Leybold D16B). The downstream end of the expansion stage was machined with a 40mm KF flange which was connected to a gate valve using a viton O-ring and a customized brace. The gate valve was then connected to the quadrupole vacuum housing which had a 40mm KF flange brazed onto it using conventional vacuum fittings. Ion optical array Once the gas and ions have entered the skimmer it is necessary to collect and focus the ions into the quadrupole mass analyser. The ion-focusing lenses are crucial for the overall sensitivity of the instrument because incorrect focusing can lead to the ion beam being scattered or accelerated to such an extent that the residence time in the quadrupole mass analyser is insufficient for effective mass analysis.The design of the ion optical lens system was aided by the use of a computer simulation program (SimIon version 4.0 Argonne National Laboratory IL USA). However this model does not account for field fringing effects and the effects of space charge in the ion beam. In order to compensate for such effects on the ion beam the entrance angle of the ions into the electrostatic array was varied and the model tested for a series of ions of different mass with differing ion kinetic energies. The ion trajectories for ions entering the optimized ion optics at+20" off-axis are shown in Fig. 15. Hence it was possible to design an ion optical array which consisted of two extraction lenses behind the skimmer and three ion focusing lenses just prior to the quadrupole. This lens system had no photon stop because the detector in the HP5970a MSD is offset and so was shaded from any photon noise.Also the low-pressure plasma studied in the present work gave rise to very little photon noise. To accommodate these ion-focusing lenses some electrical and mechanical modifications were made to the MSD. First 1 156 Journal of Analytical Atomic Spectrometry December 1996 Vol. 1 1m/z Fig. 14 EI spectra of TBuSn obtained (ci) experimentally by GC-MS; (b) from a computer library Fig. 15 SimIon plot of ion trajectories for ions with an average ion kinetic energy of 10 eV through the ion optical array designed for use in the customized LP-ICP-MS instrument with optimum voltage settings L1 -75; L2 - 10; L3 + 1; L4 -64; L5 0; entrance -45 V ramped with quadrupole voltage the existing interface was removed and the entrance to the quadrupole vacuum chamber was widened and welded u ith a 40mm KF type vacuum flange.Inside the vacuum chamber the existing ion-lenses and EI source were removed and replaced with the new set of ion lenses which were connected to the existing stabilized voltage supplies of the MSD under computer control. Details of the lenses and voltage ranges are given in Table 3. Table 3 ion optical array Details of ion-lenses and voltage ranges for the LP-ICP-MS Lens Voltage range/V Computer control L1 0 to -200 No L2 + 5 to -20 No L3 Oto +10 Yes L4 0 to -255 Yes L5 0 Yes Entrance 0 to -255 Yes Once the individual components of the new instrument had been designed they were coupled together and evacuated for an initial trial.The configuration of the final instrument is shown in Fig. 16. Optimization and Evaluation Optimization study In order to generate a useful analytical signal a solution of perfluorotributylamine (PFTBA Fluka Chemicals Gillingham Dorset UK) was contained in a vial attached to the side-arm of the low pressure plasma torch and introduced Journal of Analytical Atomic Spectrometry December 1996 Vol. 1 1 1 157Quadrupole lon Vacuum electrical feedthrough (I 0-800 m/z) opt*ics valve I 69 m/z Low pressure sampler I 219 m/z I- Fig. 16 Diagram of the custoinized L1 generator P F ~ B A solution :omputer ' ' 'ace nrlrctQtinn ?-ICP-MS instrument to the low pressure plasma via a molecular leak (Fig.16). PFTBA was chosen because the software on the MSD com- puter is typically configured to use the three fragment ion peaks from PFTBA at 69 219 and 502 m/z to tune the quadrupole and ion optics. This enabled the mass calibration and resolution settings of the quadrupole and the ion-lens voltage settings to be optimized daily by the computer. The initial operating conditions for the low-pressure plasma are given in Table 4 and Fig. 17(a) shows the three mass peaks selected for tuning the quadrupole. It can be seen from the peak shape and width that good resolution and calibration were achieved using the modified instrument. A total mass spectrum for the PFTBA is shown in Fig.17(b) and data pertaining to the major peaks are given in Table 5. Prior to conversion into an LP-ICP source the EI mass spectrum for PFTBA was acquired using the instrument and the abun- dances of the major peaks are also given in Table 5. The major difference between the data for the two sources was the higher abundance of ions at m/z 219 and 502 obtained using the EI source although the abundances can be varied fairly consider- ably using either ionization source. Once the optimization procedure for the quadrupole and ion-lenses for the MSD had been completed lenses L1 and L2 were adjusted manually to yield maximum signal for the three chosen m/z values. When the instrument had been set up in this way an optimization of the skimmer-sampler spacing and the plasma forward power was performed.Surface plots of signal intensity Table4 Operating conditions used for the initial studies with the LP-ICP-MS system Mass spectrometer Forward power/W Reflected power/W Plasma- Pressure/Torr*- Torch Interface Anal yser Ion lensjV L1 L2 L3 L4 L5 Entrance Modified Hewlett-Packard MSD 6 0 0.2 0.03 2 x 10-5 - 75 - 10 +1 - 64 0 - 45 * 1 Torr= 133.322 Pa. 502 m/z 500 505 100 60 40 2011 I I I Fig. 17 Mass spectra of (PFTBA) (a) selected mass fragments at 69 219 and 502 mjz used for quadrupole and ion-lens tuning; and (b) mass spectrum in the range 45-800 m/z versus skimmer-sampler spacing and forward power for the three fragment ions of PFTBA at 69 219 and 502 m/z are shown in Fig. 18(u) (b) and (c) respectively.The contour plots have been represented in only two dimensions in order to reveal the pertinent features. The points labelled 'A' and 'B' correspond to maxima in the signal hence the contour plots would look like twin peaks in three dimensions. The experi- ment was repeated and the observations were found to be reproducible. Maxima in signal intensity occurred at two skimming dis- tances namely 6 and 8 mm downstream of the sampler orifice with comparable signal intensities observed at each of the maxima. Gray'' has shown that as the expansion stage pressure increases the barrel shock downstream of the sampler shortens and the Mach disc gets closer to the sampling orifice. The same worker also found that for P,/P1=76 several shock regions and Mach discs could be observed downstream of the primary shock region.These observations have recently been confirmed by Luan et a1.I6 In the present study Po=0.2 Torr and P I = 0.03 Torr ( 1 Torr = 133.322 Pa) and Do = 2. 0 mm 1 158 Journal of Analytical Atomic Spectrometry December 1996 Vol. 1 1Table 5 Major peaks in the mass spectrum of PFTBA between 45 and 800 m/z obtained using LP-ICP-MS Abundance Relative abundance mlz LP-ICP-MS EI LP-ICP-MS EI 69 219 502 1 547 264 383 680 141 184 1 825 280 1100288 223 808 100 24.80 9.12 100 60.28 12.26 69 m/z (b) 3 4 5 6 7 8 9 10 11 12 13 14 15 ic1 21 9 d z 10 9 8 . 7 . 6 5 E E 4‘- 8 3 5 3 4 5 6 7 8 9 10 1 1 1 2 1 3 1 4 1 5 a 502 m/z 0 E .- E E P cn 3 4 5 6 7 8 9 10 11 12 13 14 15 Forward power/” Fig. 18 Surface contour plots showing the effect of plasma power and skimming distance on the signal intensity for (a) 69; (b) 219; and (c) 502 m/z.The points labelled ‘A and ‘B’ indicate intensity maxima hence using eqn. (l) the position of the Mach disc is calculated to be X = 3.5 mm. The presence of two maxima in the signal intensity downstream of the Mach disc observed in the present work suggests that the fragment ions were formed not in the plasma itself but in the expansion stage possibly in regions associated with one or two Mach discs downstream of the sampler orifice. At 3.5mm the theoretical position of the Mach disc a maximum in signal intensity was not observed; however the observations of Gray15 suggested that the Mach disc becomes progressively larger in relation to the barrel shock as the P,/P1 ratio is reduced so it could possibly extend out to 6 mm downstream of the sampler orifice.Visual obser- vation of the expansion stage used in this work is underway to investigate this phenomenon further. The power which yielded maximum signal was between 6 and 8 W although this depended somewhat on which fragment ion and skimming distance were chosen (Fig. 19). At a skim- ming distance of 6mm the signal intensity for the fragment ions at 219 and 502 m/z dropped rapidly above 6 W [Fig. 19(a)]. The signal intensity for the fragment ion .it 69 m/z also decreased as the power increased but not as sharply suggesting that the high mass fragments were further frag- mented adding to the signal observed at 69 m/z. At a skimming distance of 8 mm the power which yielded maximum signal intensity was very similar for the three fragment ions [Fig.19(b)]. Ion kinetic energies In order for the helium LP-ICP to be used as a universal source it is necessary to obtain information on the range of t --. 7 0 2 4 6 8 10 12 14 PowerMl Fig. 19 Plots of normalized signal intensity against plasma power for fragment ions at 69 219 and 502 m/z for two skimming distances (a) 6 mm; and (b) 8 mm Journal of Analytical Atomic Spectrometry December 1996 Vol. 11 1 159Repeller voltageN 1 .oo 0.90 3 0.80 .- & 0.70 0.60 -5 - 0.50 0.40 2 0.30 0.20 0.10 VJ a 1 9 d Z > 502dz 0 1 2 3 4 5 6 7 8 9 1 0 Energy/eV 0 1 2 3 4 5 6 7 8 9 1 0 Repeller voltageN ion kinetic energies produced by the plasma and extracted into the mass spectrometer. If ions have a wide range of ion kinetic energies it is difficult to arrive at compromise conditions for the quadrupole mass range calibration and ion-lens con- ditions.Also if the spread in ion kinetic energy at an individual value of m/z is large it is difficult to maintain good peak resolution. Ion stopping curves for the three fragment ions of PFTBA at 69 219 and 502 m/z were generated by removing the L1 and L2 ion-lenses and varying the potential on L3 from 0 to +10 V. From these the mean ion kinetic energies and the ion energy spread could be calculated. The ion stopping curves for ions extracted from the 6 W low-pressure helium ICP are shown in Fig. 20(a) and derivative plots of these curves are shown in Fig. 20(b). This shows that the mean ion kinetic energies for the three mass fragments were 1.2 1.7 and 2.1 eV for 69 219 and 502 m/z respectively at 6 W forward power but increased when the power was increased.The spread in kinetic energies was less then 1.5 eV which should result in well-resolved peaks for all three mass fragments. For ions extracted using a molecular beam type interface one would expect ion kinetic energy to increase with mass owing to the translational energy imparted by the gas sampling process; however it can be seen from Fig. 20(b) that the mass fragment at 69 m/z had the highest ion kinetic energy. This suggests that this ion was formed in a different region of the plasma an explanation which seems unlikely given the results shown in Fig. 18. A more plausible explanation is that the ions underwent different collisional ionization/fragmen- tation processes which influenced their ion energies.Ion stopping curves for a plasma power of 8 W are shown in Fig. 20(c). The derivative plots are not shown because the variability is too great to yield a useful plot. However it is evident that the spread in the ion kinetic energies increased greatly when the power was increased to 8 W. The only immediate explanation for this is that a secondary discharge existed in the expansion stage which became more intense as power was increased thereby accelerating the ions towards the skimmer. In summary the evidence seems to indicate the following. (a) Fragment ions are formed in the expansion region rather than the plasma torch. (b) Fragment ions are formed in or downstream of one or two separate Mach discs in the expan- sion region.(c) Ion energies may be influenced in the first instance by the nature of the ionization process. ( d ) A second- ary discharge may exist in the expansion region which imparts a degree of acceleration to the ions. Analytical utility Previous studies have demonstrated that low-pressure plasmas have their place as atomic ionization sources. In the present work the analytical utility of the instrument pertaining to the formation of molecular fragment mass spectra was investigated as this could be considered the more challenging application. In order to introduce analytically useful masses of analyte a gas chromatograph was interfaced with the rear of the LP-ICP torch via a heated transfer line as shown in Fig.16 with operating conditions given in Table 4. Using this set-up 1.0 pl of a 50 pg ml-1 solution of halobenz- enes in pentane was injected on-column and the gas chromato- graph ramped from 40 to 200 "C at 20 "C min-'. A total ion chromatogram (TIC) for this injection is shown in Fig. 21 and the mass spectra of the respective analyte compounds are shown in Fig. 22(a)-(c). The mass spectra obtained are similar to mass spectra obtained for the same analytes using an EI " 1 2 3 4 Ti me/min Fig.21 TIC for 50ng on-column injection of A chlorobenzene; B iodobenzene; and C dibromobenzene Fig. 20 (b) derivative plot of ion stopping curve at 6 W forward power; and (c) ion stopping curve at 8 W forward power Ion kinetic energies for ions of 69 219 and 502 m/z obtained with a helium LP-ICP-MS.(a) Ion stopping curve at 6 W forward power; 1 160 Journal of Analytical Atomic Spectrometry December 1996 Vol. 11loo00 50 60 70 80 90 100 110 120 130 60 80 100 120 140 160 180 200 (c) 236 1 ,157 98 119 f / " 60 80 100 120 140 160 180 200 220 m/z Fig. 22 Mass spectra for (a) chlorobenzene; (b) iodobenzene; and (c) dibromobenzene 50000 30000 20000 10000 1 2 3 1 2 3 Time/min Fig. 23 Selected ion chromatograms for the molecular ions of (a) chlorobenzene m/z 112; (b) iodobenzene m/z 204; and (c) dibromobenzene mjz 236 ionization source on the same instrument before conversion with the same parent ion peaks. However as shown in the power and skimming distance studies (Figs. 18) the abundance of each fragment can be altered by the plasma power.The results of selective ion monitoring (SIM) for the molecular ion peak of each analyte can be seen in Fig. 23. The S/N of such an analysis suggests that the technique could easily be used for trace analysis though the use of this instrument for such an application may be limited because the detector is con- figured in analogue mode whereas a pulse counting detector would be preferable. Conclusions The customized instrument has gone some way towards allevi- ating the problems associated with the use of commercial ICP-MS systems for low-pressure plasma work. Now that a dedicated instrument has been constructed it shows great promise for further development. Such an instrument is more economical to operate compared with conventional IC'P-MS and helium MIP systems and can potentially be operated in both atomic and molecular modes.This will also reduce capital costs because a single instrument with a single source could provide a wide range of mass spectral information. FUTURE WORK The possibility of developing the low-pressure low-power ICP into a multi-purpose source for fragmentation and qiiantifi- cation is highly promising. One of the disadvantages of low- pressure techniques is that the relatively un-robust plasmas cannot sustain liquid sample introduction thus limiting their applications. Further development of low-flow liquid sample introduction techniques (such as microbore LC and capillary electrophoresis) along with good desolvation techniques (i.e. the particle beam interface) will expand the analytical appli- cations of low-pressure low-power plasmas. The fundamental properties of the LP-ICPs used in this application need to be addressed. Knowledge of temperature and electron number density of the plasma should improve understanding of plasma behaviour at different powers and pressures so that controlled fragmentation can be achieved. REFERENCES 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 Evans E. H. Giglio J. J. Castillano T. M. and Caruso J. A. Inductively Coupled and Microwave Induced Plasma Sources for Mass Spectrometry The Royal Society of Chemistry Cambridge 1995. Heppner R. A. Anal. Chem. 1983 55 2170. Poussel E. Mermet J. M. Deruaz D. and Beaugrand C. Anal. Chem. 1988 60 923. Olson L. K. Story W. C. Creed J. T. Shen W. and Caruso J. A. J. Anal. At. Spectrom. 1990 5 471. Shen W. and Satzger R. D. Anal. Chem. 1991 63 1960. Chien B. M. Michael S. M. and Lubman D. M. Anal. Chem. 1993 65 1916. Evans E. H. and Caruso J. A. J. Anal. At. Spectrom. 1993,8,427. Castillano T. M. Giglio J. J. Evans E. H. and Caruso J. A. J. Anal. At. Spectrom. 1994 9 1335. Evans E. H. Pretorius W. Ebdon L. and Rowland S. Anal. Chem. 1994,66 3400. Kantrowitz A. and Grey J. Rev. Sci. Instrum. 1951 22 328. Campargue R. J. Phys. Chem. 1984 88 4466. Douglas D. J. and French J. B. J. Anal. At. Spectrom. 1988 3 743. Hoglund A. and Rosengren L. G. Int. J. Mass. Spectrom. Ion. Processes 1984 60 173. Olivares J. A and Houk R. S. Anal. Chem. 1985 57 2674. Gray A. L. J. Anal. At. Spectrom. 1989 4 371. Luan S. Pang H. -M. and Houk R. S. J. Anal. At. Spectrom. 1996 11 247. Paper 6/03256G Received May 9 1996 Accepted August 21 1996 Jorirnal of Analytical Atomic Spectrometry December 1996 Val. 11 1 161

ISSN:0267-9477    

DOI:10.1039/JA9961101151

出版商:RSC    

年代:1996

数据来源: RSC

摘要:

Speciation of Inorganic Selenium and Selenoaminoacids by On-line Reversed-phase High-performance Liquid C h romatograp hy-Focused Microwave Digest ion-Hydride Generation-atomic Detection Journal of Analytical Atomic Spectrometry J. M. GONZALEZ LAFUENTE M. L. FERNANDEZ SANCHEZ AND A. SANZ-MEDEL* Department of Physical and Analytical Chemistry University of 0 viedo CIJulian Claveria 8 33006-Oviedo Spain A high-performance liquid chromatographic-microwave digestion-hydride generation system coupled on-line with three atomic detectors (atomic absorption inductively coupled plasma atomic emission and inductively coupled plasma mass spectrometry) has been developed and investigated for selenium species separation and determination. Total inorganic selenium selenomethionine and selenoethionine are separated by reversed-phase chromatography prior to on-line microwave digestion of selenocompounds with a KBr0,-HBr mixture to form continuously Se" which is finally transformed into H2Se also in a continuous manner with a merging flow of sodium borohydride.Detection limits obtained for each Se species in water and urine using the three atomic detectors have been worked out and compared (i.e. for Sew DLs were 6.8 pg 1-' by AAS 30 pg 1-' by ICP-AES and 0.16 pg 1-' by ICP-MS). The integrated flow system HPLC-MW digestion-HG- atomic detection proposed here allows in a single injection reliable speciation in urine of the selenoaminoacids tested versus total inorganic selenium. Further speciation of the overlapped inorganic Sew and Sem peaks is accomplished by a second injection of the urine sample to determine only Se" by avoiding microwave heating.Results on Se speciation in human urine have shown that more speciation information of the actual species perhaps unknown present in real samples could be gathered by resorting to the use of two or more atomic detectors coupled to the same separation scheme. Keywords Selenium speciation; high-performance liquid chromatography; on-line focused microwave digestion; atomic detectors; urine samples INTRODUCTION Selenium appears in the natural selenium cycle in the form of several organic and inorganic compounds or species.' The two main inorganic species selenite and selenate are very import- ant in the biogeological and biochemical cycle of selenium exhibiting different chemical and biological properties.Organic compounds of selenium particularly selenoaminoacids also play an important role in the biological cycle of selenium and are incorporated into proteins. Thus knowledge of the difl'erent chemical forms of Se is required. The coupling of a powerful chromatographic separation with specific atomic detectors appears to be an appropriate approach for the speciation of both inorganic and organic A widely used method for the separation of selenite and * To whom correspondence should be addressed. selenate is liquid chromatography specifically either ion chro- matograph~~ or ion-pair reversed-phase chr~matography.~ The organic compound trimethylselenonium is normally separated by cation ion chromatographyY6 while reversed-phase chroma- tography has also been proposed for the characterization of selenoamin~acids.~ Unfortunately the ideal HPLC method able to separate organic and inorganic Se species in one run has not yet been developed although we have recently reported a technique to separate Se" Sevl and some common seleno- aminoacids all together by using a vesicular mobile phase of didodecyldimethylammonium bromide (DDAB) on a C18 reversed-phase column previously modified with DDAB.8 Regarding specific atomic detection for Se on-line monitor- ing flame AAS ICP-AES and ICP-MS have been pro- po~ed.~.'-'' The sensitivity of such detection approaches can be increased by post-column hydride generation with NaBH4 .'l-13 Nevertheless only Se" can directly form the volatile species; SeV1 and some organoselenium compounds (e.g.seleno- aminoacids trimethylselenonium) will not form the volatile hydride.16-18 Cobo-Fernandez et a1.I8 have proposed a method for separating trimethylselenonium Se'' and SeV1 by anion exchange chromatography with on-line determination of the element by HG-AAS after microwave oxidation of trimethyl- selenonium in the presence of persulfate and later reduction of SeV1 to Se" with HCl in a domestic microwave oven. Pitts and c o - w o r k e r ~ ~ ' ~ ~ ~ also used an on-line microwave system for reduction of SeV' to Se" by adequate acidification after separation of SeIV and Sevl by HPLC. We have recently assembled a continuous HG-AAS detection system for total Se which includes a continuous and most efficient on-line pre-treatment with HBr-KBrO assisted by on-line focused microwave digestion." Such a continuous detector allows total selenium determinations virtually inde- pendent of the nature of the investigated Se species (namely SeIV SeV1 selenomethionine selenoethionine and selenocys- tine).In a further development the analytical performance for speciation of inorganic and organic Se species of this FIA-Se 'derivatization'-HG-AAS system was evaluated using a vesicle- mediated HPLC separation.22 Most recently Quijano et aL2 reported the use of a mixed column for the speciation of selenocystine selenomethionine selenite and selenate with ICP-MS detection where the behaviour in the separation process was similar to that observed for the above-mentioned vesicle-mediated HPLC separation.As an alternative to overcome limitations of vesicle- mediated HPLC separations for high ionic strength samples,22 the potential of reversed-phase chromatography with different detectors is investigated here for inorganic and selenoamino- acids speciation in urine. SeIV SeV' selenomethionine and Journal of Analytical Atomic Spectrometry December 1996 VoZ. 11 (I 163-1 169) 1 163100 yl 1P ! sampleloop 1 8 QC-FAAS ICP-AES \ Waste 30nm Focused microwave - Reactioi n digester coil Fig. 1 Schematic diagram of the coupling of HPLC-focused MW assisted digestion-HG-(QC-FAAS/ICP-AES/ICP-MS) selenoethionine have been selected as 'model' analytes for chromatography and on-line detection with quartz tube-AAS ICP-AES and ICP-MS for selenium speciation is evaluated.As the complexity of speciation in biological samples calls for complementary technique^,^ different-principle-based detec- tion techniques could help to interpret adequately complex speciation results where separation is not complete or when detection by a given single technique is affected by interferences or perhaps is lacking the necessary sensitivity. EXPERIMENTAL Instrumentation An LKB (LKB Browman Sweden) Model 2150 HPLC pump with a Rheodyne sample injection high-pressure valve (Berkeley CA USA) equipped with a 100 pl loop was used for eluent delivery and sample introduction. The analytical column (250 x 4.6 mm id) was packed with 5 pm Spherisorb C1,-bonded silica stationary phase (Spherisorb Phase Separations Queensferry Clwyd UK). A pre-column (30x4.6mm id) was packed with 10 pm Spherisorb Cls- bonded silica stationary phase (Spherisorb Phase Separations).For AAS measurements a Unicam Model PU94OOX Atomic Absorption Spectrometer (Cambridge UK) equipped with a T-shaped quartz absorption cell (12 cm length 8 mm id) was used with a Hewlett-Packard Model HP 3394 integrator (Avondale USA) and a Unicam hollow cathode lamp. An ICP-MS Model HP 4500 (Yokogawa Analytical Systems Tokyo Japan) for ICP-MS and a Perkin-Elmer ICP-5000 spectrometer (Uberlingen Germany) for ICP-AES measurements were used. On-line microwave (MW) digestions were carried out with a Prolabo Microdigest M301 (Paris France). The reaction coil was made of PTFE (4m lengthx0.8mm id) and wrapped around a 2.5 cm diameter glass cylinder which was mounted vertically into the microwave cavity vessel.21 In order to reduce the internal pressure created by MW-heating the flow was cooled after microwave action by passing it along a cooling coil (30 cm length x 0.8 mm id) immersed in a simple room temperature circulating water-bath as shown in the set-up diagram of Fig.1. Two multi-channel peristaltic pumps (Gilson Minipuls Middleton WI USA) were used one for the addition of sample digestion reagents and the other for reduction (with NaBH,) and adequate cooling. A laboratory-made glass gas- liquid separator designed to avoid bubbles reaching the atom- izer when urine is injected was used for continuous separation of volatile selenium.25 A schematic diagram of this latter device with its dimensions is given in Fig.2. Reagents Seleno-L-methionine and seleno-DL-ethionine (from Sigma Chemical St. Louis MO USA) stock solutions containing 10mg 1-l of Se were obtained by dissolving the appropriate amount of the corresponding compounds in ultra-pure Milli-Q water (Millipore Bedford MA USA). Inorganic Se" stock solution (1000 mg 1-I) was obtained from Merck (Darmstadt Germany). Se"' stock solution (1000 mg 1 - I ) was prepared by dissolving Na,SeO (from Merck) in ultrapure Milli-Q water. Dimethyldiselenide (Aldrich Steinheim Germany) stock solution was at a concentration of 10mg 1-1 in methanol. Trimethylselenonium solution of 50 mg I-' (provided by M. A. Palacios Madrid Spain). Monomethylselenol was prepared freshly by reaction of dimethyldiselenide with NaBH in 1 164 Journal of Analytical Atomic Spectrometry December 1996 Vol.11ethanolic solution and subsequent acidification with HC1.26 All working standard solutions were freshly prepared daily by diluting the stock solutions with ultrapure Milli-Q water. Sodium tetrahydroborate (0.5% m/v) was prepared by dis- solving 2.0 g of NaBH (Probus Barcelona Spain) in 400 ml of 0.1% m/v NaOH (Merck) and filtering this solution. mol 1-') was prepared by dissolving the appropriate amount of the salt in ultrapure Milli-Q water. Hydrobromic acid 47% and hydrochloric acid 37% were from Merck. Ammonium acetate (0.1 mol 1-' pH 4.5) was prepared by dissolving the appropriate amount of the salt (Merck) in Milli-Q water acidified with glacial acetic acid (Merck) and diluting to the mark. Methanol suprapure from Teknokroma (Barcelona Spain) was used.All other chemicals were of analytical reagent grade. Potassium bromate (Merck) solution (1.5 x Procedures ( a ) Inorganic selenium selenomethionine and selenoethionine speciation Samples 100 pl in volume or the corresponding standard solutions (containing 100 pg 1- ' of SeV1 Se" selenomethionine and selenoethionine) are injected. The mobile phase consists of a 0.1 mol 1-' ammonium acetate buffer solution (pM 4.5). The eluate is first continuously mixed with 48% HBr and KBrO solutions (see Fig. 1) going through the focused MW digester (15% power). During the digestion time (around 1 min) the liquid sample rotates inside the PTFE reaction coil immersed in the microwave field. The emerging main flow is then cooled by passing through the additional PTFE coil immersed in a room temperature water-bath while being pumped by the second peristaltic pump.21 The emerging cool solution merges with 0.5% (m/v) NaBH to form volatile H,Se (see Fig.1). A continuous stream of argon drags this hydride to the atomizer through the gas-liquid separator. A Hewlett-Packard HP 3394 integrator was used for quartz cell (QC)-FAAS or ICP-AES while for ICP-MS measurements the signals were monitored with the software provided with the instrument. Optimal conditions for the chromatographic separation are given in Table 1. Optimal microwave and instrumental detec- tion conditions are summarized in Tables 2 and 3 respectively. Human urine samples were just filtered through a Millipore 0.45 pm membrane and injected (100 11 sample injecticm) for analysis.Deuterium background correction was used for AAS measurements when analysing urine samples. In all case.; peak height measurements were used as this measurement mode provided better results with our ICP-MS software. Table 1 Optimum conditions for chromatographic separation Pre-column Column Buffer carrier Buffer carrier flow rate Sample loop 100 pl C18 (30 mm length x 4.6 mm id) Spherisorb C bonded silica stationary phase (250 mm x 4.6 mm id :i pm) Ammonium acetate 0.1 mol 1- pH4.5 1 ml min-' Table 2 Optimum conditions for on-line digestion with focused MW of inorganic and organic selenocompounds Focused microwave power 15% HBr flow rate KBrO (1.5 x lo-' mol 1-I) flow rate PTFE reaction coil (length x diameter) 1.2 ml min-' 0.6 ml min- ' 4 m x 0.8 mm Table 3 Optimum conditions using ICP-MS ICP-AES and QC-FAAS detectors Hydride Generation NaBH concentration 0.5% (m/v) in NaOH 2 ml min-' 0.1% (m/v) NaBH flow rate ICP-MS Isotopes monitored "Se lase rf Power 1200 w Sample depth 5 mm Carrier gas flow rate Integration time 0.5 s rf Power 1000 w Carrier gas flow rate Wavelength 196.01 nm Integration time 0.5 s Q C- FA AS Wavelength 196.01 nm HCL current 11 mA D background correction On Carrier gas flow rate 1.29 1 min-' ICP-AES 0.58 1 min-' 0.15 1 min-' (b) Se" determination For Se" determination the MW digester was switched off and the flow was allowed to cool to room temperature.After that another 100 pl aliquot of the sample solution was injected and analysed as detailed above using QC-FAAS detection. In this case only SeIV is determined.The SeV' can be evaluated by difference between the total inorganic selenium peak as deter- mined in (a) and the Se" obtained here.21 Of course peak heights of AAS or ICP signals are sharper with MW on therefore independent calibration of the flow system for M W on and off were carried out. RESULTS AND DISCUSSION Chromatographic Separation The effect of pH of the mobile phase on the chromatographic separation was first investigated measuring the retention time of the solutes under study in the pH range 4.2-6 using a mobile phase containing 0.1 mol 1-l ammonium acetate. The results observed have been plotted in Fig. 3 which shows that selenomethionine and selenoethionine are well resolved and separated from the inorganic selenium in that pH range.Inorganic selenium (both Se" and Se") eluted in the dead volume because of its ionic nature. Selenomethionine and selenoethionine retention times increased steadily with pH of the mobile phase. It was also observed that the AAS peak signal height of selenoaminoacids decreased when the pH increased from 4.2 to 6; a compromise pH value was finally selected for subsequent work (4.5). to 0.2 mol 1-l in ammonium acetate) in the mobile phase was investigated as an indicator of the influence of ionic strength. The effect of increasing buffer concentration (from 5 x 14 I 04 4 4.5 5 5.5 6 6.5 PH Fig.3 Effect of mobile phase pH on retention times 1 inorganic selenium (100 pg 1-' of each SeIV and SeV'); 2 selenomethionine (100 pg 1-l); 3 selenoethionine (200 pg 1-') Journal of Analytical Atomic Spectrometry December 1996 Vol.11 1 165Results obtained indicated that retention times of seleno- methionine and selenoethionine tended to increase as the concentration of ammonium acetate increased. Furthermore selenium AAS peak height signals increased with increasing buffer concentration in all cases. Finally a 0.1 moll-' concen- tration of ammonium acetate was selected for further studies. The effect of the methanol modifier in the mobile phase was also studied. Results observed showed that retention times of selenomethionine and selenoethionine decreased as expected in a reversed-phase interaction. Also percentages of methanol higher than 1% brought about a decrease in AAS signals.This negative effect should be even more pronounced for plasma detection when ICP-MS was used as detector it was observed that methanol produced high carbonaceous deposits in the ICP-MS sampler. Thus the use of methanol was abandoned eventually. Table 1 summarizes the main optimized parameters of this separation as carried out by AAS detection. On-line Microwave Digestion of the Model Selenocompounds in Human Urine In a previous paper2' it was demonstrated that the HBr- KBrO mixture assisted by focused microwave heating can convert continuously into SeIV most efficiently and rapidly different selenocompounds in aqueous samples. Thus we investigated here the effect of HBr and KBrO flow rates on the on-line microwave digestion of model selenocompounds in real samples of human urine.The influence of HBr flow rate was studied (using a 4 m reaction coil length and 15% microwave power) at a constant carrier flow rate of 1 ml min-' with a 0.6 ml min-' flow rate of 1.5 x mol 1-' KBrO by injecting independently 100 p1 of spiked urine [selenate (100 pg 1-I) selenomethionine (100 pg 1-') or selenoethionine (200 pg 1-I)]. Signals were compared with those obtained for the corresponding spikes in aqueous standards containing the same concentrations of each selenocompound.21 Results on the influence of HBr flow showed that a 1.2ml min-' flow rate of concentrated HBr provided quantitative recoveries for all selenocompounds investigated. The KBr03 concentration was varied from 8 x lo- to 3 x mol 1-' using a constant flow of 0.6 ml min-' in order to keep at a minimum the dilution of the urine sample. As noted previously,2' we verified that KBr03 is needed in order to achieve complete conversion of organocompounds into Se".Constant AAS signals were obtained for KBr0 concentrations above 1.5 x mol 1-' and therefore this oxidant concentration was finally selected. The optimized conditions obtained for on-line MW digestion of the different selenocompounds in urine are summarized in Table 2. Specific Detection Optimum conditions for final Se detection by ICP-AES quartz cell-flame-AAS and ICP-MS in such an on-line H,Se gener- ation system were worked out. The results obtained in each case are summarized in Table 3 for the three atomic detectors used. For ICP-MS detection 80Se (49.7%) and 82Se (9.2%) are not suitable for detection because the first is strongly interfered by 40Ar40Ar+ and the second by 'HSIBr (detector saturation) used for selenium continuous digestion.The 74Se (0.9%) is not abundant enough for sensitive detection. In this work 77Se (7.6%) and 78Se (23.6%) have been selected as the two isotopes to be monitored because of their relative abundance and freedom from interferences in this set-up (the 40Ar37Clf inter- ference is avoided by using hydride generation) and a low baseline of around 1000 counts s-' was obtained for 77Se while 78Se showed a baseline of 12000 counts s-' [Fig. 4(c)] probably because of the 40Ar38Ar+ dimer. Analytical Performance Characteristics Typical chromatograms of an aqueous mixture of two inor- ganic and two organic Se species as obtained under optimized conditions (Tables 1 and 2) with different detectors (Table 3) can be seen in Fig.4. The selenoaminoacids are perfectly separated between them and from inorganic selenium too in about 10 min. Both inorganic Se species are eluted at the same time. Moreover Figs. 4(a)-(c) show graphically the relative sensitivity obtained with the three atomic detectors assayed for specific detection of selenium. Five independent injections of 100 p1 of a standard mixture containing known amounts of inorganic selenium (100 pg 1-' of each SeIV and SeV1 seleno- methionine and selenoethionine) were used for precision evalu- ation with AAS detection. RSDs calculated for the determi- nation of each selenocompound in the chromatogram were always better than 5%.Typical linear calibration graphs were also obtained. Both spiked aqueous and human urine samples were injected for comparison and Table4 summarizes the observed data. As can be seen an individual calibration of each chromatographic peak with its corresponding species is required. However the urine matrix only disturbs seriously the selenomethionine AAS signal. Comparative detection limits (DLs) for each species in water and urine were also worked out by on-line detection with QC-AAS ICP-AES and ICP-MS. It is clear that the DLs with ICP-AES were worse than those using QC-AAS detection by about a factor of five. As expected the ICP-MS provided much better detection limits than QC-AAS or ICP-AES (of course the on-line hydride generation technique used avoids well known polyatomic ICP-MS interferences).A comparative view of the DLs observed for the three detectors and matrices is given in Table 5. DLs of 0.16-0.66 pg I-' of selenium (as 60000 'ww/ i 1 9 30000 20000 10000 1.0 3.0 5.0 7.0 9.0 Time (c) Fig. 4 Selenium speciation in water (a) ICP-AES detection (500 pg I-' of each selenocompound); (b) QC-AAS (100 pg 1-' of each selenocompound); (c) ICP-MS (10 pg 1-' of each selenocompound); 1 2 and 3 as above Table 4 Comparative calibration graphs for different selenocom- pounds in water and in a human urine matrix using QC-AAS detection Calibration graph Selenocompound (c is expressed as pg 1-') Correlation Inorganic Se Abs= 1.06~- 1.05 0.998 Inorganic Se (urine) Abs = 0.99~ + 2.95 0.998 Selenomethionine Abs = 1.73~ + 0.3 0.999 Selenomethionine (urine) Abs = 1.13~ + 1.55 0.998 Selenoethionine Abs = 0.75~ + 0.5 0.999 Selenoethionine (urine) Abs = 0.85~ -0.15 0.999 1 166 Journal of Analytical Atomic Spectrometry December 1996 Vol.11Table 5 Detection limits (pg 1-') for inorganic selenium selenomethionine selenoethionine in the chromatogram and for just SeIV by QC-AAS ICP-AES and ICP-MS in water and urine samples @ A 0 8 ~~ ~ ~~ ~~ ~~ ~ ~~ ~~ Water Urine Inorganic Se Se-Meth Se-Eth Se"* Inorganic Se Se-Meth Se-Eth Se"* QC-AASt 6.8 5 7.9 7.9 6.1 5 6.4 7.5 ICP-AESt 30 22 41 38 24 22 42 30 ICP-MS 0.16 0.59 0.66 0.19 0.16 0.59 0.62 0.2 3 (9.2 min) * Determined with MW switched off. 100 pg I-' of each selenocompound. 5 pg 1-' of each selenocompound. 5 9 77Se) can be achieved with the most sensitive detector and such DLs virtually do not change in a human urine matrix.Moreover Fig.4 shows that only the sensitivity seem to change from one atomizer to another. Thus interferences in the gaseous phase of the HG continuous system27 used can be ruled out from these experiments. 1 Selenium Speciation Studies in Urine Samples The proposed Se speciation system was initially tested with AAS detection for inorganic selenium and selenoaminoacids speciation in real urine samples. Typical chromatograms of a human urine sample are shown in Fig. 5 which shows that both selenoaminoacids under study are completely resolved for a spiked urine. However another small peak at 7.3 rnin retention time appears. The same unspiked urine sample produced the chromatogram Fig.S(b) showing the peak at 7.3 rnin retention time while the two 'model' selenoaminoacids investigated here appear to be absent. When the more sensitive ICP-MS detector was used several unidentified Se peaks were again obtained for spiked urine (concentration of added selenoaminoacids 5 pg 1-l). As can be seen in Fig. 6(a) ICP-MS detection shows the expected model species plus two unidentified peaks 1' at 3.6 rnin and 2' at 7.3 min retention time. Figs. 6(b) and (c) show the observed peaks in unspiked urines (with MW on and off respectively). The shift in retention times between 'on' and 'off' is not related to the HPLC separation but to the effect of the MW heating which accelerates the Thus both unknown species 1' and 2' appear even with MW off or using HCl 4 mol 1-' instead of HBr-KBrO (i.e.they should have a certain \ olatile character). The features of the urine chromatogram 6(a) are general for all urines of the individuals of our research group analysed. In fact we selected three males and three females (between 24 and 37 years old) and their urines were analysed 1 (tr = 3 min) I * (4.8 min) Fig. 5 Typical chromatogram with QC-AAS detection (a). spiked human urine with 100 pg 1-' of each of Se" and SeV1 selenomethionine and selenoethionine; (b) normal unspiked human urine; 1 2 and 3 as above 2' 1.0 3.0 5.0 7.0 9.0 Time/min Fig. 6 Selenium speciation by ICP-MS (a) spiked human urine with MW switched on ( 5 pg I-' of each SeIV SeV' selenomethionine and selenoethionine); (b) human urine with MW turned on; (c) human urine with MW turned off; 1 2 and 3 as above by the on-line general procedure.Fig. 7 shows the observed results demonstrating that unknown species 1' and 2' are always present although their relative concentrations vary depending on the individual analysed. It seems apparent that peak 1 is most probably selenite (tR around 3 min). Several 4 1' 11' 8000 4000 0 2.0 4.0 6.0 Male 2 Male 3 Female 1 Female 2 Female 3 - Timelmin Fig.7 urine samples of different people Chromatograms with ICP-MS detection of six non spiked Journal of Analytical Atomic Spectrometry December 1996 V01.11 1 167Table6 Retention times for the different Se species assayed and the peaks found in urine lapped inorganic selenium peaks by carrying out a second injection of the urine sample with the MW turned off to analyse only Se"'.Retention time/min Among the three atomic detectors assayed the ICP-MS Se" 3.0 provided detection limits two orders of magnitude better than those obtained by ICP-AES and between 10-50 times lower Sevl 3.0 Selenomethionine 4.8 9.2 than those typical of QC-AAS detection. It is worth mentioning Selenoethionine TMSef 3.7 that no differential interference^^^ were observed (see Fig.4) Se species DMSe MMSe 5.9 3.9 from foreign elements in continuous HG system and DL's did not deteriorate in urine (as compared with water) for any Peak 1' 3.6 detect or investigated. Peak 2' 7.3 Selenium speciation of human urine by the proposed pro- cedures has been shown to be rather complex but more detailed information about the actual species present in real authors have reported the presence of selenium in human urine samples could be gathered by resorting to the use of two or as selenite and trimethylselenonium.s~g Trimethylselenonium more atomic detectors coupled to a separative process of those (TMSe' ) however cannot account for peaks 1' or 2'.species. We have demonstrated here the absence of interferences Experiments injecting TMSe' in our flow system demonstrated or artefacts related to the detection used (Fig. 4) ruling out that this species appears at retention times of around 3.7 min differential interference^.^^ Moreover comparison of chromato- (similar t o 1'). However this compound provides compara- tively low sensitivity in our continuous flow system and so could account only partially for the total height observed for peak 1'.On the other hand the presence of monomethylselenol (CH,-SeH MMSe) has been recently demonstrated as the major Se component in rat urine at adequate Se nutritional status.28 Therefore several experiments were again carried out in our HPLC-MW digestion-HG-ICP-MS system in an attempt to characterize the unknown peak 2'. Thus MMSe was prepared by reaction of dimethyldiselenide (CH3 -Se-Se-CH DMSe,) with NaBH,26 and both MMSe and DMSe were added to human urine as spikes. Retention times observed for all Se species spiked in human urine are summarized in Table 6 showing that peak 2' is not any of the assayed Se species and so other chromatographies and if available other standard species should be tried for its identification.Inorganic Selenium Speciation in Urine As is shown by Figs. 4-6 the ionic species Se" and SeV' appear at the same retention times in the chromatograms. In a previous work2' the possibility of using the HBr-KBrO mixture for a non-chromatographic speciation of Se" and SeV' in tap water was demonstrated. Inorganic total selenium was determined with the MW oven turned on while only Se" is determined if MW power is off. SeV' is evaluated by simple difference between total inorganic Se and Se"'. This procedure has been confirmed and applied successfully here for the determination of just Se" in human urine samples. Taking relatively small volumes (e.g. 100 pl) of urine for injection in the system to avoid matrix interferences loo+ 5% recoveries of Se" and SeV1 have been obtained as expected,21 using Q C -A A S detect i on .CONCLUSIONS Aiming at the speciation in urine of inorganic and organic Se compounds a versatile coupling of HPLC separation- microwave digestion-HG-atomic detection in an on-line flow system has been demonstrated. The speciation of the model selenocompounds studied (Se" SeV1 selenomethionine and selenoethionine) can be accomplished in a CI8 reversed-phase chromatographic column which allows the separation of inor- ganic selenium from the different selenoaminoacids in 'salty' samples such as human urine (anion-exchange columns appear to be seriously affected by the ionic strength of such samples).22 The integrated flow system proposed allows complete separation of selenoaminoacids and also speciation of over- grams with detectors of increasing sensitivity proved to be of great value in ascertaining the presence of unknown species in investigated real samples (e.g.in basal urine only one Se species was detected reliably using AAS as shown by Fig. 5 while using ICP-MS detection three different Se species appeared as shown by Figs. 6 and 7). Further work to clarify the nature and concentration of the selenocompounds not yet identified (Fig. 7) in human urine samples is currently in progress. A Prolabo instrumental loan of the Microdigest M301 and financial support from DGICYT Project PB94-1331 are grate- fully acknowledged. The authors express their gratitude to J. I. Garcia Alonso and M. Montes for their collaboration in the HPLC-MW digestion-HG-ICP-MS coupling and to the group of M.A. Palacios for the supply of the TMSe' standard solution. REFERENCES 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 Kolbl G. Kalcher K. Irgolic K. J. and Magee R. J. Appl. Organometallic Chem. 1993 7 443. Chau Y. K. and Wong P. T. S. Fresenius' J. Anal. Chem. 1991 339 640. Mufioz Olivas R. Donard 0. F. X. Camara C. and Quevauviller P. Anal. Chim. Acta 1994 286 357. Mehra H. C. and Frankenberger W. T. Chromatographia 1988 25 585. Laborda F. De Loos-Vollebregt M. T. C. and De Galan L. Spectrochim. Acta Part B 1991 46 1089. LaFreniere K. E. Fassel V. A. and Eckels D. E. Anal. Chem. 1987 59 879. Blais J.-S. Huyghues-Despointes A. Momplaisir G. M. and Marshall W. D. J. Anal. At. Spectrom. 1991 6 225. Sanz-Medel A. Aizpun B. Marchante J.M. Segovia E. Fernandez M. L. and Blanco E. J. Chromatogr. 1994,683 233. Yang K.-L. and Jiang S.-J. Anal. Chim. Acta 1995 307 109. Goossens J. Moens L. and Dams R. J. Anal. At. Spectrom. 1993 8 921. McLaughlin K. Dadgar D. Smyth M. R. and McMaster D. Analyst 1990 115 275. Nakahara T. and Kikui N. Spectrochim. Acta Part B 1985 40 21. Ebdon L. Fisher A. S. and Worsfold P. J. J. Anal. At. Spectrom. 1994 9 611. Suzuki K. T. Itoh M. and Ohmichi M. J. Chromatogr. B Biomed. Appl. 1995 666 13. Suzuki K. T. Yoneda S. Itoh M. and Ohmichi M. J. Chromatogr. B Biomed. Appl. 1995 670 63. Lan W. G. Wong M. K. and Sin Y. M. Talanta 1994 41 195. Recknagel S. Braetter P. Tomiak A. and Roesick U. Fresenius' J. Anal. Chem. 1993 346 833. Cobo-Fernandez M. G. Palacios M. A. Chakraborti D. Quevauviller P. and Camara C. Fresenius' J. Anal. Chem. 1995 351 438. I 168 Journal of Analytical Atomic Spectrometry December 1996 Vol. 1 I19 Pitts L. Worsfold P. J. and Hill S. J. Analyst 1994 119 2785. 20 Pitts L. Fisher A. Worsfold P. and Hill. S. J. J. And. At. Spectrom. 1995 10 519. 21 Gonzalez LaFuente J. M. Fernandez Sanchez M. L. Marchante Gayon J. M. Sanchez Uria 3. E. and Sanz-Medei A. Spectrochim. Acta. Part B in the press. 22 Marchante Gayon J. M. Gonzalez LaFuente J. M. Fernindez Sanchez M. L. Blanco Gonzalez E. and Sanz-Medel A. Fresenius’ J. Anal. Chem. 1996 355 615. 23 Quijano M. A. Gutierrez A. M. Perez-Conde M. C.. and Camara C. J. Anal. At. Spectrom. 1996 11 407. 24 Sanz-Medel A. Analyst 1995 120 799. 25 Thomson M. Pahlavanpour B. Walton S. J. and Kirkbright G. F. Analyst 1978 103 568. 26 Sharpless K. B. and Lauer R. F. J. Am. Chem. SOC. 1973 95 2697. 27 Wickstrerm T. Lund W. and Bye R. J . Anal. At. Spectrom. 1995 10 809. 28 Suzuki K. T. Itoh M. and Ohmichi M. in Metal Ions in Biology and Medicine John Libbey Eurotext Pans France 1996 vol. 4 p.18. Paper 6/05090E Received July 22 1996 Accepted September 20 1996 Journal of Analytical Atomic Spectrometry December 1996 Vol. I 1 1169

ISSN:0267-9477    

DOI:10.1039/JA9961101163

出版商:RSC    

年代:1996

数据来源: RSC

摘要:

Speciation of Organic Selenium Compounds by High-performance Liquid C h romatog ra p hy-l nd uct ively Coupled Plasma Mass Spectrometry in Natural Samples RIANSARES MUfiOZ OLIVAS AND OLIVIER F. X. DONARD*. Laboratoire de Photophysique et Photochimie Moltculaire Borde zux I URA 348 CNRS 351 Cours de la Libtration 33405 Talence France NICOLE GILON AND MARTINE POTIN-GAIJTIER Laboratoire de Chimie Analytique Universitt de Pau et des Pays de I'Adour Avenue de l'Universitt 64000 Pau France The importance of selenium in the environment and in biological systems is now well recognized. Its biochemical functions e.g. its anti-oxidant role and therefore cell protecting function as an essential constituent of the enzyme glutathione peroxidase has provoked a growing interest in the determination of this element.Selenoamino acids are essential for the understanding of the biogeochemical cycle of Se and TMSe+ is a known urinary metabolite present at high levels when Se is taken in excess. In this work we present a hyphenated technique (HPLC-ICP-MS) available for the separation of the organic Se compounds (selenocystine selenomethionine and trimethylselenonium ion). The choice of column and solvents has been a critical parameter. Good repeatability and excellent detection limits (less than 1 pg 1-' for each species) have been reached for standard solutions and the application to some natural samples (enriched yeast human serum and urine) has shown quite promising results. Keywords Selenoamino acids; selenium; speciation; high- performance liquid chromatography; inductively coupled plasma mass spectrometry; ion-pairing Selenium is now widely recognized to be either toxic or essential depending on the species present and their cclncen- trations.'q2 Levels of total selenium in human serum between 20-8Opg 1-' can be considered as essential and toxic when in excess.Despite the fact that several intoxications related to selenium have been reported in recent years (from skin and ocular infections to heart troubles and even cancer3) most important problems are still associated with Se deficiency. During the ~ O ' S research works on the Keshan disease in China pointed out the essential role of this element for human metab~lism.~ Other studies have revealed some heart and muscle troubles leading to organ (liver pancreas and heart) degeneration associated with Se defi~iency.~ It is now well established that it is the chemical form rather than the total concentration that determines the toxicity or the bioavailability of trace element^.^ Thus both the identifi- cation and quantification of the different species of selenium are required to improve our knowledge of their biochemical cycle and metabolic functions. However the determination of the different selenium species in environmental and biological samples still remains a difficult task.For instance there is a lack of analytical methods available to separate all selenium ~ ~~ ~~ ~~ * To whom correspondence should be addressed. Present address Laboratoire de Chimie Bioinorganiq ue et Environnement Universite de Pau et Pays de I'Adour 64000 Pau France.Journal of Analytical Atomic Spectrometry species (organic and inorganic) with the same chromatographic procedure at the low concentrations occurring in natural samples.6 Inorganic selenium species found in the environment are mainly selenite (HSe0,- ) and selenate (HSe0,-). The selenite and selenate ions are taken up by plants and converted into selenoamino acids. This absorption phenomenon is regulated by different factors such as the temperature the organic matter content of the soil the pH etc. The main selenoamino acid evidenced in plants is ~elenomethionine.~ Other selenoamino acids identified in different animals are selenocystine selenocys- tathionine as an intermediate metabolic product and seleno- cysteine. The latter species has been identified to be an essential constituent of the human enzyme glutathione peroxidase.* This enzyme has been reported to inhibit the oxidative damage caused by free radicals formed from O2 and hence protects the cells.This result has been evidenced from epidemiological studies where total selenium in blood has been measured.' These studies clearly demonstrate that selenium levels found in older people and in patients suffering from oxidative patho- logies such as cancer are lower than those measured in healthy people. Finally TMSe+ has been found as a metabolite present in the urine of patients with high Se levels but is not a major urinary product when patients are exposed to normal or low Se condition^.^ Under these conditions the TMSe + cation could be used as a tracer of Se levels of humans.Speciation of inorganic selenium compounds (selenite and selenate) has been extensively studied in natural waters sedi- ments and soils. However less attention has been paid to differentiate between the different organic species. Most separa- tion methods reported in the literature for selenoamino acids involve LC. Ion chromatography allows the separate elution of selenocystine and selenomethionine from other amino acids.' Recently a reversed-phase chromatography approach has also been used for the separation of selenoamino acids."*" The determination of the trimethylselenonium ion is generally achieved by ion-exchange chromatography with ICP-MS dete~tion.'~-'~ In this paper we deal with the development of a method for the determination of two selenoamino acids selenomethion- ine (SeMet) and selenocystine (SeCys) and the trimethylse- lenonium ion (TMSe' ).Developments presented here follow up on the separation techniques presented by Jiang and Houk for the separation of the sulfur amino acids cystine and methionine. We have therefore used a reversed-phase chroma- tography separation based on the formation of ion-pairs. In our case the counter-ion must be anionic in order to retain cationic species such as TMSe'. This approach was mainly Journal of Analytkal Atomic Spectrometry December 1996 Vol. 11 (I 171 -1 176) 1171dedicated to the optimization of the determination of organic selenium species in view of a long term stability test carried out on these species for the EC Measurement and Testing Programme.No specific attention was paid to the simultaneous separation of the inorganic species that are eluted in the void volume. All the analytical steps have been optimized to yield excellent overall sensitivity. Hyphenation between HPLC and ICP-MS appears to be an appropriate method providing both good separation capabilities for these species and the high sensitivity required to detect them at the natural concentration levels (pg 1-l-ng 1-l). An enriched yeast sample has been analysed and results have been compared with those obtained by HPLC-ETAAS.I6 The determination of organic selenium species in biological samples such as urine and serum is also presented. EXPERIMENTAL Reagents All reagents used were of the highest purity grade.DL- Selenomethionine and DL-selenocystine were of analytical- reagent grade and were purchased from Sigma (St Louis MO USA). Trimethylselenonium chloride was obtained from Complutense University (Madrid Spain) where it was synthe- sized following the procedure described by Palmer et al.I7 Stock calibration solutions (100 mg 1-' of Se) were prepared by dissolving each compound in Milli-Q water acidified with HCl (pH 2) in order to facilitate their dissolution and were stored in the dark at 4°C. Finally working solutions were prepared daily by dilution of the 100 mg 1-' stock solution. Hydrochloric acid was used to adjust the pH of the solutions. This acid was Suprapur grade from Merck (Elmsford NY USA). The eluents used for the separation were methanol ethanol and acetonitrile (Prolabo Paris France).Different sodium sulfonic salts (pentane heptane octane and lauryl sulfonate) from Sigma were added as organic modifiers to the eluents. Apparatus HPLC The HPLC pump used was a Perkin-Elmer (Norwalk CT USA) Series 410 Bio made of titanium to avoid degradation of the reagents or analytes during analysis. The six-port Rheodyne sample injection valve also made of titanium was fitted with a 100 pl sample loop. A reversed-phase analytical column (PRPl Hamilton purchased from Alltech Arlkington Heights IL USA) was used for the optimization and analysis of real samples. In order to prolong the lifetime of the analytical column a guard cartridge made of the same material as the analytical column was used for the analysis of real samples.It protects the analytical column from particles sub-particles and other contaminants that can precipitate on the top of the column bed leading to an increase in the back pressure and a diminution in the resolution efficiency. Isocratic separation was achieved with a methanol-water mixture as the mobile phase. A sulfonic salt (sodium pentylsul- fonate) at mol I-' was added to the organic mixture. The optimum pH for separation was 4.5. The outlet of the HPLC column was directly connected to the spray chamber of the ICP-MS instrument via a PEEK (polyetheretherketone) tube (id 0.5 mm). ICP-MS The Perkin-Elmer ICP-MS SCIEX ELAN 5000 was used as a detector. It was fitted with the standard cross flow nebulizer and a Ryton spray chamber. Argon flow rates were regulated by mass flow controllers.Nickel sampler and skimmer were used. Transient signals were recorded with the ELAN Graphics application included in the ELAN software and processed with the PE Chromafile software. Data processing was based on peak area. Analytical Procedure The chromatographic separation of SeCys SeMet and TMSe + standards was performed by injecting solutions containing the three compounds at 1OOpg 1-l (as Se content) of each onto the column. Detection of each eluted Se species was achieved by ICP-MS using the operating conditions shown in Table 1. Analytical results obtained for real samples (yeast extracts) were compared with those achieved by HPLC-ETAAS.I6 Sample preparation and extraction Speciation analysis of natural samples requires a preliminary extraction step which should not modify the chemical form and distribution of the element.Whereas for the arsenic and tin organic compounds there are a wide number of extraction methods for organo-selenium compounds only a few papers have been published. A process largely used for arsenic species (to extract inorganic and organic As species) consisting of a mixture of water chloroform and methanol (2+3+5) and evaporation to dryness has been applied for selenium." Other workers have attempted acid protein hydrolysis with HCl at 110 "C but degradation of the selenium compounds has been observed during this pro~edure.'~ Finally recent publications related to organolead and organotin compounds presented a new extraction procedure for organometallic compounds enzy- mic hydrolysis.20-22 This method has been studied by Gilon et a1.I6 for selenium application.Their method was an enzymic hydrolysis using a protease which is able when used in large excess to break the peptidic bonds of any protein present in the material. The procedure involved mixing 100 mg of sample with 10 mg of protease and 4 ml of phosphate citric buffer at pH 7.5 followed by incubation for 24 h at 37 "C. The extraction yield reported by these authors for yeast was 92% of the total selenium. The serum sample used was a lyophilized human reference Seronorm (No. 311089 Nycomed Pharma Norway). It was reconstituted in 3 ml of Milli-Q water. A dilution (1 + 4) was Table 1 Experimental conditions for the speciation of organic selenium compounds Chromatographic conditions- Stationary phase Hamilton PRPl (polystyrene divinyl benzene) 15 cm length 4.1 mm id 5 pm size particle Mobile phase Water-methanol (98 +2)+ mol 1 - I PH 4.5 Pump flow rate Sample volume 100 pl C5HllSO3- 1 ml min-' Interface- PEEK connection 0.5 mm id Ar flow rate 60 cm length Outer 15 1 min-' Intermediate 0.8 1 min-' Nebulization 1.0 1 min-' ICP-MS conditions- Power supply 1100 w Acquisition parameters Dwell time 250 ms Readings 650 s Analysis time 300 s Isotope monitored **Se 1 172 Journal of Analytical Atomic Spectrometry December 1996 Vol.1 1applied prior to analysis. Human urine samples were also diluted before analysis ( 1 + 4). RESULTS AND DISCUSSION Optimization of Operating Conditions Parameters such as pH solvent polarity ionic strength are described in this section.The concentration of each compound during the optimization of experimental conditions wat 100 pg 1-1 as Se. Mobile phase We evaluated the efficiency of several polar solvent mixtures such as water-acetonitrile water-methanol and water-ethanol in order to achieve a highly polar mobile phase and to make use of the good acceptance of organic solvents into the plasma.23 A preliminary set of experiments was performed to study the separation of the compounds with the different mobile phases and results were selected with regard to sensi- tivity performance. The analytical signal obtained with the ICP-MS detector differed significantly from one solvent mix- ture to another. These differences are observed in Fig. 1 which represents sensitivity response of SeCys as a function (Jf the solvent amount in the mobile phase.Best results in terms of sensitivity were obtained with the methanol-water and etha- nol-water mixtures a similar enhancement being obtaincd for both of them. These results are consistent with our previous results using FI as the sample introduction strategy.23 The different mezhan- isms suggested to explain the sensitivity response with rzgard to the solvent introduction in the plasma could also be relevant here. Three possible mechanisms were suggested (i) small droplet size formation due to the surface tension properties of the solvent ( i i ) spatial shift in the geometry of the plasma during the combustion of the solvent modifying the sample depth thus optimizing the efficiency of ion extraction and (iii) improvements in ionization mechanisms in the plasma due to the presence of carbon and oxygen atoms.Despite the small difference in polarity between the methanol and ethanol mixtures the best peak resolutions were obtained with methanol. Methanol was therefore used throughout the experiment. The methanol load also significantly affected the overall sensitivity response. This response is displayed in Fig. 1. Optimum sensitivities were obtained for a low methanol loading (2-5%). At higher values a drastic decrease in sensi- tivity was observed. This could be due to a cooling-quenching effect on the plasma induced by solvent addition at this concentration range. The less energetic plasma could then lead to reduced intensity.These observations are in good agreement with previous The optimum working conditions retained for the mobile phase composition were found to be water-methanol (98 + 2) corresponding to the best compromise between optimum sensitivity and peak resolution. lon-pairing agent The ion-pairing agent criteria were optimized in the chromato- graphic system with regard to retention conditions of the cationic TMSe' species. The counter ions tested were different alkyl sulfonate salts. These salts present a variable hydrophobic aliphatic chain which results in a strong affinity with the apolar stationary phase. The sulfonic moiety (SO3 - ) preferentially interacts with the analytes passing throughout the column. Both mechanisms take place in a dynamic equilibrium during the elution of the analytes.Under optimum pH conditions the positively charged ammonium groups of the selenoamino acids are expected to interact with the negatively charged sulfonic group. The strong cationic character of the TMSe' will allow the formation of ion-pairing in the pH range studied. The affinity of the ion- pairing with the stationary phase increases with aliphatic chain length. Several sulfonate salts of increasing carbon chain length have been tested from pentane sulfonate (5 carbon atoms) to lauryl sulfonate (20 carbon atoms). The retention was highly affected by the number of carbon atoms of the ion-pairing agent as expected. Retention time of the three Se compounds for the different ion-pairs tested are summarized in Table 2. The lauryl sulfonate agent was discarded due to strong and irreversible interactions with the stationary phase resulting in an elevated retention time of the eluting compounds especially for TMSe' .Octane heptane and pentane sulfonate salts all gave satisfactory separation conditions for the selenium com- pounds. The shortest retention time and the best definition of the TMSe' peak was obtained with pentane sulfonate hence it was retained for our routine working conditions. Ionic strength In our system the ionic strength is determined by the concen- tration of the ion-pairing agent. Various concentrations of sodium pentane sulfonate have therefore been tested ranging from lo-' to mol I-'. Table 2 Retention time of the three selenium compounds for different ion-pairing agents 12000 1 B J .. . . . . . . . 0 1 2 3 4 5 6 7 8 9 1 0 Solvent amount (%) Fig. 1 Influence of the mobile phase composition on the analytical signal (based on SeCys signal). A acetonitrile; B ethanol; and C methanol Se compound SeCys SeMet TMSe' SeCys SeMet TMSe+ SeCys SeMet TMSe' SeCys SeMet TMSe+ Retention time/min Ion-pairing agent 1.34 2.64 Pentane sulfonate* 3.79 1.50 2.96 Heptane sulfonatef 4.78 1.53 3.05 Octane sulfonatet 6.33 1.76 4.35 Lauryl sulfonate * - * Isocratic separation. t Gradient separation. Lower repeatability for retention time. Time- 3 Strong ion-pairing formation and high retention time for TMSe+ . consuming procedure. Severe peak tailing. Journal of Analytical Atomic Spectrometry December 1996 Vol. 1 1 1 173The retention ability of the column is related to the capacity factor (k').This parameter is independent of the flow rate and the column dimensions. The variations of the capacity factor are plotted versus the ion-pairing concentration in Fig. 2. Most significant variations are observed for concentration ranges between 5 x 10-'-5 x moll-'. Increasing ion-pairing concentrations increased the possibil- ity of ion-pairing with the analytes and thus increased retention of the compounds. This effect was evidenced for SeMet and TMSe' . The capacity factor of SeCys was however surpris- ingly constant for all ion-pairing concentrations tested. This result suggests that retention processes of this compound should be more related to hydrophobic interactions between the stationary phase and the alkyl chain of the analyte than to an ion-pair formation.These types of mechanisms are often encountered in reversed-phase chromatography. The low reten- tion obtained for SeCys by the stationary phase (k'< 1) could also be explained by a size exclusion phenomenon. The alkyl chain of SeCys is less available to the small particles ( 5 mm) of the stationary phase and hence reduces the hydrophobic interactions compared with SeMet. PH A titration of the two selenoamino acids with KOH allowed the protonation constants (pK,s) of these compounds to be determined.24 These values are listed in Table 3. For both selenoamino acids the zwitterionic forms appear to be the predominant species in the pH range 4.0 to 7.0. The influence of pH on the capacity factor is plotted in Fig.3. At pH 3 to 7 the selenoamino acids are present in the zwitterionic form (neutrally charged) and the TMSe' is a cationic species (positively charged). The retention time for SeCys is the shortest for the whole pH range studied. A k' value less than 1 usually indicates that there are few interactions occurring between this compound and the stationary phase. The other two compounds (SeMet and TMSe') are more efficiently retained. The retention time of SeMet is the longest for low pH conditions whereas TMSe + exhibits an opposite behaviour. The increasing difference in the k' values between SeCys and SeMet at the low pH range can be explained by the increasing formation of the monocationic forms (H,SeCys+ and H,SeMet+) which would be in equilibrium with the 5 4 3 2 - -4 /' 0 1 2 3 4 [Ion pair]/10~ mol I-' Fig.2 Influence of the ionic strength (ion-pairing concentration) on the capacity factor ( k ) for A SeCys; B SeMet; and C TMSe' (100 pg 1-' each as Se) Table 3 Values of the protonation constants (pKas) of selenoamino acids24 measured at 25 "C at an ionic strength of 0.10 mol 1-' KCl Compound P K ~ I PKa PKa ~Ka4 DL -Selenocystine DL -Selenomethionine 2.19 k0.02 9.05 f 0.01 1.68 f 0.02 2.15 -t 0.01 8.07 & 0.01 8.94 f 0.02 Fig.3 Influence of the pH on the capacity factor (k') for A SeCys; B SeMet; and C TMSe' (100 pg 1-' each as Se) zwitterionic forms. The monocationic form of SeMet has a single positive charge. In comparison SeCys presents two positive and one negative charges. The occurrence of this negative charge will minimize the electrostatic interactions of these forms with the SO3- groups when compared with SeMet which does not present any negative charge.For the pH values ranging from 5.0 to 7.0 differences in the k' are constant since both compounds exist as zwitterions. The stability of k' over the whole pH range observed for the SeCys suggests that retention mechanisms are primarily controlled by size exclusion phenomena. The cationic form of TMSe+ seems to be more efficiently retained at high pH. This result is in good agreement with the basic character of this cationic species. An optimum pH value of 4.5 was retained for further experiment because it provided good separation of the three compounds and the retention time of TMSe' was sufficiently short ( k ' = 3 ) for routine working conditions. Other parameters Other chromatographic parameters such as liquid flow rate length and id of the interface between HPLC and ICP-MS and sample loop volume have also been evaluated.Sample introduction is provided via the standard pneumatic nebuliz- ation mode. In order to have efficient sample introduction the flow rate used for the separation should match that of standard nebulization uptake conditions ( 1 ml min-'). Liquid flow rates varying from 0.6 to 1.4ml min-' have therefore been tested. Low flow rates of 0.6 and 1.0ml min-' resulted in tailing of the late eluting compounds. Other limitations appeared with higher flow rates. The PRPl polymeric column is very sensitive to pressure and increasing flow rates generate an increase of pressure seriously limiting the column lifetime.All the above reasons led us to select 1 ml min-' as the optimum value for routine working conditions. The sample loop volume was optimized. A range of sample loop volumes (50-200 pl) were tested in order to obtain sufficient sensitivity and to avoid overloading of the chromato- graphic column. The best compromise was found with a 100 p1 loop and this volume was retained for all experiments. Finally the length and id of the interface linking the chromatographic column output and the spray chamber of the ICP-MS instru- ment can significantly affect the dispersion of the analytes. This dispersion induces severe peak tailing thereby affecting peak resolution. Several ids (ranging from 0.25 to 1 mm) were tested.Best results with respect to chromatographic signal dispersion were obtained with an id of 0.5 mm. The effect of the tubing length was minimal in comparison with the role of the id. A connecting tube length of 60cm was used for the whole set of experiments. 1 174 Journal of Analytical Atomic Spectrometry December 1996 Vol. 1 1Detection Detection was performed by ICP-MS. Among the different isotopes available for selenium quantification we have selected the "Se. This isotope has been proven to yield a good S/N (quite similar to that obtained for 78Se). Likewise the chroma- togram baseline is more stable and lower for 82Se. Previous results have also allowed the best nebulization conditions to be identified. A flow rate of 11 min-' was shown to be the most efficient for sen~itivity.~~ 7.1720 v) c 8 1240- h m 2 760- 3 280 Data acquisition Data acquisition parameters are of great importance when dealing with transient signals. A good compromise must be found between the total chromatogram duration (300 s) and the mode of data acquisition values for the best peak definition. A high number of readings (650 s) and a dwell time of 250 ms have been chosen for our routine working conditions. After careful optimization of all these parameters the final experimental conditions have been summarized in Table 1. A chromatogram of the three organic compounds (100 pg 1-' each as Se) obtained under optimized conditions is shown in Fig. 4. I SeCys - 2.64 I 3.7 - Analytical Performance Calibration graphs were obtained from HPLC-ICP-MS peak areas after analysis of standard solutions containing selenocys- tine selenomethionine and the trimethylselenonium ion.Curves were linear in the concentration range studied (0-500 pg 1-I). Repeatability was estimated after six corisecu- tive injections of a standard solution containing the three organoselenium species for a concentration of 100 pg 1-' each 2200 r 1.34 I I I 0 1.086 2.172 3.258 4.344 5.430 Time/min Fig. 4 Chromatogram obtained for the standards of SeCys !.ieMet and TMSe+ (100 pg 1-' as Se) under optimum experimental coniditions Table 4 Analytical performance of the hyphenated technique I-[PLC- ICP-MS for the speciation of organic selenium compounds. Results are based on peak area measurements ~ SeCys* SeMet* Thl Se + * Detection limits/pg 1- ' 0.20 0.60 ( 1.20 Repeatability (%) 3.5 10 4.5 Dynamic range/pg 1 - ' 0-500 0-500 0 -500 * 100 pg 1-' as Se content.as Se. Detection limits (DL) were evaluated according to the following definition (DL = 30 + Mb) where 0 b is the standard deviation of the blank and Mb is the mean value of the blank. These values are displayed in Table4 and are among the lowest published to date. Determination of Organic Selenium Compounds in Real Samples Evaluation of the method was performed with comparison of results obtained with an HPLC-ETAAS system on an enriched selenium yeast sample. Selenium species were extracted after enzymic digestion hydrolysis with a total extraction yield of 92%.16 In both cases separation of the selenoamino-acids was performed under similar chromatographic conditions.Results are listed in Table 5. A typical chromatogram is displayed in Fig. 5. The retention times of the two major peaks suggest that both SeCys and SeMet are present in this sample. The total selenium content of the sample has been identified to be 1040mgkg-'. Total selenium concentrations obtained after summing the concen- tration of the different species detected by HPLC-ICP-MS results give an overall good agreement with the total selenium value. These values are slightly higher than those obtained by HPLC-ETAAS. Respective percentages are similar for both techniques within the range of chromatographic errors associ- ated with integration. In Fig. 5 separation conditions for the first peak certainly suggest that inorganic and possibly other species elute in the dead volume close to the main SeCys compounds. Two other types of biological samples were analysed by HPLC-ICP-MS (a certified serum standard and human urine).Typical chromatograms are displayed in Figs. 6(a) and (b). Noisy signals are due to the low concentration of selenium in these samples. Samples were diluted ( 1 + 4) to minimize matrix interferences. The total selenium contents were found to be 76.4 pg 1-' for serum and 105.6 pg 1-' for urine. These levels are in agreement with the certified value of 86 pg 1-' for serum and fall within the normal values reported in the literature for these sample types.25 The serum sample [Fig. 6(a)] shows a major peak eluting at the retention time of SeCys. We could assume that only SeCys species is present in this serum sample.However as SeCys is poorly retained by the column it could be that other 0 1.086 2.172 3.258 4.344 5.430 Time/min Fig. 5 Chromatogram obtained by HPLC-ICP-MS for an enriched yeast sample. Se( I) is inorganic selenium Table 5 Summary of the quantification for selenium speciation in enriched yeast. Comparison between HPLC-ICP-MS (this work) and HPLC-ETAAS16 SeCys/ SeMet/ Se'*/' Total Se Total Se content/ 1-' I % I-' I % 1- identified/pg 1- ' vg 1-' HPLC-ICP-MS 389 + 37 551f29 144 f 34 HPLC-ETAAS 365+91 436 + 42 76+ 17 1084 + 34 878 101 1040f 100 1040 f 100 * Eluted in the void volume. Journal of Analytical Atomic Spectrometry December 1996 Vol. 1 1 1 175I 1 REFERENCES 126 II 7 @ 42 8 5 0 2 2 .P 168 v) 1 26 84 1.069 2.139 3.208 4.277 5.346 1.084 2.169 3.253 4.337 5.422 V ‘ Tirne/min Fig.6 Chromatograms obtained by HPLC-ICP-MS for (a) serum and (b) human urine species co-elute with SeCys. More developments are needed in order to elucidate this point. A noisy chromatographic profile was obtained for the urine sample [Fig. 6(b)]. The main peak (1.36 min elution time) was attributed to SeCys. However no quantification of these species was achieved because of severe peak overlap. It could be that other species elute close to SeCys such as inorganic selenium compounds (HSe0,- or HSe03-) or other unknown com- pounds. The absence of TMSe+ can be related to a low or normal Se diet. Finally we can conclude that these results are quite promis- ing for speciation of organic selenium compounds.The main limitation of this system is the inability to separate inorganic species together with organic ones. Further efforts are still needed in order to provide separation of organic and inorganic Se species using the same chromatographic system. The authors gratefully acknowledge the Perkin-Elmer society (Eric Denoyer Perkin-Elmer Norwalk CT USA) for the HPLC pump Series 410 Bio supplied to the laboratory for our experiments. This work has been supported by an EEC fellow- ship from the ‘Measurements and Testing Programme’. We would like to thank its contribution to these research developments. 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 Simonoff M. and Simonoff G. Le Silinium et la Vie Masson Paris 1991. Nkve J. and Therond P. Oligoiliments en Midecine et Biologie ed.Chappuis P. Lavoisier Paris 1991. Fishbein L. Environmental Selenium and its Signijicance. Fundamental and Applied Toxicology 1983 ch. 3 p. 9. Gu B. Q. Chin. Med. J. 1983 96 251. Shibata Y. Morita M. and Fuwa K. Adu. Biophys. 1992 28 31. Muiioz Olivas R. Donard 0. F. X. Camara C. and Quevauviller Ph. Anal. Chim. Acta 1994 286 357. Fergusson J. E. The Heavy Elements Chemistry Environmental Impact and Health EfSects Pergamon Press Oxford 1990 ch. 3. Rotruck J. J. Pope A. L. and Ganther H. E. Science 1972 179 588. Shamberger R. J. Medical Implications of Se Biochemistry 1986 ch. 3 p. 105. Potin-Gautier M. Boucharat C. Astruc A. and Astruc M. Appl. Organomet. Chem. 1993 7 593. Quijano M. A. GutiCrrez A. M. PCrez-Conde M. C. and Camara C. J. Anal. At. Spectrom. 1996 11 407. Tanzer D. and Heuman K . G. Anal. Chem. 1991,63 1984. Laborda F. Chakraborti D. Mir J. M. and Castillo J. R. J. Anal. At. Spectrom. 1993 8 643. Blais J.-S. Huyghues-Despointes A. Momplaisir G. M. and Marshall W. D. J . Anal. At. Spectrom. 1991 6 225. Jiang S. J. and Houk R. S. Spectrochim. Acta Part B 1988 45 405. Gilon N. Astruc A. Astruc M. and Potin-Gautier M. Appl. Organornet. Chem. 1995 9 623. Palmer I. S. Fisher D. D. Halverson A. W. and Olson 0. E. Biochem. Biophys. Acta 1969 177 336. Martin J. L. and Gerlach M. L. Anal. Biochem. 1969 29 257. Mackey L. N. Beck T. A J. Chromatogr. 1982 240 455. Ceulemans M. Witte C. Lobinski R. and Adams F. C. Appl. Organomet. Chem. 1994 8 451. Pannier F. Astruc A. and Astruc M. Anal. Chim. Acta. 1996 327 287. Forsyth D. S. and Iyengar R. J. Organomet. Chem. 1989 3,211. Mufioz Olivas R. Quetel C. R. and Donard 0. F . X. J . Anal. At. Spectrom. 1995 10 865. Rivail da Silva M. Muiioz Olivas R. Donard 0. F. X. and Lamotte M. Appl. Organomet. Chem. in the press. Sanz Alaejos M. Diaz Romero C. Chem. Rev. (Washington D.C.) 1995 95 (l) 227. Paper 6104951 F Received July 15 1996 Accepted October 30 1996 1 176 Journal of Analytical Atomic Spectrometry December 1996 Vol. 1 1

ISSN:0267-9477    

DOI:10.1039/JA9961101171

出版商:RSC    

年代:1996

数据来源: RSC

摘要:

Investigation of Selenium Speciation in In Vitro Gastrointestinal Extracts of Cooked Cod by High-performance Liquid Chromatography-Inductively Coupled Plasma Mass Spectrometry and Electrospray Mass Spectrometry Journal of Analytical 1 Atomic 1 Spectrometry 1 HELEN M. CREWS PHILIP A. CLARKE D. JOHN LEWIS LINDA M. OWEN AND PAUL R. STRUTT MAFF CSL Food Science Laboratory Norwich Research Park C'olney Norwich UK NR4 7UQ ANDRES IZQUIERDO Dept. Quimica Analytica San Albert0 Magno sln University of C<irdoba 14004 Cordoba Spain As part of an ongoing study of Se bioavailability from the human diet Se was determined in cooked cod and in enzyme extracts of cooked cod. Total Se was measured by direct nebulization ICP-MS and standard additions following digestion with HN03 in stainless steel pressure decomposition vessels.The concentration of Se in the cod was 1.52 mg kg-' with an LOD of 0.008 mg kg-'. An in vitro gastrointestinal enzymolysis procedure was used to extract Se species from the cooked cod. HPLC using an anion-exchange column was used to separate Se standards (selenomethionine selenocystine sodium selenite and sodium selenate) and the species soluble in the gastrointestinal extract (pH 6.8). The 82Se signal was used to quantify the individual Se species. In the extracts of cooked cod approximately 12% of total Se was thought to be selenite whilst the remaining Se eluted at a retention time different to that of any of the Se standards measured. However as the retention time was close to that of selenomethionine it was suggested that the Se-containing species was organic rather than inorganic.A similar peak was also found in the enzyme blanks. In order to gain further information as to the identity of the unknown species the samples and standards were subjected to electrospray-MS (ES-MS) in both positive and negative ionization modes. Although the four standards were amenable to analysis by ES-MS more work is required to improve the sensitivity of the system. This will involve modifying the chromatography for both ICP-MS and ES-MS. Keywords Selenium; speciation; selenomethionine; selenocystine; sodium selenite; sodium selenate; inductivt ly coupled plasma mass spectrometry; anion exchange; in \ itro enzymolysis; electrospray mass spectrometry high-perf01 mance liquid chromatography The knowledge that Se is nutritionally essential is relatively recent.It was recognised as such in 1957' but its biochemical function in animals was not shown until 1973.2 The utilization of Se by animals and humans appears to be related to the chemical form of the element but methods for assessing this bioavailability differ and the literature contains conflici ing or not easily compared data.3 There are many studies on animal uptake of various forms of Se4-7 and a smaller number on human absorption and utilization'-'' of which only a r'ew are cited here. However to quote a very recent publication by re ill^,^ 'while it is useful to have information on the bioavail- ability of such compounds [inorganic and organic forms of Se] which are widely used as nutritional supplements for domestic animals and humans it is even more important to know how Se as naturally present in different foodstuffs performs with regard to bioavailability'.Data for Se species found in common foods are sparse. Plants use inorganic selenate and selenite to synthesize selenoamino acids such as selenomethionine -cysteine -cystathionine and -rnethylselenocysteine." Olson et reported that in the gluten hydrolysate of high Se wheat approximately 50% was selenomethionine. In seleniferous cab- bage 15% of the Se in water extracts was also selenomethionine but eight other compounds were also found.13 Recently Se has been found as selenomethionine in soybean lectinI4 and associ- ated with soybean whey proteins and globulin polypeptide^.'^ Piepponen et a1.16 found that the element was present in mushrooms as mostly non-protein compounds.Garlic grown in a Se-enriched medium was found to be superior to regular garlic in mammary cancer prevention in an animal m0de1.I~ Animal tissues can contain selenocysteine as a constituent of a class of proteins known as selenoproteins. To date" three such proteins have been characterized glutathione peroxidase type I 5'-deiodinase and plasma selenoprotein P. In canned tuna variable amounts of di- quadri- and hexavalent forms of Se were present and it was thought that storage may effect the nature of the chemical species." In aqueous samples selenite selenate selenomethionine and selenocystine were separated by anion exchange using a pH gradient.20 In environmental water samples,21 selenite selenate and organoselenium species including trimethylselonium have been separated using a mixture of anion and cation exchange.Anion-exchange chromatography was used to separate methyl- selonium selenite and selenate in water and urine samples.22 A review of methods for speciation analysis of organic Se compounds has been published recently.23 However for protein and food samples more complex procedures are necessary to isolate the often very low levels of species present and retain their integrity. Sathe et ~ 1 . ' ~ used hydroponically grown soybean intrinsically radiolabelled with 75Se. The flour was defatted and then subjected to anion-exchange chromato- graphy protein fractionation and sodium dodecyl sulfate polyacrylamide gel electrophoresis.The use of element specific chromatographic detectors for speciation studies is being increasingly adopted.24 Gas chroma- tography with AED has been used to isolate and identify organoselenium compounds in garlic onion and broccoli2' and in human breath after ingestion of garlic.26 Selenium speciation has been reported using HPLC-HGAF with on-line microwave reduction27 and for aqueous and clinical samples Journal of Analjfical Atomic Spectrometry December 1996 Vol. 11 ( 1 177-1 182) 1 177using HPLC-ICP-MS.28-31 Volatile Se species have been investigated by GC-ICP-MS with ID analy~is.~' Electrospray-MS has been used to study Cd and Zn incor- poration in reconstituted metal lot hi one in^.^^ The technique was assessed34 for speciation measurements of ionic species in solution and for the determination of organometallic com- pounds and the authors concluded that whilst not ideal for speciation studies it could become a valuable additional tool.Most re~ently,~' the same group has published quantitative information for S042- in waste waters. Brown et com- pared electrospray (ES) and ICP sources using the same mass spectrometer for detection and concluded that the LODs for 10 elements were only 2-3 orders of magnitude higher using ES. In this paper as part of an ongoing study of Se bioavailability from the human diet we report the measurement of Se in cooked cod samples and the subsequent in vitro enzymolysis of the sample to broadly simulate human gastrointestinal digestion. The gastrointestinal extracts were analysed by HPLC-ICP-MS using anion-exchange chromatography.In addition a preliminary investigation of the ES-MS response for some Se standards (seleno-DL-cystine seleno-L-methionine sodium selenate and sodium selenite) and the gastrointestinal extracts is reported. EXPERIMENTAL Instrumentation A VG Plasma Quad (PQ I1 Turbo Plus) inductively coupled plasma mass spectrometer (VG Elemental Winsford Cheshire UK) was used for total Se determinations and HPLC-ICP-MS. Samples were introduced via a fixed cross-flow nebulizer and a water-cooled double-pass Scott-type spray chamber. Typical operating conditions are shown in Table 1. All Se isotopes except *'Se were monitored for information and Se was quantified using 82Se. For HPLC-ICP-MS the PQ I1 was directly coupled to the chromatographic system via the column outlet tubing.The system consisted of a Dionex DX500 HPLC system with a Dionex ASM autosampler (Dionex Sunnyvale MA USA). Full loop injections (50 pl) were made using a Dionex LC-10 chromatography organizer fitted with a Rheodyne 9010 injec- tion valve (Rheodyne Cotati CA USA). Anion-exchange separations were performed on a Polysphere IC AN-2 column (120x4.6mm id) preceded by a guard column (25x4.6mm id) of the same material (Merck Darmstadt Germany). Electrospray-MS was carried out using a VG Platform 2 API mass spectrometer (Micro-Mass Altrincham UK) with a pneumatically assisted standard probe. The nebulizer gas was Table 1 Instrument conditions ~~ ~~ Typical PQ II Turbo Plus conditions- Rf power 1350 W Outer gas flow Intermediate gas flow Nebulizer gas flow Spray chamber temperature 10 "C Mass range 73-83 m/z No.of channels 2048 Measuring mode Collector type Pulse Reflected power <1 w 13.5 1 min-' 1.5 1 min-' 0.8 1 min-' Peak jump (time resolved acquisition for HPLC) Source conditions for VG Platform 2 API- Capillary voltage 3.00 kV Cone voltage 25 V Source temperature 140 "C HV lens voltage 0.10 kV Oflset voltage 5v N2. Both positive- and negative-ionization modes were investigated. The source conditions are shown in Table 1. Reagents and Reference Materials Milli-Q-water (Millipore Bedford MA USA) was used ( 18.2 Ma). Nitric acid (Aristar) and Se (1000 ppm) standards were obtained from Merck (Poole Dorset UK). Hydrochloric acid (Primar) was obtained from FSA Laboratory Supplies (Loughborough UK).Standard Reference Material 841 8 Wheat Gluten came from NIST (Gaithersburg MD USA). Bile salts (B8756) and porcine enzymes (Pepsin P7000; a-amylase A6880 and Pancreatin P1750) and sodium bicarbonate were obtained from Sigma (Poole Dorset UK). The mobile phase for HPLC-ICP-MS was 5 mmol 1-' sal- icylate [sodium salicylate and salicylic acid ( AnalaR Merck)] adjusted to pH 8.5 with TRIS (Aristar Merck). The Se standards seleno-DL-cystine (Se-cys) seleno-L-meth- ionine (Se-met) sodium selenate and sodium selenite were purchased from Sigma. Working standards were made daily using the HPLC mobile phase as diluent. SAMPLE PREPARATION Preparation of Cooked Cod Cod (240 g wet mass) including flesh bones and skin was cooked in a Saisho MW 2000 microwave oven (Matthews NC USA).The cod had been stored frozen and was defrosted using the Autodefrost programme following the manufacturers instructions. The fish was then cooked as for consumption on high (loo% 650 W) power for 4 min. The fish was judged to be cooked when the flesh could be flaked and separated from the bone. When cool the skin and bones were removed. The flesh was weighed (84g) and then homogenized using a Stomacher homogenizer (Seward Laboratory London UK) prior to analysis. In oitro Gastrointestinal Enzymolysis This procedure has been described previo~sly.~' Briefly the method uses pepsin and HC1 to broadly simulate gastric digestion in the stomach and pancreatin amylase and bile salts for simulating digestion in the small intestine. Subsamples of cooked cod (5 g) were incubated at 37 "C for 4 h with 10 ml gastric juice (1% m/v pepsin in 0.15 moll-' NaC1 acidified with HC1 to pH 2.0) in plastic screw top containers in an oven fitted with a shaking tray (Gallenkamp Loughborough UK). After adjusting the pH to 4.8 with saturated NaHCO 10ml gastrointestinal juice ( 1.5% pancreatin 0.5% amylase and 0.15% bile salts m/v in 0.15 moll-' NaCl) was added.The pH was then adjusted to 6.9 and the samples incubated for a further 4 h at 37 "C. Reagent blanks were taken through the entire process. In addition some blanks were spiked prior to enzymolysis with selenite (50 pg) to determine the recovery of the added Se at the end of the gastric and gastrointestinal stages. At the end of each stage samples were removed and ultracentrifuged at 10°C for 2 h at 2500 rpm in an MSE Centrikon T-1045 centrifuge (Fisons Instruments Crawley Sussex UK).The supernatants were decanted into polycarbon- ate-polypropylene Nunc vials (25 ml Life Technologies Paisley Scotland) and stored in a refrigerator (+4 "C) until required for analysis. Total Se Determinations Total Se was measured by direct nebulization ICP-MS of acid digests following digestion of samples and reference materials in stainless steel pressure decomposition vessels (50 ml capacity). Duplicate samples of cooked cod (0.25 g) CRM 1 178 Journal of Analytical Atomic Spectrometry December 1996 Vol. 11(0.25 g) and gastric and gastrointestinal supernatants (2 ml) were digested with HNO ( 5 ml) overnight for 6 h in an air circulating oven (Gallenkamp Loughborough UK) at 1 SO "C.After this they were cooled the digests transferred to Nunc vials and made up to 10ml. Reagent blanks were taken through the entire procedures. Calibration was by the method of standard additions (50 pg as selenite). Selenite (5 pg) was added to some of the samples prior to acid digestion for determination of the analyte recovery. HPLC-ICP-MS Using the Dionex DX500 system the mobile phase 5 mmol 1-' salicylate (pH 8.5) was delivered at a flow ra?e of 0.75 ml min-'. These were found to be optimal for our ICP- MS and were based on a method used at the National Food Agency Copenhagen Denmark (Pedersen and Larsen verbal communication). Standard solutions of Se-met Se-cys selenite and selenate were used to prepare a mixed working standard containing all four compounds.The working standard was diluted (1 + 1) with the mobile phase to provide Se at concen- trations (pg ml-') of 1.02 (Se-met); 1.26 (Se-cys) 1.24 (selenite) and 0.99 (selenate). This standard provided approximately 50 ng as Se for each of the Se species in a 50 pl injection. Subsamples of the gastrointestinal supernatants for Flanks and cooked cod were diluted 1 + 1 with the mobile phahe and immediately injected onto the column (50 p1 injection). In addition some supernatants were spiked with the mixed stan- dard immediately prior to injection onto the column. The recovery of Se from the column was estimated by comparing the response from on column injections of samples with post column injections flow injected into the mobile phase immedi- ately after the column.The retention time of peak; was noted and the peak areas calculated using the time resolved analysis mode. Electrospray-MS The standard Se species (Se-met Se-cys selenite and selcnate) and the gastrointestinal supernatant of cooked cod were examined by ES-MS in both positive- and negative-ionization modes using standard operating conditions. Firstly the same column and mobile phase used for HPLC- ICP-MS were used. Both 10 and 20 pl injection volume.; were tried. The HPLC mobile phase flow rate was 0.75 ml niin-' but the eluant was split post column so that approximately 0.012ml min-' was directed to the mass spectrometer. This was done to reduce the amount of buffer entering the source. It does not result in a concomitant linear reduction in sensi- tivity as the technique is primarily concentration not mass dependent.Secondly aqueous solutions of standards and gastrointesti- nal extract were analysed by FI-ES-MS using 10 pl injections into H20-acetonitrile (9 + l) pH 7.0 carrier flow rate 0.1 ml min-'. RESULTS AND DISCUSSION Total Se Determinations in Cooked Cod and I n Vitro Supernat ants Table 2 shows the total Se concentrations in cooked cod and in the supernatants from the gastric and gastrointestinal stages of the in vitro enzymolysis. The quality assurance data for these analyses included the measurement of NIST SRM 8418 Wheat Gluten and spiked samples. The values obtained for SRM 8418 were 2.99 and 2.44 pg g-' (certified value 2.58k0.19 pg g-I). Wheat gluten was used in these analyses since it is a proteinaceous matrix with a Se concentration similar to that expected in the cod.For future work NRCC DORM-2 Dogfish Muscle (certified Se value 1.40k0.09 pg g-I) will be used. At the time of the study reported here DORM-2 was not available to us and we had no DORM-1 (certified Se value 1.62k0.12 pg g-'). Two other fish-based CRMs were available to us NRCC DOLT-1 Dogfish Liver and TORT-1 Hepatopancreas but the Se level in both of these is between 6 and 7.5 pg g-' and therefore we chose to use wheat gluten. The mean percentage recovery for Se added as selenite prior to acid digestion was 96+6% ( n = 6 ) . The recovery of Se as selenite added prior to the start of the in uitro enzymolysis was 102+7% ( n = 3 ) . This data was accepted as satisfactory.The value for Se in the cooked cod is comparable with other Se data for raw fish obtained in this laboratory (Baxter Crews Lofthouse and Robb unpublished data). The LOD was 0.008 pg g-' based on 3s blank expressed as sample. The recoveries of endogenous Se from the cooked cod in the gastric (pH 2.0) and gastrointestinal (pH 6.8) supernatants were 50 & 10% (n = 5) and 60f 14% (n =4) respectively. These recoveries may reflect either insoluble Se bound to the residue or an equilibrium between the solid and liquid phases. Other workers have reported incomplete recovery of amino acids after enzymic hydr~lysis.~~*~' Ibe et aL3* noted that an in vitro enzymolysis procedure based on the method reported here released between 25 and 40% of histidine from raw and cooked beefburgers when compared to hydrolysis at 110 "C with HC1.Gilon et aL3' noted that degradation of seleno-compounds has been noticed using this procedure. Our aim was to try and treat the cooked cod in a way which might partially mimic the in uiuo situation. However future work in this laboratory will compare the use of HCl hydrolysis with the in vitro enzymol y sis. HPLC-ICP-MS of Se Standards and Gastrointestinal Extracts of Cooked Cod Table 3 and Figs. 1-6 show the data obtained following anion- exchange chromatography of Se standards and gastrointestinal extracts. Figs. 1 (a) 1 (b) and 2 show the 82Se profile for Se-met (peak A) Se-cys (peak B) selenite (peak C) and selenate (peak D). In Figs. l(u) l ( b ) and 2 the amount of Se per standard injected onto the column was 0.2 0.5 and 50ng respectively. Figs.l(a) and l(b) were obtained on a different day to Fig. 2. These traces demonstrate the sensitivity of the technique for Se. The precision data for retention times presented in Table 3 relates to standards where approximately 50 ng Se per standard were injected onto the column. The % RSDs for the standards are below 5% and are acceptable. The recovery of Se off the column varied for the standards. Monitoring "Se the two inorganic standards gave similar responses in terms of counts per pg Se. However making the assumption that the inorganic standard response for Se was loo% then the Se from the Se-met and Se-cys gave responses which were approximately Table 2 Selenium in cooked cod and supernatants following in vitro treatment with gastric and gastrointestinal enzymes Sample ug Se g-' cooked cod ( 5 s ) % Se extracted from cooked cod by enzymes Total Se in cooked cod (n = 2) Se from cooked cod in gastric supernatant (n = 5 ) Se from cooked cod in gastrointestinal supernatant (n = 4) 1.52f0.07 0.77 f 0.08 0.93f0.13 - 50.5 61.3 Joitrnal of Analytical Atomic Spectrometry December 1996 Vol.11 1 179Table 3 Mean retention times for ''Se response for standards and in uitro gastrointestinal extracts of blanks and cooked cod following anion-exchange chromatography-ICP-MS. See text for chromato- graphic conditions. Peak indentification as in Figs. 1-6 Sample Se-met* Se-cys* Selenitet Selenate Gastrointestinal blankt Gastrointestinal extract of Gastrointestinal blank cooked cod + standard Gastrointestinal extract* of cooked cod + standard Mean retention time (n = 6) +_ s Peak min % RSD A 2.82 k0.06 2.37 B 4.10 k0.12 3.02 C 6.66 f0.13 1.99 D 10.93 k0.35 3.17 E 2.18 k0.0 0.0 G 7.62 k0.26 3.44 E 2.43 k0.40 16.40 C 7.02 f0.33 4.8 1 E 2.18 (2.18 2.18) F 2.46 (2.46 2.46) G 7.83 (7.57 7.79) H 11.61 (11.35 11.87) E 2.18 (2.18 2.18) C 6.76 (6.82 6.71) G 7.48 (7.57 7.40) H 11.55 (11.81 11.29) * Standard approximately 50 ng on column.+ n=3. n = 2 therefore 2 plus individual values given. * I I I + 2 4 6 8 1 0 1 2 Time/min Fig. 1 HPLC-ICP-MS of Se standards (a) 4ngml-' and (b) 10 ng ml-' using anion-exchange chromatography. See Table 3 for peak notation and text for chromatographic conditions hC Ti me/m i n Fig. 2 As Fig.1 but using 1 pg ml-' standard 40 and 55% respectively of that due to selenite or selenate. This could be due to either suppression of the Se signal for the organic standards or retention of the Se from the organic standards by the chromatography system. Limited data for post column FI standards suggests that there is suppression of the Se signal from the organic standards but the effect is not clear and is being investigated further. Figs. 3 and 4 show the 82Se profiles for the in uitro gastroin- testinal enzyme blank and the cooked cod. They are very similar. Peak G in the enzyme blank has not been identified. In the extracts of cooked cod (Fig. 4) approximately 12% of 2 4 6 8 10 12 14 Ti me/m i n Fig. 3 HPLC-ICP-MS of in uitro gastrointestinal enzyme blank.Peaks E and G unidentified see text for explanation and chromato- graphic conditions .- E v) C a c w - i C 2 4 6 8 10 12 14 Timdmin Fig. 4 HPLC-ICP-MS of in uitro gastrointestinal extract of cooked cod. Peak E unidentified see text for explanation and chromato- graphic conditions Time/min Fig. 5 HPLC-ICP-MS of in uitro gastrointestinal enzyme blank plus standard mixture. Peak E unidentified see text for explanation and chromatographic conditions I 'IF 2 4 6 8 10 12 14 Ti me/mi n Fig. 6 HPLC-ICP-MS of in uitro gastrointestinal extract of cooked cod plus standard mixture. Peaks E F G and H unidentified see text for explanation and chromatographic conditions 1 180 Journal of Analytical Atomic Spectrometry December 1996 Vol. 1 1total Se was thought to be selenite (peak C) whilst the remaining Se (peak E) eluted at a retention time differmt to that of any of the selenium standards measured.The unknown peak E is also present in the enzyme mixture and is possibly due to incomplete hydrolysis of proteins to their amino a ~ i d s . ~ ' ~ ~ This demonstrates the need for careful interpretation of unknown peaks when using enzymes as peak E originates from the reagents and not from the sample. In Figs. 5 and 6 the profiles for the enzyme blank and cooked cod spiked with the four standards immediatelj prior to injection are shown. Whilst selenite is still seen (peak C) in the cooked cod extract it is not present in the spiked enzyme blank. The other three standards are not apparent in either the cooked cod extract or the enzyme blank as judged by the retention times given in Table 3.The unidentified peaks E F G and H are shown in Figs. 5 and 6. (Peak X was found in the spiked gastrointestinal blank on one occasion and is xhown in Fig. 5 for information). However the appearance oi these unidentified peaks is directly related to the addition Iof the standards and it may be that the nature of the enzyme extracts has caused either shifts in retention time for the standards themselves or conversion of the standards to other species. The data for the enzymolysis supernatants were generated over several days. The working standards were made daily. The supernatants were produced in advance and refrigerated prior to use. In an attempt to minimize speciation changes caused by defrosting and then refreezing they were not frozen.They gave reproducible data as tabulated in Table 3 for 2 or 3 consecutive days but after that the profiles showed signs of change. For example the peaks eluting in the region of the inorganic standards disappeared completely. Selenium species are known to be difficult to stabilize and this work serves to emphasize the difficulty of working with real samples as opposed to freshly prepared standards. 248 357 Electrospray-MS The samples and standards were subjected to preliminary ES-MS analysis to assess whether or not this approach could provide any further information on the nature of the Se species in the enzyme extracts. Using the same column and mobile phase as that used for our HPLC-ICP-MS analyses w.is not as anticipated successful.The high salt content of the mobile phase was problematic although splitting the eluant flow considerably reduced this effect. However the complexity of the spectra combined with the low levels of Se did not permit any useful information to be obtained from the chromato- graphed standards and samples. Fig. 7 shows negative-ion scans for selenite and selenslte and positive-ion scans for Se-cys and Se-met following FI of the aqueous standards. The levels of Se in the aqueous enzyme supernatants were too low to be detected. The levels of Se injected into the source to provide the scans shown in Fig. 7 correspond to 27 20 1.1 and 7.5 pg for selenite selenate Se-cys and Se-met respectively. Normally for organic. com- pounds the instrument is capable of detecting 0.1 (ideal amount) to 0.01 pg of analyte injected into the source.Thus under the conditions we were using which were not optimized in any way for these compounds the sensitivity was 2 to 3 orders of magnitude worse. The better sensitivity for the organic species relates to the protonated amine groups two are possible on Se-cys and one is possible for Se-met. The presence of the two protonated sites facilitates ionization resulting in a further improvement in sensitivity. For selenite no positive-ion spectrum was produced but the negative-ion spectrum (Fig. 7) shows a base peak at nz/z 129 (HSe0,-) and an isotope distribution consistent with that for a single Se atom in the molecule. For selenate (Fig. 7) again only a negative-ion spectrum was observed with a base peak 183 351 120 n 2 .0 5 337 /I 1 lY I I I1 239 395 102 0 0 0 115 120 160 200 240 280 320 360 400 Fig. 7 Electrospray-MS spectra of aqueous standards of (a) sodium selenite; (b) sodium selenate; (c) selenocystine; and ( d ) selenomethionine. See text for experimental conditions and explanation of spectra at m/z 145 (HSe0,-) and an isotope distribution consistent with that for a single Se atom in the molecule. Se-cys (Fig. 7) did not produce a spectrum in negative-ionization mode but a positive-ion spectrum with a base peak at m/z 337 [HOOC- CH(NH,)+CH2-Se-Se-CH,-CH(NH,)-COOH] was record- ed. The isotope distribution was characteristic of and con- sistent for two selenium atoms in the molecule and with the structure for Se-cys with a protonated amine group.Se-met (Fig. 7) produced spectra in both positive- and negative- ionization modes but the former was more sensitive. A base peak was recorded at m/z 198 [CH,Se-CH,CH,CH( NH3)+- COOH] with an isotope distribution consistent with that for a single Se atom in the molecule. In addition a fragment ion at m/z 181 was noted consistent with [M + H - NH,] + ion. There is also evidence of a [2M+H]+ at m/z 395. The spectrum was consistent with the structure for Se-met with a protonated amine group. For future studies the mobile phase will be modified to be compatible with our anion-exchange requirements and those of ES-MS. This is likely to involve the use of ammonium acetate which is already in use for ES-MS studies in this laboratory. The standards could be infused into the source at a low flow rate to produce a steady state ion source which will permit optimization of source conditions for each analyte.Cone voltage switching combined with ionization polarity switching would enable scans of mixtures of the inorganic and organic compounds to be analysed. Currently a standard pneumatically assisted ES probe is used. A triaxial probe which has an additional outer sleeve through which solvent can be added would aid the prevention of salting up. However to echo the conclusions of Horlick's group3' this preliminary study has illustrated the complex mixture of requirements that must be satisfied to interpret the qualitative data obtained. It has also given us a glimpse of the potential power that a combination of techniques can provide for studying speciation.Jomnal of Analytical Atomic Spectrometry December 1996 Vol. 11 1 181CONCLUSIONS The speciation of Se in foods for human consumption is not well understood. The difficulties in studying an element in real samples have been illustrated including the need to try and preserve the speciation under experimental conditions. The HPLC-ICP-MS methodology for anion exchange separation of the Se standards Se-met Se-cys selenite and selenate appears to be reproducible in terms of the precision of retention times. For unknown samples it is imperative that they are analysed as rapidly as possible. Preliminary work with ES-MS indicates that aqueous standards are amenable to the technique but more work is required to optimize instrumental and chromatographic conditions.This work has been funded by the Ministry of Agriculture Fisheries and Food. A.I. thanks the Spanish Government for support (Grant No. PF 95 30486679). REFERENCES 1 2 3 4 5 6 7 8 9 10 11 12 13 Rotruck J. T. Pope A. L. Ganther H. E. Swanson A. B. Hafeman D. G. and Hoekstra W. G. Science 1973 179 585. Burk R. F. Annu. Rev. Nutr. 1983 3 53. Reilly C. Selenium in Health and Food 1996 Blackie London UK 1996 p. 273. Douglas J. S. Morris V. C. Sores J. H. Jr. and Levander 0. A. J. Nutr. 1981 111 2180. Alexander A. R. Whanger P. D. and Miller L. T. J. Nutr. 1983 113 196. Bell J. G. and Cowey C. B. Aquaculture 1989 811 61. Suzuki K. J. Itoh M. and Ohmichi M. Toxicology 1995 103 157. Young V. R. Nahapetian A. and Janghorbani M. Am. J. Clin. Nutr.1982 35 1076. Butler J. D. Thomson C. D. Whanger P. D. and Robinson M. F. Am. J. Clin. Nutr. 1991 52 748. Smith A. M. Chen L-W. Thomas M. R. Am. J. Clin. Nutr. 1995,61,44. Shrift A. in Organic Selenium Compounds Their Chemistry and Biology ed. Klayman D. L. and Gunther W. H. H. Wiley- Interscience New York 1973 p. 863. Olson 0. E. Novacek E. J. Whitehead E. I. and Palmer I. S. Phytochemistry 1970 1181. Hamilton J. W. J. Agric. Food Chem. 1975 23 1150. 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 Sathe S. K. Mason A. C. Rodibaygh R. and Weaver M. J. Agric. Food Chem. 1992 40 2084. Wang Z. Xie S. and Peng A. J. Agric. Food Chem. 1996 44 2754. Piepponen S. Pellinen M. J. and Hattula T. in Trace Element Analytical Chemistry in Medicine and Biology ed.Peter B. and Peter S. de Gruyter Berlin 1984 pp. 159-166. Ip C. Lisk D. J. and Stowesan G. S. Nutr. Cancer. 1992,17,279. Ip C. and Lisk D. J. Nutr. Cancer. 1993 20 129. Cappon C. J. and Smith J. C. J. Appl. Toxicol. 1982 2 181. Quijano M. A. Gutierrez A. M. PCrez-Conde M. C. and Camara C. J. Anal. At. Spectrom. 1996 11 407. Tanzer D. and Heumann K. G. Anal. Chem. 1991 63 1984. Laborda F. Chakraborti D. Mir J. M. and Castillo J. R. J. Anal. At. Spectrom. 1993 8 643. Pyrzyiiska K. Analyst 1996 121 77R. Uden P. C. Anal. Proc. 1993 30 405. Cai X-J. Block E. Uden P. C. Zhang X. Quimby B. D. and Sullivan J. J. J. Agri. Food Chem. 1995 43 1754. Cai X-J. Block E. Uden P. C. Quimby B. D. and Sullivan J. J. J. Agri. Food Chem. 1995 42 1752. Pitts L. Fisher A. Worsfold P. and Hill S. J. J. Anal. At. Spectrom. 1995 10 519. Houk R. S. Shum S. C. K. and Wiederin D. R. Anal. Chim. Acta. 1991 250 61. Shum S. C. K. and Houk R. S. Anal. Chem. 1993,65 2972. Suzuki K. T. Itoh M. and Ohmichi M. J. Chromatogr. Biomed. Appl. 1995 666 13. Yang K-L. Jiang S-J. Anal. Chim. Acta. 1995 307 109. Gallus S. M. and Heumann K. G. J. Anal. At. Spectrom. 1996 11 887. Yu X. Wojciechowski M. and Fenselau C. Anal. Chem. 1993 65 1355. Agnes G. R. Stewart I. I. and Horlick G. Appl. Spectrosc. 1994 11 1347. Stewart I. I. Barnett D. A. and Horlick G. J. Anal. At. Spectrom. 1996 11 877. Brown F. B. Olson L. K. and Caruso J. A. J. Anal. At. Spectrom. 1996 11 633. Owen L. M. W. Crews H. M. and Massey R. C. Chem. Speciation Bioavailability 1992 4 89. Ibe F. I. Blowers S. D. Anderson D. and Massey R. Food Addit. Contam. 1994 11 403. Gilon N. Astruc A. Astruc M. and Potin-Gauter A. Appl. Organomet. Chem. 1995 9 632. Paper 6/06801 D Received October 4 1996 Accepted October 14 1996 1 182 Journal of Analytical Atomic Spectrometry December 1996 Vol. 1 1

ISSN:0267-9477    

DOI:10.1039/JA9961101177

出版商:RSC    

年代:1996

数据来源: RSC

摘要:

Approaches to the Determination of Meta I lot h io nei n(s) by High = pe r f o rman ce Li q u id C h roma t og ra p h y-Q ua r tz Tu be Atomic Absorption Spectrometry Journal of Analytical Atomic Spectrometry YANXI TAN PATRICK AGER AND WILLIAM D. MARSHALL* Department of Food Science and Agricultural Chemistry Macdonald Campus of McGill 21,111 Lakeshore Road Ste-Anne-de-Bellevue Qutbec Canada H9X 3 V9. E-mail marshall@agradm.lan.rncgill.ca HING MAN CHAN Center for Nutrition and the Environment of Indigenous Peoples (CINE ) Macdonald Campus of McGill 21,111 Lakeshore Road Ste-Anne-de-Bellevue Qutbec Canada H9X 3 V9 The operation of a quartz T-tube interface for coupling liquid flowing streams with AAS when optimized for Ag detection provided low to sub-nanogram responses for flow injections into aqueous carrier streams.Limits of detection were not greatly different to those for Cd Cu or Zn but were dependent on the mobile phase composition. The interface was used to monitor the Ag Cd Cu or Zn bound to individual metallothionein (MT) isoforms which had been partially resolved by size-exclusion or ion-exchange HPLC. Commercial MT isolates from rabbit liver were readily resolved into 4-5 Cd- Cu- and/or Zn-containing fractioris within 10-15 rnin using a solvent programme which increased the tris( hydroxymethy1)methylamine content over time. In addition to the major Cd-MT-I and Cd-MT-I1 isoforms [retention time (tR) = 5.84 and 7.58 min] two rapidly eluting fractions (tR = 2.22 and 3.21 min) collectively accounted for 21-24% of the Cd-response to the MT mixture and the MT-I enriched standards.By contrast the MT-I1 standard contained only trace amounts of Cd bound to the tR=2.22 rnin fraction and virtually none bound to the tR=3.21 rnin fraction. Zn was associated predominantly with the MT-I1 isoform and with a more retained fraction which did not contain the other metals. Cu which represented only a small fraction (< YO) of the metal content of these polypeptides was present in greater proportion in the MT-I isoform than in the MT-I1 isoform and was prevalent in the early-eluting fraction with tR=3.25 min. An Ag-saturation procedure rapidly and efficiently replaced the bound Cd Cu and Zn in these polypeptides but modified the chromatographic behaviour of the products appreciably. Keywords Metallothionein; metallothionein isoforms; high- performance liquid chromatography; atomic absorption spectrometry; metal binding; silver-saturation Metallothioneins'-* are a group of thermally stable low mol- ecular mass cysteine-rich metal-binding proteins which are characterized by a unique amino acid sequence spectroscopic features characteristic of tetrahedral metal-thiolate (-mercap- tide) complexes and a strong avidity for Ag' Cu' Cd Elg and Zn.Although first isolated from equine kidney cortex this class of metal-binding polypeptide is widely distributed throughout the animal kingdom and non-translat ionally biosynthesized analogues have been identified in more than 85 species of plants yeasts and fungi. The most conserved features among the mammalian structural sub-group of this class of protein are the 20 cysteines (cys) (of a total of 60-62 amino acid residues) which collectively bind seven bivalent met a1 ions * To whom correspondence should be addressed.in two separate metal-thiolate oligonuclear clusters (M,Cys and M4Cysll) buried as 'mineral cores' in the interior of the two globular domains formed by the carboxyl- (a-domain) and amino-terminal (/?-domain) halves of the polypeptide chain.' Two other structural motifs have been identified for mono- valent metals (M12CysZ0 and Ml8CysZ0). Mammalian (class 1) metallothioneins occur in two principal isoforms (conven- tionally labelled as MT-I and MT-I1 based on their order of elution from a DEAE-Sephadex column) although several sub- isoforms have also been separated by HPLC.'*" Whereas the average apparent association (binding) constant for Zn'l at pH 7.0 is in the range 1011-1012 1 mol-' Cd" Cu' and Ag' are bound more tightly by several orders of magnitude and follow the general order found for inorganic thiolates Hg"> Ag' z Cu'> Cd"> Zn1J.8*'1 Despite binding constants which at pH 7 are nearly equal to EDTA in avidity and their position within the recesses of the enveloping protein cluster- bound metal ions are fairly mobile in exchange interactions with other metals in solution with other ligands such as glutathione and with other metallothionein molecules due principally to a cleft within each domain of the enveloping polypeptide.Elevated levels of this class of polypeptide are accumulated in uiuo in response to exposure to a variety of metal salts as well as oxidative stress heat solvents and steroids.Conventional atomic absorption and emission spectrometric methods have been used extensively for the study of metal- bound metal lot hi one in^^^^'^ but can suffer from a lack of sufficient sensitivity for environmental monitoring of these analytes in certain tissues. We have recently des~ribedl~*'~ a novel silica T-tube interface design for coupling HPLC with AAS and have evaluated the device for As15 and Se'5916 speciation in marine tissues and plants. In addition to As and Se the interface provided sensitive responses (low to sub- nanogram limits of detection) to Cd Cu Hg Pb and Zn.14 Relative to previous designs the principal advantages were considered to be the compatibility with volatile buffers con- tained in either aqueous or organic mobile phases the low cost and the robust character of the device which can be operated over several months without an appreciable change in response.The objectives of the present work were 3-fold (i) to explore size-exclusion and ion-exchange HPLC as separ- atory techniques for the determination of individual isoforms of mammalian and molluscan metallothioneins; (ii) to develop a sensitive method to quantify these polypeptides by detecting the Cd Cu or Zn metal@) bound to their surfaces; and (iii) to explore Ag-saturation as a means of determining the total amounts of these polypeptides. Journal of Analyrical Atomic Spectrometry December 1996 Vol. 11 (1 183-1 187) 1 183EXPERIMENTAL HPLC-AAS The mobile phase was provided by a Varian (Palo Alto CA USA) Model 9010 HPLC pump and delivered at 1 ml min-' to the column via a manual 6-port rotary injection valve.The column eluate contained in a 25 cm x 0.05 mm section of silica capillary transfer line (SGE Industries Austin TX USA) was nebulized by thermospray effect into an enclosed pyrolysis chamber containing a diffused flame maintained by separate flows of O2 and H2 to the base of the chamber. The combustion products were entrained by the hot gases directly into the upper optical tube of the interface which was interposed within the optical beam of a Varian SpectrAA 5 atomic absorption spectrometer equipped with a deuterium background correc- tion system. Maximum responses to aqueous AgNO flow- injected into an aqueous carrier were achieved with flow rates of 895 and 240 ml min-' for H and 02 respectively added through separate gas entry ports to the base of the pyrolysis chamber.Optimum Ag-response was provided by applied voltages of 16.8 33.6 and 16.8 V of rectified current applied to separately energized heating coils surrounding the thermospray tube the combustion tube and the optical tube respectively. Spectrometer operating parameters are recorded in Table 1. 80 Size-exclusion HPLC Size-exclusion HPLC was performed on a 30 cm x 7.8 mm column of TSK Gel (G3,000PWxL Supelco Bellefonte PA USA) eluted with 30 mmol 1-' TRIS buffer (pH 8.9) delivered at 1 ml min-'. 60 40 > E 2 0 1 8 Ion-exchange HPLC Ion-exchange HPLC was performed on a 7.5cmx7.8mm column of Progel (Supelco) eluted with a two-step solvent programme.The concentration of TRIS buffer (pH7.5) was increased from 10 to 100 mmol 1-' over 5 min then further increased to 150 mmol 1-' over 15 min. The column was subsequently regenerated with 10 mmol 1-l TRIS buffer for 15 min prior to the next injection. L > Chemicals Metallothionein (MT) metallothionein-I (MT-I) and metallothionein-I1 (MT-11) (each nominally 5.7% m/m Cd and 0.7% Zn isolated from rabbit liver and essentially salt-free) haemoglobin and TRIS were purchased from Sigma (St Louis MO USA). Silver Saturation Metallothionein standard (0.2-1 mg in 0.2 ml of 10 mmol I-' TRIS buffer) was incubated with 0.1 ml of 0.5% m/v AgNO at 20 "C for 30 min with occasional vortex-mixing. Table 1 AA spectrometer operating parameters and response characteristics Lamp operating 328.1 4 9.3/1.4t Analyte Wavelengthlnm current/A LOD*/pmol 23.1 Ag Cd 228.8 6 c u 324.8 12 20.1 Zn 213.9 6 16.8 * LOD for flow injection into 10 mmol 1-l TRIS buffer determined LOD for flow injection into distilled de-ionized water determined as 3sb/s.as 3sb/s. Haemoglobin (0.2ml of a 1% m/v solution in lOmmoll-' TRIS buffer) was added to the crude incubation mixture thoroughly mixed then heated to 80-100°C for 5 min. The crude reaction mixture was then centrifuged for 10min at 10 000 rev min-' and the resulting crude supernatant fraction was directly analysed by HPLC. RESULTS AND DISCUSSION Initial studies were directed to defining optimum interface operating parameters for the detection of Ag contained in aqueous media.For these studies analyte solutions were introduced into 1.0 ml min-' of distilled water and transferred directly to the interface in a flow-injection mode. Stepwise univariate analysis of the influence on Ag-response of the flow rates of (i) O2 and (ii) H to the pyrolysis chamber and (iii) the skin temperatures of the thermospray tube the pyrol- ysis chamber and the optical tube in the vicinity of the separately energized heating coils indicated that maximum peak area response to AgN0 was obtained with flow rates of 340 and 895 ml min-' for O2 and H2 respectively and 720 1300 and 1120°C for the temperatures of the thermospray tube the pyrolysis chamber and the optical tube respectively. Arbitrarily 1 ml min-' of mobile phase was chosen to optimize the chromatography.Typical responses are presented in Fig. 1. The optimum H, O2 ratio for Ag detection (2.6) although slightly hydrogen-rich was appreciably less than the optimum ratio for other volatile metals (4.1-5.7).14 Based on the conven- tional definition ( 3 4 9 the limit of detection (LOD) for Ag was determined to be 0.15 ng (1.4 pmol) for distilled water (Table l) but was dependent on the composition of the mobile phase. The LOD in 10mmoll-l TRIS buffer was Long (9.3 pmol). The LODs of Cd Cu and Zn in TRIS buffer were also increased appreciably relative to distilled water (Table 1). Previous trials had also demonstrated that the total analyte metal response to the co-injection of MT standard and aqueous inorganic Cd Cu or Zn standard was additive.No matrix effect was detected.I2 140 1 A 120 1 R2 = 0.9692 250 I 200 25 ng a 5 R2 = 0.9932 75 ng 50 ng I l l 50 ng f00 ng 100 ng Fig. 1 into A 10 mmol I-' TRIS buffer; or B distilled water Quartz T-tube AAS response to AgNO standard flow-injected 1 184 Journal of Analytical Atomic Spectrometry December 1996 Vd. 1 IPreliminary chromatographic trials were conducted with a molluscan MT fraction which had been saturated with Ag' in the presence of 2-mercaptoethanol treated with haemoglobin to complex the excess of free and/or loosely bound metal ion and then heated to denature the haemoglobin which was subsequently removed by centrifugation. The supernatant frac- tion was then subjected to size-exclusion HPLC-AAS (Fig. 2). Although three fractions were clearly evident in the chromatog- ram of the Ag-labelled MTs [retention time (tR)=5.89 6.28 and 7.80min1 it was not possible to assign the fractions to individual isoforms.A separate chromatogram of the Ag-haemoglobin complex (Fig. 2A t R = 5.22 min) indicated that a maximum of 6% of the Ag-response in the chromatog- ram of the MT fraction could have been due to this potential interferent. Similarly the injection of 2-mercaptoethanol alone failed to release any column-bound Ag-analyte (Fig. 2C I and the Ag-mercaptoethanol complex (Fig. 2D) was appreciably more retained than any of the fractions in the Ag-labelled MT fraction. Separate Cd- or Cu-specific chromatograms of the supernatant fraction failed to detect the presence of bound or free residues of these metals in the Ag-labelled product mixture.Exchange was thus rapid and virtually complete. Ion exchange was also evaluated as a separatory technique. Commercial standards of MT (containing both MT-J and MT-11) isolated from the liver of rabbits which had been exposed to Cd as well as separate fractions enriched in zither MT-I or in MT-I1 from the same source were separated using a two-step gradient which increased the concentration of TRIS over time. Cd-specific chromatograms of these materials are presented in Fig. 3. The chromatograms in Fig. 3 were cor- rected for changes in the baseline with changing mobile phase (Fig. 3D). In addition to the MT-I and MT-I1 isoforms (tR= 5.85 and 7.58 min respectively) which were only partially 7 5.89 Fig. 2 Ag-specific size-exclusion chromatograms of A 1 1 Ag-haemoglobin complex; B Ag-labelled molluscan MT fraction; C 2-mercaptoethanol control; and D 1 1 Ag-mercaptoethanol complex 800 - 7.58 600 > E Fig.3 Cd-specific ion-exchange chromatograms of A MT-I-MT-I1 mixture (0.5 pg); B sample enriched in MT-I1 (0.5 pg); C sample enriched in MT-I (0.5 pg); and D solvent blank. Chromatograms A B and C are background (D) corrected separated two other fractions were also detected (tR = 2.22 and 3.18 min). The MT-I enriched material contained the same four fractions although this isolate was enriched in the MT-I isoform relative to the MT-I1 isoform (Table 2). By contrast the MT-I1 fraction contained only trace amounts of Cd bound to the t R = 2.22 min fraction none of the tR = 3.18 min fraction and smaller amounts of Cd bound to the MT-I isoform (16% Table 2).The identities of the t,=2.22 and 3.18 min fractions which collectively accounted for 21-2470 of the total Cd-response in the mixture and MT-I enriched fractions are not known. These fractions might represent oligomers of MT or degradation products. Although short-term Cd- Cu- and Zn-specific chromatograms of solutions of these materials were highly repeatable longer-term storage of standards in 10 mmol 1-1 TRIS buffer at 4 "C resulted in a progressive loss of signal presumably reflecting the loss of analyte metal from the protein surface. For Cd monitoring deuterium background correction was unnecessary. The Zn content of these materials was much lower than the Cd concentration and increases in the baseline signal over the course of the solvent programme were appreciably worse (Fig.4A). The Zn-response (at 213.9 nm) of the detection system was reduced appreciably by the TRIS buffer and short- term noise proved to be more severe. Efforts to circumvent the dependence of the Zn-response on the mobile phase and to minimize the short-term shot noise were only moderately successful. The response of the detector at 307.7 nm proved to be too insensitive for our purposes; however deuterium back- ground correction reduced the change in the background signal (over the course of the solvent programme) by about 50% while reducing the Zn signal by less than 5%. The Zn-response to 100 pg of the MT-I-MT-I1 mixture (Fig. 4D) indicated that Table 2 Proportions of components (YO composition as calculated by background-corrected relative peak areas from metal-specific chromotograms) in MT mixture and in fractions enriched in either MT-I or MT-I1 AAS Cd-response AAS Cu-response AAS Zn-response t,/min Mixture MT-I MT-I1 Mixture MT-I MT-I1 Mixture MT-I MT-I1 2.22 3 4 5 1 3.18 18 20 32 48 8 6 3 5.85 38 45 16 48 35 56 18 14 5 7.58 40 32 84 15 15 44 38 44 32 9.43 34 36 60 Jourrial of Analytical Atomic Spectrometry December 1996 Vol.11 1 185t 1.98 F 7.58 A 9.43 2001- j i 1:; 0 5 10 15 20 25 Time/min A 30 Fig. 4 Zn-specific ion-exchange chromatograms of A solvent blank; By MT-I enriched sample ( 125 pg); C MT-I1 enriched sample (100 pg); and D MT-I-MT-I1 mixture (125 pg). Chromatograms B C and D are background (A) corrected and for clarity chromatogram D is displaced vertically by 40 mV this analyte was bound predominantly to two fractions the MT-I1 isoform (tR = 7.58 min) and a later-eluting fraction (tR = 9.43 min) which had not been' observed in the Cd-specific chromatograms.The MT-I enriched fraction (Fig. 4B) con- tained four fractions ItR = 3.18 5.84 (MT-I) 7.58 (MT-11) and 9.43 rnin]. For the MT-I1 enriched material (Fig. 4C) more of the Zn was bound to the t,=9.47 rnin fraction than to the MT-I1 isoform (tR=7.58 min). Only trace amounts of the t R = 3.18 rnin fraction were detected and only minor amounts of Zn (5%) were present in the MT-I impurity (Table 2). Cu-specific chromatograms (Fig. 5 ) for the ion-exchange separation of these same solutions were appreciably different from the Cd- or Zn-responses.Only trace amounts of Cu were detected in any of these materials. Chromatograms A B and C represent responses to 100 pg of protein versus Cd-responses to 0.5 pg of protein in Fig. 3. Approximately 50% of the Cu in the MT mixture was associated with the MT-I isoform and a further one-third with the tR=3.18 min fraction (Fig. 5C). The MT-I1 isoform in this sample accounted for only 15% of the bound Cu. This is consistent with a previous report that 3.18 * c : 5.81 Fig. 5 Cu-specific ion-exchange chromatograms of A sample enriched in MT-I (100 pg); B sample enriched in MT-I1 (100 pg); C MT-I-MT-I1 mixture ( 1 0 0 pg); and D solvent blank Ti me/mi n Fig. 6 Ion-exchange chromatograms of Ag-labelled A MT-I-MT-I1 mixture (100 pg); B fraction enriched in MT-I1 (40 pg); C fraction enriched in MT-I (40pg); and D solvent blank.Chromatograms A B and C are background (D) corrected MT-I contains more Cu than the MT-I1 is0f0rm.l~ Interestingly for the MT-I enriched sample (Fig. 5A) about half of the bound Cu (48%) was associated with the t R = 3.25 min fraction. The MT-I1 sample (Fig. 5B) contained appre- ciably less Cu which was approximately equally distributed between the MT-I and MT-I1 isoforms. No free Cu ion was detected in any of these chromatograms. In order to explore the proposed Ag-saturation technique the mixed standard and the fractions enriched in each isoform were separately Ag-saturated using the procedure described above. The crude supernatant fraction from the haemoglobin separation was analysed without further purification by ion- exchange HPLC using the TRIS buffer-based solvent pro- gramme (Fig.6 ) . Perhaps not surprisingly there was little separation of the components under these chromatographic conditions. The size-exclusion separation of Ag-saturated mol- luscan MTs had suggested that the denaturationlcentrifugation step efficiently removed haemoglobin from the crude reaction mixture. Cd7-MT is known to bind up to 18 Ag ions and then to unwind with the loss of the tertiary structure of the polypeptide.' Additionally the addition of four units of positive charge (18 Ag versus 7 Cd/Zn) would be anticipated to change the chromatographic behaviour of the Ag-substituted prod- uct(s) and result in less anionic retention. The isoforms of Ag-saturated denatured products can be anticipated to be difficult to separate in that these polypeptide chains vary by only a single amino acid residue. In terms of operating characteristics the relatively low operating temperatures of the interface resulted in an appreci- able influence of the mobile phase on the response to analyte metals.Not only were the LODs increased by the TRIS buffer (Table l) but also changes in the mobile phase composition contributed appreciably to changes in the Zn-background signal over the course of the solvent programme (Fig. 4A) and modestly to the Cd-signal (Fig. 3D) but had no influence at the longer wavelengths used for Cu and Ag (Figs. 5 D and 6D). Only for Zn monitoring did this become a limiting factor in the current application. CONCLUSIONS Whereas the Cd in the MT-I-MT-I1 mixture was approxi- mately equally distributed between the two isoforms the Cu in this mixture was predominantly associated with the MT-I 1 186 Journal of Analytical Atomic Spectrometry December 1996 Vol.11isofonn and the Zn was predominantly associated with the MT-I1 isoform. The MT-I enriched sample contained appreci- able Zn and Cd but less Cu bound to the MT-I1 isoform (accounting for approximately 44 33 and 12% respectively of the total Zn- Cd- or Cu-response). By contrast the MT-I1 enriched sample contained appreciably less metal-bound MT-I isoform (16% of the total Cd and 5% of the total Zn but no Cu). Three other fractions were also detected among the chromatograms. Whereas the two early-eluting fractions (tR = 2.22 and 3.18 min) contained all three metals the tR = 9.43 min fraction contained only Zn.These fractions remain unidentified. Although these preliminary studies were only partially suc- cessful in identifying rapid procedures to monitor/determine the metal content associated with separate isoforms (sub- isoforms) of mammalian MTs the quartz T-tube interface provided a sensitive yet inexpensive alternative to conventional atomic absorption or emission techniques. As reported pre- v i o ~ s l y ~ ~ - ~ ~ the operation of the device was compatible with conventional flows (1 ml min-') of HPLC column eluate con- taining volatile buffer. Moreover the low LODs for a variety of metal analytes including Ag offers the possibility for monitoring metal-binding biopolymers in a variety of environ- mental media by detecting the metal@) bound to their surfaces.Whereas the commercial substrates used for these studies were saturated with respect to metal loadings isolates from environmental matrices might be undersaturated and suscep- tible to partial oxidation (during isolation) to form disulfide bridges. To monitor analyte biopolymer levels in these isolates the Ag-saturation technique (performed in the presence of 2-mercaptoethanol to reactivate metal-binding sites on the protein) seems to be a promising route. However further studies of the stoichiometry of the reaction and the chromato- graphic behaviour of the Ag-labelled products will be required. Financial support from the Natural Science and Engineering Research Council of Canada (NSERC) is gratefully acknowledged.REFERENCES 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 Hamer D. Annu. Rev. Biochem. 1986,55 913. Nath R. Kambadur R. Gulati S. Paliwal V. K. and Sharma M. CRC Rev. Food Sci. Nutr. 1988 27 41. Rauser W. E. Annu. Rev. Biochem. 1990 59 61. Bremner I. and Beattie J. H. Annu. Rev. Nutr. 1990 10 63. Methods in Enzymology ed. Riordan J. E. and Vallee B. L. Academic Press San Diego/New York 1991 vol. 205 pt. B. Robbins A. H. McRee D. E. Williamson M. Collett S . A. Xuong N. H. Furrey W. F. Wang B. C. and Stout C. D. J. Mol. Biol. 1991 221 1269. Proceedings of the Third International Conference on Metallothionein Tsukuba Japan 1992 ed. Suzuki K. T. Imura N. and Kimura M. Birkhauser Verlag Basle 1993. Stillman M. J. Coord. Chem. Rev. 1995 144,461. Kagi H. R. J. in Proceedings of the Third International Conference on Metallothionein Tsukuba Japan 1992 ed. Suzuki K. T. Imura N. and Kimura M. Birkhauser Verlag Bade 1993 Vasak M. in Methods in Enzymology ed. Riordan J. E. and Vallee B. L. Academic Press San Diego/New York 1991 Otvos J. D. Liu X. Li H. and Basti M. in Proceedings of the Third International Conference on Metallothionein Tsukuba Japan 1992 ed. Suzuki K. T. Imura N. and Kimura M. Birkhauser Verlag Basle 1993 pp. 57-74. Klaassin C. D. and Lehman-McKeeman L. D. in Methods in Enzymology ed. Riordan J. E. and Vallee B. L. Academic Press San Diego/New York 1991 vol. 205 pt. B pp. 190-198. Suzuki K. T. in Methods in Enzymology ed. Riordan J. E. and Vallee B. L. Academic Press San Diego/New York 1991 Tan Y. Momplaisir G.-M. Wang J. and Marshall W. D. J. Anal. At. Spectrom. 1994 9 1153. Momplaisir G.-M. Lei T. and Marshall W. D. Anal. Chem. 1994,66 3533. Lei T. and Marshall W. D. Appl. Organomet. Chem. 1995,9,149. Chan H. M. and Cherian M. G. Toxicologist 1993 13 165. pp. 29-55. VO~. 205 pt. B pp. 41-44. V O ~ . 205 Pt. B pp. 198-205. Paper 61064801 Received September 20 1996 Accepted September 25 1996 Journal of Analytical Atomic Spectrometry December 1996 Vol. 1 1 1 187

ISSN:0267-9477    

DOI:10.1039/JA9961101183

出版商:RSC    

年代:1996

数据来源: RSC

摘要:

Speciation of Some Metals in River Surface Water Rain and Snow and the Interactions of These Metals With Selected Soil Matrices Journal of Analytical Atomic Spectrometry J. Y . LU,* C. L. CHAKRABARTIT M. H. BACK AND A. L. R. SEKALY Ottawa-Carleton Chemistry Institute Department of Chemistry Carleton University 11 2.5 Colonel By Drive Ottawa Ontario Canada K1 S 5 B6 D. C. GREGOIRE Geological Survey of Canada 601 Booth St. Ottawa Ontario Canada K l S OE8 W. H. SCHROEDER Atmospheric Environment Service Environment Canada 490.5 Dujrerin Street Downsview Ontario Canada M3H 5 T4 Some metal species in river surface water rain and snow were characterized by size and molecular weight using filtration and ultrafiltration and by their chemical reactivity on Chelex cation-exchange resin using both column and batch techniques.Trace metal species were determined in samples of atmospheric precipitation and wet-deposition of trace metal species through runoff into the Rideau river (Ottawa Ontario Canada) was estimated by determining the concentrations of metal species in the rain water before and after contact with selected soils in the laboratory. Two types of soils Plainfield Sandy soil and Green Belt soil (Ottawa Ontario Canada) were used as models for studying the transformation of metal species from rain water to soil. Aluminium Cd Cu Ni Pb and Zn complexes in the samples of rain and snow were found to have relatively low molecular weight (< 1000). A significant fraction of the metal species in the sample of Rideau river surface water were found to have a relatively high molecular weight (> 3000).Kinetic studies of the metal speciation revealed that the metal species in the sample of snow were similar to those in the model solutions studied earlier by the present authors and were probably complexes of inorganic ligands or small organic ligands and that the metal species in the Rideau river surface water had low dissociation rates and were probably metals bound to strong binding sites of humic compounds and/or bound to or trapped in colloidal materials. Most metal species in the sample of rain water were transformed from simple small and labile species originally present in the rain water to larger non-labile species (the metals were probably bound to humic compounds and/or colloidal materials) after the rain water was equilibrated with the soils investigated.Key words Trace metals; speciation; precipitation; rain; snow; river water; Chelex cation-exchange resin; ultrajiiltration; soils; wet-deposition; inductively coupled plasma mass spectrometry; electrothermal atomic absorption spectrometry In recent years considerable attention has been focused on the deposition of toxic air pollutants in terrestrial and aquatic ecosystems where they can enter the food chain and thereby pose a potential threat to humans. The effect of acid rain on the quality of surface waters has long been recognized.'-" Rain * Present address Atmospheric Environment Service Environment Canada 4905 Dufferin Street Downsview Ontario M3H 5T4 Canada. t To whom correspondence should be addressed.is one of the major sources of input of heavy metals to surface waters. More importantly acid rain lowers the pH of soil leading to dissolution of heavy metals present in soils thereby accelerating their mobilization from soils to surface waters. Both bioavailability and toxicity are functions of the tendency of a metal to react as quantified by the free metal ion activity under pseudoequilibrium condition^.^-^ Our previous studies have shown that the metal species found in atmospheric precipitation are smaller and more labile than those in Rideau river surface Further studies are necessary on where and how metal species typical of atmospheric precipitation are transformed to the chemical/physical forms predominant in river surface waters. The chemistry of metal ions in freshwaters and soils is affected by the binding of metal ions to organic macromolecules collectively called humic compounds which are essentially a mixture of compounds of different molecular weights (MWs).Humic compounds are fractionated into several fractions which include humic acid and fulvic acid which are defined operationally by the method of fractionation. The general terms humate humic acid or humic compounds are used to designate all fractions humic fulvic and hydrophilic acids with the understanding that a differentiation is made between humic acid and fulvic acid by calling the fraction of higher MW humic acid and the fraction of lower MW fulvic acid. However these two acids should properly be viewed as two 'extremes' of a continuum of compounds rather than two different types of compounds.Humic compounds are homolo- gous complexants which have three unique properties. They are (i) polyfunctional; (ii) polyelectrolytes (oligoelectrolytes); and (iii) can undergo conformational changes resulting from electrostatic interactions among the various functional groups on a single molecule which may make the coordination properties of the molecule highly dependent on the extent of cation binding and on the ionic strength of the solution. All these properties strongly affect the thermodynamic stability and kinetic lability of their metal humates.' Humic compounds are also heterogeneous which makes the determination of thermodynamic stability and kinetic lability of their metal complexes difficult and subject to uncertainties.Kinetic studies of metal speciation can not only differentiate chemical species according to their kinetic parameters but can also give information on the distribution of the chemical species in the system at any time during the kinetic process. The results obtained can therefore be used to estimate the bioavailability of a metal if the kinetic model can be con- structed to represent the process of biological uptake. Chelex Journal of Analytical Atomic Spectrometry December 1996 Vol. 11 (1 189-1 201) 1 189100 cation exchange resin which has been used by others"-15 for metal speciation was used in this study. Since the Chelex- batch-labile fraction of cadmium and its toxicity in salmon have been found to be well correlated,I6 further investigation of the Chelex batch technique seems justified.This work forms a part of our program for developing a comprehensive scheme of metal speciation based on size frac- tionation of dissolved metals by ultrafiltration and character- ization of each size fraction by its chemical reactivity as manifested by the dissociation kinetics of its metal com- p l e ~ e s . ~ ~ ~ ~ - ~ ~ The objective of this research is to characterize the trace metal species in river surface waters rain and snow by (i) size using filtration through a 0.45 pm filter and ultrafiltration; ( i i ) by their chemical reactivity at Chelex cation- exchange resin sites using both the column and the batch technique; and ( i i i ) studying transformation of the metal species in samples of rain and snow after they have been in contact with samples of two selected soils in the laboratory.Since homologous complexants such as humic compounds are ubiquitous in freshwaters and soils and because their polyfunc- tionality polyelectrolyte nature and conformational changes play an important role on metal complexation fate and cycling in the aquatic environment,' a fulvic acid (which accounts for most of the dissolved humic compounds in river surface waters) was used as a model complexant for Al Cu Ni Cr Pb and Cd. A variety of analytical techniques and methods have been used for determining trace metal speciation in natural waters.24 In general some of these techniques use direct detection and measurement of a particular trace metal species without pre- vious ~ e p a r a t i o n ; ~ ~ some separate the metal species prior to their determination;26 some differentiate the metal species based on their kinetic properties e.g.the rate of dis- s o ~ i a t i o n . ' ~ - ~ ~ ~ ~ - ~ ~ Th ese techniques and methods have been widely applied to the determination of speciation of Cd Cu Pb and Zn in natural Simultaneous kinetic analysis of multicomponent systems is well e ~ t a b l i s h e d ~ ~ . ~ ~ * ~ ' and has been widely used. Kinetic analysis applied to metal speciation has been reported by several a ~ t h o r s . ~ ~ ~ ~ ~ ~ * ~ ~ - ~ ~ ~ ~ The model for kinetic studies of metal speciation using the Chelex batch technique has been described elsewhere,23 hence only an outline of this model and the mathematical equations derived are given below.THEORY Kinetic Model Following the kinetic model proposed by Olson et a1.35*36 we have developed the following kinetic model to study the dissociation kinetics of ML where M is a metal ion and L is a macromolecular polyfunctional complexant such as a humic compound having binding sites of different chemical affinities (the charges on M and L have been omitted for simplicity). In the complexant L the metal M is bound to multiple sites in L all of which dissociate independently and simultaneously at a rate that depends on the nature of the functional group its position on the macromolecules and the residual charge. The dissociation is taken to be first-order or pseudo-first-order. Consider an aqueous mixture of n components in which each component designated MLi exists in equilibrium with its dissociation products k k-1 MLi M+Li (slow) (1) where k and k - are the rate constants for the forward and the backward reaction.In the Method of Chemical Competition with a Solid Phase,42 in which we have used Chelex 100 as the competing cation-exchange resin Chelex reacts with M as follows k2 k - 2 Chelex+M Chelex-M (fast) (2) We now make two assumptions ( i ) that reaction ( 2 ) is much faster than reaction (1); and (ii) that [Chelex] >>[MI. Because [Chelex] is large and [Chelex]>>[M] [Chelex] can be con- sidered as constant and reaction (2) as pseudo-first-order. Since k is large as has been determined by ICP-MS,8v17 and with Chelex in sufficient excess the condition k [Chelex] >>k- [ Li] holds and the overall reaction Chelex + MLi +Chelex-M + Li (3) is assumed irreversible.If the rate limiting step in reaction (3) is assumed to be the dissociation of MLi then (4) The two assumptions made above have been validated as follows the first assumption by our previous studies using ICP-MS7*8*17*23 and the second assumption by taking a large excess of Chelex 100 resin over the very low concentrations of metal ions present in the test samples. If each complex ML undergoes independently and simul- taneously a first-order or pseudo-first-order dissociation reac- tion the sum of the concentrations of all components remaining in the solution at time t can be described as n C(t)= 1 C exp(-kit) ( 5 ) i = l where Cio is the initial concentration of ML the ith compo- nent.The sum of the total concentration remaining C(t) is measured using ICP-MS by monitoring the loss of metal ion over time.7,8,'7,23 EXPERIMENTAL Reagents Chelex 100 cation-exchange resin (Bio-Rad 100-200 mesh sodium form) was equilibrated with NaOAc-HOAc buffer at pH 5.0. Because dryness of the resin affects its properties and changes the pH of the sample during kinetic measurements the resin was kept in the above buffer solution until the time of the kinetic measurements when it was separated from the buffer solution. The NaOAc-HOAc buffer was prepared as follows 59.0 ml of pure concentrated acetic acid (A.C.S. Anachemia Ville St. Pierre Canada) was added to about 11 of ultrapure water. The pH of this solution was adjusted to 5.0 using 6 moll-' NaOH (reagent grade Fisher Scientific) and the solution was diluted to 2 1 with ultrapure water.The Laurentian soil fulvic acid supplied by Dr. D. S. Gamble Agriculture Canada Ottawa Canada was prepared by him following known procedure^^^.^^ from a sample of a podzol collected from the Laurentian forest preserve of Lava1 University Quebec Canada. Ultrafiltration of this fulvic acid revealed that about 37% had MW>30000 38% had MW < 1 0 0 0 ~ 15% had MW 1000-5000 and the rest ( N loo/,) had MW 5000-30000.8 This fulvic acid was used without any further treatment. The Plainfield Sandy Soil and the Green Belt Soil used in this study were also supplied by Dr. D. S. Gamble. The composition and properties of these soils46 are listed in Table 1. ICP-MS-2 Metal Standard (Delta Scientific) containing Al Cd Cr Cu Fe Ni Pb Zn and other metals (10 mg 1-l each) was used.Stock solutions each containing a single metal (1000 pg m1-l) of Cd Cu Ni and Pb were prepared by dissolving an appropriate quantity of CdO (Analyzed reagent J. T. Baker Philipsburgh NJ USA) copper metal (99.9% 1 190 Journal of Analytical Atomic Spectrometry December 1996 Vol. 11Table 1 Composition and properties of the soils used in this work Green Belt Plagioclase (% of soil) 45 Quartz (YO of soil) 15 Microcline (% of soil) 8 Clay (Y of soil) 27 Cation exchange capacity (meq per 8.8 100 g soil) Organic carbon (YO) 0.8 Final pH of the batch experiment* 5.8 Weight of metals in soil (pg g-' of NA soil)* Plainfield Sandy Minor Dominant Minor 7 NAt 0.7 6.5 A1 19530 Cu770 Ni 2240 Cr 3500 * Measured in this laboratory.NA-not available. pure) nickel metal powder (Spex Industries Metuchen NJ USA 99.999%) and Pb(NO,) (Fisher Scientific A. C. S. Reagent Grade) in nitric acid with heating and diluting the solutions to the appropriate volume with ultrapure water; the final solutions contained 1 % (v/v) ultrapure HNO,. Stock solution (lOOOpgml-') of A1 was prepared by dissolving 0.5000g of A1 powder (Spex) in 20ml of HC1 (ULTREX- brand)-H,O (1 l ) with heating and diluting the solution to 500ml with ultrapure water. A standard solution (1000 pg ml-l) of Zn was purchased from Merck Chemicals (Poole Dorset UK). Stock solution of Cr (100 pg ml- ') was prepared by dissolving 0.1923 g of CrO in water acidifying the solution with 10 ml of concentrated HNO and diluting to 1OOOml with ultrapure water.Stock solution of Fe (100 pg ml-I) was prepared by dissolving 0.100 g of pure iron wire in a mixture of 10 ml of aqueous 50% HC1 and 3 ml of concentrated HNO with heating followed by the addition of 5ml of concentrated HNO and dilution to 1000ml with ultrapure water. Ultrapure water of resistivity 18.2 MIR cm was obtained direct from an ultrapure water system Milli-Q-Plus (Milford MA USA). Screw-cap bottles (Teflon and polyethylene) were used as reactors and containers for storage of samples and reagents. These bottles were pre-cleaned following the pro- cedure described in our previous paper.2o The reactors (made of Teflon) were equilibrated with the test solutions before making the kinetic measurements.Samples To test the usefulness of the Chelex batch technique for metal speciation a series of model solutions was prepared by spiking ultrapure water with an appropriate volume of the following solutions metal standard fulvic acid (0.1000 g 1-I). Since both the speciation of the metals and the ionization of the COOH functional group in the Chelex molecule depend on the pH the pH must be held constant. The pH of both the model solutions and the test solutions described below was main- tained at 5.0f0.5 so that this pH was close to the pH of the samples of rain and snow (which ranged from 3.4 to 5.4) and also allowed adequate ionization of the COOH functional group of the Chelex molecule for proper functioning of the Chelex cation-exchange resin.The pH (measured using a Fisher Scientific Accumet 925 pH/ion meter) of all model solutions was adjusted with dilute HN03 or dilute NaOH solutions as appropriate and the test solutions were equilibrated over- night. The pH (5.0 & 0.5) was confirmed after equilibration. Ultrafiltration was then carried out and the fractions collected were set aside at least overnight for equilibration before the kinetic measurements were carried out. A 5 1 sample of Rideau river surface water (pH 8.0) was collected from a site at Carleton University using a pre-cleaned polyethylene water sampler. Immediately after collection the sample was filtered through a 0.45 pm (Gelman Ann Arbor MI USA) filter. The particulate phase retained by the filter was subjected to high-pressure-bomb acid digestion2' and its metal contents determined by ETAAS.The filtrate which contained the dissolved phase was subjected to fractionation by ultrafiltration and the ultrafiltered fractions were then passed through the Chelex column each fraction separately and the metals in each fraction quantified by ETAAS. Since the concentrations of the metals of interest in the Rideau river surface water were too low for the kinetic study the sample of the river surface water (unfiltered) for the kinetic study was spiked with standard solutions of Al Cd Cry Cu Fe Ni Pb and Zn to contain a few pg 1-' of each metal and the pH of the spiked sample was adjusted to 5.0 & 0.1. After equilibrating for 2d the spiked unfiltered sample was filtered through a 0.45 pm filter to separate the particulate matter.The filtrate was set aside for 1 d for further equilibration and was then subjected to ultrafiltration for fractionation of the species in the dissolved phase. These ultrafiltrated fractions were equilib- rated at least overnight before the kinetic measurements were carried out. The pH was confirmed after the equilibration. A sample of snow was collected with a pre-cleaned polyethyl- ene sample-collector at a site on the roof-top of the Chemistry Building at Carleton University. The sample was held in the sample-collector at room temperature until the snow melted. The snow-melt was then immediately filtered through a 0.45 pm filter. The particulate matter retained by the filter was subjected to high-pressure-bomb acid digestion2' and its metals content determined by ETAAS.The filtrate from the above filtration was used for determining the metal speciation in the dissolved phase. The concentrations of major ions in the dissolved phase of the snow sample were Na (34 pg l-') Ca (383 pg 1-I) and Mg (26 pg 1-I). The snow sample for the kinetic study of metal speciation was prepared as follows the snow-melt (filtered through a 0.45 pm filter) was spiked with standard solutions of Al Cd Cr Cu Fe Ni Pb and Zn to contain a few pg I-' of each metal. The pH of the spiked test solution was adjusted to 5.0 & 0.1. The spiked test solution was equilibrated overnight and was then filtered through a 0.45 pm filter. The filtrate was ultrafiltered and the ultrafiltered fractions were equilibrated at least overnight before the kinetic measurements were carried out.Experimental Procedure Fig. 1 shows a multiple-criterion fractionation scheme devel- oped for this work. Determination of the total concentration of metals in the sample in the particulate phase and in the dissolved phase of the samples Analysis of the digested particulate phase and the filtrate by ETAAS gave the metal concentrations in the particulate phase and in the dissolved phase respectively. The sum of these two concentrations gave the total concentrations of metals in the water sample. The total concentrations of metals in the unfil- tered samples were also independently determined by analysing the digested unfiltered sample. The results from the two methods agreed well. Determination of metal speciation in the dissolved phase The filtrate from the 0.45 pm filter was used for determining metal speciation employing the following techniques ultrafiltration Chelex 100 column and Chelex 100 batch.Ultrafiltration was carried out using an Amicon (Lexington MA USA) ultrafiltration stirred cell model 8200 equipped Jozirnal of Analytical Atomic Spectrometry December 1996 Vol. 11 11910,45 pin FILTER I 1 PARTICULATE PHASE - I DISSOLVPD PHASE I SOIL BATCH FILTRATION AND ULTRAFILTRATION WET ACID BOMB DIGESTION + GFAAS I I ULTRAFILTERED I FRACTIONS CHELEX COLUMN CHELEX-BATCH + GFAAS + ICP-MS I CHELEX COLUMN EACH FRACTION + GFAAS KINETICALLY DISTINGUISHABLE COMPONENTS Fig. 1 Multiple-criterion fractionation scheme with an Amicon disc membrane filter (diameter of 62mm).When filtering a sample the first 10% and the last 50% of the ultrafiltrate were d i s ~ a r d e d . ~ ~ . ~ ~ In the Chelex column technique the water sample was passed through a Chelex 100 column (id 0.8 cm containing 5 g of Chelex 100 resin) at a flow rate of 2 ml min-'. The eluate which contained the Chelex-column-non-labile species was collected and the Chelex-column-non-labile species were determined by ETAAS. Subtraction of the Chelex-column- non-labile species fraction from the total dissolved metal species fraction gave the Chelex-column-labile metal species fraction. In the Chelex batch technique appropriate amounts of pre- treated7 Chelex 100 cation-exchange resin ( 1 % m/m) were added to the dissolved phase (the filtrate from the 0.45 pm filter) of the sample. The mixture was stirred with a Teflon- coated magnetic stirring bar.The kinetics of dissociation of the metal complexes in the sample were determined using ICP-MS following the procedure given in our earlier paper^.^.^^ The instrumental conditions and the data acqui- sition protocol for the ICP-MS procedure are listed in our earlier paper^.^.^^ The experimental data are analysed using the Iterative Deconvolution Method.23 The pH of the sample was monitored throughout the Chelex batch procedure. The Chelex batch technique had been validated and the opti- mum experimental conditions established earlier using model solutions.8 The present study was done using the optimum experimental conditions given in our earlier paper^.^.^ The soil batch technique An appropriate amount of the soils (5 g 1-') was added to the dissolved phase of the rain water sample. This mixture was equilibrated using a mechanical shaker.At pre-selected inter- vals the shaker was stopped and the mixture was centrifuged filtered and then subjected to fractionation by ultrafiltration. Each ultrafiltered fraction was passed through the Chelex 100 column. The concentrations of metal species in the filtrates from 0.45 pm filters the ultrafiltered fractions and the eluates from the Chelex 100 column were determined by ETAAS. RESULTS AND DISCUSSION Since pH (because of its effect on ionization of polyelectrolytes and the ligand acid and also because of its effect on metal hydrolysis) and dissolved organic carbon (DOC) (because of complexation reactions with metals) have strong effects on metal speciation and since conductivity reflects the ionic concentration of the dissolved solute they are important parameters of water quality. Fig.2 shows the pH and the conductivity of the samples of Rideau river surface water rain and snow collected between April 1992 and March 1994. Table 2 lists the DOC concentrations of some of these samples. The results show that the rain and snow samples collected over this two year period were all acidic with pH values ranging from 3.4 to 5.3 for rain and 4.7 to 5.4 for snow samples. By comparison the pH of Rideau river surface waters was remarkably constant at 8.0k0.1 over these two years because of the buffering action brought about by a large number of co-existing phases in the rocky environment of Rideau river and the equilibrium partial pressure of C 0 2 in the atmosphere which exerted a chemostatic control in steady-state systems such as river waters.Conductivity values for the samples of Rideau river surface water were much higher than those for the samples of rain and snow because of the higher concen- trations of charged inorganic and organic species in the Rideau 14 e I " 200 .g 150 5 6 4 2 50 0 n n 100 Snow Rain . River wate; " Fig. 2 The pH (filled) and conductivity (open) of Rideau river surface water rain and snow Table 2 Dissolved organic carbon (DOC) concentrations in samples of Rideau river surface water rain and snow Date of sampling Sample type DOC/pg ml- November 15 1993 Rain 3.4 December 12 1993 Rain 1.6 February 24 1994 Snow 1 .o February 25 1994 Snow 0.7 March 10 1994 Snow 1.1 March 10 1994 Rideau river surface water 6.6 1 192 Journal of Analytical Atomic Spectrometry December 1996 Vol.11river surface water. The concentration of DOC in the sample of Rideau river surface water was also considerably higher than that in the samples of rain and snow. Chelex Column Technique Table 3 shows the metal species in the samples of rain water collected on 3 d in August 1992. The total concentrations of Al Cu and Ni in the rain water were highest on the first day. Most of the dissolved species of the metals except Cr were Chelex- column-labile. The Chelex-column-non-labile Cr species were probably negatively charged CrV' species (e.g. Cr04'-) which being repelled by the negatively charged (ionized) functional group COOH of the Chelex 100 cation exchange resin were not taken up.Table4 shows the metal species determined in the sample of snow collected in April 1992. Compared with the results for the sample of rain collected in August of the same year as shown in Table 3 the total concentrations and the percentages of Al"' and Ni" species in the particulate phase in the sample of snow were much higher than those in the sample of rain water. Between 56-100% of the Cu" species in the sample of rain water but only 20% of the Cu" species in the sample of snow were in the dissolved phase. The results of ultrafil- tration show that all Al"' species in the sample of snow had MW < 5000 and 74% of them had MW < 500. All Al"' species having M W > 500 were Chelex-column-non-labile but only 29% having MW < 500 were Chelex-column-non-labile.The Chelex-column-non-labile Al"' species were probably colloidal Fe"' species and Al"' species strongly bound to or trapped in colloidal material. The Chelex-column-labile species were prob- ably Al-aquo -hydroxo and other inorganic complexes. All dissolved Cu" species had MWG 10000 of which 489'0 had MW G 500. All Cu" species having MW > 500 and 60% of the Cu" species having MW < 500 were Chelex-column-labile. The total concentration of Fe in the sample of snow was 345 pg l-' of which only 3% was in the dissolved phase. All the dissolved Fe species had MW < 500 and 29% of them were Chelex- column-labile. The Chelex-column-labile Fe species were prob- ably inorganic Fe complexes and complexes of small organic complexants.The Chelex-column-non-labile Fe species were probably colloidal Fe"' species and Fe"' species strongly bound to or trapped in colloidal material. Similar to the distribution of the A1 species all dissolved Ni species had MW <55(w)O of which 63% had MW < 500. Fifty-eight percent of the smallest fraction of the Ni species (MWx500) was Chelex-column- non-labile. The smallest MW (<500) fractions of the Al Ni and Cu species were probably complexes formed by inorganic complexants such as carbonates bicarbonates sulfates hal- ides nitrates 0x0-and -hydroxo complexes and/or complexes of small organic complexants. Table4 also shows that in the dissolved phase the largest size fraction in which the metal species are present in the snow sample follows the order of Fe"<Ni"<Cu" which is the natural order of complex stability called the Irving-Williams series of stability.48 The largest size fraction (MW 5000-10000) of the Cu species and the size fraction (MW 500-5000) of the Cu and of the Ni species were probably complexes of relatively large organic complexants whereas the only size fraction (MW<500) of the Fe species present in the sample were probably formed by inorganic complexants such as carbonates bicarbonates sulfates halides and nitrates and/or complexes of small organic complexants. Table 5 shows that A1 and Cu species were present in the sample of Rideau river surface water.Comparison of the results presented in Tables 4 and 5 reveals that significant amounts of A1 and Cu species had larger MWs in the sample of Rideau river surface water suggesting that they were complexes of large organic complexants such as humic compounds and/or were bound to colloidal material.Sixty per cent. of the dissolved A1 species in the sample of Rideau river surface water but none in the sample of snow had MWb5000. Seventy per cent. of the dissolved species of Cu in the sample of Rideau river surface water but none in the sample of snow had MW 2 10 000. The percentages of the Chelex-column- labile A1 species were almost the same 54% of the dissolved species in both the samples of Rideau river surface water and snow. But all of them in the sample of snow and only a small amount of them in the sample of Rideau river surface water had MW < 500.For Cu the percentage of the Chelex-column- labile species was smaller in the sample of Rideau river surface water than in the sample of snow. Similar to Al a large percentage of the Chelex-column-labile Cu species in the sample of snow but only a small percentage in the sample of Rideau river surface water had MW<500. Since Chelex- column-labile complexes are moderately labile species the Chelex-column-non-labile fractions also include slowly labile metal complexes (Chelex-batch-labile complexes) and the metal bound to colloidal material and inert and very stable complexes. Table 3 Metal species in the samples of rain water collected iii August 1992 and determined by ETAAS [Metal species]/pg 1-' (%)* August 28 1992 (pH 4 .7 F Total In the particulate phase In the dissolved phase Chelex-column-labile Chelex-column-non-labile August 29 1992 (pH 3.4)- Total In the particulate phase In the dissolved phase Chelex-column-labile Chelex-column-non-labile August 31 1992 (pH 3.8)- In the dissolved phase Chelex-column-labile Chelex-column-non-labile A1 53.6 (100.0) 37.4 (69.8) 16.2 (30.2) 14.2 2.00 46.0 (100.0) 34.2 (74.3) 11.8 (25.7) 11.8 0.00 10.1 8.36 1.65 c 11 42.1 ( 100.0) 18.6 (44.2) 23.6 (45.8) 20.5 3.04 8.57 (100.0) 8.57 (100.0) 7.84 0.73 0.00 (0.0) 6.10 5.30 0.80 Ni 36.8 (100.0) 9.02 (24.5) 27.8 (75.5) 26.3 1.55 10.5 (100.0) 6.46 (61.5) 4.04 (39.5) 1.66 2.38 13.8 13.0 0.75 Cr 2.05 (100.0) 0.84 (41.0) 1.21 (59.0) 0.61 0.60 1.38 (100.0) 1.24 (89.8) 0.14 (10.2) 0.09 0.05 3.22 1.38 1.84 Pb -t -t -t -t -t 351.96 (100.0) 0.27 (13.4) 1.69 (86.6) 1.69 0.00 -t -7 -t Cd -t -t -t -t -t 0.067 (100.0) 0.014 (20.9) 0.053 (79.1) 0.05 1 0.002 -t -t -t * n = 3 standard deviation < 0.5%.Not determined. Journal of Analytical Atomic Spectrometry December 1996 Vol. 1 1 1 193Table 4 Metal species in the sample of snow collected on April 12 1992 (pH 4.7) [Metal species] f standard deviation/pg 1- ' (YO)* Element Fraction A1 Total metal In the particulate phase In the soluble phase Molecular weight > 5000 500-5000 < 500 Total metal 601 f 18 (100.0) 594+ 18 (98.8) 7.0 f 0.8 ( 1.2) Chelex-column-labile Chelex-column-non-labile 3.8 f 1.0 3.2 k 0.2 ND 1.8 & 1.2 5.2 f 0.4 ND ND 3.7 +_ 0.4 ND 1.8 f 0.2 1.4k0.2 c u Total metal In the particulate phase In the soluble phase Molecular weight >50000 10 000-50 000 5000- 10 OOO 1000-5000 50&1000 < 500 24.6 f 0.4 ( 100.0) 19.8 k0.6 (80.5) 4.8 f0.2 (19.5) 4.0 f 0.4 0.8 f0.2 ND ND 0.9 f 0.4 0.9 f 0.2 1.0 f0.3 2.0 f 0.5 ND ND 0.9 f 0.2 0.9 f 0.2 1.0f0.3 1.2f0.6 ND ND ND ND ND 0.8 f 0.1 345 f 12 (100.0) 336 f 13 (97.0) 9.0f 1.0 (2.6) Fe Total metal In the particulate phase In the soluble phase Molecular weight > 50 000 10 000-50 OOO 5000- 10 000 10W5000 500-1000 < 500 2.6 f 0.8 6.4 5 1 .I ND ND ND ND ND 9.0 & 0.6 ND ND ND ND ND 2.6 +_ 0.8 ND ND ND ND ND 6.4 f 0.3 Ni Total metal In the particulate phase In the sohble phase Molecular weight > 50 000 10 000-50 000 5000- 10 000 50&5000 < 500 94 & 26 ( 100.0) 89 f 27 (94.7) 5.2 f 0.3 (5.3) 2.4 f 0.8 2.8 f 0.7 ND ND ND 1.9 k 0.4 3.3 f0.5 ND ND ND 1.0 & 0.6 1.4 & 0.5 ND ND ND 0.9 f 0.2 1.9 f 0.0 * n=3.ND-not detectable. Table 5 Metal species in the sample of Rideau river surface water collected on February 26 1992 (pH 8.0) [Metal species] f standard deviation/pg 1-' (YO)* Total metal 98f2 (100.0) 77 f 1 (78.6) 21 f 1 (21.4) Chelex-column-labile Chelex-column-nonlabile Element Fraction A1 Total metal In the particulate phase In the dissolved phase Molecular weight >100000 500&100 OOO 500-5000 < 500 11.31frl.8 9.4 f 0.4 9.9 f 0.7 2.8 & 0.3 5.8 f 0.8 2.2 f 0.3 3.4 f 0.4 1.7k0.3 5.7 f 0.9 0.5f0.1 6.5 f 0.8 1.1 f0.2 0.1 fO.O 1.7f0.3 c u Total metal In the particulate phase In the dissolved phase Molecular weight > 100 000 10 OOO-100 000 500-10 000 < 500 9.5 f 1.4 (100.0) 0.9 & 0.0 (9.5) 8.6f 1.4 (90.5) 4.8 & 1.8 3.8 f0.6 2.3 + 0.9 2.7 f 1.0 1.4 f 0.7 2.2 f 0.2 1.7 k 0.6 2.4f 1.0 0.6 & 0.2 0.2 f 1.0 0.6 f 0.2 0.3 f O .l 0.8 f 0.4 2.0 & 0.2 The value shown after f in the above tables are standard deviations of 3 successive replicate determinations. * n=3. 1 194 Journal of Analytical Atomic Spectrometry December 1996 Vol. 11Chelex Batch Technique Fig. 3 shows the uptake of Al"' by the Chelex resin from size fractionation of (a) the model solution containing metal ions and fulvic acid; (b) the sample of snow; (c) the sample of Rideau river surface water. The overall rates of uptake of Al"' from the sample of snow were initially the fastest and those from the sample of Rideau river surface water were the slowest. The overall rate of uptake can be affected either by the presence of different metal species and/or by the distribution of these metal species.The kinetic parameters of Al"' species in the various size fractions of the model solution the sample of snow and the sample of Rideau river surface water are listed in Tables 6-8 respectively. Table 6 shows the concentrations of Al"' species in the various size fractions of the sample 39 47 and 14% of the dissolved species of Al'" had MW<3000 3OOo-lOo OOO and > 100 000 respectively. Three kinetically distinguishable components of Al"' as shown in Table 6 were resolved for the model solution. The fastest Component had a rate constant similar to that of Al"' aquo ion7 and probably consisted of Al"' aquo ions and other very labile species of Al"'.The other two kinetically distinguishable components were probably Al"' bound to binding sites of different affinities " 0 0 500 lo00 1500 2Ooo 2500 in the fulvic acid. The similarity in the value of the dissociation rate constants in the different size fractions of each kinetically distinguishable component suggests that as a result of chemical and steric differences in their neighbouring groups coordi- nation sites on the fulvic acid may well have a continuous range of affinities for metals.49 Table 7 shows that in the sample of snow 80% of all dissolved species of Al"' had MW < 1OOO. Two kinetically distinguishable components of Al"' were resolved for the sample of snow.The faster component prob- ably consisted of Al"' aquo ions and other very labile species of Al"'. The slower component had a very small dissociation rate constant ( s-') and because most of the dissolved species of Al"' had MWs < 1O00 this very slow component probably consisted of Al"' bound to or trapped in colloidal material and/or Al"' polynuclear 0x0- and -hydroxo species. Table 8 shows only one kinetically distinguishable component of AIIn present in the sample of Rideau river surface water. Because this component had a very small dissociation rate constant and relatively large MW it was probably Al"' bound to strong (high affinity) humate binding sites and Al"' species in colloidal state and/or bound to or trapped in colloidal material. various size fractions of a.model solution b. snow and c. Rideau river surface water. Fig. 3 Uptake of aluminum by the Chelex 100 resin from Orfiltrate-from 0.45 pm filter; 0 MW<100000; 0 MW<10000; 0 MW<3000; and A MW<1000' Table 6 Kinetically distinguishable components of metals in different size fractions of a model solution* containing metal ions and fulvic acid (pH 5.0) Element A1 Cd c u Ni Pb Zn Fraction [M],,,,/pmoll-' Clo ('YO) k x s - ' CZo(%) k x s - l C3'(%) Dissolved species MW<100000 MW<3000 Dissolved species MW<100000 MW<3000 Dissolved species MW<100000 MW<3000 Dissolved species MW<100000 MW<3000 Dissolved species MW<100000 MW<3000 Dissolved species MW<l00000 MW<3000 1.5 1.3 0.59 8.2 x lo-' 7.9 x 7.4 x lo- 1.5 x lo-' 1.4 x lo-' 1.1 x lo-' 1.2 x lo-' 1.2 x lo-' 1.1 x lo-' 4.2 x lo-' 4.0 x lo-' 3.6 x lo- 1.2 x lo-' 1.2 x lo-' 1.2 x lo-' 16+35 22+ 36 36+ 26 99f 3 9 9 f 3 9 9 f 3 67f.5 67f 5 6 8 f 4 96+3 99+4 9 8 f 5 9 6 f 2 9 5 f 2 9 5 f 2 9 9 f 3 9 9 f 3 95f 3 1.5k0.6 1.0 f 0.3 2.7 f 0.7 2.3 + 0.0 2.9 + 0.0 2.5 + 0.0 1.8 f 0.06 2.2f0.1 2.1 f 0.07 2.5 f 0.09 2.8 f 0.0 2.9 f 0.04 2.0 & 0.0 2.6 f 0.0 2.4 f 0.0 2.3 + 0.05 2.8 f 0.02 2.6 + 0.04 18f27 2.7 f 0.8 28 f 46 1.2f0.6 28 + 27 4.1 k0.8 16f 11 3.6 f 0.3 16f 11 3.5 & 0.4 1 4 f 9 3.3 f 0.3 66f 11 49f41 36f 14 1 7 f 4 17f 5 1834 k3 x 10-41 s-1 1.3 kO.1 0.5 f0.5 1.3f0.2 0.54 f 0.07 0.25 f 0.08 0.14f0.08 * [MI = {[Al"'] = 1.5) + { [Cd"] =0.099} + {[Cu"] =0.16f + {[Ni"] =0.17} + {[Pb"] =0.048} + { [Zn"] =0.15} =2.1 p o l 1 - I ; [FA] [MI = 1.2.C:= the initial percentage of the ith kinetically distinguishable component. Values after k signs are standard deviations of non-linear regression analysis. Since the total uncertainties of the analytical method including that of the regression analysis are greater than the values of the standard deviations shown above only two significant figures in the values of the rate constant are justified. Journal of Analytical Atomic Spectrometry December 1996 Vol. 11 1 195Table 7 Kinetically distinguishable components of metals in the sample of snow* (pH 5.0f0.5) ~ Element Species A1 Dissolved MW<10000 MW < 1000 clo (%) 43f5 52f6 54f7 k x s-' l.7 f 0.08 :I .7 f 0.07 2.1 f O . l C20 (%) k x lop3/ s-' C30 (%) 57f2 48f3 46f3 k3 x s-l 0.72 & 0.2 0.95 1 0.4 1.2 & 0.4 Cd Dissolved MW<lOOOO MW < 1000 0.12 0.10 0.10 99 f 4 98 f 4 98f4 2.5 & 0.05 2.1 f 0.04 2.4 & 0.05 cu Dissolved MW < 10000 MW < 1000 0.57 0.49 0.49 83 f 2 72f7 72f7 2.7k0.1 2.4 & 0.1 2.6f0.1 10f7 3.6 f 0.2 7+4 18f 11 3.7 f 0.3 10&6 20 f 26 6.6 f 1 8 f 6 1.2 f 0.1 1.1 kO.1 1.010.6 Ni Dissolved MW<10000 MW < 1000 0.99 0.55 0.55 98k4 96f4 96f4 2.6 f 0.04 2.0 & 0.04 2.4 f 0.03 90f2 86f3 8 8 f 2 5 f 5 3.8 k 0.2 4&2 8f 10 5.8 k 0.4 6 f 2 8+9 3.7 & 0.3 4+3 Pb Dissolved MW<10000 MW < 1000 0.13 0.13 0.13 3.1 f0.03 3.0 f 0.05 3.0 f 0.03 2.8 f 0.7 1.4k0.4 0.74 f 0.7 97f4 94&5 94f5 2.6 f 0.05 :1.7f0.05 2.5 f 0.0 Zn Dissolved MW < 10 000 MW < 1000 0.63 0.45 0.44 * C =the initial percentage of the ith kinetically distinguishable component.Values after f signs are standard deviations of non-linear regression analysis. Since the total uncertainties of the analytical method including that of the regression analysis are greater than the values of the standard deviations shown above only two significant figures in the values of the rate constant are justified. Table 8 Kinetically distinguishable components of metals in the sample of Rideau river surface water (pH 5.0f0.5) Element Fraction A1 Dissolved species MW < 100 000 MW < 3000 c,o (%) kl x lo-,/ s-' C; (%) k x S - ' C3O ("/o) k3 x S - ' * * - - 100 4.2 f 0.03 100 4.0 f 0.03 0.96 0.95 0.71 Cd Dissolved species MW<100000 M W < 3000 0.67 0.67 0.63 * - 96f3 92+9 -* 1.9 f 0.2 1.7 k 0.5 Cu Dissolved species MW< 100000 MW < 3000 0.37 0.37 0.29 * - 0.42 f 0.01 0.27 f 0.01 * - 0.78 f 0.02 0.59 f 0.03 * - 22f3 25+6 * - 5.4 f 0.2 4.5 f 0.3 * - 44f3 46f5 * - 34f 3 29f6 Ni Dissolved species MW < 100 000 M W < 3000 0.16 0.16 0.16 * - 64f3 54f7 -* 1.1 k0.03 0.28 k 0.02 * - 28+2 32+9 * - 1.2 f 0.02 0.74 f 0.05 - * 8 f 2 14f8 * - 9.7 f 0.2 7.7 f 0.5 Pb Dissolved species MW<100000 MW < 3000 0.28 0.26 0.26 * - 86f5 87 f 4 -* 0.49 f 0.01 0.49 f 0.01 * - 6f 13 6120 * - 1.6 f 0.09 1.2 & 0.02 * - 8 f 4 7+3 * - 4.3k0.1 4.4 f 0.2 Zn Dissolved species MW<100000 MW < 3000 0.13 0.13 0.13 * - 99f2 100f9 * - 2.8 k 0.02 1.7 f 0.4 * Not measured.C = the initial percentage of the ith kinetically distinguishable component. Values after f signs are standard deviations of non-linear regression analysis.Since the total uncertainties of the analytical method including that of the regression analysis are greater than the values of the standard deviations shown above only two significant figures in the values of the rate constant are justified. Fig. 4 shows the uptake of Cu" by the Chelex 100 resin from the size fractionations of (a) the model solution containing metal ions and fulvic acid; (b) the sample of snow; and (c) the sample of Rideau river surface water. There was no significant difference in terms of kinetic behaviour for the Cu" species in different size fractions of the model solution. The following difference in the overall rate of Cu" uptake in different size fractions of the samples of snow and Rideau river surface water was observed. The curves (a) (b) and (c) reveal that the overall rate of Cu" uptake from the sample of snow was the fastest and that from the sample of Rideau river surface water was the slowest.This was the same behaviour as was observed for Al"'. The results from our data analy- sis using the Iterative Deconvolution Method23 are listed in Tables 6-8 which show that 73 and 78% of the dis- solved Cu" species in the model solution and the sample of Rideau river surface water respectively had MW < 3000 whereas 86% of the dissolved Cu" species in the sample of snow had MW < 1000. Three kinetically distinguishable com- ponents of Cu" species were resolved for all the samples tested; the fastest component had a dissociation rate constant similar to that for the uptake of Cu" aquo ions by the Chelex 100 resin,8 suggesting the presence of Cu" aquo ions and/or other very labile Cu" species.Considering the fact that most of the dissolved species of Cu" in the samples of rain and snow had low MW the slowest component which accounted for a relatively small percentage was probably Cu" species in col- loidal state and/or bound to or trapped in colloidal material. 1 196 Journal of Analytical Atomic Spectrometry December 1996 Vol. 11C Fig. 4 Uptake of copper by the Chelex 100 resin from size fractions of a model solution; b snow; and c Rideau river surface water. 0 filtrate from 0.45 pm filter; 0 MW<100000 0 MW<10000; 0 MW<3000; A MW<1000. Plots for the uptake of Cd" Ni" Pb" and Zn" by the Chelex 100 resin from the various size fractions of the model solution the samples of snow and Rideau river surface water are not shown here as they are similar to those of Cu" species preqented above. Fig.5 illustrates the Iterative Deconvolution Met hodz3 for analysing kinetic data for Pb" species having MW < 100000 in the sample of Rideau river surface water. An increase in the number of components from 1 to 3 resulted in a significant improvement in the distribution of the weighted residuals. Further increase in this number to 4 caused no further improve- ment in the distribution. The number of kinetically distinguish- able components of Pb" having MW < 100000 in the sample of Rideau river surface water was therefore 3. The results of this kinetic analysis are listed in Tables 6-8.The size distri- butions of these metals were similar to that of Cu" species i.e. most of dissolved species of these metals had low MWs in all the samples tested. One kinetically distinguishable component was resolved for the Cd" and the Zn" species in all samples with dissociation rate constants comparable to those for the uptake of the Cd" and the Zn" aquo ion^,^,^ probably suggest- ing the presence of metal aquo ions and very labile spezies of Cd" and Zn". One kinetically distinguishable component of Ni" was resolved in the model solution and the sample of snow and three in the sample of Rideau river surface water. The slower components in the sample of Rideau river surface water were probably Nil' species in colloidal state and/or bound to or trapped in colloidal material.One kinetically distinguishable component of Pb" species in the model solution and tbree in both the samples of snow and of Rideau river surface water were resolved. The dissociation rate constants for the kin- 80 40 - u o !I 4 0 x 40 r 80 2 E O CJ) .- 4 0 80 40 0 4 0 -80 0 2000 4000 6OOO 8000 loo00 Time/s Fig.5 Plot of weighted residuals as a function of time for kinetic analysis of the uptake of Pb" by Chelex 100 resin from Rideau river surface water. Bottom trace 1-component-fitting; middle trace 2-component-fitting; and top trace 3-component-fitting. etically distinguishable component in the model solution and the fastest component in the sample of snow were comparable to that for the uptake of Pb" aquo ions by the Chelex 100 resin.The dissociation rate constant for the fastest component of Pb" species in the sample of Rideau river surface water was about one order of magnitude smaller than that for the fastest component of Pb" species in the model solution and in the sample of snow; probably the Pb" was bound to large organic complexants and/or was in colloidal state and/or bound to or trapped in colloidal material. Table 6 presents the kinetically distinguishable components of a mixture of six metals found in different size fractions of the model solution containing these metal ions and fulvic acid at pH 5.0. The ionic strength of the mixture was entirely due to these metal ions fulvic acid and the acid added to the mixture to adjust the pH to 5.0. The dissociation rate constants reported in Table 6 are weighted averages of many similar rate constants measured within the analytical window of the analytical method the weighting factors are the relative abundances of the binding sites involved in the kinetic measurement.Dissociation rates of complexes formed with homologous complexants such as humic compounds can be considered a decreasing continuous function of the meta1:ligand ratio. The binding sites on humic compounds are occupied by metals in order of decreasing strength (affinity) of the binding sites. Because of the polyfunc- tionality and polyelectrolyte nature of fulvic acid the thermo- dynamic stability of its complexes with metal ions is determined by the sum of the chemical free energy and the coulombic free energy of the complex formation reaction.49 These metals are all competing for the binding sites (both strong and weak) of fulvic acid and the metal that forms the thermodynamically strongest complex with a given binding site will fill that binding site.Of these six metals A1 and Cu form the strongest complexes with fulvic It should be noted that A1 is present in the model solution as the largest component of the mixture of the six trace metals giving it an advantage of the dominant metal mass to drive the complexation reaction by mass action. Of these six trace metals only A1 forms a trivalent ion and the coulombic energy for a trivalent A13+ is consider- ably higher than that for the other metals which form divalent metal ions. Fulvic acid has strong (high-affinity) binding sites which are much less abundant (-1%) and are called minor binding sites and weak binding (low-affinity) sites which are much more abundant ( ~ 9 9 % ) and are called major binding sites.The strong (high-affinity) binding sites are filled first by A1 and Cu forming stable complexes; the weak binding sites are filled forming weak complexes only after the strong (high- affinity) binding sites have been filled. The strong complexes will be less labile than the weak complexes as can be seen from the following rea~oning.~' Rate constants for the complex Journal of Analytical Atomic Spectrometry December f996 Vol. I 1 1197dissociation can be related to the complex formation rate constant through detailed balancing kf M+L ML kd where kf and kd are formation rate constant and dissociation rate constant respectively of the metal complex ML and K is the stability constant.If the formation rate constant for a reaction of a single metal with a series of ligands (in this case the major and the minor binding site of fulvic acid) show little variation being mostly dependent on the rate constant for water exchange of the metal the dissociation rate constant should show an inverse correlation with the stability constant K.50 Table 6 confirms the above trend. Aluminium has the largest fraction (component C,') forming a very-weakly-labile (kdz10-4 s-') complex whereas Cd Ni Pb and Zn are mostly and Cu is over two-thirds distributed in the more labile fraction (C,'). However the generally high affinity of fulvic acid binding sites for Cu binding49 is reflected in the slowest component (C,') i.e.-18% of the total copper is bound to the strong (high-affinity) binding sites. Cadmium Ni Pb and Zn have each only one kinetically distinguishable component which is expected to be a relatively weak complex of fulvic acid and hence by the above reasoning is found to be relatively more labile. From the distribution of the MWs of various ultrafiltered fractions of the fulvic acid given earlier it can be concluded that fulvic acid can form metal complexes of various sizes. In Table 6 the dissolved metals of the ultrafiltered fractions both above and below MW 100000 are probably metal complexes of the higher MW fractions of fulvic acid whereas those below MW 3000 are probably metal complexes of the lower MW fractions of fulvic acid and metal-aquo complexes.It is interes- ting to note that whereas for A1 only about 40% of the dissolved phase is in the low MW fractions most or almost all of the dissolved phase of the other metals is in the low MW fractions. The above unusual characteristics of A1 compared with these other metals are probably the result of its being bound to and/or trapped in large colloidal materials and/or its ability to form very strong complexes with fulvic acid and/or its ability to form large polynuclear aquo complexes. Table 7 presents the kinetically distinguishable components in a sample of snow (pH 5.0). Eighty per cent. of the dissolved phase of A1 was lower than MW 1000 and was divided almost equally between a slowly-dissociating fraction (C,') and a moderately labile fraction (C,').These fractions were probably a mixture of various Al-aquo complexes and A1 complexes of inorganic ligands such as sulfates phosphates nitrates hal- ides 0x0- and -hydroxo complexes and/or complexes of small organic ligands. The dissolved fractions of Cd Ni and Zn were mostly lower than MW 1000; the metal species in these fractions were moderately labile and were probably aquo complexes and complexes of inorganic ligands and small organic ligands as described above for Al. The dissolved fractions of Cu and Pb species mostly consisted of species of lower than MW 1000 and were moderately labile species. These metal species were probably aquo complexes and com- plexes of inorganic ligands,and small organic ligands as described above for Al.Polyelectrolyte eflects of humic compounds49 Humic acid includes macromolecular compounds and because these macromolecular compounds are relatively large their polyelectrolyte effects are much stronger than those of the relatively small fulvic acid. As shown in Table 2 the concen- tration of DOC (the majority of which is due to humic compounds) is far greater in Rideau river surface water than in rain and snow (if one disregards the sample of rain collected on November 15 1993 which shows an anomalously large concentration of DOC). Humic acid is a polyelectrolyte and forms polyanions. Coulombic forces are effective over relatively large distances and affect the free energies of ions even in dilute solutions such as unpolluted river water.Humic acid is a weak acid and pH has a dramatic effect on the extent of binding by humic acid as in the case of any weak acid ligand whose apparent affinity for a metal increases with pH. This effect is exacerbated by the increase in coulombic attraction of the metal ion to the binding sites as the humates become deprotonated with increasing pH. When a cation reacts with an acidic functional group of a molecule with multiple func- tional groups (a polyanion such as a humic acid) the chemical free energy of interaction is augmented by the long-range coulombic attraction emanating from all the neighbouring non-reacting negative sites. At relatively high pH this polyelec- trolyte effect can strengthen greatly the binding of metal ions by polyanions.The change in free energy of reaction (AGO) between a metal M"+ and the functional group L"- M"+ +Lm- 'LMn-m) 7 is the sum of the chemical free energy change (&hemo) due to the reaction with the functional group itself and the cou- lombic free energy change (AGcoulo) due to electrostatic inter- actions between M"' and all the charges on the p ~ l y a n i o n ~ ~ AGO = AGchem + AGcoul O Copper is one of the most reactive metals in the aquatic environment. In freshwaters copper is 99.9% bound to hum- ates of which 93% is bound to malonate-type ligand in the large humic Both the high affinity of the humates for copper and its (large and pH-dependent) sensitivity to ionic strength (see Table 9) are due to polyelectrolyte effects of humic compounds. At the low CU, LT ratios present in the sample of Rideau river surface water (where CU and LT are the total Cu and total humic acids respectively) the larger humates (i.e.the larger humic acid fractions) whose copper affinities are mostly enhanced by coulombic attraction account for the bulk of copper binding. The metal speciation in the Rideau river surface water sample (Table 8) was significantly different from that in the snow sample (Table 7). In Table 8 the entire dissolved fraction of A1 (C,') was a relatively inert fraction (dissociation rate constant s-') suggesting that the A1 was probably bound to high-affinity humate sites i.e. forming strong com- plexes with humic acid in the Rideau river surface water. About 74% of the dissolved fraction of A1 was MW<3000 suggesting that the metal species were probably A1 complexes of small fulvic acid inorganic complexants and small organic complexants.Probably because the strong (high-affinity) bind- ing sites humic acid in the Rideau river surface water were filled by A1 which formed strong complexes with these humate sites,50 not many of the high-affinity humate sites were left unoccupied for complexation by the other trace metals. However the competition among the other trace metals prob- ably favoured Cu which is known to form strong complexes with the high-affinity humate sites.49 In Table 8 the entire dissolved fractions of Ni Pb and Zn were species of MW < 3000 suggesting that these metal species too were probably complexes of small fulvic acid inorganic complexants and small organic complexants.The fact that almost the entire dissolved fraction of Cd and Zn (C,') were moderately labile (rate constant -lop2 s-') was probably a reflection of the fact that they form weak complexes with weak (low-affinity) humate binding sites and following the same reasoning pre- sented earlier are more labile than the relatively inert com- plexes formed by strong (high-affinity) humate binding sites 1 198 Journal of Analytical Atomic Spectrometry December 1996 Vol. 11(C30 fractions having dissociation rate constant % s-I). These relatively inert complexes may also be a result of the restrained diffusion of M and ML in and out of the gel-like structure of of humic acid macromolecules( where M is the metal L is the complexant and ML is the metal complex). Fig.6 shows Al"' leached from the Green Belt and Plainfield Sandy soils as a function of time. Under the same experimental conditions more Al"' was leached from the Green Belt soil than from the Plainfield Sandy soil. This difference might have arisen from differences in the soil composition and in the equilibrium pH values. The lower pH in the batch experiment with the Green Belt soil (Table 1) probably resulted in more Al"' being leached from the soil. Fig. 7 shows the changes in the concentrations of metals in the sample of rain water as a function of time after it has remained in contact with the Plainfield Sandy soil. The curve for Pb showed almost a continuous increase in the Pb leached from the soil with increasing contact time. Cadmium was initially adsorbed by but was later leached from the soil.Within the experimental uncertainty there was no significant leaching of Ni from the soil nor adsorption of Ni by the soil after a period of 2 h. Chromium was initially leached rapidly followed by a slow leaching. Copper was adsorbed by the soil from the sample of rain water in the first few minutes after its contact with the soil but after some time of contact a slow leaching occurred. The results of the metal speciation experiments in the sample of rain water before and after contact with the Plainfield Sandy soil and the Green Belt soil are presented in Tables 9 and 10 respectively. Tables 9 and 10 show the effect of the rain water 400 pi 0 7 3 0 0 - c *. B 0 loo L o o o o O C 0 50 100150200250300350400450500550 Time / min Fig. 6 Aluminum leached from the soils by rain (pH 3.8) as a function of time. [soil] = 5 g 1-'.A Green Belt soil; B Plainfield Sandy soil. 7.0 7 ~:~~ 4.0 3.0 ~- 2 ~ 3 . 6 ~ ~ - - ~ 0 . ~ 2 ~ 2.0 2.0 2 1.0 - 3.2 0.038 2.8 0.034 2.4 0.030 2.0 0.026 1.6 0.022 0 40 80 120 160 0 40 80 120 160 Ti me/m i n Fig. 7 Concentrations of the dissolved metal species in rain (pH 3.4) as a function of contact time with Plainfield Sandy soil ( 5 g 1- ') sample before (pH 3.4 and 3.9 respectively) and after (pH 6.5 and 5.8 respectively) the rain water samples were equilibrated with the above soil samples. The reason for the observed increase in the pH of the rain water after equilibration with the soils was that the soils functioned as cation exchangers exchanging the soil cations for the hydrogen ions of the samples of rain water thereby increasing their pH.That the concentration of the Chelex-column-labile Cu species (i.e. the moderately labile Cu species) in the rain water sample was reduced from the initial 4.12 to O.OOpgl-' meant that the non-lability of the Cu species increased dramatically after contact of the rain water sample with the soil. It has been reported in the literature that water-leaching of soil largely carries away fulvic acid and relatively little humic acid.51 The above decrease suggests that the Cu in the rain water sample with its low ionic strength and at pH 6.5 forms strong com- plexes with fulvic acid which by the reasoning given earlier are non-labile. Because the coulombic attraction factor in the effective stability constant of the copper humate complex at pH 6.5 (the maximum ionic effect is observed at pH 7) is inversely proportional to the ionic strength the effective affinity of the fulvic acid sites for Cu binding increases dramatically at the low ionic strength of the rain water sample.49 The other metals Al Ni Cr Pb and Cd followed the same trend as that of Cu though the concentrations of Chelex-column-labile A1 and Pb complexes did not decrease to zero.The above results are different from those of the Rideau river surface water sample at pH 5 (Table 8) which because of its much higher ionic strength did not show a preponderance of Chelex- column-non-labile metal complexes except for A1 (100% of the A1 formed strong complexes with humic compounds which were non-labile because of the reasoning given earlier).Contrary to the results shown in Table9 the Cd and Zn complexes in Table 8 were found to be Chelex-column-labile. In Tables 9 and 10 substantial fractions of the metal species in the sample of rain water before its contact with the soils had MW -c 500 and were Chelex-column-labile whereas most of the metal species in the sample of rain water after its contact with the soils had MW>500 and were Chelex-column-non- labile. These results suggest that the metal species of MW < 500 in the rain water before its contact with the soils were probably metal-aquo ions and complexes of either inorganic com- plexants and/or of small organic complexants and the metal complexes of MW>500 after its contact with the soils were probably complexes of relatively larger organic complexants (e.g.soil fulvic acid) metals in the colloidal state and/or bound to or trapped in colloidal materials derived from the soils. The observed increase in the concentrations of the Chelex-column- non-labile metal species after the rain water came in contact with the soils was probably due to the metals being bound to or trapped in colloidal materials derived from the soils. CONCLUSIONS The most significant finding of this research project is that metals such as Al Cu Ni and Pb at the low metal ligand ratio present in unpolluted freshwaters form relatively inert complexes (dissociation rate constant N lo-' s- ') with humic compounds as a result of the strong (high-affinity) humate binding sites binding the metal ions.Table 8 shows that -100% of the dissolved A1 species have a dissociation rate constant of =lO-'s-' whereas only about 22-25% of the dissolved Cu species have a dissociation rate constant N s-'I. Nickel and Pb show a similar trend but to a much lesser extent a reflection of their lower intrinsic complexation ability i.e. these metals do not form strong complexes with humic compounds. The above observation of the relative non-lability of metal complexes in unpolluted freshwaters in which the concentrations of the metals are very low (< lo-' mol 1-') is Journal of Analytical Atomic Spectrometry December 1996 Vol. 1 1 1 199Table 9 contact with the Planfield Sandy soil sample (contact time 36 h) Metal species in rain water samples* A Dissolved phase; B ultrafiltered fraction having MWc500 before (pH 3.4) and after (pH 6.5) Element A1 c u Ni Cr Pb Cd Before After Before After Before After Before After Before After Before After (A) Total dissolved phase/pg 1-' (B) Ultrafiltered fraction having MW < 500/pg 1- ' Chelex-column-labile 5.85 8.99 x lo-' 4.32 0.00 1.25 0.00 9.00 x lo-' 0.00 1.47 6.40 x lo-' 3.10 x lo-' 0.00 Chelex-column-non-labile 0.00 3.01 x lo-' 3.60 x lo-' 1.80 x lo-' 1.80 2.26 5.00 x lo-' 2.8 1 0.00 1.03 0.00 8.10 x Chelex-column-labile - 1.60 3.28t 0.00 0.00 0.00 7.80 x 1.30 x lo-' 3.10 x lop2+ 2.20 x lo-' - - Chelex-column-non-labile - 2.00 0.00 1.27 2.83 5.10 x lo-' 2.10 x lo-' - - - - - * Collected on August 29 1992.Total concentration of the metal species having MW < 500.Table 10 Metal species in rain water samples* A Dissolved phase; B ultrafiltered fraction having MW < 500 before (pH 3.9) and after (pH 5.8) contact with the Green Belt soil sample (contact time 17.5 h) (A) Total dissolved phase/pg 1-' (B) Ultrafiltered fraction having MW < 500/pg I-' Element A1 Before c u Before Ni Before Cr Before After After After After Chelex-column-labile 5.89 8.25 2.84 3.13 x lo-' 1 .oo 0.00 0.00 0.00 Chelex-column-non-labile 5.71 8.25 x lo-' 2.44 6.45 x lo-' 2.38 3.70 x lo-' 3.43 3.60 x 10 Chelex-column-labile 3.00 x lo-' 4.58 1.04 5.63 0.00 0.00 0.00 0.00 Chelex-column-non-la bile 4.03 4.17 3.72 4.00 3.28 3.70 x lo-' 3.43 2.50 x lo-' * Collected on October 16 1992. contrary to the reported formation of labile complexes in the l i t e r a t ~ r e ~ ~ .~ ~ which was based on the study of model solutions containing relatively high metal ligand (fulvic acid) ratios. The results from the soil batch experiments which revealed changes in the concentration and speciation of metals in the rain water after it was equilibrated with the soils suggest that the larger metal complexes and the slowly-dissociating kinetically dis- tinguishable components of the metals in the dissolved phase of the Rideau river surface water were (probably at least partially) derived from surface runoff that was associated with rain events. Another significant finding is that labile metal species in a low-ionic-strength medium such as in rain water after equilibration with soils form strong complexes with soil fulvic acid with the result that the previously labile metal complexes in rain water samples become non-labile after rain water comes in contact with soils. The authors are grateful to the Natural Sciences and Engineering Research Council of Canada and Environment Canada Atmospheric Environment Service for financial sup- port of this project.REFERENCES 1 Schofield C. L. Acid Rain Proceedings of ASCE National Convention American Society of Civil Engineers New York 2 Schindler D. W. Can. J. Fish. Aquat. Sci. 1980 37 373. 3 Johnson D. W. Water Air Soil Pollut. 1981 16 243. 4 Anderson M. A. and Morel F. M. M. Limnol. Oceanogr. 1982 27 789. 5 Sunda W. G. and Guillard R. R. J. Mar. Res. 1976 34 511. 6 Anderson D. M. and Morel F. M. M. Limnol. Oceanogr.1978 23 283. 1979 pp. 55-69. ' 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 Lu Y. Chakrabarti C. L. Back M. H. Gregoire D. C. and Schroeder W. H. Anal. Chim. Acta 1994 293 95. Chakrabarti C. L. Lu Y. Gregoire D. C. Back M. H. and Schroeder W. H. Environ. Sci. Technol. 1994 28 1957. Buffle J Complexation Reactions in Aquatic Systems An Analytical Approach Ellis Horwood Chichester 1988 pp. 354-358,380-383. Pai S. C. Whung P. Y. and Lai R. L. Anal. Chim. Acta 1988 211 257. De Mora S. J. and Harrison R. M. Anal. Chim. Acta 1983 153 307. Liu Y. and Ingle. J. D. Jr. Anal. Chem. 1989 61 525. Figura P. and McDuffie B. Anal. Chem. 1979 51 120. Figura P. and McDuffie B. Anal. Chem. 1980 52 1433. Campbell P. G. C. Bisson M. Bougie R. Tessier A. and Villeneuvve J.-P.Anal. Chem. 1983 55 2246. Buckley J. A. Yoshida G. A. Wells N. R. and Aquino R. T. Water Res. 1985 19 1549. Lu Y. Chakrabarti C. L. Back M. H. and Schroeder W. H. Intern. J. Environ. Anal. Chem. 1995 60 313. Chakrabarti C. L. Cheng J. Lee Wai Fai Back M. H. and Schroeder W. H. Environ. Sci. Technol. 1996,30 1245. Cheng J. Chakrabarti C. L. Back M. H. and Schroeder W. H. Anal. Chim. Acta 1994 288 141. Chakrabarti C. L. Lu Y. Cheng J. Back M. H. and Schroeder W. H. Anal. Chim. Acta 1993 267 47. Chakrabarti C. L. Lu Y. Cheng J. Gregoire D. C. Back M. H. and Schroeder W. H. Proceedings of International Conference on Heavy Metals in the Environment (Toronto Canada Chakrabarti C. L. Cheng J. Lu Y. Back M. H. and Schroeder W. H. Proceedings of International Conference on Heavy Metals in the Environment Toronto Canada 1993 vol. 2 pp. 227-230. Lu Y. Chakrabarti C. L. Back M. H. Gregoire D. C. Schroeder W. H. Szabo A. G. and Bramall L. Anal. Chim. Acta 1994 288 131. Trace Element Speciation Analytical Methods and Problems ed. Batley G. E. CRC Press Inc. Boca Raton Florida 1989 1993) V O ~ . 1 pp. 242-245. pp. 43-1 16. 1200 Journal of Analytical Atomic Spectrometry December 1996 Vol. 1125 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 Florence T. M. Analyst 1986 111 489. Jandik P. Capillary Electrophoresis of Small Molecules and Zons VCH New York 1993. Pardue H. L. Anal. Chim. Acta 1989 216 69. Willis R. G. Woodruff W. H. Frysinger J. R. Margerum D. W. and Pardue H. L. Anal. Chem. 1970 42 1350. Schechter I. Anal. Chem. 1992 64 727. Mottola H. A. Kinetic Aspects of Analytical Chemistry John Wiley and Sons New York 1988 pp. 122-146. Martinotte W. Queirazza G. Guarioni A. and Mori G.. Anal. Chim. Acta 1995 305 183. Reddy K. J. Wang L. and Gloss S . P. Plant Soil 1995 171 53. Groschner M. and Appriou P. Anal. Chim. Acta 1994 297 369. Mach M. H. Water Air Soil Pollut. 1996 90 269. Olson D. L. and Shuman M. S. Anal. Chem. 1983 55 1103. Olson D. L. and Shuman M. S. Geochim. Cosmochim. Acta 1985 49 1371. Lavigne J. A. Langford C. H. and Mak M. K. S. Anal. Chem. 1983,59,2616. Langford C . H. and Gutzman D. W. Anal. Chim. Acta 1992 256 183. Schechter I. Anal. Chem. 1991 63 1303. Larsson J. A. and Pardue H. L. Anal. Chim. Acta 1989,224,289. Laios I. Fast D. M. and Pardue H. L. Anal. Chim. Acta. 1986 180,429. 42 43 44 45 46 47 48 49 50 51 Buffle J. Complexation Reactions in Aquatic Systems An Analytical Approach Ellis Horwood Chichester 1988 pp. 600-602. Griffith S. M. and Schnitzer M. Soil Sci. 1975 120 126. Schnitzer M. and Skinner S. I. M. Soil Sci. 1968 105 392. Li J. Langford C. H. and Gamble D. S. J. Agri. Food Chem. in the press. Cathum S. J. Gamble D. S. Kodama H. and Bowman B. T. personal communication. Wheeler J. R. Limnol. Oceanogr. 1976 21 846. Irving H. and Williams R. J. P. Nature 1948 162 146; J. Chem. SOC. 1953 3192. Morel F. M. M. and Hering J. G. Principles and Applications of Aquatic Chemistry John Wiley & Sons New York 1993 Morel F. M. M. and Hering J. G. Principles and Applications of Aquatic Chemistry John Wiley & Sons New York 1993 Reuter J..H. and Perdue E. M. Geochem. Cosmochim. Acta 1977 41 325. pp. 378-391. pp. 378-401. Paper 6106481 G Received September 20 1996 Accepted October 18 1996 Journal of Analytical Atomic Spectrometry December 1996 Vol. 1 1 1201

ISSN:0267-9477    

DOI:10.1039/JA9961101189

出版商:RSC    

年代:1996

数据来源: RSC

摘要:

Investigations Into Chromium Speciation by Electrospray Mass Spectrometry IAN I. STEWART AND GARY HORLICK Department of Chemistry University of Alberta Edmonton Alber :a Canada T6G 2G2 Journal of Analytical Atomic Spectrometry Chromium is an element whose toxicity is oxidation state dependent. Chromium (VI) is considered toxic to humans whereas C P is considered to be an essential nutrient. It therefore becomes important to be able to distinguish between the two oxidation states in solution. In this study it is shown that electrospray mass spectrometry (ESMS) has the potential to provide useful information on chromium species in solution samples. The various solution forms of both oxidation states may be observed directly and are presented. In addition other more uncommon species of both oxidation states such as polymeric C p and the polyanion trichromate are discussed.Finally the ability of ESMS to monitor other aspects of the solution chemistry of both chromium oxidation states is explored. Keywords Speciation; chromium; electrospray ; mass spectrometry From a toxicological standpoint it is well known that different species of the same element will follow different metabolic pathways. Thus the toxicity of an element can and will depend on its speciation in a sample. Speciation as a term is one yet to be wholly defined; however in the literature it is commonly referred to as the specific form (monatomic or molecular) or configuration in which an element can occur. The element chromium is one of many elements that fall into this category where its toxicity is species dependent.In solution chromium primarily exists in either the Cr"' or CrV' oxidation states. For the most part chromium(m) is non toxic and even considered to be an essential nutrient for most mammals. Chromium(v1) on the other hand is a powerful oxidizing agent and is considered to be quite toxic. There are of course multiple species possible for each oxidation state which are usually dependent on such factors as matrix concentration and pH. The determination of chromium has been performed by numerous flame and plasma spectroscopic techniques with detection limits ranging from the ppb to the sub-ppb levels.' The most sensitive technique to date is however ICP-MS with detection limits for the most recent commercial instruments being reported in the 0.001-0.1 ppb range.2-4 ICP-AES typi- cally exhibits higher detection limits.However in recent reports on the 'end-on' or axially viewed plasma by Ivaldi and Tyson' detection limits as low as 0.068 ppb using ultrasonic nebuliz- ation were reported. The only drawback to these techniques is that when they are used 'alone' they provide only the total chromium content and speciation information is lost. Therefore when speciation is critical other methods must be incorporated. One solution to this problem is through the incorporation of chromatographic methods. A variety of separation tech- niques have been coupled to the ICP such as HPLC',6 ion chromatography ( IC),7 and capillary electrophoresis (CE)8 for example. It should be pointed out that the list of coupling ICP with chromatographic techniques and others is quite extensive and dates back to the 1970s.Mention should also be made of coupled FAAS measurements by Sperling et a1.' who report ppb to sub-ppb detection limits for both Cr"' and CrV' species. The goal of these techniques is obviously not to provide complete separation and subsequent determination as differ- entiation is only required between same element species. Therefore rapid separation times may be achieved at the cost of chromatographic resolution. However in order to properly realize the high sensitivity that techniques such as ICP-MS or ICP-AES are capable of the interface must be designed to provide a means to yield complete nebulization/atomization with a high transport efficiency into the plasma. For the most part these techniques provide rapid determinations at the ppb level for Cr"' and CrV1 species.The speciation in the above cases tends simply to differentiate between the two oxidation states with potentially ambiguous information provided on the actual chemical species present. It would therefore be advantageous to determine specifically the species present in solutions. Recently the use of electrospray mass spectrometry (ESMS) for elemental and speciation work has been investigated by Agnes and c ~ - w o r k e r s ~ ~ - ~ ~ and a brief review on the subject of the electrospray of inorganic solutions has been given by Stewart and Horlick." The detection limits reported for a variety of elements seem encour- aging and have been reported in the ppb to sub-ppb 1e~els.I~ In terms of speciation the technique shows great promise as solution information tends to be preserved rather than destroyed during analysis.Evidence has however been pre- sented where in special cases species may undergo electrochemi- cal reactions at the solution capillary interface thus creating new ions.16 In general however electrospray is a relatively simple ion source that does not typically create ions but merely takes pre-existing ions in solution and transfers them to the gas phase. This fact is critical as information on a particular target analyte such as oxidation state and molecular form will be preserved. Investigations of chromium solution species in both the Cr"' and CrV1 states will be provided with some fundamental discussion of the processes that are involved in the generation and sampling of these ions.BACKGROUND The electrostatic spraying of liquids is by no means a new phenomenon researchers have been studying it since the turn of the century.17 It was not however until Yamashita and Fennl8,l9 reported the results of coupling ES to MS in the 1980s that its potential as an analytical tool was realized. Since then the field has literally exploded with the majority of the focus being applied to biological studies. Its use as a probe for inorganic solution species has not taken off as fast. As stated earlier the primary focus in the past was on total element analysis and it was not until recently that researchers have been commenting on the need to focus more on species dependent determinations.Also evidence in the literature to suggest that ES is capable of achieving the same levels of detection is scarce. Electrospray has been described extensively in the litera- t ~ r e ~ ~ - ~ ~ and so only a qualitative description will be given here. Simply put ES is an electrostatic spraying process where a liquid surface becomes disrupted when an intense electric field is applied resulting in a spray of charged droplets. In ES a solution of some minimum conductivity (typically a meth- anolic electrolyte solution in the order of 10-5-10-3 moll-') Journal of Analyrical Atomic Spectrometry December 1996 Vol. 11 (I 203-1 21 4 ) 1203is pumped ( N 1-10 pl min-') through a stainless steel capillary (1 100 pm id) which is held at some high potential relative to a counter electrode (the mass spectrometer sampling orifice).The capillary to counter electrode separation is usually between 0.5-3.0 cm and the potential difference in this region is usually 2:2-4 kV. This spraying phenomenon at the capillary tip affords the transfer of solution ions to the gas phase where they may be sampled by a mass spectrometer. A negatively biased capillary results in negatively charged droplets and ultimately negative gas phase ions the opposite is true for a positively biased capillary. The formation of the spray the charging of the droplets and formation of gas phase ions from such droplets are described in some detail in the literature. In particular a review article by Kebarle and Tang28 summarizes most of the current thought on the processes involved in ESMS. EXPERIMENTAL Instrumentation The electrospray source and mass spectrometer used have been previously described." The high pressure sampling region of the modified Perkin-Elmer SCIEX ELAN Model 250 ICP-MS instrument is based on a sampling plate-skimmer configur- ation.The mass spectrometer was further modified by removing the shadow or first photon stop in the ion optics as well as the second stop located in the Bessel box. This led to an increase in sensitivity of one to one and a half orders of magnitude. As both cations and anions were being studied the mass spectrometer was operated in both positive and negative ion modes. In the negative ion mode the ES tip was typically biased at - 3000 V the front plate at - 600 V the sampling plate was varied to minimize or maximize collisionally induced dissociation (CID) in the high pressure region (typically between -2 and -60 V) and the skimmer was held at -2 V.For positive ion mode the potentials were of opposite polarity with the sampling plate typically run at higher potentials (5-100 V). As the potential on the skimmer was held constant the potential drop and hence the total collisional energy (being a function of the potential difference) were proportional to the sampling plate voltage (see ref. 29 for discussion). The curtain gas used for these experiments was nitrogen at a flow rate of about 1.3 1 min-'. The ES needle tip had a 100 pm id and was operated at a flow rate of 1.0-2.0 pl min-l via a syringe pump (Harvard Apparatus 22).The tip position although optimized for each set of experiments was usually set 5 mm from the front plate and 0-1 mm off axis. It was found that careful selection of flow rate (usually a low flow rate) and applied capillary potential led to stable signals without the use of discharge suppressing gases for negative ion mode. Measurements of pH were taken by a Fisher Accumet Mini pH Meter Model 955. The probe (Corning general purpose combination) was standardized through pH4 7 or 10 buffer solutions depending on the range studied. UV/VIS absorption measurements were acquired on a Hewlett-Packard 84542A Diode Array Spectrometer. The cell had a 1.0cm path length. Absorption spectra were acquired with 1 s integration times over the wavelength range of 190-820 nm.Reagents and Solutions All solutions were prepared by dissolving the ACS grade salts in distilled deionized water to form a stock solution except where specified. Aliquots of the stock solution were then diluted with HPLC-grade methanol or HPLC-grade aceto- nitrile to the desired concentration. These procedures resulted in the solutions being primarly methanol or acetonitrile with water content in the range of 0.5-2.0% by volume depending on the sample. RESULTS AND DISCUSSION Chromium ( III ) The chemistry of chromium(rr1) is quite extensive as it may exist in a wide number of complexes all of which with little exception are hexacoordinate. The chemistry of Cr"' as a metal salt dissolved in aqueous solutions is not as simple or uninteres- ting as one might initially anticipate.The aqua ion is quite acidic and will readily hydrolyse in solution. This in itself is not unique as many other transition metals such as iron and vanadium hydrolyse in aqueous solutions. Chromium is unique however in its kinetic inertness where ligand exchange in the inner coordination sphere is quite slow and can have half lives in the range of several hours.30 These attributes lead to some quite complex solution chemistry. When chromium is dissolved in aqueous solution it readily hydrolyses to an extent which is determined by its pKa values:31 Under certain conditions (of concentration pH and tempera- ture) the hydrolysis products can polymerize to yield dimers trimers tetramers etc. For example one scheme for forming the dimer is given below by eqn.(3) CrOH2+ + CrOH2+ +CT~(OH),~+ (3) The above reaction has a reported rate constant of about k =2.0 x 1 mol-I s-' under experimental condition^.^^ The above dimer also has an approximate stability constant These systems have been studied quite extensively by Marty and c o - w o r k e r ~ ~ ~ - ~ ~ and also Rai et have studied aqueous chromium chemistry. There is still some debate as to what extent polynuclear chromium species are present especially in systems of low total chromium content. In most natural water samples or samples collected under neutral to acidic pHs chromium would probably only be present in dilute concen- trations and so would exist primarily as the aqua [CI-(H,O)~~+] or hydroxo species [Cr(OH)(H20)52f]. For the most part the normal working concentrations for ES are between and lo-' mol I-'.Chromium(n1 ) Solutions The ES mass spectra of chromium(n1) solutions may vary in complexity. Electrospray when operated in the stable cone-jet mode will ultimately produce gas phase ions independent of the sampling process. It is the sampling interface that allows some flexibility in obtaining either simplified or complex mass spectra. The use of a sampling plate skimmer configuration similar to that of most early commercial ICP-MS instruments to sample the atmospheric pressure gas phase ions allows for this. The use of this configuration allows for efficient high pressure CID which has been discussed with respect to the ES of metal and so only a cursory discussion will be given here.Simply put the 100pm orifice in the sampling plate entrains both curtain gas and gas phase ions into the jet expansion region immediately behind the sampling plate (nozzle). Under stable conditions the gas phase ions (usually 1204 Journal of Analytical Atomic Spectrometry December 1996 VoZ. 1 Isolvated) drift i g immediately in front of the sampling plate are entrained in L. steady or constant manner. It is presimed that under stable conditions (constant applied capillary and front plate potential solution flow rate etc.) the species distribution of these 'drift ions' is static and is unaffected by slight changes on the sampling plate. However a potential drop between the sampling plate and skimmer accelerates the entrained ions relative to the neutral curtain gas (N2) which ultimately produces energetic collisions the energy of which is a function of the potential drop between the sampling plate and skimmer.The result of the energetic collisions as recorded by the mass spectrometer is the result of collisional frequency with the background gas (which is a decreasing function and is directly proportional to the number density) and acceler- ational energy due to mean free path (which is inhersely proportional to the number density). The net effect is hat a large potential difference affords complete stripping of a sol- vated metal ion down to its bare form whereas a small potential drop results in minimal stripping and therefore preserves solution information. A mass spectrum of a chromium solution is given in :ig.1. Here a relatively mature solution of chromium chloride (aged 2 5 months) was diluted with methanol (-6 ppm) an 1 run under conditions that resulted in very efficient CID (i.e.. large potential difference between sampling plate and skimm :r) by ESMS. Similar to an ICP mass spectrum Fig. 1 consist i of a 'bare' or atomic chromium isotope packet signal cente ed at -m/z 52 with an associated oxide peak at -m/z 66 The MO' M + ratio in this case is 2 2 % . This ratio is sanipling condition dependent and may be made lesser or @*eater depending on the magnitude of the potential drop or stverity of the CID conditions. Another species to note in the rec irded mass spectrum is Fe' which may originate from simple impurity or contamination from the stainless steel cay illary.Also not identified on the mass spectrum may be other contaminants from the stainless steel such as nickel or zinc or even additional chromium. Such contamination is espl cially noted with a freshly cut capillary and is minimized aft :r use due to surface passivation. Clearly such contamination vould have to be minimized or eliminated in order to obtain ac urate determinations at trace levels. Casual inspection of Fig. 1 may give the impressioii that similar to an ICP mass spectrum very little solution infor- mation may be obtained other than the fact that chrom um is present. If one considers the fact that ES run in a positi ie ion mode will 'liberate' only positive solution ions then or e can assume that this mass spectrum is due to a positive sclution ion chromium species.Further in natural waters under r eutral to mildly acidic conditions this would imply a Cr"' ion IS the Cr"' species exists predominantly in anionic form. In either 600] (I) 500 ' 400 8 I53cr+ 0 300- 'E .- E 200- a c. - 0- 30 40 50 60 70 80 90 100 110 120 130 m/z Fig. 1 methanol-l%H,O. Flow rate 2.00 p1 min-' AT/= 145 V Mass spectrum of CrC1,.6H20 moll-' (-6.6 r pm) in case this may be confirmed by changing the sampling con- ditions to gentler ones in order to preserve the solution form. The pH of the lop2 mol I-' CrCI stock solution (before dilution) was measured and found to be 3.02. This is indicative of solution phase hydrolysis as would be predicted by the K value given in eqn. (1). Indeed calculation using the above constant for a mol I-' stock solution yields [Cr"] = 9.31 x lop3 mol 1-' [Cr(OH)2+] =6.91 x l o u 4 moll-' and pH = 3.1 6 units.One would therefore predict that the above stock solution consists primarily of Cr3' ion with some Cr(OH)'+ ion. A 100-fold dilution in methanol will have an unknown effect in that we are diluting it in a solvent of lower dielectric and solvating properties. It might be fair to say that methanol being a protic solvent should have at least similar solvating properties to water and that the major difference might lie in a potential for increased complexation with the counter ion. This should be small for dilute solutions of electrolytes. Solvolysis should still occur; however the associ- ated constants for these events are unknown. It will be assumed that for this case methanol just acts to dilute the sample.A second consideration that should be discussed is that of gas phase solvolysis or charge reduction. This topic has been discussed in the literature especially with regards to metal ions.29*36 A bare 3+ metal ion which is unsolvated in the gas phase will be quite unstable. These ions typically are much more stable when solvated by a medium in which a portion of this charge may be delocalized. Consider the case where such a metal centre is 'liberated' from a droplet with an accompanying solvation sphere of some size or number; if these solvent ligands are gradually stripped off there will come a point where the solvation sphere no longer provides the stability of the bulk solution and so the metal ion becomes less stable.Eventually there will come a point where the solvent is stripped down to a minimum number such that the metal centered complex is no longer stable and thus a 'charge reduction' reaction occurs which is represented by eqn. ( 5 ) below M(CH30H),3+ -$ M(OCH,)(CH,OH) - 2 + + CH,0H2+ ( 5 ) In effect what happens is that in an attempt to better delocalize its charge the metal centre forms a more localized covalent bond with one of the immediate inner sphere ligands. This effectively reduces the net charge but not the oxidation state associated with the metal centre. The above can be expressed explicitly in a thermochemical cycle.36 A rough rule of thumb is that when the ionization potential (IE) of the metal ion (for a particular charge state) is greater than the ionization potential of the ligand charge reduction becomes increasingly favored as solvation is lost and this is typically aided or hindered by the ligand's ability to form a covalent bond (i.e.M-OCH or M-OH) with the metal centre. The ionization potential for chromium and various ligands is given in Table 1. The third ionization potential of chromium is about three times in excess of the ionization potential for methanol this coupled with the fact that chromium readily forms C r - 0 bonds indicates that charge reduction will be favored unless excessive solvation is Table 1 Ionization potentials of selected species Ionization potential*/eV I I1 I11 IV v VI Chromium 6.76 16.49 30.95 50 73 91 Methanol 10.84 Water 12.6 Acetonitrile 12.2 - ~ ~~ ~~~ * CRC Handbook of Chemistry und Physics 1993 edn.301 rnal of Analytical Atomic Spectrometry December 1996 Vol. I f 1205present. The ability of methanol to stabilize a Cr(OCH3)2+ species is however likely for some minimum ligand number. Therefore under gentle sampling conditions species like Cr(CH,OH),3+ are unlikely to be observed (even though they may exist in solution) whereas solvated species like Cr(OCH3)(CH30H):+ would be expected exclusively. The above example has assumed that the only species present in solution is the 3' ion where the solvent consists solely of methanol. Where water is present Cr(OH)2+ species of mixed ligand spheres will also exist and this will be observed below. Fig. 2 shows a CID profile for chromium perchlorate solu- tions diluted in methanol.The three spectra were acquired under different conditions ranging from a fairly gentle [Fig. 2(u)] to a fairly harsh set of conditions [Fig. 2(c)]. Under the gentle conditions of Fig. 2(u) (AV= 20 V) it is observed that the dominant ion is the CrOH2+ ion with a mixed water- methanol solvation sphere ranging in total ligand number from 5-7. This spectrum indicates the oxidation state of chromium as it exists in solution i.e. Cr'". The species distributions are typically separated by 9 m/z units indicative of a 2' ion separated in mass by 18 units or 1 H20 ligand. This CrOH2+ ion is the predicted ion that we should observe based on the above discussion. It has been assumed that the ion is CrOH" and not Cr(OCH3)2+ which may or may not be entirely true and will be addressed in a later section.Regardless of which ligand has formed the covalent bond gentle conditions indicate that the ion observed is chromium in the (111) formal oxidation state and the overall charge of the species is 2 + consistent with predictions. Figs. 2(b) and 2(c) are the resultant mass spectra acquired with increased potential differences. They illustrate the typical steps which the chro- mium 2+ species of Fig. 2(u) goes through as it gets stripped down to a bare Cr+ ion as observed in Fig. 1. The proposed steps based on the various species recorded by the mass spectrometer are summarized in the schemes below where Schemes 1 and 2 are for hypothetical systems consisting of pure methanol and water solvation spheres respectively. It is assumed in all cases that the starting distribution of species is 20 - 10 - 7 0- 8 - 6 0 40 .- g 20 c = o 20 0 50 60 70 80 90 100 110 120 130 140 150 160 m/z Fig.2 Mass spectra of Cr(C104)3-6H,0 lop4 mol 1-1 in methanol-1%H20. Flow rate 2.00 pl min-'. (a) AV=20 V; (b) AV= 35 V; (c) AV= 55 V the same regardless of the potential difference applied. As there is a certain number of energetic collisions which occurs in the expansion the final product is the sum total of these where the less energetic collisions (more frequent) occur at the beginning of the expansion and the more energetic (less frequent) colli- sions occur at the end. Thus a stepwise collision sequence is developed with the result of the final collision being recorded at the MS detector for a given potential difference.Therefore depending on the magnitude of the potential difference the final product will be one of the products shown on the right hand side of Schemes 2 or 3. These schemes are representative of pure systems and a variety of CID products are possible when mixed solvent spheres are present initially. The final step proposed in Scheme 1 is observed under CID conditions that are less intense as compared to the final step in Scheme 2. In either case the Schemes show possible mechanisms to explain how the original Cr"' gas phase ion becomes 'stripped' down to the Cr' species observed in the mass spectrum. [Cr(OCH,)( HOCH,),I2 + AE - Cr(OCH3)2(HOCH3)+ + CH30H2+ Cr(OCH,)2(HOCH3)+ - Cr(OCH3)2+ + CH30H Cr(OCH,)2+ + Cr(OCH2)+ + CH,OH Cr(OCH,)+ + Cr+ + OCH2 Scheme 1 Cr(OH)*(H20)32+ 5 Cr(OH) *(H20)+ +H30+ Cr(OH) *(H20)+ - Cr(OH)2+ +H20 Cr(OH)2+ -+ CrO+ +H20 CrO+ -+ Cr+ +O Scheme 2 As stated the schemes above show possible mechanisms to account for the decomposition products.Support for some of the steps in the above schemes may be inferred with caution from the When the same chromium perchlorate solution is diluted in acetonitrile similar species are observed. A CID profile is given in Fig. 3. Fig. 3(u) is acquired under relatively gentle conditions (AV= 20 V) and the expected CrOH2+ ion is observed. Here the stabilizing ligand will be H20 rather than CH,CN which may be expected as both ligands have similar ionization potentials. However aceto- nitrile cannot form a stabilizing covalent bond similar to the M-0 bond that water can and therefore acts to hinder the charge reduction process or 'protect' the charge.It should be noted that under the gentle conditions of Fig. 3(u) the CrOH2+ ion was observed with mixed solvation spheres of water and acetonitrile however in all cases there were only two distributions observed; one containing 3 acetonitrile ligands [Cr(OH)(CH,CN) -(H20)x2+] and one containing 4 acetonitrile ligands [Cr(OH) (CH,CN) ( H20),2+ ] with accompanying water. Under conditions gentler than those of Fig. 3(u) these two distributions were observed but with corre- spondingly more water ligands. This observation was con- sistent and made on several different occasions. In addition during CID stripping it was the extraneous water ligands that were lost in preference to one of the 3 or 4 acetonitrile ligands.Both of the 2' solvated chromium species [Cr (OH)(CH,CN) ( H20)x2+ Cr(0H) (CH,CN) . ( H20),,2+ ] 1206 Journal of Analytical Atomic Spectrometry December 1996 Vol. 1 1(4 H30+CH3CN Cr(OH)(CH,CN),(H,O),?+ Cr(oH)(CH3CN),(H20)Pt 400 200 100 0 I + I 4 0 g 400 1 (c) H30+CH3CN Cr(OH),(CH,CN)+ 1 00 0 100 Ca(OH),+ Ca(CH&N)+ 0 '9 50 60 70 80 90 100 110 120 130 140 150 160 m/z Fig. 3 Mass spectra of Cr(C10,)3.6H,0 mol 1- ' in acetonitrile-l%H,O. Flow rate 2.00 pl min-'. (a) AV=20V; ( h AV= 30 V; (c) AV=45 V; and ( d ) AV= 65 V are typically stripped down to a Cr(OH)(CH3CN)32+ s 3ecies before further charge reduction reactions occur [Fig 3(b)] however there is some minor contribution by hydrated s 7ecies [Cr(OH)(CH,CN) - ( H,0)2f].The CID processes of chro- mium in acetonitrile as shown in Fig. 3 are illustra ed in Scheme 3 below where the first part represents the major process and the second represents the minor process Cr(OH)(CH,CN),'2 CrO(CH,CN),+ + H+(CH CN) CrO(CH,CN),+ - CrO' + 2CH3CN or Cr(OH)(CH,CN) .( H20)+ -% Cr(OH),(CH,CN),+ +H'(CI IJN) Cr(OH),(CH,CN),+ - Cr(OH),+ + 2CH,Ch Cr(OH),+ - CrO+ +H,O Scheme 3 Not shown here is the decomposition of the Cr-0' bond which is readily observed and is consistant with the last step in Scheme 2. Chromium(m) acetonitrile solutions typically produce 'cleaner' spectra than the corresponding methanol solutions. Complexation of Cr"' with a large organic chelating agent may serve to clean up the spectra even further. From the above discussion it has been demonstrated that chromium(m) may be observed where its oxidation state may be verified conclusively and its solution form may be inferred. Polymeric Species A series of solutions was prepared in an attempt to investigate polymeric chromium species under certain conditions where such species are thought to exist.Four stock solutions of l o p 2 moll-' Cr(C1O,),-6H2O were prepared where to the first no base was added to the second moll-' NaOH was added to the third lo- mol 1-' NaOH was added and to the fourth lo- moll-' NaOH was added. The solutions were then allowed to age for 10 months at room temperature at which time their pHs and UV/VIS absorption spectra were acquired. All four solutions were acidic with the solution with no base and lop4 moll-' base having pHs both of 2.79 and the lo- mol-' base and l o p 2 moll-' base having pHs of 2.89 and 3.23 respectively.The colours of the solutions ranged from violet blue to a green colour with increasing pH. According to Stunzi and Marty3' the ratio of the extinction coefficients of the two d-d band maxima at -420 nm and =580 nm provide an easy identification and check on the purity of the oligomers. The ratio of the band maxima cmax( ~ 4 2 0 nm) E,,,( 2 580 nm) for the different species are monomer= 1.17 dimer= 1.18 trimer= 1.60 and tetramer= 1.95 (easy distinction between monomer and dimer seems dubious). The spectral data collected in Table 2 provide evidence for species other than the monomer being present especially for the solution with moll-' base added.These solutions were then diluted with both acetonitrile and methanol and then electrosprayed. ESMS analysis did not yield any con- clusive evidence that would support the presence of poly- meric species in these solutions (there were however some unexplained peaks which may or may not be due to the pres- ence of polymeric species). The stability constant of the dimer (5.5 x mol I-') given in eqn. (4) is highly dependant on the conditions in which it was measured. Under the dilute conditions of these experiments it is difficult to predict the total concentration of the dimer present. In the literature however polymeric species have been reported under the dilute conditions associated with natural waters.40 With this in mind these experiments were carried out as it was not known quantitatively how much dimer (or other oligomer) was present initially nor at what rate the reverse reaction occurred in the electrospray solvents; methanol or acetonitrile.No direct evi- dence was found to support the presence of polymeric species at the concentrations examined. Kinetic Inertness of Chromium (111) It is known that one of the unique characteristics of chrom- ium (III) is its relative kinetic inertness towards exchange. As Table 2 Absorption data Solution emax/l mol-' cm A. (ly420 nm) ~ ~ ~ ~ / l mol ' cm I i. ( rr 580 nm) Ratio at final pH mol I-' Cr(CIO,) with- No base 14.30 11.83 1.21 moll-' NaOH 15.05 12.44 1.21 rnol I - ' NaOH 15.46 12.64 1.22 lo-' mol I - ' NaOH 21.60 15.99 1.35 J urnal of Analytical Atomic Spectrometry December 1996 Vol.11 1207a result of this many different Cr'" complexes can be isolated and studied. This will have direct consequences on the resultant ES mass spectra of such solutions. When stock solutions of chromium(II1) (which are not strictly controlled by low pHs) are examined it is fair to say that the investigation is of the solution at that particular time as some time later whether it is a couple of hours or months the solution composition will change. This can act to complicate a determination because for chromium(~rr) it may take several hours or days before a state of semi-equilibrium is reached or perhaps a full equilib- rium is never established. To illustrate this case a relatively mature stock solution of CrCl .6H20 was diluted in methanol and then run immediately by ES the spectra were acquired under identical conditions as a function of time and are given in Fig.4. The first spectrum [Fig. 4(a)] acquired about 5 rnin after dilution consists for the most part of fully water solvated species [Cr(OH),*(H,O),+] as well as some mixed ligand species. For simplicity only the 1 + region of the mass spectrum is being shown. However under these conditions 2' species are also present [i.e. Cr(OH)*(H20),2+] at correspondingly lower m/z. Examination of Figs. 4(b) and 4(c) at 30 and 90 rnin after dilution respectively show marked contrasts where the latter contains species that are indicative of almost exclusive solvation by methanol and the former contains species that are intermediate between water rich and water poor species.Upon re-examination of Fig. 2 at this point it should become clear why Fig. 2(b) contains almost exclusively methanolic species such as Cr(OCH,),+ as opposed to hydrolytic species Cr(OH),+. The series of spectra in Fig. 2 were of course collected in chronological order where Fig. 2(a) was acquired about 20 min after dilution and Fig. 2(b) was acquired about 50 min after dilution. The above solution (Fig. 4) was a chromium chloride solu- tion prepared in water and allowed to mature for 5 months.When it was diluted with methanol the dominant ion 5 150 - 8 C 304 I I / x = 4 100- 80 - 60 - 40- 8 8 20- s v 'CI 20 10 0 (4 o - e l Cr(OCH3),(CH30H),+ Cr(OCH,),(CH,OH)(H,O)+ 0 0 30 m a 9 100-(b) .- +-' ([I 5 80- r 60 - 40 - 20 - 10 0 0 - e 110 120 130 140 150 160 170 180 190 200 210 m/z Fig.4 Mass spectra of CrC13.6H,0 moll-' in methanol- l%H,O. AV=25 V. A lo-' moll-' stock solution was prepared and allowed to age for 5 months in water the solution was then diluted and spectra were acquired (a) after 5 min (b) after 30 min and (c) after 90 rnin observed was a solvated 2 + species or 1 + species depending on the sampling conditions. A fresh stock solution of the same CrCl -6H20 salt was then prepared in methanol and allowed to stand for 12 h at which time it was diluted to moll-' with methanol and observed by ES as given in Fig. 5. The spectrum was acquired under moderately harsh CID condit- ions (AV=40V). There is a marked contrast between the two experiments. The data presented in Fig. 5 is a relatively simple spectrum which consists almost exclusively of the mono and di-chloride species Cr(OCH3)C1(CH30H),+ and CrC12(CH30H),+ with some contribution by the uncom- plexed Cr(OCH3)2(CH30H),f species where n is some arbi- trary ligand number.The bar graphs in Fig. 6 illustrate the relative isotopic contribution of the two chloro complexes observed in Fig. 5. From the above two examples it should be apparent that when methanol is used as an ES solvent for Cr"' species the kinetic inertness of the metal centre is maintained. In addition methanol also seems to yield increased com- plexation with counter ion. C r(OCH,)( I) (CH,OH )+ P Cr(OCH,)(CI)(CH,OH),+ 20 40 60 80 100 120 140 160 180 200 220 240 260 m / Z Fig. 5 Mass spectrum of CrCl3.6Hz0 moll-' in methanol. AV=40 V.A moll-' stock solution was prepared in methanol and then set aside for 12 h before it was diluted to the final concen- tration with methanol and run Cr(OCH3)CI+ 5 -Q 115 116 117 118 119 120 121 122 123 124 125 119 120 121 122 123 124 125 126 127 128 129 m/z Fig. 6 Relative isotopic distribution of (a) CrCl(OCH,)+ and (b) 47rC1 + 1208 Journal of Analytical Atomic Spectrometry December 1996 VoE. 11When the same chronological experiments were condu Gted with acetonitrile as the solvent the results were not as diam- atic. In Fig. 7 two mass spectra acquired 8 min [Fig. '(a)] and 90min [Fig. 7(b)] after the dilution of a mi ture Cr(CI04j,.6H20 stock solution are presented. For the inost part there are no major differences and the mass spectra :on- sist in both cases of the two distributions Cr( I H ) (CH,CNj .(H20),2+ and Cr(OH)(CH,CN) -(H20)y2 as discussed previously.The only real difference is that the sp xies seem to have shifted to lower water content i.e. the chan ;e in intensity of the x = O and x = 2 species. Investigations 01 this phenomenon using absorption spectrophotometry indic ated that the spectra acquired immediately upon dilution o ' the same stock solution with acetonitrile and 1 h later showtd no extreme changes. The 2 d-d absorption maxima rem;lined constant at i=414nm and ,l=578nm and there was only ~ 5 % increase in the molar absorptivities over this time This minimal change is consistent with that of the ESMS eTperi- ments where little change was observed over the given time period; however these results should be viewed with ca ition as lower concentrations of the chromium perchlorate sol^ tions were used.From the above discussion it was shown that Cr"' sol1 tions could be investigated by ESMS. Where both the oxidatior state and the indirect solution form (owing to gas phase stabilii ation reactions) the ion could be determined by selecting the a ,)pro- priate conditions. The characteristic chemistry of Cr"' in ique- ous solutions was also discussed especially with regards o the ES process. It was found that specific characteristics of the metal centre such as its kinetic inertness could be ob.erved directly which is important to consider as it will have direct consequences on the solution composition at any given time. The use of ES to probe the solution chemistry of chrom um is a unique and informative method which allows direct and indirect information to be gained on a variety of chrc mium complexes with minimal sample volumes.2oo (a) Cr(OH)(CH3CN),(H2O)x2+ Cr(OH)(CH,CN),(H,O)y 2+ 1 Chromium( V I ) The chemical form of CrV' in aqueous solutions is highly dependant on its concentration and the solution pH. Recently Tandon et al.,l presented data on the effects of pH and concentration on CrV' species in solution that were later corrected by Shen-Yang and Ke-An.42 From the various equilibrium constants listed below Kl H2Cr0 HCr0,- + H + K2 HCr0,- Cr0,-2 + H+ 2HCr0,- C Cr20,-2+ H2( which are reported for conditions of 0 K 3 onic strength,42 the pH and concentration dependance is obvious. Fig. 8 presents a graphical interpretation of the above constants between the pH ranges of 2 and 12 for and mol I - ' total chromium concentrations.It should be noted that the effect of total ionic strength on the equilibrium constants is known to be rather large and therefore these representations should be viewed with caution. In terms of the equilibrium distribution of CrV' species in solution it does seem that the above equilibria [eqns. (6)-(8)] are fairly well accepted; however they have been criticised in that there has never been direct evidence to support all of the postulated species in aqueous solution. Michel and cowork- have done extensive studies on these systems using ers43.44 Raman spectroscopy a technique which should allow for the direct determination of such species if they existed. In particular they focussed on the existence of the HCr0,- ion in aqueous solutions and they were able to conclude that 'the presence of the entity HCr0,- in acidic CrV' solutions was very doubtful'.They then went on to determine their own equilib- rium constants with this knowledge in mind. Therefore it would seem that the solution chemistry of CrV' is still somewhat undefined and is still open for debate. 150 1 00 Y h 50 3 m o 2 8 0 1 x = 1 I x = 2 I I Y = O 1 ] x = 2 90 100 110 120 130 140 150 160 170 1;lO m/Z Fig. 7 Mass spectra of Cr(C104),-6H,0 mo 1-' in acetonitrile-l%H,O. Flow rate 2.00 yl min-I AV= ' 5 V. A moll-' stock solution was diluted and spectra were acquired (a) after 8 min and (h) after 90 min K = 1.6 (6) ~ = 3 . i x 10-7 ( 7 ) K3=34 (8) g 110 g90 *s 100 8 80 6 70 0 60 40 50 ! K 2 3 4 5 6 7 8 9 i o i i i 2 PH Fig.8 Effect of pH on the species distribution of Cr"' in aqueous solutions for the total concentration of (u) moll-' and (h) mol I - ' J w n a l of Analytical Atomic Spectrometry December 1996 Vol.1 1 1209Chromium(v1) Solutions With this limited knowledge the solution chemistry of CrV' solutions was studied by ESMS. It will be assumed that for the sake of comparison the above equilibria [eqns. (6)-(8)] are valid. One of the first experiments undertaken was to examine a mixture of Na,Cr,O and HN03 (each moll-') in methanol. The result given in Fig. 9 is not surprising and it emphasizes a very important point that must be taken under consideration when using ES. In this case as would be expected CrV' will react with methanol in fact CrV* is used rather routinely in organic synthesis to oxidize alcohols to aldehydes or ketones.Fig. 9(u) shows the chromate ester which is an intermediate step in the oxidation of the alcohol to in this case an aldehyde and is represented in Scheme 4 below 0 It I1 0 CH30H + HCr0,- + H E CH3-O-Cr-OH + H20 Scheme 4 When the potential difference in the sampling region is increased to 40 V [Fig. 9(b)] the chromate ester decomposes to the HCr0,- species probably via the loss of CH20. Although this presents a potential problem it also opens up new avenues to explore the solution reactions of CrV' especially with regards to oxidation products and intermediates. This however is not the focus of this paper. In order to better study these systems a more suitable ES solvent must by used.Acetonitrile was selected and it was found to be suitable for the most part; however it too suffers some limitations as will be discussed later. The same stock solution of lo- moll-' Na2Cr207 and HN03 was diluted 100 fold in acetonitrile and was examined by ES. The results are presented in Fig. 10. Fig. 10 is a CID profile where Fig. lO(u) was acquired at AV=25 V Fig. 10(b) and at AV= 55 V and Fig. 1O(c) at AV=75 V. The spectrum in Fig. 1O(u) is a solvent stripped spectrum where conditions were such that all the N 0 m/Z Fig. 9 CID profile of Na,Cr,O and HNO each methanol-1 %H,O. (a) A V= 25 V; and (b) AV= 40 V moll-' in solvent ligands were stripped from the molecular ions with minimal decomposition.In this spectrum we see the character- istic ions NO3- Cr207,- and HCr0,- which are more or less indicative of the ions present in the stock solution. At this point it is necessary to consider some fundamental aspects of the process to better understand or interpret results in particu- lar those pertaining to equilibrium discrepencies. In the above experiment the stock solution had a total CrV' content of 2 x mol l-' and a pH of 2.15. Based on the equilibrium constants given in eqns. (6)-( 8) the relative ratio of [Cr2072-] [HCr04-] is ~ 0 . 7 7 . For a similar solution of only moll-' total CrV' this ratio is e0.24. Now if these stock solutions are diluted 100 fold in aqueous media with the decrease in concentration would also come an increase in pH.The net result would be a shift to an almost complete solution content of the HCr0,- species under mildly acidic conditions. This shift in equilibrium species is not observed when the stock solution is diluted in acetonitrile as shown in Fig. lo@). In fact the actual distribution is more representative of the stock solution which might be expected. The comparison to dilution in acetonitrile an aprotic solvent is not wholly appropriate. There is only N 1% water after dilution and so the equilibria as described by eqns. (6)-(8) will not strictly apply. The second consideration is the proposed droplet preconcentration that occurs as a result of solvent evaporation during a droplet's lifetime.'* Consider a charged droplet of radius R the free charge will occupy a surface layer of some thickness and the interior of the droplet will consist primarily of bulk solution ions.Now if the droplet starts evaporating both the surface area and the droplet volume will decrease where the former contributes to a higher surface charge density and the latter contributes to an increase in bulk concentration. The rate of droplet evaporation determines the rate at which the surface charge density increases and ultimately to what point the droplet is stable before a fission event occurs. While this is occuring the bulk concentration is simultaneously increasing and so the solution composition of the bulk inside the droplet will begin to differ from that of the original solution. As a rough guide consider the following scheme 50 I /I - 0- 'Y) 20 60 100 140 180 220 260 70 80 90 100 110 120 130 140 150 160 170 70 80 90 100 110 120 130 140 150 160 170 m/Z Fig.10 CID profile of Na2Cr,0 and HN03 each acetonitrile-1%H20. (a) AV=25 V; (b) AV=55 V; and (c) AV=75 V moll-' in 121 0 Journal of Analytical Atomic Spectrometry December 1996 Vol. 1 1(a) Dilution Scheme 5 Thus any increase in concentration will favour an increase in Cr207,-. Under acidic conditions the proton becomes concen- trated as well and thus serves to shift the equilibrium 10 the right. To further complicate this the low initial water content present in the solution ( N 1 YO) may act to further retard the reverse reaction going from extreme right. Ultimately new equilibria will depend to a large extent on the solvent media and its effect.The above discussion has been focused on bulk concentration effects and does not take into account any intensity differences due to the actual gas phase ion production which will probably act to complicate the matter further. This particular aspect of a particular ion's intensity related to its solution concentration has been discussed in some detail by Kebarle and Tang.28 The above considerations have to be understood before a real assessment of the ion intensities may be made. This study is however a qualitative one and its goal is to determine whether or not ions that are present in solution may be observed in the gas phase. Figs. 1O(b) and lO(c) illustrate the decomposition pathway of these species where scheme 6 given below is probably representative of the types of decomposition reactions occuring HCr0,- - CrO,- +OH HCr0,- - CrO,- + H Cr0,- --+ Cr0,- + 0 Cr0,- - Cr0,- + 0 Scheme 6 Two experiments were performed to explore the concen- tration and pH dependence of CrV' species under conditions of ES.The first experiment was a concentration profile where the amount of lo- moll-' Na2Cr04 stock solution diluted in acetonitrile was varied. The stock solution had a pH of 7.76 which from Fig. 8(a) indicates that the dominant ions are Cr042- and HCrO,-. In Fig. 11 spectra are presented of (a) 0.5 x lo- mol 1-' (b) lo- moll-' and (c) 2.0 x lo- mol I-' Na2Cr0 respectively. The dilution of the stock solution would favour a shift in the equilibrium to the left as represented in Scheme 5. The spectra were acquired under identical con- ditions and it was found that the ratio of Cr2072- to HCrO,- increased with concentration.For the three concentrations going from low to high the ratios are 7.0 9.0 and 18.O% respectively. This of course would not be expected based on the equilibrium constants because all bulk concentrations would indicate exclusively the chromate ion however it may be understood from the above discussion of droplet preconcen- tration. The dominant ion expected when the aqueous equilib- ria would be obeyed [based on Fig. 8(u)] is the chromate ion Cr042- which is not observed. Considering that the ion is generated by ES then it would probably be 'liberated' from the droplet with an accompanying solvation sphere of n i.e. Cr04,-( L) (where L = ligand either water or acetonitrile). If the solvation sphere contains water then a competitive reaction might occur between ligand loss and charge separation as 200 "1 HCrO I q 100 20 40 60 80 100 120 140 160 180 200 220 m/z Fig.11 Concentration profiles of Na,CrO in acetonitrile. All spectra acquired under identical conditions (AV=25 V). (a) 0.5 x mol I-'; (b) 1.0 x lo- mol 1-'; and (c) 2.0 x moll-' illustrated in Scheme 7 below AE Cr042-(H20) -+ HCr04-(H20)n-1 +OH- Scheme 7 Charge separation occurs readily in ions that are not able to effectively stabilize the charge without solvation. For example it is obvious based on Fig. 9 that the large dichromate ion is able to exist in the gas phase as a stable ion. This might not be and probably is not the case for the smaller chromate ion in a similar case the sulfate ion (SO,,-) is unable to sustain a bare gas phase 2- charge and will charge separate below a ligand number of 4 (see ref.15). Under gentle conditions of A V=O V a chromate distribution is not observed and the only distri- bution observed is the HCr0,-(L) distribution of some low ligand number. It is difficult to say whether the conversion of Cr0,2- to HCr0,- occurs in the gas phase or in the solution phase. It would seem reasonable that in order to form more Cr,072- as observed in Fig. 11 there must also exist a corre- spondingly greater amount of the precursor HCr0,- which would indicate that it must be present in solution to some degree. The second experiment explored the pH dependence of the CrV1 solutions. Three stock solutions of lo- moll-' Na,CrO each were prepared.The first contained lo- moll-' NaOH the second no additive and the third lo- moll-' HNO,. The corresponding pHs were found to be 11.65 7.76 and 3.57. The ES mass spectra of all three solutions diluted 200 fold in acetonitrile are given in Fig. 12 where each was acquired under identical conditions. There is a corresponding increase in the ratio of Crz072- to HCr0,- going from high to low pH. The ratios for the three solutions were 4.0 7.0 and 38.0% respect- ively. Again this trend may be expected based on Scheme 5 and the above discussion on droplet preconcentration. The large jump from 7.0 to 38% as exhibited in the difference between Figs. 12(b) and 12(c) may be due to the effective preconcentration of both the CrV1 ion and the H+ ion.Journal of Analytical Atomic Spectrometry December 1996 Vol. 11 121 1I v) C v) 500 3 400 300 c 200 c 8 .- E 2 loo - C O 0 Q) c 400 - HCr04- (c) HCr04- 105 110 115 120 125 d z Fig.12 pH profiles of 0 . 5 ~ moll-' Na2Cr04 in acetonitrile-0.5%H20 with varied pH control. All spectra acquired under identical conditions (AV=25 V). (a) 0.5 x mol 1-' NaOH added; (6) no pH adjustment; and (c) 0.5 x mol I-' HN03 added Trichromate Ion The existence of the trichromate ion (Cr3OlO2-) has only been detected at high concentrations ( N 1 mol I-' or greater Na2Cr04) and at pHs less than 0. Based on the above knowl- edge where dilution of Cr"' solutions in acetonitrile does not have the same effect as it would with water it was thought that this system could be studied or at least probed.A stock solution of 1 moll-' Na2Cr04 with a corresponding 1 moll-' HN03 concentration was prepared which had the characteristic dark orange-red colour. From this solution 10 pl was diluted to 100 ml in acetonitrile keeping the effective water content down to <tl%. This dilution gives an overall concentration of moll-' in acetonitrile and the corresponding mass spec- trum acquired under gentle conditions of this solution is given in Fig. 13. In this mass spectrum the characteristic Cr2072- Cr,07(Cr,CN) 2- *O01 90 . i i o Cr207 I 130 1 :H3CN);- ~ r 0,:- 3 170 190 210 m/z Fig. 13 Mass spectrum of the dilution of a concentrated chromate solution. 10 pl of a stock solution of Na2Cr04 ( 1 moll-') and HNO (1 moll-') diluted to 100 ml in acetonitrile.AT/= 17 V and HCr0,- ions are observed as well as the HCr20,- ion at m/z 217 and the trichromate ion (Cr3OlO2-) at m/z 158. It was found that the trichromate ion may also be observed from solutions prepared from more dilute stock solutions (ie. lop2 moll-' with moll-' HN03). When these are diluted in acetonitrile to give overall concentrations of 22.0 x moll-' Na2Cr04 and 2.0 x mol I-' HNO there does appear to be evidence for the trichromate ion. The intensities however are not as great as those observed in Fig. 13(a). Again this may be understood as a result of precon- centration of the Cr"' species in the droplets especially under conditions (higher concentration and lower pH) favourable to the existence of the Cr20,2- ions as given by reaction Scheme 8 3Cr2072- + 2H' SCr3Olo2- + H 2 0 Scheme 8 Chlorochromate Ion For the most part Cr"' in aqueous solution will exist as species dictated by the equilibria given above [eqns.(6)-(8)]. These equilibria may be perturbed by the presence of certain matrix ions which can give rise to the existence of other species. Notably the presence of halides such as chloride under concen- trated conditions can give rise to chlorochromate complexes as illustrated in Scheme 9 below CrO,(OH)- + H+ + C1- +Cr03C1- + H 2 0 Scheme 9 This reaction is reversible as it may be hydrolysed in water. 0 ther complexes include sulfato complexes which can be formed upon reaction with sulfuric acid or acid sulfate. To see if the chlorochromate species may be observed by ES a solution of Na2Cr04 HNO and KCl each 0.5 x moll-' in acetonitrile-1 % H20 was prepared. The mass spectrum is presented in Fig.14 and it clearly shows the presence of the chlorochromate species at m/z 135 as well the other character- istic equilibrium species. Michel and M a c h i r ~ u x ~ ~ also observed the chlorochromate species by Raman spectroscopy and it was determined that it became the dominant species in solution when HC1 was in excess of Cr"' in concentrated solutions. The data presented in Fig. 14 is from dilute equal molar solutions and reflect the effects of droplet precon- centration. It becomes important to consider these species when solution control for such purposes as chromatographic separation is required. 200 r I 150 4- C 3 8 9 100 25 .- cr) v) C Q) - = 50 1) Cr0,CI- Fig.14 Mass spectrum of chlorochromate. Na2Cr04 HNO and KC1 each 0.5 x lop4 moll-' in acetonitrile-l0/0H2O AV=25 V 121 2 Journal of Analytical Atomic Spectrometry December 1996 V01.1 ISolution Decomposition The very nature of Crv' is that of a strong oxidizer. The oxidation strength of the species is pH dependant where under acidic conditions it behaves as a stronger oxidant than under basic conditions as given by eqns. (9) and (10) below:30 Cr207-2+14H++6e-=2CrC3+7H20 E0=1.33 V (9) ( 10) Cr0,-2+4H20+3e- =Cr(OH),+5OH- Eo= -0 13 V In acetonitrile CrV' is not completely stable and will react with the solvent over time. The rate of reaction is dependant on the pH of the solution the more acidic the solution the quicker the decomposition. When CrV' is allowed to mature in xeto- nitrile the solution turns to a pink-violet colour indicalive of a reaction.The mass spectra of a solution allowed to age 4 d in acetonitrile mol 1-1 Na,CrO,) is given in Fig. 15 where Fig. 15(u) is acquired under gentle conditions and Fig. 15(b) is acquired under harsh conditions and shows the CID decomposition products of the precursors [Fig. 3 5(u)]. The three major reaction products observed are at m/. 126 142 and 158 and are tentatively assigned Cr0,CN- Cr(02)2CN- and CrO(O2),CN- respectively. This assign- ment is supported in part by the decomposition products observed in Fig. 15(b); however these assignments should be viewed with caution. The purpose of the experiment is to illustrate the potential shortcomings of acetonitrile as an ES solvent for CrV' species.The decomposition seems LO be accelerated in the cases of solutions (initial stock solutions) of lower pH. In choosing an appropriate solvent for CrV' deaermi- nation by ESMS it would seem highly unlikely that any organic solvent would be immune to oxidation and therefore the best solvent might be the one with the slowest solution decompostion. Overall the ES mass spectra of the CrV' species seem to be reflective of the stock solution concentration; however the species distribution is understandably shifted based on the above discussion. The mass spectra presented can be used effectively to show directly the existence of CrV' species in 400 - .- v) C g 300- - 200 - (b) CrO,CN- 90 110 130 ' 150 ' 170 ' 190 m/Z Fig. 15 Reaction products of 0.5 x moll-' Na2Cr04 left to age in acetonitrile-0.5%H20 for 4 d.CID profile (a) AV=25 V; and (b) AV=40 V solution although the equilibria involved with such species are complicated. The one species that was not directly observed was the Cr04'- ion this may be due to a gas phase (ie. charge separation) or a solution phase conversion to the HCr0,- ion. As a corollary of this it is difficult to say that the HCr0,- ion exists in solution; however it cannot be argued that it exists in the gas phase. For acidic solutions where Cr042- is expected to be absent the HCr0,- ion is observed as expected and so probably does not form as a result of gas phase charge separation. The evidence that it exists however is more favourable than not based on the results obtained.CONCLUSIONS ESMS is able to offer other avenues for exploring certain aspects of solution chemistry not wholly realized by other techniques. In terms of speciation ES may not be the ideal source; it does however set a precedent. In order to achieve an ideal speciation tool development efforts must be focused on tech- niques that retain solution information rather than destroying it while at the same time providing a means of unambiguous determination. Knowing the complexities of solution chemistry and the resultant diversity of species possible in certain systems prudence in interpretation is cautioned. Financial support by the Natural Sciences and Engineering Research Council of Canada (NSERC) and the University of Alberta are gratefully acknowledged. REFERENCES 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 29 30 Parsons M.L. Major S. and Forster A. L. Appl. Spectrosc. 1983 37 411. Technical Summary report Perkin-Elmer-SCIEX ELAN 5000. Technical Summary report VG Elemental Limited System VG Plasma Quad PQ2 Turbo Plus. Technical Summary report Hewlett Packard HP 4500. Ivaldi J. C. and Tyson J. F. Spectrochim. Acta Part B 1995 50 1207. La Ferniere K. E. Fassel V. A. and Eckels D. E. Anal. Chem. 1987 59 879. Tomlinson M. J. Wang J. and Caruso J. A. J . Anal. At. Spectrom. 1994 9 957. Olesik J. W. Kinzer J. A. and Olesik S. V. Anal. Chem. 1995 67 1. Sperling M. Xu S. and Welz B. Anal. Chem. 1992 64 3101. Agnes G. R. and Horlick G. Appl. Spectrosc. 1992 46 401. Agnes G. R. and Horlick G. Appl.Spectrosc. 1994 48 655. Agnes G. R. Stewart I. I. and Horlick G. Appl. Spectrosc. 1994 48 1347. Agnes G. R. and Horlick G. Appl. Spectrosc. 1994 48 649. Agnes G. R. and Horlick G. Appl. Spectrosc. 1995 49 324. Stewart I. I. and Horlick G. TrAC Trends Anal. Chem. (Pers. Ed.) 1996 15 80. Van Berkel G. J. McLuckey S. A. and Glish G. L. Anal. Chem. 1992 64 1586. Zeleny J. Phys. Rev. 1917 10 1. Yamashita M. and Fenn J. B. J. Phys. Chem. 1984 88 4451. Yamashita M. and Fenn J. B. J. Phys. Chem. 1984,88 4671. Smith D. P. H. ZEEE Trans. Ind. Appl. 1986 63 527. Hayati I. Bailey A. I. and Tadros Th. F. J. Colloid. Interface Sci. 1987 117 205. Hayati I. Bailey A. I. and Tadros Th. F. J . Colloid. Interface Sci. 1987 117 222. Taflin D. C. Ward T. L. and Davis E. J. Langmuir 1989,5 376. Gomez A. and Tang K. Phys. Fluids. 1994 6 404. Iribarne J. V. and Thomson B. A. J. Chem. Phys. 1976,64,2287. Thomson B. A. and Iribarne J. V. J. Chem. Phys. 1979,71,4451. Dole M. Mack L. L. Hines R. L. Mobley R. C. Ferguson L. D. and Alice M. B. J . Chem. Phys. 1968 49 2240. Kebarle P. and Tang L. Anal. Chem. 1993,65 972A. Stewart I. I. and Horlick G. Anal. Chem. 1994 66 3983. Cotton F. A. and Wilkinson G. Advanced Inorganic Chemistry. Journal of Analytical Atomic Spectrometry December 1996 Vol. 11 121 331 32 33 34 35 36 37 38 A Comprehensive Text Interscience New York 5th edn. 1988 p. 687. Stunzi H. and Marty W. Inorg. Chem. 1983 22 2145. Rotzinger F. P. Stunzi H. and Marty W. Inorg. Chem. 1986 25 489. Spiccia L. and Marty W. Inorg. Chem. 1986 25 266. Spiccia L. Stoeckli-Evans H. Marty W. and Giovanoli R. Inorg. Chem. 1986 26 474. Rai D. Sass B. M. and Moore D. A. Inorg. Chem. 1986,26,345. Blades A. T. Jayaweera P. Ikonomou M. G. and Kebarle P. Int. J. Mass Spectrom. Ion Processes 1990 101 325. Cassady C. J. Freiser B. S. McElvany S. W. and Allison J. J. Am. Chem. Soc. 1984 106 6125. Kang H. and Beauchamp J. L. J. Am. Chem. SOC. 1986,108,7502. 39 Azzaro M. Breton S. Decouzon M. and Geribaldi S. Int. J. Mass Spectrom. Ion Processes 1993 128 1. 40 Saleh F. Y. Mbamalu G. E. Jaradat Q. H. and Brungardt C. E. Anal. Chem. 1996 68 740. 41 Tandon R. K. Crisp P. T. Ellis J. and Baker R. S. Talanta 1984 31 227. 42 Shen-Yang T. and Ke-An L. Talanta 1986,33 775. 43 Michel G. and Machiroux R. J. Raman Spectrosc. 1983 14 22. 44 Michel G. and Cahay R. J. Raman Spectrosc. 1986 17 79. Paper 6/03380F Received May 14 1996 Accepted October 8 1996 121 4 Journal of Analytical Atomic Spectrometry December 1 !)96 Vol. I 1

ISSN:0267-9477    

DOI:10.1039/JA9961101203

出版商:RSC    

年代:1996

数据来源: RSC

摘要:

Arsenic Speciation by Liquid Chromatography Coupled With lonspray Tandem Mass Spectrometry Journal of Analytical Atomic Spectrometry JAY J. CORR SCIEX 71 Four Valley Drive Concord Ontario Canada L4K 4L.8 ERIK H. LARSEN National Food Agency of Denmark Institute of Food Chemistry and Nutrition 19 Msrkhsj Bygade DK-2860 Saborg Denmark Ionspray mass spectrometry a well established organic analysis technique has been coupled to high-performance liquid chromatography for speciation of organic arsenic compounds. The ionspray source and differentially pumped interface of the mass spectrometer were operated in dud modes for elemental and molecular analysis. Tandem mass spectrometry was employed to increase selectivity. Dual mode detection was applied to demonstrate the utility of the technique for analysis of nine organoarsenic standards including four dimethylarsinylriboside derivatives (arsenosugars).Structural fragmentation patterns showing molecular dissociation through an expected common product ion were obtained for the four arsenosugars. Molecular mode detection was utilized for qualitative verification of speciation analysis by high-performance liquid chromatography coupled to inductively coupled plasma mass spectrometry of organoarsenicals in plaice and for analysis of an arsenosugar in extracts of oyster and mussel. Keywords Arsenic speciation; organoarsenic species; arsenosugars; ionspray mass spectrometry; electrospray mass spectrometry Arsenic occurs in the environment and in biological systems in a number of different organic and inorganic molecular f o m s or species.Since the toxicity bioavailability and transport of arsenic is dependent on the species present it is essential to determine selectively the individual arsenic species present in the sample studied. Inorganic arsenic as arsenite (As"') and as arsenate (As') in the oxidation states + 3 and + 5 respectively are the most toxic arsenic species. The methylated arsenicals such as monomethylarsonic acid (MMA) dimethylarsinic acid (DMA) and trimethylarsine oxide (TMAO) are lower in toxicity. The quaternary arsonium compounds arsenocholine (AsC) arsenobetaine (AsB) and the tetramethylarsoniuni ion (TMAs) are essentially non-toxic. The toxicity of the numerous naturally occurring dimethylarsinylriboside derivatives' has yet to be adequately evaluated.These compounds are referred to as arsenosugars since their molecular structure contains a pentose moiety. The formulae of the arsenic species of interest to this study are given in Table 1. A variety of analytical techniques have been employed in the numerous arsenic studies reported in the literature.2i3 Since a typical biological sample may contain several arsenic species at low concentrations the necessary requirements of an analyt- ical technique are sensitivity and selectivity. Combinations of separation techniques (for species differentiation) with atomic spectrometric detectors (for selective and sensitive detection) are included in most hyphenated instrumental systems for arsenic speciation. The most frequently employed element- specific detection methods for HPLC separations have been AAS and ICP-AES or ICP-MS.Speciation studies using HPLC coupled to ICP-MS have revealed the presence of two arsenosugars in shellfish!-6 Capillary electrophoresis (CE) has recently been demonstrated to be effective as a separation technique coupled to ICP-MS for detection of arsenic standard substances. 7* For speciation analysis in general in addition to the advan- tages of high selectivity and sensitivity HPLC-ICP-MS offers on-line real time detection of elements common to the analyte species and is insensitive to concomitant sample matrix con- stituents. However species identification is based on chromato- graphic retention times compared with those of available standard substances. Consequently chromatographic peaks which do not correspond in retention time to those of the available standard substances indicate the presence of species of unknown identity.Since no molecular information is avail- able in conventional ICP-MS and hence no structural infor- mation is attainable such species cannot be identified with an atomic spectrometric detector. Even if standard substances were available complete chromatographic resolution of poss- ible co-eluting species would remain to be necessary for their unambiguous identification and quantification. With atomic spectrometric detection in speciation research there is no qualitative or quantitative verification possible even for peaks corresponding to known species. Table 1 Formulae of the nine organoarsenic species studied ~~ ~~~ Compound Formula Dimethylarsinic acid (DMA) (CH3)2As0(0H Trimethylarsine oxide (TMAO) (CH3)3AsO Tetramethylarsonium ion (TMAs) (CH,).lAs + Arsenobetaine ( AsB) Arsenocholine (AsC) ( CH3)3A~+CHzCH10H Arsenosugars 10-13 (CH3),As +CH2CO0 - 0 bH 6 H 10 R=SO3H 11 R=OH 12 R=OS03H 13 R = 0 II AH AH &-I 0-P -0-CH,-CH-CH Journal of Analytical Atomic Spectrometry December 1996 Vol.11 ( 1 21 5-1 224) 121 5Soft ionization sources for mass spectrometry have been employed to detect inorganic arsenic and organoarsenicals as their molecular ions in standards and real samples. Reports on the utilization of field desorpti~n,~-" thermospray,12 atmos- pheric pressure chemical i~nization'~ and electro~pray'~ mass spectrometry displayed molecular arsenic species but offered few characteristic fragment ions to provide structure eluci- dation except in the last two studies in which collision-induced dissociation (CID) by tandem mass spectrometry (MS-MS) was used.Fast atom and desorption chemical ionization ( DCI)I6 mass spectrometry yielded signifi- cantly more extensive structurally characteristic fragmentation with the more detailed spectra of DCI-MS including both molecular ions and the atomic arsenic ion at m/z 75. At the other extreme the spectra for electron impact mass spec- trometry ( EI-MS)I7 possessed severe fragmentation yet offered little or no elemental arsenic. A technique that offers molecular as well as elemental detection capabilities would greatly improve the state-of-the- art in speciation analysis.A low-pressure ICP ion source with MS detection has been demonstrated for the purpose of allowing dual mode molecular as well as atomic detection of certain organometallic and organohalide species separated by GC.'8*19 For the same purposes radiofrequency (rf ) glow discharge (GD) has been examined as an ion source for MS detection of GC separations.20 An alternative approach to modifying a harsh atomic ionization technique is to operate or modify a soft molecular detection technique to permit both molecular and elemental analysis. Kebarle and c o - ~ o r k e r s ~ ' - ~ ~ first demonstrated this possibility with electrospray-MS (ES-MS) in their extensive studies of singly doubly and triply charged uncomplexed metal ions in solution. Further ES-MS (or IS-MS) studies involving uncomplexed metal ions in solu- tion have since been presented by Horlick and c ~ - w o r k e r s ~ ~ - ~ ~ and by a few other g r o ~ p s .~ ' - ~ ' These reports included studies of alkali and transition l a n t h a n i d e ~ ~ ~ ' ~ ~ speci- ation of inorganic anions,27*28y34 solvent-derived metal correlations between ES-MS spectra and the distribution of pre-formed ions in solution,32 isotope ratio measurernent~,~~ quantitative aspects,26 and the effects of system operating parameters on the nature of the ions dete~ted.~' HPLC has been coupled to ES-MS-MS for molecular mode analysis of organoarsenic species and quantification of AsB in the reference material DORM-136 (National Research Council of Canada). The use of ES-MS elemental mode analysis for detection of metal cations and inorganic anions separated by CE37,38 and for inorganic 0x0-anions separated by anion-exchange chr~matography~~ has been demonstrated.Electrospray and ionspray (IS) are closely related atmos- pheric pressure ionization techniques which produce gas-phase analyte ions directly from pre-existing ions in solution and hence are particularly suitable for providing the interface between liquid separation techniques and MS. ES-MS(-MS) and IS-MS(-MS) are well established analytical techniques which have been employed extensively often coupled to HPLC for the analysis of organic molecules such as proteins peptides and other biomolecules. Briefly the ES and IS techniques involve pumping an analyte-containing solution through a capillary maintained at high potential relative to the sampling plate of a mass spectrometer.A mist of highly charged droplets is dispersed from the capillary and drifts under the influence of the applied electric field towards the mass spectrometer sampling plate. In transit the droplets evaporate at atmospheric pressure until a critical droplet diameter is reached at which ions are transferred into the gas phase by a very low energy process which induces no fragmentation. Compounds contain- ing one or more charge sites are observed as singly or multiply charged ions as they existed in solution. IS is the pneumatically assisted version of ES. IS utilizes a high velocity coaxial jet of gas to shear droplets from the liquid stream exiting the capillary probe. The gas velocity and geometry of the sprayer nozzle affects droplet size and aids in formation of small droplets particularly at higher liquid flow rates where ES is not effective.Although ES and IS are routinely used analytical tools the details of the individual mechanisms in the over-all process are the subject of continued debate and are not fully under- stood.39 The aim of this paper is to employ the IS technique for arsenic speciation analysis in biological samples. The paper will detail the coupling of cation-exchange HPLC to IS-MS(-MS) detection for analysis of organoarsenic species. Dual mode elemental and molecular analysis will be presented using standard mixtures which include arsenosugars and for extracts of plaice oyster and mussel as samples. EXPERIMENTAL Chromatographic System Cation-exchange chromatography was performed using a silica-based Chrompack (Raritan NJ USA) Ionospher C analytical column (100 x 3 mm id 5 pm particle size) with sulfonic acid functional groups.A Shimadzu (Kyoto Japan) LC-1OAD HPLC pump isocratically supplied 1 ml min-' of the mobile phase which consisted of 15 mmol 1-l of pyridinium formate in water-methanol (80+ 20) adjusted to pH 2.7 with formic acid.40 Optimum IS sensitivity is usually obtained at low flow rates (typically 5-10 pl min-') and the efficiency of the ionization process decreases minimally for flow rates up to approximately 50 pl min-'. Hence the post-column effluent was split using a Valco (Houston TX USA) zero dead volume tee such that only 50 plmin-' was allowed to exit the IS capillary. The remainder of the effluent flowed through the appropriate length of Teflon tubing to waste.A Rheodyne (Cotati CA USA) 8125 injector with a 20 pl stainless-steel sample loop was used. Ionspray Source The needle probe of the IS source consisted of solution- carrying 0.1 cm id stainless-steel capillary tubing interior to 0.4 cm id coaxial tubing of the same material. The outer tubing supplied the nitrogen nebulizing gas at 0.95 1 min-' for molecu- lar mode operation and at 1.441 min-' for elemental mode. The solution-carrying tube protruded approximately 2 mm from the nebulizer tube. A Valco tee was used to hold the tubes in place and facilitate admission of the nebulizing gas. Solution transfer lines from the analytical column to the splitter tee and IS tee were 125 pm Teflon tubing.The IS probe was oriented at a severely oblique angle to the sampling plate of the mass spectrometer such that the edge of the aerosol cone formed by the IS device was sampled into the mass spec- trometer. This was empirically found to result in increased stability in the IS signal and to permit sampling of ions that exhibited less ion-solvent clustering which will be discussed in more detail in a later section. The two modes of analysis required different potentials to be applied to the IS probe 4.4 kV for molecular mode and 5.0 kV for elemental mode. The IS assembly was located in a mildly exhausted housing. For infusion experiments (no HPLC pump used) analyte solution was delivered by a Harvard Apparatus (Southnatick MA USA) Model 22 syringe pump at 5 pl min-'.Mass Spectrometer A triple quadrupole PE-SCIEX (Thornhill Ontario Canada) API 300 mass spectrometer (Fig. 1) utilizing a differentially pumped atmosphere-to-vacuum interface was modified to permit elemental analysis as well as the standard molecular 121 6 Journal of Analytical Atomic Spectrometry December 1996 Vol. 11Cumtin Plate orifice Skimmer Fig. 1 Schematic diagram of the IS system and mass spectrometer. The declustering potential is the voltage difference between the orifice and the skimmer. For MS-MS experiments nitrogen collision gas was introduced into Qz analysis. Ions from the IS source were sampled into the MS system through a dry nitrogen curtain gas between the curtain plate (or sampling plate) and the orifice which assisted rn the declustering of ion-solvent adducts and the desolvation of IS droplets as well as preventing contaminants from entering the vacuum system.The pressure in this region was minimally higher than atmospheric pressure. The ions underwent 3 free jet expansion through the orifice into the differentially pumped region in which pressure was in the Torr range before passing through a skimmer and into an rf-only quadrupole (Qo) maintained at pressures in the milliTorr range. The potential difference between the orifice and the grounded skimmer is referred to as the declustering potential and is equivalent to the orifice potential. Although a number of source and interface parameters were adjusted to select a particular mode of operation the declustering potential was the most critical. The major modification to the PE-SCIEX API 300 mass spec- trometer was the installation of a power supply capable of exceeding the usual orifice potential limit of 200 V.In the Qo region collisional focusing caused energetic cooling of the ions and forced them onto the axis of the MS system as described by Douglas and French.41 There were two forms of molecular mode operation detec- tion of fully intact molecular species and MS-MS. In elemental analysis mode and in the fully intact molecular mode the instrument was operated as a single quadrupole device with only the first quadrupole (Q1) employed as a mass analyser. In these modes of operation only rf potentials were applied to the second and third quadrupoles (Q2 and Q3 respectively) causing them to transport but not mass analyse ions.For MS-MS analyses Q1 was set to allow transmission of only the molecules of interest the precursor ions and the enclosed rf-only Q2 quadrupole was filled with nitrogen collision gas to create a high-pressure collision cell.42 Molecules of selected masses impinged upon the collision gas with an energy given by the dc offset potential difference between Q1 and Q2 multiplied by the charge of the precursor ion exiting Q1. CID of the precursor ion would occur resulting in characteristic fragment or product ions. The third quadrupole (Q3) was operated as a mass analyser for the purpose of monitoring specific fragment ions [multiple reaction monitoring (MKM)] or to scan the full mass range. The mass spectrometer was operated with resolution such that in full scan mode the valleys between peaks differing by 1 Da were less than 10% of the more intense peak.Full bidths at half height were less than 0.6 Da. Chemicals The aqueous standard substances used to prepare standard solutions were AsB AsC bromide TMAs iodide TMAO MMA disodium salt DMA sodium salt and hydrogenarsenate (As') disodium Aqueous mixtures of these standards were prepared at 2 pg ml-' each (as species concentration) for HPLC-IS-MS-MS experiments. For infusion experiments the individual standard substances were prepared at 1 pg ml-' in water-methanol (50+ 50) with 20 pl of 0.1 mol 1- ' HCl added to each 5 ml of solution (to aid signal stability). The four dimethylarsinyl-riboside derivatives (arsenosugars) were kindly donated by Dr.Kevin Francesconi University of Odense Odense Denmark. Extracts of NFA plaice (National Food Agency of Denmark) which is a flounder-type fish NIES No. 6 Mussel (National Institute of Environmental Studies Tsukuba Japan) and NIST SRM 1566a Oyster Tissue (National Institute of Standards and Technology Gaithersburg MD USA) were prepared by extraction with methanol-chloroform-water as described by Larsen et aL4' The crude water-methanol phases containing the arsenic analytes were injected without further purification into the LC-IS-MS-MS system. HPLC-grade chemicals were purchased from Aldrich (Milwaukee WI USA) or Fisher Scientific (Nepean Ontario Canada). Distilled de-ionized water was produced in-house with a Waters Milli-Q purification system (Millipore Bedford MD USA).RESULTS AND DISCUSSION Dual Mode Detection of DMA IS or ES mass spectra of inorganic cations generally contain high degrees of ion-solvent clustering which may dominate the spectra. These clusters and their relation to the inorganic ionic species in solution have been studied in detail by a number of Results have been presented in which relatively clean full scan spectra of transition and alkali metals were dominated by singly charged declustered metal ions.25-30,34935 The most critical factor in obtaining such spectra was the use of a significantly elevated declustering potential producing more energetic collisions between ion-adduct mol- ecules and nitrogen gas in the differentially pumped interface region. Extensive inorganic IS studies conducted at SCIEX have shown that it is possible to obtain extremely clean full scan spectra showing only singly charged bare metal elements (and perhaps low levels of the corresponding oxides) regardless of the charge of the analyte in solution by applying a sufficiently high declustering potential.Kebarle and c o - ~ o r k e r s ~ ' - ~ ~ have explained the observance of singly charged metals as being due to gas phase charge reduction. This mode of ionization was applied in this work for detection of elemental arsenic. Three modes of detection (fully intact molecular MS-MS and elemental) are demonstrated for infusion at 5 pl min-l of a 1 pg ml-' standard solution of DMA as shown in Fig. 2. Fig. 2(a) shows full scan molecular mode detection of the protonated DMA molecule at m/z 139.The spectrum was obtained by scanning the mass analysing quadrupole Q1 from 60 to 150 m/z in 0.1 u steps with 5 ms dwell times. A declus- tering potential of 25 V provided sufficiently energetic collisions in the differentially pumped region to eliminate DMA-adduct clusters and resulted in no fragmentation of the DMA mol- ecule. For most organometallic species observation of the uncomplexed yet unfragmented molecule occurs only for a very narrow range of orifice potentials. Optimization of the molecular signal included adjustment of the declustering poten- tial the curtain gas flow the IS potential and nebulizer gas flow and the ion optics up to Q1. The background signal intensity across the full scan mass range is significant due to the soft mode of detection used.While it is possible to optimize the system for maximum DMA.H+ signal it is not possible simultaneously to obtain selective fragmentation of other molecules which therefore contribute to the background mass spectrum. Fig. 2(b) demonstrates the MS-MS mode of detection and the availability of molecular structure information in this Journal of Analytical Atomic Spectrometry December 1996 Vol. 11 121 791 6 34 0 I 0.4 I I ; 4 60 80 100 1 20 140 0 d Z Fig. 2 Detection of DMA in three modes (a) molecular with 25 V declustering potential; (b) MS-MS with 16 eV collision energy; and (c) elemental with 250 V declustering potential. Sample introduction of 1 pg ml-' DMA or 540 ng ml-' As was by infusion at 5 pl min-' mode.All instrumental parameters up to and including Q1 were identical with those used in Fig. 2(u) including the declustering potential. In this mode Q l was not scanned but was operated to allow selective transmission of only the DMAsH' molecule. The dc quadrupole offset potential differ- ence between Q1 and Q2 was 16 V providing a collision energy of 16eV for the singly charged DMA-H' molecule with the nitrogen gas in Qz. A collision energy of 16eV was chosen since it gave the highest intensity for the characteristic product ion (CH3),AsO+ at m/z 121 corresponding to the loss of HzO. Ion path parameters downstream of the collision cell were optimized for maximum sensitivity for this product ion while retaining mass resolution. Maximum sensitivity for the product ion at m/z 121 was achieved at a collision energy for which total dissociation of the precursor ion DMAeH' did not occur.This is a common phenomenon in MS-MS. Small increments in collision energy permit access to further dis- sociation channels spreading the total ion signal over an extended range of fragmentation products. The other MS-MS product ion in Fig. 2(b) is AsO' at m/z 91 corresponding to further loss of two methyl groups. Increasing the collision energy caused more substantial dissociation with the lower mass product ions increasing in intensity at the loss of higher mass fragments. Under extreme CID conditions bare As+ at m/z 75 was observed but the conditions necessary to achieve such extensive fragmentation were detrimental to signal inten- sity and mass resolution rendering the fragmentation spectrum of little use other than to verify the presence of As'.An important difference between the molecular and MS-MS modes of detection is the extremely low background in MS-MS. The selectivity of the MS-MS process offers low background levels particularly for MRM experiments in which precursor and product ion masses are selected causing it to be the mode utilized most often in organic analysis. Elemental mode detection of DMA is demonstrated in Fig. 2(c) which is a full scan single-mass spectrum. The domi- nant features in the spectrum are a peak for As+ at m/z 75 and for AsO' at m/z 91. Otherwise the spectrum is extremely clean although at lower masses the background is increased due to fragmentation of other substances in the solution.The spectrum was acquired with a declustering potential of 250 V. The curtain gas flow was increased by 52% to 1.441 mind' for two purposes to aid in solvent cluster removal and desolvation of IS droplets leaving more of the collision energy in the interface available for molecular fragmentation; and to shorten the mean free path of the molecules and their fragments in the interface region. Experiments conducted at SCIEX have indicated that the degree of fragmentation achievable is equally a function of the number of collisions the molecules experience and the energies of the collisions. Elemental fragments passing through the skimmer and into Qo have experienced a substan- tial electric field and thus possess kinetic energies too large for quadrupole mass analysis with good mass resolution.This is illustrated in published elemental spectra which indicate low mass r e s o l ~ t i o n ~ ~ - ~ ~ or good mass resolution obtained by adjustment of ion optics parameters at the expense of signal inten~ity.~' Collisional focusing4' in Qo of this mass spec- trometer cooled the energetic elemental ions such that they trickled out of Qo and into the analysing quadrupole Q1 with approximately 1 eV of kinetic energy allowing adequate mass resolution. This same collisional focusing forced the elemental ions onto the axis of the mass spectrometer as they were energetically cooled allowing for increased transmission and hence higher sensitivities than in other studies. Fig. 2 illustrates the effects of the declustering potential at two extremes molecular and elemental detection.Moderate declustering potentials may be selected which produce substan- tial but not total dissociation. The declustering potential may be carefully selected to optimize production of a particular fragment ion on which MS-MS may then be performed. Operation in such a mode provides another level of information on molecular structure by in effect offering another degree of MS. Optimization of the LC-IS-MS-MS System ICP-MS signal intensities of arsenic species have been demon- strated to increase with the addition of carbon as methanol in aqueous analyte solutions.43 It was proposed that elevated populations of carbon ions or carbon-containing ions in the plasma improved ionization of elements with lower ionization energies than carbon.Addition of 3% v/v methanol resulted in optimum sensitivity enhancements by a factor of 3-4 with no significant increase in background noise although the average background signal for arsenic increased by a factor of 2. The resulting increase in signal-to-noise by a factor of 3-4 allowed detection of AsC in samples such as shrimp for which it was previously not detectable. An analogous situation exists for the ES and IS ionization techniques where 100% aqueous solutions are among the most difficult from which to extract ions. For a particular compound to be observed from an ES or IS source the compound must be ionized in the solvent being sprayed; hence observation of a compound depends on the pK of the compound and the pH of the solution.The efficiency of transfer of a particular ion from the liquid to the gas phase depends on the free energy of solvation of that ion in the particular solution. Since these techniques depend on ion evaporation at atmospheric pressure methanol is routinely added to solvents to aid in the ion evaporation process. The lower surface tension of methanol-containing droplets facilitates disruption of the bulk liquid surface allowing for more rapid evaporation of solvent. Hence the sprayed droplets more readily achieve the critical electric field on their surface that is necessary for transfer of ions to the gas phase. The result is increased sensitivity and stabilization of signal. LC-IS-MS-MS experiments were performed with different amounts of methanol in the 15mmol 1-' pyridinium-ion 121 8 Journal of Analytical Atomic Spectrometry December 1996 Vol.111.2- h rA Y = 1.0- E -2 0.8- 8 2 v W 0 10 20 30 MeOH in mobile phase (96) Fig. 3 Effect of methanol content in the 15 mmol 1- ' pyridiniu in-ion mobile phase of pH 2.7. Peak areas are background-subtracted mobile phase. MRM was employed to monitor the effects of methanol addition on the chromatographic peaks for DMA AsB TMAO AsC and TMAs. Background-subtracted peak areas are plotted as a function of methanol content in the mobile phase in Fig. 3. Signal increases were observed for all ions for addition of methanol up to 30% v/v. The effect of methanol addition is clearly species-dependent as expected since the rate of ion emission from the IS droplets depends on the solvation energies of the individual ions in the methanol- containing solvent.In contrast to ICP-MS detection as the methanol content was increased in IS-MS-MS detection the background signal and background noise decreased for the detected fragments. This is consistent with a more stable spray being formed with the addition of a polar solvent. Retention times of the various arsenic species were affected by the methanol content of the mobile phase. Addition of methanol at 10% v/v resulted in retention time reductions of approxi- mately 10% compared with the experiment with no methanol addition. Conversely although addition of 30% v/v methanol yielded the most substantial signal increase retention times were increased by approximately 20% and tailing of the last eluting peaks became more pronounced.As a compromise between sensitivity and retention time a mobile phase contain- ing 20% v/v methanol was chosen for the remainder of the experiments presented in this paper. The approximate compound-dependent improvements for this mobile phase composition over the aqueous mobile phase were background signal reduction by a factor of 2.5; background noise reduction by a factor of 1.5; and signal enhancement by a factor of 5. Retention times were slightly increased compared with those obtained when no methanol was added. LC-IS-MS(-MS) of Five Arsenic Standard Substances Fig. 4(a)-(c) demonstrates three modes of detection (fully intact molecular MS-MS and elemental respectively) for cation- exchange separation of seven co-injected arsenic species.The arsenic species were at concentrations of 2 pg ml-' each (as the molecule) corresponding to injections of 40ng of each molecule or approximately 20 ng of As for each species. Effluent splitting prior to the IS probe resulted in 2 ng of each molecule or approximately 1 ng As for each species reaching the detector. AsV and MMA which co-eluted with the void volume of the HPLC system were included in the standards mixture but were not detectable in positive ion mode under the experimental conditions used. Chromatograms were acquired with 250 ms dwell times. Using continuous infusion of AsB in the mobile phase at 20 p1 min-l the IS-MS(-MS) system was optimized for detection of AsB in each of the three n n A Time/min Fig. 4 LC-IS-MS(-MS) detection of five co-injected arsenic species in three modes (a) molecular with 35 V declustering potential; (b) MS-MS with 25 eV collision energy; and (c) elemental with 250 V declustering potential.(a) and (b) are the summed or TIC chromatog- rams of the five separate chromatograms corresponding to the five arsenic species. Protonated molecules were detected for DMA AsB and TMAO in molecular mode and in MS-MS mode the precursor ions were protonated molecules for these species modes of detection. In each of the three modes the 15 mmol I-' pyridinium-ion mobile phase was responsible for sensitivity suppression due to matrix effects by approximately a factor of 6 as compared with infusion of AsB in the 50% v/v methanol solvent previously described. However the low pH of the mobile phase promoted cation formation by protonation of the arsenic molecules.Fig. 4(a) is the total ion current (TIC) chromatogram for fully intact molecular detection. Only the first quadrupole was utilized as a mass analyser and the declustering potential was set to 35V. The acquisition of this chromatogram was accomplished by individually monitoring the masses corresponding to DMA-H' (m/z 139) AsB.H+ (m/z 179) TMAO.H+ (m/z 137) AsC' (m/z 165) and TMAs' (m/z 135). The TIC chromatogram is the sum of the signal intensities recorded at these five masses. Fig. 4(b) is the TIC chromatog- ram for MS-MS detection of five separate CID reactions with a collision energy of 25 eV. For the individual chromatograms the fragment with the highest sensitivity under these CID conditions was monitored.All instrumental parameters up to and including Q1 were identical with those used in Fig. 4(a). Fig. 4(c) shows the chromatogram for single-MS elemental mode detection of As+ at m/z 75 using a declustering potential of 2.50V. Even though an intrinsic interference such as the ArCl' interference in ICP-MS was not present a background signal of 1300 counts s-' at m/z 75 existed. This background level may be further reduced as will be discussed later. With the experimental conditions used for Fig. 4(c) the calculated detection limit for TMAs in this standard solution in elemental mode is 14 pg (reaching the detector). The three chromatog- rams appear similar but there are important differences some of which indicate factors that must be considered when Journal of Analytical Atomic Spectrometry December 1996 Vol.11 121 9employing molecular ionization techniques for analysis of more than one species. In Fig. 4 the three modes of detection reveal differences in relative signal intensities for the various arsenic species. Fig. 3 presents a graphical representation of this observation in MS-MS mode. For any given solvent the ES or IS responses for particular compounds are expected to differ. Such com- pound-dependent variations in response are generally not observed with ICP-MS detection and are due to factors affecting analyte cation formation. The pK of the compound and the pH of the solution may determine the degree of ionization of a particular analyte. However the ionization behaviour of permanent cations such as TMAs and AsC is not pH-dependent.The free energy of solvation of the analyte in the particular solution determines the relative ease with which the ion will be transferred to the gas phase. Hence for any particular solvent and a common set of spraying conditions differences in relative responses of analytes are to be expected. This complicates optimization procedures and therefore an analysis which targets one particular species may significantly improve detection for that analyte. Comparison of the fully intact molecular mode Fig. 4(a) and MS-MS mode Fig. 4(b) reveals significant differences in the relative intensities of the various arsenic species for the two different modes. Since both chromatograms correspond to identical operating conditions in the IS and interface regions the differences are attributable to dissociation in the Qz collision cell. MS-MS conditions including the collision energy were optimized for AsB detection and thus do not represent optimum conditions for some of the other arsenic species.In particular at the selected collision energy of 25 eV DMA AsC and their respective CID product ions undergo a greater degree of dissociation than the remaining three species. This collision energy caused their common characteristic frag- ment ions at m/z 121 to dissociate significantly. Optimum production of the m/z 121 product ions from DMA and AsC occurs at a collision energy of 16 eV. The most intense fragment ion from TMAO was also not optimized at this collision energy.The signals for AsB and TMAs are factors of 7 and 5 respectively lower than in the fully intact molecular mode. These are typical values for optimized fragments in MS-MS although the optimized signal for an individual fragment is extremely compound-dependent. For targeted analysis of any one of these five arsenic species the intensity of the MS-MS signal for the optimized major characteristic fragment ion is within a factor of 5 of that of the fully intact molecular ion signal. The chromatograms for the fully intact molecular mode Fig. 4(a) and elemental mode Fig. 4(c) display fairly well matched relative intensities. The elemental mode sensitivities for DMA TMAO and TMAs are greater than 80% of the corresponding molecular mode sensitivities while the same comparison yields AsB and AsC sensitivities of 50 and 25% respectively. DMA TMAO and TMAs are readily dissociated in the interface region at lower declustering energies while AsB and AsC require significantly more energy to fragment to the bare As atom and present greater difficulties in dissociating the oxide formed in the interface region.The declustering potential may be increased further to promote greater pro- duction of As’ from AsB and AsC but the necessary adjust- ments of the source interface region and first lenses of the mass spectrometer are then unfavourable for transmission of arsenic ions liberated at different energies from the other three arsenic species. Again detection of several species simul- taneously demands compromise conditions because of the species-dependent ionization of the IS or ES technique.Operation of the declustering potential at 350-400 V in elemen- tal mode presented certain advantages. Although the analyte signal was reduced by a maximum of 20% due to defocusing 1 220 Journal of Analytical Atomic Spectrometry December effects in the interface region the background signal was reduced to 500 counts s-l at m/z 75 which increased the signal-to-background ratio by a factor of 3 with a correspond- ing reduction in the background noise resulting in a 3-fold improvement in detection limits. The extracted ion chromatograms for the individual MS-MS transitions are shown in Fig. 5. Fig. 5(a) presents the same TIC chromatogram as Fig. 4(b). Fig. 5 demonstrates the selectivity available with a molecular detection technique.Each of the chromatograms is the result of the monitoring of a structurally significant CID fragment of the precursor molecule after only that particular precursor molecule had been transmitted through the first analysing quadrupole. For example Fig. 5(b) corresponds to Q1 allowing only transmission of ions of m/z 139 (DMA.H+) and 4 3 detecting only m/z 121 fragment ions. The individual chromatograms indicate no interference from any of the other arsenic species and therefore demonstrate that with a molecular detection method not only is molecular structural information available but also knowledge of it may be used to reduce the necessity of complete chromatographic resolution. The advantages of this are shorter analysis times and with IS or ES the possibility of reducing ionic strengths of mobile phases thus reducing matrix effects and therefore promoting increased sensitivity and reduced detection limits.With the experimental conditions used for Fig. 5(f) the calcu- lated detection limit for TMAs in this standard solution in MS-MS mode is 2 pg (reaching the detector) an improvement by a factor of 7 compared with the elemental mode. If elemental mode detection was operated with a declustering potential of 350V in order to reduce detection limits as discussed 41(bl 2 DMA n mAo 04 I I 1 I ’ L 1 l--kL 0 40r 20 0 1 1 1 1 0 2 4 6 8 10 Time/min Fig. 5 Molecular mode LC-IS-MS-MS detection of five co-injected arsenic species (a) TIC chromatogram; (b) DMA Q1 m/z 139 Q3 mjz 121; (c) AsB Q1 m/z 179 Q3 m/z 120; ( d ) TMAO Q1 m/z 137 Q3 m/z 122; (e) AsC Q1 m/z 165 Q3 m/z 121; and (f) TMAs Q1 m/z 135 Q3 m/z 120.The collision energy was 25 eV. The precursor ions were protonated molecules for DMA AsB and TMAO 1 !)96 Vol. 11previously the MS-MS detection limit would remain approxi- mately a factor of 2 lower. Dual Mode Detection of Arsenic Species in Plaice Extract Application of dual mode detection to a real sample of NFA plaice extract is demonstrated in Fig. 6. Fig. 6(a) is the LC-MS elemental mode chromatogram for m/z 75 with a declustering potential of 250V. The true background signal is approxi- mately 800 counts s - '. The chromatogram has been expanded in this region to allow observation of the minor arsenic constituents detected. The background signal undulates due to substantial amounts of other non-arsenious compounds elut- ing from the analytical column and causing suppression of ionization in the IS source.HPLC-ICP-MS analysis4" was used to determine the arsenic species concentrations as 94 pg g-' for AsB 0.18 pg g-' for TMAO 0.064 pg g-' for AsC and 0.24 pg g-' for TMAs. Fig. 6(b)-(e) shows the LC-MS-MS extracted ion chromatograms for four of the arsenic species determined by HPLC-ICP-MS to occur in plaice The LC-MS-MS chromatograms for AsB TMAO and TMAs verify identification of the obvious peaks in the LC-MS elemental mode chromatogram while the LC-MS-MS peak for AsC is near the detection limit under these experimental conditions which included a collision energy of 25 eV. From the elemental mode chromatogram it is not possible to identify AsC in plaice.Although absolute signal peak areas are lower in the MS-MS mode than in elemental mode signal-to-noise ratios and hence detection limits are better in MS-MS mode. For example with these experimental conditions the calculated detection limits for TMAs in plaice extract using the quantifi- cation results of HPLC-ICP-MS,40 are 0.070 pg g- ' for h A 30 0 200 0 = 2 4 Timdmin 6 8 10 Dual mode analysis of extract of plaice (a) elemental mode m/z 75 with 250 V declustering potential; 25 eV collision energy MS-MS transitions (b) AsB Q1 m/z 179 Q3 m/z 120; (c) TMAO Q1 m/z 137 Q3 m/z 122; (d) AsC Q1 m/z 165 Q3 m/z 121; and (e) TMAs Q1 m/z 135 Q3 m/z 120. In MS-MS mode the precursor ions were protonated molecules for AsB and TMAO elemental mode and 0.009pg g-' for MS-MS mode.For analysis of real samples such as plaice the fully intact molecular mode is often of limited value since the sample may contain many non-arsenious substances which appear at masses corre- sponding to those of the arsenic species of interest. For example the molecular mode chromatogram for m/z 135 yielded a number of peaks only one of which possessed a retention time corresponding to TMAs. In such an analysis the onus is then placed on retention time matching as in HPLC-ICP-MS. However molecular mode analysis is very valuable for the identification of unknown chromatographic peaks for which standard substances are not available and would be instrumen- tal in the development of such standards. As shown above the MS-MS form of molecular mode analysis has verified results previously reported by HPLC-ICP-MS.40 LC-IS-MS-MS of Arsenosugars LC-IS-MS-MS was used to probe the structure of four arsenosugars and produce the first characteristic CID fragment ion spectra using an ES or IS source.The usual optimization of source and interface parameters during infusion experiments was not performed since phthalate compounds in the standard solution dominated molecular mode spectra and interfered with the process. The methodology involved injecting a single arsenosugar standard while performing LC-MS in elemental mode monitoring m/z 75 to determine the retention time. A second injection was made while performing LC-MS in mol- ecular mode monitoring the mass corresponding to the pro- tonated arsenosugar species to verify the retention time.Finally LC-MS-MS was performed on subsequent injections with Q3 operating in full scan mode to detect all CID fragments of the protonated molecular species. For three of the samples collision energies of less than 25 eV resulted in low degrees of fragmentation; hence collision energies of 25 and 35 eV were used with little loss of information. Q3 was scanned in 0.5 u steps with 1 ms dwell times from m/z 70 to a mass larger than that of the fully intact protonated molecule. The resulting TIC full scan chromatograms possessed peaks at times correspond- ing to those of the LC-MS retention times. Full scan LC-MS-MS traces corresponding to these peaks were extracted to yield fragmentation information.Concentrations of arsenosugars in the standard solutions were 1.76 pg ml-' for arsenosugar 10 (35 ng injected 1.76 ng detected); 0.16 pg ml-' for arsenosugar 11 (3 ng injected 0.16 ng detected); 4.09 pg ml - ' for arsenosugar 12 (82 ng injected 4.09 ng detected); and 3.20 pg ml-' for arsenosugar 13 (64 ng injected 3.20 ng detected). LC-IS-MS-MS spectra for arsenosugar derivative 10 are shown in Fig. 7. The protonated molecular species appears at m/z 393 in Fig. 7(u). The product ion at m/z 375 represents loss of H 2 0 from the protonated molecule. The fragment at m/z 296 is due to loss of OS03H from the protonated molecule and m/z 279 represents subsequent loss of OH. The m/z 237 fragment is the base structure of the dimethylarsinyl derivative- pentose moiety and is observed in the CID spectra of all the arsenosugars presented in this paper.In Fig. 7(b) the fragments at m/z 97 and m/z 80 are OS03H and SO respectively. The identity of the fragments at m/z 167 and m/z 149 are not obvious but possibilities are (Me,AsOCH,)(OH)(CH,) for m/z 167 and (Me,AsOCH,)(CH,) for m/z 149. It was not possible to detect bare As+ fragments at m/z 75 even under extreme CID conditions for this or any other arsenosugar. CID spectra for LC-IS-MS-MS of arsenosugar derivative 12 are displayed in Fig. 8. The structure of compound 12 is very similar to that of compound 10 the difference being one oxygen atom; hence common fragments are detected. In Fig. 8(u) the protonated molecule was not observed at m/z 409 because the collision energy of 25 eV was sufficient for complete Joirrnal of Anulytical Atomic Spectrometry December 1996 Vol.11 1221OH+ OH 6- OHOH mass409 5 - OH+ n W & m 2 X 329 Fig.7 LC-IS-MS-MS CID spectra for arsenosugar 10 (a) 25eV collision energy; (b) 35 eV collision energy. Q1 was set to transmit only precursor ions of mass 393. Q3 was scanned from m/z 70 to m/z 400. The structure of the protonated arsenosugar molecule is given in (a) 4 - II Me2As-CH2 ,o ,O CH CH -CH2-OH x 2.4 1.8 1.2 0.6 90 329 I 360 Fig.8 LC-IS-MS-MS CID spectra for arsenosugar 12 (a) 25eV collision energy; (b) 35 eV collision energy. Q1 was set to transmit only precursor ions of mass 409. Q3 was scanned from m/z 70 to m/z 420. The structure of the protonated arsenosugar molecule is given in (a) although it is not detected at this collision energy dissociation of the precursor molecule. Loss of SO produced a fragment at m/z 329 which is identical with arsenosugar 11.The fragments at m/z 237 and m/z 97 have been described above. LC-IS-MS-MS fragmentation spectra for arsenosugar 13 are shown in Fig. 9. The protonated molecule is observed at Me2As-CH2 ?++ /o /O-CHz-CH -CH2-0- WI P-!O-CH2-CH -CHp I I OH OH I OH I I OH 391 I 242.5 1500 I 329 I i Fig.9 LC-IS-MS-MS CID spectra for arsenosugar 13 (a) 25eV collision energy; (b) 35 eV collision energy. Q1 was set to transmit only precursor ions of mass 483. Q3 was scanned from m/z 70 to m/z 500. The structure of the protonated arsenosugar molecule is given in (a). Loss of the section of side chain to the right of the vertical line in addition to loss of a proton yielded the product ion at m/z 391 m/z 483 in Fig.9(a). Loss of OHCH,(CHOH)(CH,OH) results in the product ion at m/z 391. The fragment at m/z 329 identical with arsenosugar 11 is due to loss of ROH (mass 154) corresponding to compound 13 (Table 1). Identities of the fragments in Fig. 9(a) at m/z 242.5 and m/z 227.5 are not known. The common base fragment is observed at m/z 237 in Fig. 9(b) as well as DMA.H+ at m/z 139. The latter is a result of the pentose moiety being fragmented from the base product ion at m/z 237. LC-IS-MS-MS experiments for aisenosugar 11 were only performed for a collision energy of 25 eV which resulted in a single product ion at m/z 237. Hence the CID spectrum is not shown. A thermospray mass spectrum of arsenosugar 11 presented in the literature12 indicated no fragment ion at m/z 237.For the conditions used in that experiment the fragments reported were at m/z values of 311 221 and 177. Although possible dissociation channels were assigned to these fragments some of the dissociation routes proposed are not as intuitive as the dissociation channels observed by IS-MS detection. Dual Mode Detection of Arsenosugar 11 in Oyster and Mussel LC-IS-MS-MS was employed for verification of the HPLC- ICP-MS detection6 of arsenosugar 11 in NIST SRM 1566a Oyster Tissue and NIES No. 6 Mussel. The mobile phase employed was the same as that described previously except that the pyridinium ion concentration was adjusted to 4 mmol 1-'. Fig. 10 qualitatively demonstrates dual mode verification of the presence of arsenosugar 11 in NIST SRM 1566a.Fig. 1O(a) shows an expanded view of the LC-MS elemental mode chromatogram (500 ms dwell time) for m/z 75. Identification of several of the peaks was not attempted since the intent was only to verify the presence of arsenosugar 11 in real samples. Fig. 10(b)-(d) shows the LC-MS-MS chromatog- rams (250 ms dwell times) for a collision energy of 25 eV. The transition monitored for detection of arsenosugar 11 was Q1 m/z 329 to Q3 m/z 237. The responses for AsB and arsenosugar 1222 Journal of Analytical Atomic Spectrometry December 1996 VoL. 11arseuosugar 11 100 : 0 - 50 EL W . . . . A AsB 60- 01 1 1 I arsenosugar 11 40 6 8 10 Time/min Fig. 10 Dual mode analysis of extract of NIST SRM 1566a Oyster Tissue for arsenosugar 11 (a) LC-IS-MS in elemental mode at m/z 75 with 350 V declustering potential; and LC-IS-MS-MS mode (b) DMA Q1 mfz 139 Q3 m/z 121; (c) AsB Q1 m/z 179 Q3 m/z 120; and ( d ) arsenosugar 11 Q1 m/z 329 Q3 m/z 237.The chromatograms (b)-(d) were recorded using 25 eV collision energy for the MS-MS transitions. In MS-MS mode the precursor ions were protonated moleculzs 11 were similar in agreement with the HPLC-ICP-MS result,6 which indicated an arsenosugar 11 concentration of 1.13 pg g-' As. Under identical operating conditions the peak area corresponding to arsenosugar 11 in NIES No. 6 Mussel was found to be 50% of the peak area from Fig. 10. This is in agreement with the HPLC-ICP-MS results stating a concen- tration of 0.69 pg g-' As. With these experimental conditions the calculated detection limit for arsenosugar 11 in plaice extract using the quantification results of HPLC-ICP-MS,6 is 0.064pg g-' As for MS-MS mode.The HPLC-I<:P-MS analysis indicated detection of arsenosugar 13 close to the void volume of the HPLC system in the oyster and mussel samples but monitoring of the LC-IS-MS-MS transition Q1 m/z 483 to Q3 m/z 327 corresponding to arsenosugar 13 yielded no indication of this species in either sample. It is possible that this is the result of conversion of arsenosugar 13 to arsenosugar 11 by hydrolysis. However the arguments given above regard- ing the relative sensitivity of arsenosugar 11 in the oyster sample compared with AsB in the same sample and cornpared with arsenosugar 11 in the mussel sample render this possibil- ity improbable.Lack of detection of arsenosugar 13 in this sample by LC-IS-MS-MS was probably due to matrix effects from substances co-eluting with the void volume of the HPLC system and causing suppression of ionization in the IS source rendering the detection of species present at low levels and eluting close to the solvent front impossible. CONCLUSIONS The coupling of cation-exchange HPLC to IS-MS(-MS) and the dual mode analysis possible with the IS technique has been employed for verification of arsenic species detected by HPLC- ICP-MS in extracts of plaice oyster and mussel and the first IS or ES CID mass spectra for four arsenosugars have been obtained. Although detection limits for the various arsenic species in the crude extraction phases studied are not as low as those offered by HPLC-ICP-MS further effort would certainly permit detection at lower levels.The results indicate that IS-MS-MS may be valuable as a complementary tech- nique to ICP-MS for speciation analysis since it affords molecular detection capabilities not present in atomic spectro- metric techniques. This could be invaluable in the production of standard substances and for the identification of species for which standard substances are not available. E.H.L. thanks the H.C. Oersted Travel Grant (Royal Danish School of Pharmacy) for financially supporting his travel to Canada. REFERENCES 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 29 30 31 32 33 34 Francesconi K. A. and Edmonds J.S. Oceanogr. Mar. Biol. Annu. Rev. 1993 31 11 1. Cullen W. R. and Reimer K. J. Chem. Rev. 1989 89 713. Donard 0. F. X. and Ritsema R. in Techniques and Instrumentation in Analytical Chemistry Vol. 13 Environmental Analysis Techniques Applications and Quality Assurance ed. Barcelo D. Elsevier Amsterdam 1993 p. 549. Shibata Y. and Morita M. Appl. Organomet. Chem. 1992,6 343. Le S. X. C. Cullen W. R. and Reimer K. J. Enuiron. Sci. Technol. 1994 28 1598. Larsen E. H. Fresenius' J. Anal. Chem. 1995 352 582. Olesik J. W. Kinzer J. A. and Olesik S . V. Anal. Chem. 1995 67 1. Liu Y. Lopez-Avila V. Zhu J. J. Wiederin D. R. and Beckert W. F. Anal. Chem. 1995,67 2020. Norin H. and Christakopoulos A. Chemosphere 1982 11 287. Luten J. B. Riekwel-Booy G. and Rauchbaar A.Enuiron. Health Perspect. 1982 45 165. Luten J. B. Riekwel-Booy G. van der Greef J. and ten Noever de Brauw M. C. Chemosphere 1983 12 131. Cullen W. R. and Dodd M. Appl. Organomet. Chem. 1989,3,401. Siu K. W. M. Gardner G. J. and Berman S . S. Rapid Commun. Mass Spectrom. 1988 2 69. Lawrence J. F. Michalik P. Tam G. and Conacher H. B. S. J . Agric. Food Chem. 1986 34 315. Lau B. P.-Y. Michalik P. Porter C. J. and Krolik S. Biomed. Enuiron. Mass Spectrom. 1987 14 723. Cullen W. R. Eigendorf G. K. and Pergantis S . A. Rapid Commun. Mass Spectrom. 1993 7 33. Beauchemin D. Bednas M. E. Berman S. S. McLaren J. W. Siu K. W. M. and Sturgeon R. E. Anal. Chem. 1988,60 2209. Evans E. H. Pretorius W. Ebdon L. and Rowland S. Anal. Chem. 1994 66 3400. Castillano T. M. Giglio J.J. Evans E. H. and Caruso J. A. J. Anal. At. Spectrom. 1994 9 1335. Olson L. K. Belkin M. and Caruso J. A. J. Anal. At. Spectrom. 1996 11 491. Blades A. T. Jayaweera P. Ikonomou M. G. and Kebarle P. Znt. J. Mass Spectrom. Ion Process. 1990 101 325. Blades A. T. Jayaweera P. Ikonomou M. G. and Kebarle P. Znt. J. Mass Spectrom. Ion Process. 1990 102 251. Blades A. T. Jayaweera P. Ikonomou M. G. and Kebarle P. J. Chem. Phys. 1990 92 5900. Jayaweera P. Blades A. T. Ikonomou M. G. and Kebarle P. J. Am. Chem. SOC. 1990 112 2452. Agnes G. R. and Horlick G. Appl. Spectrosc. 1992 46 401. Agnes G. R. and Horlick G. Appl. Spectrosc. 1994 48 649. Agnes G. R. and Horlick G. Appl. Spectrosc. 1994 48 655. Agnes G. R. Stewart I. I. and Horlick G. Appl. Spectrosc 1994 48 1347. Stewart I. I. and Horlick G. Anal. Chem. 1994 66 3983. Agnes G. R. and Horlick G. Appl. Spectrosc. 1995 49 324. Douglas D. J. in inductively Coupled Plasmas in Analytical Atomic Spectroscopy eds. Montaser A. and Golightly D. W. VCH New York 1992 p. 613. Cheng Z. L. Siu K. W. M. Guevremont R. and Berman S . S. J. Am. SOC. Mass Spectrom. 1992 3 281. Cheng Z. L. Siu K. W. M. Guevremont R. and Berman S . S. Org. Mass Spectrom. 1992 27 1370. Corr J. J. and Douglas D. J. in Proceedings of the 41st ASMS Conference on Mass Spectrometry and Allied Topics Sun Francisco Journal of Analytical Atomic Spectrometry December 1996 Vol. 11 1223CA American Society for Mass Spectrometry San Diego CA USA 1993 p. 732a. 35 Ketterer M. E. and Guzowski J. P. Anal. Chem. 1996 68 883. 36 Siu K. W. M. Guevremont R. Le Blanc J. C. Y. Gardner G. J. and Berman S. S. J. Chromatogr. A 1991,554,27. 37 Huggins T. G. and Henion J. D. Electrophoresis 1993 14 531. 38 Corr J. J. and Anacleto J. F. Anal. Chem. 1996 68 2155. 39 Kebarle P. and Tang L. Anal. Chem. l993,65,972A. 40 Larsen E. H. Pritzl G. and Hansen S . H. J. Anal. At. Spectrom. 1993 8 1075. 41 Douglas D. J. and French J. B. J. Am. SOC. Mass Spectrom. 1992 3 398. 42 Thomson B. A. Douglas D. J. Corr J. J. Hager J. W. and Jolliffe C . L. Anal. Chem. 1995 67 1696. 43 Larsen E. H. and Sturup S. J. Anal. At. Spectrom. 1994,9 1099. Paper 6103381 D Received May 14 1996 Accepted September 25 1996 1224 Journal of Analytical Atomic Spectrometry December 1996 Vol. 11

ISSN:0267-9477    

DOI:10.1039/JA9961101215

出版商:RSC    

年代:1996

数据来源: RSC