Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Evaluation of electrothermal vapourization as a method of sample introduction for the ICP-MS and determination… Chan, Sanny 1993

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata


831-ubc_1993_spring_chan_sanny.pdf [ 5.15MB ]
JSON: 831-1.0061807.json
JSON-LD: 831-1.0061807-ld.json
RDF/XML (Pretty): 831-1.0061807-rdf.xml
RDF/JSON: 831-1.0061807-rdf.json
Turtle: 831-1.0061807-turtle.txt
N-Triples: 831-1.0061807-rdf-ntriples.txt
Original Record: 831-1.0061807-source.json
Full Text

Full Text

EVALUATION OF ELECTROTHERMAL VAPOURIZATION AS AMETHOD OF SAMPLE INTRODUCTION FOR THE ICP-MSANDDETERMINATION OF TRACE LEVELS OF TITANIUM, GALLIUMAND INDIUM IN THE CENTRAL PACIFIC GYREbySANNY CHANA THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTSFOR THE DEGREE OF MASTER OF SCIENCEinTHE FACULTY OF GRADUATE STUDIESDepartment of ChemistryWe accept this thesis as confirmingto the required standard.THE UNIVERSITY OF BRITISH COLUMBIADEC 1992 / JAN 1993Sanny Chan, 1993(Signature)In presenting this thesis in partial fulfilment of the requirements for an advanceddegree at the University of British Columbia, I agree that the Library shall make itfreely available for reference and study. I further agree that permission for extensivecopying of this thesis for scholarly purposes may be granted by the head of mydepartment or by his or her representatives. It is understood that copying orpublication of this thesis for financial gain shall not be allowed without my writtenpermission.Department of  ChemistryThe University of British ColumbiaVancouver, CanadaDate c.-Tattiiard 13 ) DE-6 (2/88)iiAbstractThe analysis of trace metals in the ocean is a relatively new field of study.Since the elements of interest often exist at picomolar levels, this work demandsanalytical methods with ultra low detection limits. The inductively coupled plasmamass spectrometer (ICP-MS), with its improved detection powers, has proven to be avaluable tool. However, problems resulting from the method of sample introduction,such as low transport efficiency, still prevail.The first objective of this work was to evaluate an electrothermal vaporization(ETV) device as a method of sample introduction for the ICP-MS. Absolute precisionsobtained ranged from 2-10% RSD after modifications to improve the performance ofthe ETV were made. Once optimized, improvements for both sensitivity and detectionlimits resulting from increased transport efficiency and matrix separation were found tobe more than two orders of magnitude for most of the elements studied. Severalelements suffered from isobaric interferences, and therefore exhibited improvementsin sensitivity and detection limits of only one order of magnitude. In general, detectionlimits were in the range of 0.057-72 fmol. Acceptable absolute precision was obtainedfor multielement determination of three isotopes. Better precision was obtained forisotope ratio measurements. Five or more isotopes may be simultaneously analysedwhen using isotope ratio techniques. The use of freon to form volatile metal halideswas instrumental in the analysis of refractory metals. Signals obtained from refractorymetals, previously erratic and poorly defined in graphite furanace atomic absorptionspectroscopy (GFAAS), were well defined with large increases in sensitivity as a resultof freon addition. Matrix effects were observed using the ETV-ICP-MS in the analysisof seawater samples, thus requiring the use of standard additions or isotope dilution.The second objective of this thesis was to determine trace levels of Ti, Ga and Inin the central Pacific gyre. Little is known about Ti and Ga distributions in the oceanand virtually nothing is known about In behavior in seawater. Dissolved titaniumIIIexhibited elevated surface values (-100 pM), a subsurface minimum (-50 pM) and abottom maximum (-230 pM). Enhanced Ti concentrations at 400-1000 m, correlatingwith the mid-depth oxygen minimum in this region were observed. This Ti distribution,in combination with the limited published data, suggests both atmospheric and bottomsources, removal via scavenging throughout the water column, and a mobilization inthe 02 minimum.The dissolved Ga distribution shows intermediate surface values (-17 pM), asubsurface increase (-20 pM), an intermediate depth minimum (5-10 pM) and abottom maximum (-30 pM). On combining the present data with the previous datapool, the observed trends suggest that Ga has an atmospheric input. Sub-surface andintermediate water concentrations may be a result of vertical processes combiningscavenging removal and regeneration, or from horizontal advection. The dissolved Indistribution was similar to Ga, with intermediate surface values (-0.3 pM), a subsurfacemaximum (-0.45 pM), a mid-depth minimum (-0.12 pM), and higher concentrations atdeep waters (-0.28 pM). Because this is the first reliable profile of indium to beproduced, there are no other data to compare with. A comparison of In distributionwith Ga show some similarities. The In/Ga ratio, however, changes linearly with depthsuggesting that the two elements are controlled by different input and removalprocesses or rates in the water column. All three elements demonstrate enrichmentwith respect to Al by comparison with their crustal abundances. Although the -Rimenrichment (11 times) may be explained by preferential removal of Al, the degree ofGa/Al and In/Ai enrichments (750 and 1500 times) suggest that the source may not be ofcrustal abundance.Work to improve the existing chelating resin, a TSK 8-hydroxyquinoline, hasafforded a 30-fold improvement of resin capacity. Chelating capacity obtained for thisresin is 34 ± 3 ..t.mole Cu(II) / g resin. A new resin was synthesized by coupling5-amino 8-hydroxyquinoline to a new solid support, Affi-Prep®, gave similar chelatingcapacities.ivTable of ContentsPageAbstract^ iiTable of Contents^ ivList of Tables ixList of Figures^ xiList of Abbreviations^ xiiiAcknowledgements xvDedication^ xviChapter 1Introduction^ 11.1 History of the ICP-MS^ 11.2 Operation of the ICP-MS 21.2.1 System Overview^ 21.2.2 Sample Introduction 41.2.3 The ICP^ 51.2.4 Analyte Excitation & Ionization^ 71.2.5 Ion Extraction^ 71.2.6 Ion Lens and Mass Analyser^ 91.2.7 Ion Detection^ 121.3 Advantages and Limitations of the ICP-MS^ 131.4 Alternate Methods of Sample Introduction 151.4.1 Increased Transport Efficiency^ 151.4.2 Reduction of Molecular Interference 16V1.4.3 On-Line Matrix Separation^ 171.4.4 Speciation Studies 171.4.5 Electrothermal Vaporization Sample Introduction^ 181.5 Trace Metals in the Ocean^ 211.5.1 Background 211.5.2 Determination of Trace Metals^ 231.5.3 Marine Geochemistry of Ti, Ga and In in Seawater^251.6 Aims of This Study^ 261.7 References and Footnotes^ 28Chapter 2Evaluation of Electrothermal Vaporization as a Method ofSample Introduction for the ICP-MS^ 322.1 Introduction^ 322.2 Experimental 322.2.1 General Instrumentation^ 322.2.2 Data Acquisition^ 332.2.3 Solutions 342.2.4 Electrothermal Vaporization (ETV) Device^ 342.2.4.1 ETV Instrumentation^ 342.2.4.2 ETV Modifications 362.2.4.3 ETV Optimization^ 372.2.5 Nebulizer Mode^ 372.2.5.1 Nebulizer Instrumentation^ 372.2.5.2 Nebulizer Optimization 382.3 Results and Discussion^ 382.3.1 Electrothermal Vaporization Optimization^ 382.3.1.1 Modifications To Obtain Better Precision^38vi2.3.1.2 Effects of Ash Temperature^ 402.3.1.3 Effects of Vaporization Temperature^402.3.1.4 Effects of Nebulizer Gas Flow 412.3.1.5 Acquisition Parameters - Effects of Dwell Time^432.3.1.6 Acquisition Parameters - Effects of Acquisition Time^452.3.1.7 Optimization of Torch Position and Ion Lens Tuning ^462.3.1.8 Effects of Freon Addition^ 482.3.2 Analytical Figures of Merit^ 512.3.2.1 Evaluation of Sensitivity and Precision^512.3.2.2 Evaluation of Linearity^ 542.3.2.3 Comparison of Sample Throughput^542.3.2.4 Evaluation of Multielement Capabilities 552.3.2.5 Isotope Ratio Accuracy and Precision^572.3.3 Isobaric Interference - Problematic Elements 602.3.3.1 Titanium^ 602.3.3.2 Aluminum 642.3.3.3 Zirconium^ 672.3.3.4 Iron 692.3.4 Analysis of Seawater Samples^ 722.4 Conclusions^ 732.5 References 75Chapter 3Determination of Trace Levels of Titanium, Galliumand Indium in the Central Pacific Gyre^ 763.1 Introduction^ 763.2 Experimental 783.2.1 Study Site^ 78vii3.2.2 Seawater Sample Collection^ 783.2.3 Reagents and Solutions 783.2.4 Column Preparation^ 803.2.5 Seawater Sample Processing^ 803.2.6 Spike Recovery Tests^ 813.2.7 ICP-MS Operating Conditions 823.3 Results and Discussion^ 823.3.1 Attempts to Remove Isobaric Interferences^ 823.3.2 The Method of Isotope Dilution^ 833.3.3 Method of Standard Additions 843.3.4 Matrix Effects^ 853.3.5 Titanium Distribution^ 863.3.6 Gallium Distribution 883.3.7 Indium Distribution^ 913.3.8 Vertical Advection-Diffusion Model^ 913.3.9 Interelement Comparisons^ 943.3.10 Comparison of Ti, Ga and In with Aluminum^963.3.10.1 Surface Waters^ 973.3.10.2 Deep Waters 983.4 Conclusions^ 1003.5 References 103Chapter 4Resin Development^ 1044.1 Introduction^ 1044.2 Experimental 1054.2.1 Reagents^ 1054.2.2 Synthesis of TSK 8-Hydroxyquinoline Resin^ 105VIII4.2.3 Affi-Prep® Coupling^ 1064.2.4 Capacity Test on Affi-Prep® Resin^ 1074.2.5 Capacity Tests on TSK 8-Hydroxyquinoline Resin^ 1094.3 Results and Discussion^ 1104.3.1 Studies on 8-Hydroxyquinoline Immobilization onto TSK^ 1104.3.2 Studies on Affi-Prep® Resin^ 1134.4 Conclusions^ 11 44.5 References 115Chapter 5Summary^ 116ixList of TablesPageTable 2.1^Typical operating conditions for ETV and nebulizer modes.^33Table 2.2^Effects of dwell time on Isotopic ratio accuracy, precisionand sensitivity and absolute precision^ 44Table 2.3^Comparison of absolute sensitivity and detection limitsfor pneumatic nebulizer and ETV modes.^ 52Table 2.4^Precision obtained for In by ETV and Nebulizer modes^53Table 2.5^Linear dynamic range obtained with pneumatic nebulizerand ETV.^ 54Table 2.6^Results of multielement analysis for five isotopes by ETVand nebulizer modes.^ 56Table 2.7^Precision and accuracy for Hf, Zr and Ga for nebulizerand ETV modes^ 59Table 2.8^Ga/In Isotope ratio precision obtained using theETV and nebulizer modes^ 60Table 2.9^SiF+ interfering species with increasing Freon^60Table 2.10 Signals observed for m/z 24 - 27 in ETV modewith and without Freon^ 64Table 2.11 Possible carbon-containing interferences for aluminum^65Table 2.12 Possible doubly-charged interferences for m/z 26 and 27. ^66Table 2.13 Effect of Freon on Zr m/z 90 - 93.^ 67Table 2.14 Possible NiSi and CuSi interferences. 67Table 2.15 Comparison of isotope ratios expected forvarious interferences and the isotope ratios observedfor masses 90-93.^ 68Table 2.16 Comparison of Ar0+ signals obtained by nebulizerand ETV modes^ 69xTable 2.17 Comparison of Ga concentrations obtained bycalibration curve and standard additions for bothmethods of sample introduction.^ 72Table 3.1^Column washes for matrix effect experiment^85Table 3.2^Table of estimated enrichment factors of Ti, Ga and Inwith respect to Al^ 97Table 3.3^Speciation of elements in seawater at pH 8.2^ 100Table 4.1^Resin capacities obtained from Affi-Prep® resin. 113xiList of FiguresPageFigure 1.1^Overview of the ICP-MS system^ 3Figure 1.2 Schematic of the double-pass spray chamber^3Figure 1.3 Schematic of the ICP torch^ 5Figure 1.4^Degree of ionization calculated from the Saha equation^8Figure 1.5^Ion extraction and ion lens configuration. ^ 9Figure 1.6^Quadrupole mass analyser and axis orientation.^ 10Figure 1.7^Electrothermal vaporization device developed by Park^ 19Figure 1.8 Electrothermal device produced by modifying an HGA-300. ^20Figure 2.1^ETV original furnace design.^ 35Figure 2.2^Modifications to tubing configuration leading fromthe ETV unit to the ICP-MS.^ 36Figure 2.3^Modifications to the ETV unit. 37Figure 2.4 Expanded vaporization stage for the variousramping profiles. ^ 39Figure 2.5 Vaporization temperature effects on Hf signal.^41Figure 2.6^Representative of heating profile used in this study^42Figure 2.7 Nebulizer Gas optimization on a 1 ppb solution of Hf^43Figure 2.8 Representative signal obtained from the ETV for Th^45Figure 2.9 Thorium signals obtained by the ETVwith and without Freon^ 49Figure 2.10 Freon effects on Hf Signal for a 1 ppb solution..^50Figure 2.11 Freon effects on Hf background.^ 50Figure 2.12 Zirconium background as a result of Freon Flow^51Figure 2.13 Optimized log - logcalibration curve for Hf using ETVsample introduction.^ 55xi iFigure 2.14 Blank spectra for m/z 45 - 50 in ETV mode.^61Figure 2.15 Corrosion resistant torch.^ 63Figure 2.16 Effect of Freon flow rates on SiF+ signal at m/z 48with a corrosion resistant torch^ 63Figure 2.17 Representations of Fe blank solutions obtainedwith the nebulizer and ETV modes^ 70Figure 2.18 Drawing of furnace and quartz tube. 71Figure 3.1^Map indicating study site for this study and sitesthat will be used for comparison.^ 79Figure 3.2 Titanium distribution at AV 10 87Figure 3.3 Titanium and 02 distribution at AV 10^ 88Figure 3.4^Gallium distribution at AV 10. 89Figure 3.5 Contours generated by combining data obtainedin this study with those previously obtained^90Figure 3.6^Indium distribution at AV 10.^ 91Figure 3.7 Temperature vs. salinity plot for depths below 1000 mat AV 10.^ 93Figure 3.8 Titanium vs. Salinity in the deep Pacific ocean^93Figure 3.9^Gallium - Salinity and Indium - Salinity Plots forthe deep Pacific Ocean^ 94Figure 3.10 Plots of Ga/In ratios vs. depth at AV 10.^ 95Figure 4.1^Reaction sequence for the coupling of 8-hydroxyquinolineto the TSK-Gel Resin^ 107Figure 4.2 Affi-Prep0 coupling reaction sequence^ 108Figure 4.3 Poly-Prep Chromatography column 109Figure 4.4 Model compound (ethyl p-amino benzoate)reaction sequence.^ 110XIIIList of Abbreviationsac^alternating currentamu atomic mass unitASV^anodic stripping voltammetryAV research vessel "Aleksandr Vinogradov"CPS^counts per secondCSV cathodic stripping voltammetryDC^direct current (plasma)dc direct currentDDI^distilled deionizedDIN direct injection nebulizerDSI^direct sample insertionEDTA ethylene diaminetetraacetateETV^electrothermal vaporizationeV electron voltFAAS^flame atomic absorption spectroscopyFAES flame atomic emission spectroscopyg^gramGC gas chromatographGFAAS^graphite furnace atomic absorption spectroscopygP+ statistical weight for state p+gq^statistical weight for state qh Planck's constantHPLC^high performance liquid chromatographi.d. inner diameterICP^inductively coupled plasmaICP-AES^inductively coupled plasma atomic emission spectroscopyICP-MS inductively coupled plasma mass spectrometryK^degrees Kelvink Boltzmann's constantkg^kilogramKHP potassium acid phthalatekW^kilowattM molarm/z^mass to charge ratiomA milliamperexivmbar^millibarMCA multichannel analyserme^mass of an electronMHz megahertzmin^minutemL millilitreI-tm^micrometermm millimeterwhole^micromolemmole millimoleMS^mass spectrometryMS2 megaohmN^normalityn e^free electron number densityng nanogramnM^nanomolarnmole nanomoleNMR^nuclear magnetic resonancenp+ species number density in state p+n q^species number density in state qo.d. outer diameterpg^picogramPJ peak jumpingpM^picomolarppb parts per billionppm^parts per millionRF radio frequencyRSD^relative standard deviations secondSIM^single ion monitoringSSMS spark source mass spectrometryTi^ionization temperatureV voltW^wattAEpq^energy difference between between states p+ and qXVAcknowledgementsAs always, there are many people who must be thanked for their direct orindirect contributions in making my stay here a fruitful experience. I would like to beginby acknowledging the members of the group, Helen Nicolidakis, Brad McKelvey,Robert Mugo, Lucila Lares and Lu Yang for their efforts. In particular, Lucila Lares andLu Yang are acknowledged for their support that made during the bleak portions of thiswork more bearable. Drs. Adrian Wade, Joan Power, and Albert Leung are alsothanked for providing inspiration when the chips were down. I would also like toacknowledge "Piers Central", and in particular Mr. Rene Lemieux and Dr. ChristineRogers without whose help and insightful discussion, the resin work would not havebeen possible. A very special thanks goes to Dr. Lucio Gelmini who has given upmuch time and sanity in the preparation of this manuscript.I would also like to acknowledge the technical and support staff in thedepartments of Chemistry and Oceanography at the University of British Columbia.Mr. Brian Snapkauskas and Mr. Ron Marwick are thanked for their "No problem, thiswill just take ten minutes to fix - I can do it now" attitude. Mr. Bert Meuller is thanked forhis work in maintaining the ICP-MS. The efforts of these gentlemen have beeninstrumental in keeping the research ball rolling.Many thanks go to my friends, Cameron Forde, Eric Brouwer, and LucioGelmini whose support on and off the volleyball court have made the duration of mystay here much more enjoyable.Last, but certainly not least, Dr. Kristin Orians is thanked for her inspirationduring the course of this work.To my Mother and Fatherxvi"There may be work without results,But never will there be results without work."Unknown1Chapter 1Introduction1.1 History of the ICP-MSThe original concept of the inductively coupled plasma mass spectrometer(ICP-MS) can be traced back to A. L. Gray in 1970 at the UK branch of AppliedResearch Laboratoriesl. It had become apparent, especially to geologists, thatinductively coupled plasma atomic emission spectroscopy (ICP-AES) suffered frommatrix interferences from line-rich elements such as calcium, aluminum and iron. Inaddition, rare earth elements exhibit complex spectra and generally have poordetection limits in atomic emission spectroscopy. A better method of analysis wasclearly needed. The proposed new method of analysis was a mass spectrometricmethod capable of accepting solid samples and providing multielement analysis atlevels down to ng/g levels with an analysis time of several minutes. This criterion wasbased on the performance of spark source mass spectrometry (SSMS)2 which was thetechnique used in low level, multielement work at the time. However, spectra obtainedfrom SSMS were complicated due to non uniform energy transfer resulting in theformation of molecular fragments. These spectra required time consuminginterpretation as well as long sample analysis times. Furthermore, throughput forSSMS was about one to two samples per day. In spite of these drawbacks, SSMSgave uniform and low detection limits across the periodic table.Atmospheric pressure plasmas appeared to be a good ion source for severalreasons. High gas temperatures (5000 K) could be obtained in the case of a DCplasma. In addition, long analyte residence times of a few milliseconds in the plasmaresulted in efficient energy transfer to the analyte. However, the ions produced couldnot be mass analysed until they were transferred to a high vacuum. A feasibility studyof an atmospheric ion source into a high vacuum mass analyser was performed using2a small DC plasma, which had a small tail flame that was easily accessible to ionextraction. This first system gave high sensitivity, 104-105 counts s -1 per ppm Co, withalmost no background interference. Sample introduction was facilitated by anultrasonic nebulizer and gave isotope ratio measurements with <0.5% RSD 3 .Unfortunately, the DC plasma suffered from matrix effects and had lowionization efficiencies for elements with high ionization energies (>9 eV). Inefficientionization was due to the small amounts of sample solution that actually reached thehigh temperature plasma core. This is a phenomenon common in DC plasmas. Highlevels of NO, from the acid sample matrix with ionization energies of 9.4 eV dictate theionization equilibrium for analytes with high ionization energies 6 .The most attractive answer to the above limitations was the inductively coupledplasma, ICP, first proposed in 1975. In 1977, interest from the Ames Laboratory in theUS resulted in a collaboration with the British Geological Survey to further thedevelopment of this technique. By June of 1978, Houk et al. (Ames Laboratory)produced the first spectra of the major ions, Ar+, H+, and 0+ from the ICP. Later thatyear, the first spectra of analyte ions (Mg+, Cr+, Mn+, Co+, Cu+, Rb+, Ag+) at 50 gg/mLconcentration were produced and resulted in the first publication on ICP-MS 3 .The first commercial ICP-MS instrument was introduced at the 1983 PittsburghConference by Sciex with the Elan® followed closely by the VG PlasmaQuad® fromVG Isotopes in May of that same year. In 1988, over 175 ICP-MS instruments were inuse world wide. This number has risen to over 600 instruments in operation in 1992.1.2 Operation of the ICP-MS1.2.1 System OverviewA schematic of the ICP-MS system is shown in Figure 1.1, and the primarymethod for sample introduction, the Meinhard nebulizer and spray chamber, is shownin Figure 1.2.El2ifs^from spraychamber•••38^7Figure 1.1 Overview of the ICP-MS system. (1) channel electron multiplier (2) quadrupole (3) ion lens(4) skimmer cone (5) RF coils (6) ICP torch (7) torch box (8) sampler coneFigure 1.2 Schematic of the double-pass spray chamber (1) solution up-take (2) nebulizer (3) spraychamber (4) nebulizer gas inlet.A peristaltic pump pumps the sample into the nebulizer where it is transformed to theaerosol form. Droplets not condensed on the double-pass spray chamber walls aretransported to the ICP by the argon carrier gas.In the ICP, the aerosol droplets are desolvated, vaporized, atomized andionized. Some of the ions are extracted into a high vacuum region first by a samplingcone, followed by a skimmer cone. The ions then pass through a series of electrodes,or lenses, which focus the ions into the mass spectrometer. Only ions with theselected mass-to-charge ratio are transmitted through the quadrupole mass analyserto turbo pumps4and subsequently detected by a channel electron multiplier. Pulses from the detectionsystem are fed to a microprocessor-controlled multichannel scaler data acquisitionunit. Another microprocessor looks after the overall instrument operation. The overallsystem is controlled by a personal computer.1.2.2 Sample IntroductionThe most common form of sample introduction is pneumatic solutionnebulization, by a Meinhard/concentric nebulizer (Figure 1.2). Gas, typically argon,flows quickly through an opening that concentrically surrounds a capillary tubecausing a pressure reduction at the tip. This has the effect of drawing the samplesolution from the capillary tube, and due to the high velocity of the gas, causes theformation of droplets with diameters between 1 to 50 gm.Sample solution may be introduced in two ways. Sample may be transported tothe tip of the nebulizer by a peristaltic pump or at natural rates governed by the speedof the gas flow. Under normal use, the sample is transported by a peristaltic pump toinsure signal stability.Aerosol (10-50 gm) from the nebulizer is separated according to size in thespray chamber. The most common design of spray chamber is the double-pass spraychamber based on the design of Scott et al. 4 (Figure 1.2). Aerosol droplets from thenebulizer enter the chamber (approximately 2.5 cm i.d.) and subsequently through aconcentric tube of larger diameter at 180° incident to the initial direction of motion.Only the droplets that are not condensed or impacted onto the spray chamber walls(typically <10 gm) are transported to the ICP. The larger droplets are prevented fromreaching the plasma since they may not completely desolvate in the plasma. Theselarger droplets would only contribute to water loading in the plasma. In this type ofsystem (nebulizer and spray chamber), the unused portion constitutes 99% of thesample solution and is normally pumped away from the spray chamber and discarded.5Solution nebulization is only one of several methods of sample introduction.Alternate methods will be presented later.1.2.3 The ICPThe ICP torch (Figure 1.3) is made of three concentric quartz tubes surroundedby a copper induction coil. This coil is connected to a high frequency generatortypically operating at 27.12 MHz with output powers of 1-2 kW. Argon flows through allthree tubes. The gas sheath in the outer quartz tube has functions as both the supportgas for the plasma and the coolant gas for the quartz tube. ire^NebulizerGasir^CoolAuxiliaryGas^GasFigure 1.3 Schematic of the ICP torch.Sample is introduced through the central channel by the aerosol gas. Onceinside the plasma, several processes take place. The analyte aerosol is dried,vaporized, atomized and ionized in the central channel. These processes are madepossible by the high gas temperatures (7000-10000 K 5-8) of the plasma.The auxiliary gas in the middle gas channel serves to push the plasma awayfrom the injector tip, thus preventing a torch melt down.To form the plasma, a spark from a Tesla coil is used to produce seed electronsand ions in the support gas to make the gas electrically conducting. Once this occurs,the plasma forms. The high frequency alternating currents in the induction coilgenerates magnetic fields with their lines of force parallel to the tube walls. Seedelectrons and ions are accelerated in a circular flow. When the current in the induction6coil reverses direction, the magnetic field and the eddy currents also reverses. Theaccelerated ions and electrons collide with the support gas to cause further ionizationas well as intense heating.Some characteristics of an ideal ion source are listed below 7 .1. Complete and single ionization of all elements2. Inert chemical environment3. No background interference4. Accept solution, solids, gases5. Reproducible ionization conditions6. Inexpensive7. Ease of operationUnfortunately, no atomic source has yet been developed with all of the abovelisted characteristics. However, the ICP does possess many of these favorableattributes. Analyte residence time in the ICP is relatively long, approximately 2 to 3milliseconds. Combined with the high temperature from the plasma, the ICP givesalmost complete ionization of the analyte. Argon plasmas provide chemically inertionization environments, thus, atomic ions generally will not react to form molecularspecies. These combined properties result in few molecular interferences over most ofthe mass range. Aqueous and gaseous samples are readily analysed by the ICPwhile solids may also be analysed with some difficulty. It should be noted that somebackground interferences are found with the ICP originating from the Ar support gasand the air which is entrained into the plasma. These interferences include 40Ar+,41 ArH+, 56 (ArO)+, 160+ , 320 2 + , 17(OH)+ , 28(N 2 )+ , 29(N2H)+, 80 (Ar2)+. In suchinstances, alternate isotopes, which do not suffer from interferences can besubstituted. Unfortunately, an ICP-MS is quite an expensive technique: the instrumentis costly ($500,000), and its maintenance and operating costs are substantial. Inaddition, considerable skill and training is required to become an efficient andknowledgeable user.71.2.4 Analyte Excitation & IonizationThe function governing the ion population in the ICP is the Saha equation 56 :/-np+ne _ 2gP+ (2nmekTi)1.5^6,EpqNexp k--k-Trj^n q^gqwhere n e is the free electron number density, np+ and n q are the ion and atomconcentrations respectively, gp and g q are the statistical weights of the two states pand q, m e is the mass of an electron, h is Planck's constant, AEpq is the energydifference between levels p and q (in this case, the ionization energy), Ti is theionization temperature and k is Boltzmann's constant. Thus, according to the Sahaequation, analyte ion density at the sampling aperture in the central channel of theplasma depends locally on:i. Analyte atom concentration in the plasmaii. Ionization temperature, Ti, of the plasmaiii. Electron population n e of the plasmaiv. Ionization energy of the elementv.^Statistical weights between the atom and ion energy statesIn addition to the above processes, the analyte atom concentration in theplasma also depends on the sample introduction transport efficiency. Recently, Grayand Date 8 have determined that the ionization temperature, is 9000 K in an ICP bymonitoring the response (from the mass spectrometer) to known concentrations ofelements with a large difference in ionization energies. A curve generated for theSaha equation (Figure 1.4) with Ti at 9000 K shows that for over 75% of the periodictable, the degree of ionization in the ICP is almost 100%.1.2.5 Ion ExtractionThe ion extraction system used (Figure 1.5) consists of two extraction apertureswhere the analyte ions are taken from atmospheric pressure to a high vacuum8100 ^80---t00=2"E'o40 -15t6 20-00I^I^I3^5^7 9 11^13^15Ionization Energy (eV)Figure 1.4^Degree of ionization calculated from the Saha equation for 59 elements between Li - U.Adapted from Ref. 6.60 -0environment where mass analysis may be performed. Ions are extracted from thecentral channel by the sampling aperture which ranges from 1.0 mm to 1.2 mm. Theseapertures are drilled to the tips of cones made from metals of high conductivity such asaluminum, copper, nickel or platinum.Behind the sampling aperture, gas pressures are reduced to about 5 mbar by arotary pump. The gas expanding into the lower pressure region reaches velocitiesexceeding the speed of sound in a distance of less than one aperture diameter. In thisregion, the temperature drops rapidly and reactions which could change the gascomposition are effectively frozen out. The free jet formed is bounded by a shockwave known as the barrel shock. A second shock wave, called the Mach disc, occursabout 10 mm behind the sampling aperture. Beyond the Mach disc, the flow becomessubsonic and remixes with the surrounding gas.The skimmer aperture (typically from 0.7 mm to 1.0 mm) is positioned at adistance less than the Mach disc (- 6.5 mm) behind the sampling cone. Once past theskimmer aperture, the extracted gas enters a region where the pressure is low enoughthat the mean free path (distance between collisions) is longer than the system length.At this point, the ions can be focused by the ion lenses.Ion Lens Stack2Figure 1.5^Ion extraction and ion lens configuration. (1) photon stop (2) skimmer cone (3) samplercone1.2.6 Ion Lens and Mass AnalyserThe function of the ion lens is to focus as many ions as possible from the ioncloud behind the skimmer into an axial beam at the quadrupole mass analyserentrance. A photon stop in the ion lens stack prevents photons emitted from theplasma to reach the channel electron multiplier detector.The quadrupole mass analyser operates as a tunable variable bandpass massfilter where ions are resolved (0.5-1.0 amu resolution) on the basis of the mass-to-charge (m/z) ratio. This method of separation is different from conventional massspectrometers, such as magnetic sector and time of flight mass spectrometers, whereresolution is achieved on the basis of momentum and kinetic energy respectively. Thequadrupole resolution is independent of momentum and kinetic energy, therefore, unitmass resolution is retained even when the sampling ion population possesses a widerange of velocities along the quadrupole axisl 1 . Advantages of the quadrupole overmagnetic sector mass analysers include lower costs, a more compact size, shorterscan times and higher durability.9quadrupole rodsto detectorfrom ICP10Although a thorough, quantitative description of quadrupole operation isbeyond to the scope of this thesis, it is useful to understand, in a qualitative manner,the operations of a quadrupole mass analyser.The quadrupole mass analyser consists of four relatively short, parallel, circularelectrodes (Figure 1.6) arranged symmetrically around the ion beam. A combinationof dc and ac (RF) potentials are applied to each rod. Rods opposite to each other areelectrically connected and are attached to the same side of the dc potential. Thus, onepair of electrodes is connected to the positive side of the dc source while the other pairis connected to the negative terminal. The RF ac potentials are also applied at 180°out of phase with respect to each pair of electrodes. These combined fields cause theions to oscillate about the central axis.Figure 1.6 Quadrupole mass analyser and axis orientation.Consider the quadrupole electrodes in the X-Z plane, when only the RF acpotential is applied. These electrodes spend 1/2 of the ac cycle at a positive potentialand the other 1/2 of the cycle at a negative bias. During the positive 1/2 cycle, thepositively charged ion beam is repelled by the positive bias on the electrodes, and isin essence, focused towards the centre of the Z-axis. Alternately, during the negative1/2 cycle, the positively charged ion beam is accelerated towards the negativelybiased electrodes. Whether the ion will be removed during its time spent in thequadrupole will be governed by the time the ion needs to reach the negatively biased1 1electrode (during the negative ac cycle) which, in turn, is dependent on the magnitudeof the applied potential, the duration of the applied potential, and the m/z ratio of theion. The ion velocity in the Z-axis is also important. Although the applied fields do notaffect the velocity in the Z-axis, and this velocity is not involved in mass resolution, theion must be traveling slow enough to spend at least one ac cycle in the quadrupole.Now consider the application of a positive dc potential to the electrodes on theX-Z plane. If an ion is very heavy, it would only feel the effects of the average potentialapplied, that is, the dc potential. The small periods of time that the electrodes are at anegative bias will have little or no affect on the heavy ions. Conversely, a very light ionwill be affected much more by the changing ac potential (in comparison to the heavierion) and will cause it to collide with the electrode during the negative 1/2 cycle. Thus,ions below a given m/z ratio will be eliminated by the defocusing action of the negativeportion of the ac potential. Thus, the electrodes in the X-Z plane act as a high-passmass filter.Consider the electrodes in the Y-Z plane whose applied potentials are equal inamplitude by opposite in sign to those on the X-Z plane. In this instance, the acpotential acts to focus the lighter ions onto the Z-axis and prevent them from strikingthe dc negatively biased electrodes. Since the heavy atoms are affected by theaverage value of the applied negative potential, they will be eliminated above somecritical m/z ratio. In this instance, the electrodes in the Y-Z plane act as a low-passmass filter.In order for an ion to travel from the source to the detector, it must be stable inboth the X-Z and Y-Z planes. Thus, the amplitude of the RF and dc potential determinethe transmitted mass. In addition, it can be shown that the resolution is determined bythe ratio of the RF and ac potentials. One of the limitations of the quadrupole massanalyser is that the resolution attainable is about 1 amu which is not enough toseparate an isobaric oxide peak from an elemental peak (e.g. atioAriso+ from 56Fe+).121.2.7 Ion DetectionIons transmitted through the quadrupole are detected by a channel electronmultiplier. Typically, a continuous dynode channel electron multiplier is used.Channel electron multipliers are made from heavily lead-doped glass and arerelatively small (1 mm id x 70 mm). The tube is curved for reasons that will beaddressed shortly.A high voltage is applied (-3000 V to -3500 V) to the multiplier. The appliedvoltage has two functions. The first is to replenish the charge on the channel wall. Thesecond is to accelerate the secondary electrons, initially with low energy, fast enoughsuch that upon collision with the surface, they will create efficiently secondaryelectrons. Positively charged ions from the mass analyser are attracted to the negativevoltage in the funnel of the multiplier. The multiplier is placed off axis to reduce thechance of detecting stray photons.Once a positive ion strikes the funnel, at least one secondary electron isemitted. The secondary electrons accelerate down the tube, due to the appliedpotential, until it impacts with the wall. Collisions with the wall free other secondaryelectrons and the avalanche builds exponentially. With high electron fluxes, such asthose found in the exit of the multiplier, residual gas in the multiplier can be ionized.These gaseous ions have the potential to accelerate back to the input of the multiplierand begin another avalanche of electron. This is called ionic feedback. The curvatureof the multiplier allows these positively charged ions to travel only a short distancebefore encountering the multiplier wall. At this stage of the multiplier, the gain is smalland the effects are negligible. The cloud of -10 7-108 electrons (originating from eachpositive ion) leaving the multiplier are gathered by a collector electrode.. There are numerous advantages to using a channel electron multiplierincluding excellent signal-to-noise ratios, low dark counts (<0.5 counts s -1 ), stabledynode surface that can be exposed to air without degradation, low power13requirements for operation, tolerance to high operating pressure (up to 10 -5 mbar) anda reasonable long life. Linear response is typically 5 to 6 orders of magnitude.Two modes of operation are possible: pulse counting mode and analog mode.In pulse counting mode, the output pulses above a set discriminator level are counted.In essence, individual ions may be counted giving the highest possible sensitivity. Inthe analog mode, the applied voltage is only -1500 V. The electrons are collected bythe normal collector electrode, but the current is then amplified by a linear amplifier.The response of the amplifier is fast enough that the output accurately reflects thecurrent.1.3 Advantages and Limitations of the ICP-MSSome of the advantages of the ICP-MS system includes:1. Low background level2. High sensitivity3. Simple spectra4. Isotope ratio capabilities5. Fast sample throughput (when using nebulizer)6. Multielement capabilitiesBackground levels obtained on the ICP-MS are very low, -10-50 counts s -1 . Coupledwith high sensitivities obtained from pulse counting, the ICP-MS system is able toachieve very low detection limits. The spectra obtained from the ICP-MS are verysimple due to the high energy inert ICP ion source, which also effectively desolvates,volatilizes, atomizes and ionizes most elements. Very few molecular species areformed resulting in few isobaric interferences. ICP-MS is a very fast, sequentialmultielement technique that is capable of isotope measurements. Isotope ratiomeasurements are important for isotope dilution techniques as well as in stableisotope tracer experiments which are often used in biological uptake studies. Thespeed of this technique is due mostly to the ability to ionize the sample at ambient14pressures rather than at low pressures as is common with other mass spectrometrictechniques such as SSMS.With this in mind, some of the limitations of the system are also presented.These include:1. Limited solution introduction2. Matrix induced analyte enhancement/suppression3. High initial and maintenance costsAs with inductively coupled plasma atomic emission spectroscopy, ICP-AES,sample introduction poses one of the greatest limitations to the overall system. Themost common form of sample introduction is solution nebulization which suffers fromlow transport efficiency and the introduction of large quantities of 0, N, S, and CI fromthe water or acid matrix into the ICP. For example, Ar0+ and ArN+ interfere with themajor isotopes of Fe. Tan and Horlick have studied extensively the molecularinterferences that originate from various acid matrices 10 . In addition to low transportefficiencies, the nebulization and sampling systems cannot tolerate high salt matricesdue to the clogging of the nebulizer and the sampling cone orifice. Concentratedmineral acid matrices are also undesirable as they lead to accelerated conedeterioration. Typically, sample introduction is limited to dilute, well-behaved solutionsand gaseous samples. Solid samples require different methods of sampleintroduction and will be discussed later.Signal suppression observed in the ICP-MS is higher than that observed inICP-AES system and has been found to be mass dependent. It has been proposedthat the increased analyte suppression is related to space charge effects 11 . Spacecharge effects originate from the electrostatic interactions in the ionic beam. Once theion beam has been extracted, coulombic interactions cause the positively chargedions to repel each other. The lighter ions are repelled from the central ion beam morethan the heavier ions, resulting in lower overall extraction15efficiency for lighter elements. High initial investment and the high maintenance costsare other disadvantages of the ICP-MS system.1.4 Alternate Methods of Sample IntroductionThe method of sample introduction has often been the weak link in elementalanalysis. Solution nebulization, the most commonly form of sampling introduction, isplagued with low transport efficiency and molecular oxide interferences. Attempts toremedy these problems, as well as extending the types of samples which may beanalysed have been the focus of much recent research effort.1.4.1 Increased Transport EfficiencyThe transport efficiency of a Meinhard nebulizer and spray chamber system is<1%. Much of the solution is removed in the spray chamber because the dropletsformed by the nebulizer are too large.Wiederin et al. has recently developed a direct injection nebulizer (DIN) thatworks on the same principle as the pneumatic nebulizer12 . The DIN is capable ofproducing a very fine aerosol which can be introduced directly into the plasmaresulting in very high transport efficiencies. Low flow rates (20 gliminute) are usedand so only small sample sizes are required. Oxide to ion ratios were found toincrease and the system facilitated on-line standard addition to reduce samplepreparation time 13 .A direct sample insertion (DS!) device, developed by Karanassios and Horlick,also gives very high transport efficiencies 14 . This device is essentially a small cupmade of graphite, Mo or Ta. The analyte is carried by the cup directly to the base ofthe plasma, through the aerosol injector tube, where it is dried and vaporized, by theheat of the plasma. The analyte vapor is transported directly into the plasma resultingin 100% transport efficiency. Another advantage of this device is that both solutionand solid (in the form of a powder) samples may be analysed with this device.16An electrothermal vaporization device (ETV), developed by Shen and Caruso,also improves the analyte transport efficiency to the ICP-MS 15 . This device is verysimilar to the graphite furnace used in GFAAS where the sample is dried, ashed andvaporized in a graphite tube. The analyte vapor is then transported to the ICP-MS foranalysis. As with the DSI device, both solutions and powdered solid samples may beanalysed.1.4.2 Reduction of Molecular InterferenceOne of the major sources of interference originates from the acid matrix which isa source of oxygen for the molecular oxide interferences. Advances have been madein this area to modify the method of sample introduction such that the analyte istransferred to the plasma in the absence of oxide interference.Slight modifications to the normal (nebulizer) mode of instrument operationincluded a cryogenic desolvation system which is effective in reducing the oxidemolecular interferences by several orders of magnitude. This is achieved by "freezingout" the matrix prior to aerosol introduction into the ICP 16 . Several studies have shownthat introducing small amounts of a molecular gas, such as N 2 17-20 , xe21 or an organicsolvent 20 into the plasma gas also reduces molecular oxide and chlorideinterferences, prevalent interferences in Fe+ (Ar0+) and As (ArCl+) analyses. Forexample, addition of Xe (ionization energy 12.1 eV) shifts the ionization equilibriumsuch that it is lower than the ionization potential of molecular oxides (ionization energy-15 eV). In effect, the presence of Xe 'caps' the maximum ionization energy within theplasma and prevents the ionization of molecular species. These modificationsmaintain the favorable characteristics of the nebulizer, namely high sample throughputand simplicity of operation.Alternately, changing the method of sample introduction can be used toeliminate oxide molecular interferences. Karanassios' and Horlick' s 14,22,23 directsample insertion device was shown not only to offer 100% transport efficiency, but also17to reduce signals attributed to molecular oxide and hydride interferences. The ETVhas been shown to reduce Ar0+ interference with 56Fe by removing the matrix prior tovaporization. Laser ablation sample introduction have been utilized by several groupsto vaporize solid samples directly into the plasma 24 .1.4.3 On-Line Matrix SeparationOn-line preconcentration and matrix separation have also been studied byseveral groups. Caruso et al. used on-line anodic stripping voltammetry (ASV) topreconcentrate copper and cadmium 25 . Samples are passed through the ASV cellwhere desired analytes are stripped from a predetermined volume of sample at a fixedpotential. Output from the ASV cell was directed to waste. Once the sample volumehad completely passed through the ASV cell, the cell was washed and the flow thenredirected to the ICP and the potential applied was such that the analyte was released.The ASV method of sample introduction has the dual effect of concentration as well assimultaneously eliminating matrix effects such as high concentrations of Na and U.McLaren et al. has developed an on-line preconcentration system using achelating resin in micro-columns for simultaneous determination of Fe, Mn, Co, Ni, Cu,Zn, Cd, Pb in seawater such that accurate analyses of these trace metals may beperformed in less than ten minutes 26 . This greatly reduces the sample preparationtime and the sample size required for determining these metals in high salt matrices.1.4.4 Speciation StudiesTypically, analysis with ICP-MS gives information on total analyte concentrationwithout any information on analyte speciation. Combining the separation technologyof HPLC with detection by ICP-MS allows speciation studies to be carried out. Oneelement that has been studied using this technique is As. Some species of arsenic(As'll, Asv) are highly toxic while others (methylarsonic acid, dimethylarsonic acid,arsenobetaine) are much less toxic. Since normal mode ICP-MS can only18determination total As content, it cannot accurately produce information regarding theAs toxicity of a sample.Arsenic is monoisotopic and has an isobaric interference from ArC1+ at m/z 75.The use of HPLC as a method of sample introduction also allows one to separatechloride containing species from the As signal27-30 . High sensitivity of the ICP-MS hasallowed these As studies to be performed for samples with very low concentrations orfor systems that are sample limited (such as human urine or dogfish analyses). Otherelements that have been analysed using HPLC-ICP-MS include Zn 31 and Cr32 .1.4.5 Electrothermal Vaporization Sample IntroductionOne of the approaches to sample introduction that overcomes several of theshortcomings of solution nebulization, such as low transport efficiency and molecularinterferences, is electrothermal vaporization (ETV) ICP-MS. Electrothermalvaporization devices have found application in ICP-AES since 1974. 33 In ETV-ICP-AES, the furnace vaporizes the analyte. Analyte vapor is then transported to the ICPwhere atomization and ionization occurs. This is different from GFAAS where theanalyte is vaporized and atomized in the graphite furnace. Until recently, littleattention has been paid to the use of ETV as a method of sample introduction for anICP-MS. The addition of Freon (CHF3) forms volatile metal halides within the furnace.The metal halides are then introduced to the plasma where they are atomized andionized. The use of Freon has opened up the use of ETV-ICP-MS for analysis ofrefractory metals.To date, only two ETV designs have been used in research. The first is simply acommercial electrothermal device modified from GFAAS to allow for analyte transportto the ICP. The second is a "home-made" design by Park et al. first developed at theUniversity of Toronto 34 . Both of these systems will be considered in some detail.The "home-made" system (Figure 1.7) consisted of a Re filament encased in aglass envelope. Temperature control for the Re filament is achieved using a variable1 9voltage transformer. An 0-ring seals the glass envelop to an Al base. Argon isallowed to flow through the device and carry the analyte to the ICPFigure 1.7^Electrothermal vaporization device developed by Park et al. 35 (1) nebulizer gas inlet(2)-Teflon® block (3) electrodes (4) metal filament (5) quartz cover (6) aluminum baseIt was found that a considerable fraction of the sample was lost as a result ofcondensation on the glass walls. However, if the analyte vapor was cooled to formmieroparticles, condensation losses become negligible. Cooling of the analyte vapormay be accomplished by introducing the carrier gas in a tangential fashion. Theenvelop volume above the filament was optimized to 5 mL. Larger envelop volumesresulted in temporal peak broadening due to excessive dilution. Smaller volumes giverise to high aerosol condensation on the glass envelop.This device was used to analyse metals, such as Mo, W35 , platinum groupmetals36 , TI 37 , rare earth elements 38 , Cu, Cd, and Ni34 . Matrix separation of Fe fromthe oxide interference, Ar0+, was evident. Using this system, they found that matrixeffects found using the ETV was attributed to interelement compound formation on thefilament rather than plasma loading effects39 A comparison of ETV with DSI sampleintroduction system found that although both systems only required small sample20volumes, the ETV provided a more controlled environment for matrix separation 40 .Typically, relative detection limits using the ETV are improved by an order ofmagnitude over the nebulizer ICP-MS and precision ranges from 8-19% RSD forabsolute determinations and 1-2% RSD for isotope ratio measurements 35 .Alternately, Caruso and co-workers 15,41 have modified a commercial Perkin-Elmer HGA-300 electrothermal unit for atomic absorption such that an argon carriergas may be used to transport the analyte to the ICP (Figure 1.8).Figure 1.8 Electrothermal device produced by modifying an HGA-300. (1) front adaptor (2)-coolingblock (3) temperature sensor (4) graphite furnace (5) Lvov platform (6)-carrier-gas inlet(7) rear adapter (8) drying ventSample introduction was achieved by inserting the tip of a pipet into the furnacethrough the opening for the rear adapter and depositing the required amount ofsolution onto the platform. This device successfully showed the reduction of isobaricoxide and chloride molecular species for Fe and As respectively. Detection limits of1.5 pg and 0.2 pg were obtained for As and Fe respectively with absolute precisionsbetween 5-8%15.211.5 Trace Metals in the Ocean1.5.1 BackgroundThe introduction of the GFAAS and, more recently, ICP-MS has provided thechemical oceanographic community with extremely useful methods for the analysis oftrace metals. Prior to these analytical methods, detection limit requirements for theanalysis of trace metals in seawater could not be achieved by standard instrumentalmethods, such as flame atomic absorption (FAAS), flame atomic emission (FAES) orinductively coupled plasma atomic emission spectroscopy (ICP-AES). Even with largeconcentration factors, detection limits of FAAS and FAES spectroscopy were not lowenough to detect the trace amounts of analyte present in seawater. GFAAS providesthe very low detection limits that are required for trace metal detection afterpreconcentration, however, this method is limited to relatively volatile elements.Detection limits for refractory metals and metals which formed refractory carbides andoxides remain high. The recent addition of ICP-MS allows analysis of the refractorymetals not readily determined by GFAAS. Owing to the high temperatures present inthe ICP, most elements are completely ionized. Combined with the increasedsensitivity of the MS, the ICP-MS has proven to be a powerful tool for the analysis oftrace metals. The much superior detection power of ICP-MS for most of the elements,coupled with recent advances in trace metal clean sampling methodology, permitsnearly simultaneous multielement determination with far better precision and accuracyfor more metals than those previously obtained using more traditional methods.Trace metals are defined as those present in sea water at concentrations lessthan 1 ppm. Trace element behavior can be, in a general sense, divided into threecategories. Some trace metals behave conservatively (Figure 1.9 A), that is, theconcentration varies only with salinity due to very slow removal from the oceans. Slowremoval from the water column leads to concentrations which are high, relative to thecrustal abundance. This type of element, therefore, is accumulated in sea water until asteady state is achieved. Conservative elements, usually high in concentration,Afi22have long residence times in the oceans and include most alkali metals (such aslithium, caesium) as well as some oxy-anions of Mo and U 42 .Concentration^Concentration^ ConcentrationFigure 1.9 Three representative oceanic distributions. (A) Conservative (B) Nutrient (C) Scavenged typeprofilesSome trace metals exhibit nutrient type behavior (Figure 1.9 B). Theseelements are not necessarily nutrients, but they do show a strong correlation withmacronutrient distributions. Such distributions typically show surface depletion, due tobiological uptake, and increasing concentrations with depth from release duringdecomposition of the organic material or dissolution of skeletal material. Someelements which show nutrient type behavior include Cd and Zn42 .Trace metals can also show scavenged or reactive behavior (Figure 1.9C).These elements are rapidly removed, typically by adsorption onto sinking particles,resulting in spatial and temporal variability as well as short oceanic residence times.Trace metals displaying scavenged behavior have distributions that strongly reflect thelocation of external sources, with concentrations in waters distant from the sourcemarkedly depleted. Such sources originate at ocean boundaries. Thus, scavengedmetals are valuable as tracers for elucidating the transport and mixing mechanisms inthe oceans.23One example of a good tracer is aluminum. The dominant input of Al in theNorth Pacific is from aeolian (atmospheric) inputs 43 . Evidence supporting this premiseis the low dissolved aluminum concentration in the low salinity surface waters of theCalifornia current (0.3 nmol/kg), where estuarine input is high, and increasingconcentrations into the North Pacific subtropical gyre (5 nmol/kg). The observed Algradient is consistent with decreasing input of aeolian particulate matter from west toeast, as atmospheric dust is carried from Asia by the westerlies towards NorthAmerica. Riverine input of dissolved aluminum has been found to be removed inestuarine and coastal regions, with little or no net input to the open ocean. Aluminumconcentrations in the California current range from 0.3 to 5 nmol/kg.Dissolved Al concentrations vary in the deep waters by 2 orders of magnitudefrom ocean to ocean. Thus, it is useful for tracing movement and mixing of deepwaters. Vertical distributions of Al show a mid-depth minimum with subsequentincreases in concentration in deep waters. These observations suggest a deep watersource and removal throughout the water column. The proposed removal mechanismis particle scavenging, which is intensified in coastal regions. Aluminum sources tothe deep ocean could be from sediment surface remineralization processes and/ordiffusion out of the sediments.1.5.2 Determination of Trace MetalsAs mentioned previously, the analysis of trace metals in seawater has beenlimited by their presence at extremely low levels and the limited instrumentalcapabilities available. Large improvements in sensitivity and detection limit madepossible by the ICP-MS allow many trace metals to be determined with accuracy.Seawater, however, cannot be nebulized directly into the ICP-MS. Tracemetals in seawater are present at such low levels that preconcentration is stillrequired. In addition, high salt content (-3.5%) of the sea water matrix can not be24tolerated by the nebulizer or the sampling orifice. Thus, the desired analyte must beisolated from the sea water prior to analysis.One of the most common methods of preconcentration is via solid-liquid phaseextraction. Seawater must be pumped through a chelating resin at optimizedconditions. The chelating resin selectively adsorbs the desired analytes, after whichthe analyte may be eluted using a relatively small volume of strong acid.Several different chelating resins are used for such extraction andpreconcentration processes. These include Chelex-100, silica immobilized 8-hydroxyquinoline and vinyl polymer (TSK) immobilized 8-hydroxyquinoline resin.Chelex-100, a commercially available resin has been utilized in extracting may of thefirst row transition elements 44,45 . The advantages of using Chelex include high resincapacities (700 p.mol Cu(II)/mL of resin) and commercial availability. However, Chelexsuffers from undesirable swelling and contracting with changing pH. Chelex alsoretains significant amounts of Ca and Mg, major ions present in seawater, even afterthorough rinsing.Another commonly used resin is the silica immobilized 8-hydroxyquinolineresin, first developed by Sturgeon et al. at the National Research Council of Canada46 .Again, most of the first row transition elements as well as Cd and Pb, have beenconcentrated and analyzed using this resin. Unfortunately, this resin is notcommercially available and has lower chelating capacities than that of Chelex-100.The silica immobilized 8-hydroxyquinoline resin exhibits good physical stability as wellas faster exchange kinetics when compared to Chelex-100.A third resin that has shown promise was first synthesized by Landing et al.,where 8-hydroxyquinoline was immobilized onto a vinyl polymer resin 47 . This solidphase is a highly porous, mechanically and chemically stable hydrophillic resin gel.Capacity of this resin has been reported to be as high as 20-30 .tmol Cu 2A-/mL of resin.TSK 8-hydroxyquinoline resin has been previously used to preconcentrate Al, Cd andmost of the first row transition elements from organic rich freshwater samples47.251.5.3 Marine Geochemistry of Ti, Ga and In in SeawaterLittle is known about the behavior of dissolved titanium, gallium and indium inseawater owing to difficulties in their analysis. Recent studies using adsorptivecathodic stripping voltammetry have shown dissolved titanium concentrations of 0.2-67 nM in estuarine waters 48 . However, the detection limit for this method was 0.03 nMand is inadequate for determination of Ti in open surface ocean waters. Investigationsusing solid-liquid extraction-preconcentration similar to those described above,followed by ICP-MS analysis, show that titanium exists at concentrations of 5-300 pMin the open ocean with a concentration range in the water column of over 2 orders ofmagnitude49 . Dissolved titanium has a vertical distribution similar to that of copper, anelement controlled by a complex combination of nutrient type cycling and scavenging,which suggests that Ti may also be controlled by such complex mechanisms. Thislarge variability within the water column could lead to the use of titanium as a valuabletracer of ocean mixing and transport.Recent studies of gallium by solid-liquid extraction followed by graphite furnaceatomic absorption spectroscopy (GFAAS) analysis, suggest complex input andremoval mechanisms govern gallium distributions in the oceans50,51 . Galliumconcentrations found in these studies were between 2-30 pM. Earlier studies usingFe(OH)3 co-precipitation followed by spectrophotometric analysis showed values overan order of magnitude higher than what is now believed52 .Early investigations for dissolved indium were carried out with similar solid-liquid extraction followed by very time consuming (irradiation was carried out forseveral weeks) neutron activation analysis 53 and thermal ionization massspectrometry with isotope dilution 54 . Estimate values for indium concentrations inseawater from these studies ranged from 1-35 pM.261.6 Aims of This StudyThe desire to understand the factors that control the distribution of trace metalsin the ocean have led to the search for sensitive and accurate techniques for theirdetermination. The presence of these metals at extremely low concentrations as wellas the complex seawater matrix in which they exist has proven to be an immensechallenge to the analytical chemist. Clean techniques, such as those developed byBruland and coworkers 55 , have provided contamination control during sampling andstorage, so that under proper processing and analysis, accurate and meaningfulresults may be obtained.The sensitivities made possible by the ICP-MS have led to significantadvancements in trace metal analysis. However, many of the second and third rowtransition metals and many of the main group elements are still difficult to detect withICP-MS, even with concentration factors of 1000. Since pneumatic nebulization has atransport efficiency of <1%, an increase in transport efficiency could conceivablyincrease sensitivity by two orders of magnitude. Changing the sample introductionmethod to electrothermal vaporization has previously demonstrated increasedtransport efficiency and, therefore, sensitivity. Chapter 2 of this thesis is devoted toevaluating the capabilities of an ETV device for low level determinations. Theincreased sensitivity and lowered detection limits should make significantcontributions to the determination of trace metals in seawater.Ultra-trace metal analysis in seawater is a relatively new field of study. Little isknown about the behavior of many elements in seawater due to the above mentionedinstrumental limitations. Very little is known about the behavior of Ti and Ga and evenless is known about In in the oceans. Titanium and Ga distributions in the easternNorth Pacific region have demonstrated a bottom source and scavenged behavior inthe water column. Gallium is seen to have an atmospheric source in the same region.Chapter 3 will focus on a method for determining of Ti, Ga and In in seawater. Thismethod will be used to investigate these elements in the Central Pacific gyre. These27data, combined with previous studies from the Western Pacific gyre, will be used tofurther elucidate the control mechanisms of these elements in the Pacific ocean.The ability to preconcentrate trace metals from seawater is of vital importance inthis work. To date, many of the preconcentrations have been performed by solid-liquidextraction with an 8-hydroxyquinoline immobilized resin that required in-housesynthesis. Difficulties encountered in successfully synthesizing this resin led to thethird portion of this thesis. Chapter 4 describes attempts to improve the synthetic routeas well as the development of a possible new ion exchange resin.281.7 References and Footnotes1. Gray, A.L. J. Anal. At. Spectrom. 1986, 1, 403-405.2. In spark source mass spectrometry , the analyte is ionized by a highpotential (-30kV) radio frequency spark under high vacuum. The ions areaccelerated toward the mass analyser, situated adjacent to the spark source,by a dc potential.3. Houk, R.S.; Fassel, V. A.; Flesch, G. D.; Svec, H. J.; Gray, A. L.; Taylor, C. E.Anal. Chem. 1980, 52, 2283-2289.4. Scott, R. H.; Fassel, V. A.; Kniseley, R. N.; Dixon, D. E. Anal. Chem. 1974,46, 75-80.5. Skoog, D. A. Principles of Instrumental Analysis 3rd ed.; CBS CollegePublishing: New York, 1985; Chapter 10.6. Date, A. R.; Gray, A. L. Applications of Inductively Couples Plasma MassSpectrometry; Chapman and Hall, New York, 1989; Chapter 1.7^Ingle, J. D.; Crouch, S. R. Spectrochemical Analysis; Prentice-Hall, NewJersey, 1988; Chapter 8.8. Gray, A. L.; Date, A. R. Analyst 1983, 108, 1033-1050.9. Miller, P. E.; Denton, M. B. J. Chem. Ed. 1986, 63(7), 617-622.10. Tan, S. H.; Horlick, G. Appl. Spectrosc. 1986, 40(4), 445-460.11. Hieftje, G. M. Spectrochim. Acta 1992, 478(1), 3-25.12. Wiederin, D. R.; Smith, F. G.; Houk, R. S. Anal. Chem. 1991, 63, 219-225.13. Wiederin, D. R.; Smyczek, R. E; Houk, R. S. Anal. Chem. 1991, 63,1626-1631.14. Karanassios, V.; Horlick, G. Spectrochim. Acta 1989, 44B(12), 1345-1360.15. Shen, W. L.; Caruso, J. A.; Fricke, F. L.; Satzger, R. D. J. Anal. At. Spectrom.1990, 5, 451-455.2916. Alves, L. C.; Wiederin, D. R.; Houk, R. S. Anal. Chem. 1992, 64,1164-1169.17. Ford, M.; Ebdon, L.; Hill, S. J. Anal. Proc. 1992, 29, 104-105.18. Branch, S.; Ebdon, L.; Ford, M.; Foulkes, M.; O'Neill, P. J. Anal. At.Spectrom. 1991, 6, 151-154.19. Beauchemin, D.; Craig, J. M. Spectrochim. Acta 1991, 46B(5), 603-614.20. Evans, E. H.; Ebdon, L. J. Anal. At. Spectrom. 1990, 5, 425-430.21. Smith, F. G.; Wiederin, D. R.; Houk, R. S. Anal. Chem. 1991, 63,1458-1462.22. Karanassios, V.; Horlick, G. Spectrochim. Acta 1989, 44B(12), 1361-1385.23. Karanassios, V.; Horlick, G. Spectrochim. Acta 1989, 44B(12), 1387-1396.24. Pang, H.; Wiederin, D. R.; Houk, R. S.; Yeung, E. S. Anal. Chem. 1991, 63,390-394.25. Pretty, J. R.; Evans, E. H.; Blubaugh, E. A.; Shen, W. L.; Caruso, J. A.;Davidson, T. M. J. Anal. At. Spectrom. 1990, 5, 437-443.26. McLaren, J. W.; Adatsuka, K.; Azeredo, M. A.; Lam, J. W. H.; Berman, S. S.J. Anal. At. Spectrom. In press.27. Sheppard, B. S.; Caruso, J. H.; Heitkemper, D. T.; Wolnik, K. A. Analyst1992, 117, 971-975.28. Sheppard, B. S.; Shen, W. L.; Caruso, J. A.; Heitkemper, D. T.; Fricke, F. L. J.Anal. At. Spectrom. 1990, 5, 431-435.29. Beauchemin, D.; Siu, K. W. M.; McLaren, J. W.; Berman, S. S. J. Anal. At.Spectrom. 1989, 4, 285-289.30. Thompson, J. J.; Houk, R. S. Anal. Chem. 1986, 58(12), 2541-2548.31. Owens, L. M. W.; Crews, H. M.; Hutton, R. C.; Walsh, A. Analyst 1992, 117,649-655.3032. Roehl, R.; Alforque, M. M. At. Spectrosc. 1990, 11(6), 210-215.33. Nixon, D. E.; Fassel, V. A.; Kniseley, R. N. Anal. Chem. 1974, 46, 210-213.34. Park, C. J.; Van Loon. J. C.; Arrowsmith, P.; French, J. B. Can. J. Spectr.1987, 32(2), 29-36.35. Park, C. J.; Hall, G. E. M. J. Anal. At. Spectrom. 1987, 2, 473-480.36. Gregoire, D. C. J. Anal. At. Spectrom. 1988, 3, 309-314.37. Park, C. J.; Hall, G. E. M. J. Anal. At. Spectrom. 1988, 3, 355-361.38. Shibata, N.; Fudagawa, Norike, Masaaki, K. Anal. Chem. 1991, 63,636-640.39. Park, C. J.; Van Loon, J. C.; Arrowsmith, P.; French, J. B. Anal. Chem.1987, 59, 2191-2196.40. Hall, G. E. M.; Pelchat, J. C.; Boomer, D. W.; Powell, M. J. Anal. At.Spectrom. 1988, 3, 791-797.41. Carey, J. M.; Evans, E. H.; Caruso, J. A.; Shen, W. L. Spectrochim. Acta1991, 46B(13), 1711-1721.42. Burton, J. D.; Statham, P. J. Phil. Trans. R. Soc. Lond., A 1988, 325,127-145.43. Orians, K. J.; Bruland, K. W. Earth Planet. Sci. Lett. 1986, 78, 397-410.44. Kingston, H. M.; Barnes, I. L.; Brady, T. J.; Rains, T. C. Anal. Chem. 1978,50, 2064-2070.45. Sturgeon, R. E.; Berman, S. S.; Desaulniers, J. A. H.; Mykytiuk, A. P.;McLaren, J. W.; Russell, D. S. Anal. Chem. 1980, 52, 1585.46. Sturgeon, R. E.; Berman, S. S.; Willie, S. N.; Desaulniers, J. A. H. Anal.. Chem. 1981, 53, 2337-2340.47. Landing, W. M.; Haraldsson, C.; Paxeus, N. Anal. Chem. 1986, 58,3031-3035.48.^Skrabal, S. A.; Ullman, W. J.; Luther, G. W. Mar. Chem. 1992, 37, 83-103.3149. Orians, K. J.; Boyle, E. A.; Bruland, K. W. Nature 1990, 348, 322-325.50. Shiller, A. M. Geochim. Cosmochim. Acta 1988, 52, 1879-1882.51. Orians, K. J.; Bruland, K. W. Nature 1988, 332, 717-719..52. Culkin, F.; Riley, J. P. Nature 1958, 181, 179-180.53. Chow, T. J.; Snyder, C. B. Earth Planet. Sci. Lett. 1969, 7, 221-223.54. Matthews, A. D.; Riley, J. P. Anal. Chim. Acta 1970, 51, 287-294.55.^Bruland, K. W.; Franks, R. P.; Knauer, G. A.; Martin, J. H. Anal. Chim. Acta1979, 105, 233-245.56^Blades, M. W. In Inductively Coupled Plasma Emission Spectroscopy Part 2;Boumans, P. W. J. M., Ed.; John Wiley & Sons: New York, 1987.32Chapter 2Evaluation of Electrothermal Vaporization as a Method ofSample Introduction for the ICP-MS2.1 IntroductionThe introduction of the ICP-MS in 1983 has allowed great advances in tracemetal analysis. Detection limits obtained by ICP-MS rivaled those of GFAAS forrelatively volatile elements and surpassed GFAAS for refractory elements.Geochemists and chemical oceanographers, require even lower detection limits,reduced isobaric interferences from the acid matrix and reduced sample sizerequirements. Clearly, modifications to the conventional system are required.An electrothermal vaporization device, used as a method of sampleintroduction, appears to be well suited to these demands. The reported increasedtransport efficiency of this technique should increase sensitivity thereby loweringthe detection limits. Controlled drying and ashing steps should allow selective acidmatrix removal and the inherently small volumes required by the ETV (10 - 50 iit)is ideal for use with low level, sample limited analyses.This chapter describes the optimization and evaluation of an ETV device forthe purposes of ultra-trace elemental detection with applications to seawateranalysis.2.2 Experimental2.2.1 General InstrumentationThe ICP mass spectrometer used for this work was a VG PlasmaQuad® PC)2Turbo Plus (VG Elemental, Winsford, Cheshire, U.K.). A quartz torch made in theFassel configuration was used for much of the work in the nebulizer and ETVmodes of operation. Coolant, auxiliary and nebulizer gases were controlled by33mass flow controllers. The sampling interface consisted of nickel sampling andskimmer cones with aperture diameters of 1.0 mm and 0.7 mm respectively.Vacuum in the intermediate and analyser chambers is maintained by turbomolecular pumps. Detection of the transmitted ions was performed by a channelelectron multiplier and subsequent amplifiers. Typical operating parameters areshown in Table 2.1.Parameter Standard Settings UsedCool Gas 13.75 L/minAuxiliary Gas 0.500 L/minNebulizer Gas 0.800-0.825 L/minIncident Power 1350 WReflected Power -2 W (ETV), - 8 W (nebulizer)Plate Current 0.69 APA grid 0.22 mAPA Volts 0.4 mVPA Filament 0.64 mVExtraction Lens -100 V -^-212.5 VCollector Lens -18.49 VL1 -2.11 VL2 -77.9 VL3 10.68 VL4 -44.1 VPole Bias -4.68 VTable 2.1^Typical operating conditions for ETV and nebulizer modes.2.2.2 Data AcquisitionData acquisition from the ICP-MS was performed using an on-lineCOMPAQ-386 personal computer equipped with VG Plasmavision® softwaresupplied by VG Instruments.34Pulses from the signal handling electronics during an acquisition werecounted into memory channels of the multichannel analyser (MCA) andsubsequently transferred to the computer once the acquisition had beencompleted.Two modes of data acquisition were used in this work. The first mode waspeak jumping (PJ) in which ion counts were acquired, sequentially, at specifiedmasses. The second mode was single ion monitoring (SIM) where only one masswas sampled over time.2.2.3 SolutionsAll reagents were prepared using distilled deionized (DDI) water (1 . 8 MO,Nanopure, Barnstead). The acid used was environmental grade nitric acid (HNO3)(Anachemia). Standard solutions were prepared by serial dilution of 1000 ppmatomic absorption standards (Johnson Matthey Inc., Seabrook, N.H.) with theappropriate acid matrix in acid leached polyethylene bottles.2.2.4 Electrothermal Vaporization (ETV) Device2.2.4.1 ETV InstrumentationThe electrothermal vaporization device was a MicroTherm II, also suppliedby VG Elemental (Figure 2.1). The ETV furnace was a Perkin Elmer-type graphitetube (Buck Scientific) mounted between two carbon contact blocks whichthemselves were recessed in two nickel plated copper contact blocks. Water(15°C) was circulated at 2L/min through channels machined in the blocks. Furnacetemperature was regulated by a thermister at temperatures below 650°C and by anoptical pyrometer at higher temperatures. A sheath of argon gas maintainedaround the furnace prevented oxidation of the graphite tube during the heatingcycle. This sheath was contained by a quartz ring situated around the graphitetube with the flow of argon at 1-2L/min. A small hole in the quartz ring facilitated35sample introduction as well as the insertion of the carbon rod to seal the graphitefurnace. A sliding probe carrier housed the nebulizer gas inlet. The system washeld together by a pencil spring pack which pushed the sliding furnace blocktowards the stationary furnace block to facilitate contact between the graphitefurnace and the carbon contacts. Analyte vapor produced during the vaporizationstep is then carried through the quartz tube leading from the furnace, then throughTeflon transport tube which leads directly to the injector tube in the ICP torch.Freon-2,3 (CHF3) gas (Matheson), used for refractory metals, was regulated by amass flow controller (Edwards 1105).To ICP1Figure 2.1 ETV original furnace design. 1) Cooling water inlet 2) Cooling water outlet 3) Screwretaining spring pack 4) Argon gas inlet 5) Support rod 6) Sliding probe carrier 7) Slidingfurnace block 8) Quartz shroud 9) Graphite tube 10) Carbon bush/contact11) Stationary furnace block 12) ETV cabinet 13) Quartz tubeSample introduction into the graphite furnace was accomplished byinserting the tip of a pipet (Eppendorf®, NY, USA) into the furnace opening anddepositing, typically, 254 of solution onto the furnace wall. A graphite pencil wasthen used to seal the opening of the furnace in order to prevent sample loss duringvaporization.362.2.4.2 ETV ModificationsSeveral modifications were made to the EN device (Figure 2.2 and Figure2.3). The Swagelock fittings provided by VG Instruments for connecting the quartztube from the furnace to the Teflon® transport tubing leaked. This posed severe airleak problems when trying to light the plasma. A "jam fitting" of a tapered quartztube (5/32" id, 1/4" o.d.) against the Teflon® transport tubing inside a piece of largerethyl vinyl acetate tubing (1/4" id, 3/8" o.d.) was found to eliminate the air leak.Other modifications included the addition of a stronger spring to push theheating blocks together. It was found that the pencil spring provided with the ETVwas not adequate in applying enough pressure to give good contact between thegraphite furnace and the heating blocks. The new spring was used to push thesliding gas housing against the sliding furnace block (Figure 2.3).Vinyl Acetate Tubing Quartz TubePFA tubing^Teflon® tapeFigure 2.2 Modifications to tubing configuration leading from the ETV unit to the ICP-MS.The general mode of ICP-MS operation was also altered during ETVoperation. It was found that the seal in the quartz ball-and-socket joint whichconnected the tubing to the torch was compromised after threading the stiff Teflon®tubing through the various holes provided in the torch box and the hood to connectthe tubing to the ETV device. Again, air leaks were evident which made lighting theplasma very difficult.37Figure 2.3 Modifications to the ETV unit. (14) Additional springTo remedy this problem, the instrument hood was left open and an aluminum sheetequipped with a welder's shield (to allow viewing of the plasma) :was placed infront of the torch box to protect the operator from the UV light emitted from theplasma. ETV OptimizationUnivariate optimization parameters included torch position, nebulizer gasflo.v, ash time, ash temperature, vaporization temperature, and acquisition time.Various species were evaluated as a tune species and will be discussed later.2.2.5 Nebulizer Mode2.2.5.1 Nebulizer InstrumentationA four-line peristaltic pump (Gilson) was used to mobilize the carrier flow,sample uptake and waste drainage for the nebulizer mode. The carrier flow ratewas 0.6 mL per minute. Sample uptake is manually controlled by a six-way flowinjection valve fitted with a 300 41... flow loop. Typically, the carrier is 1% HNO3.The carrier bypasses the sample loop during uptake of the six-way valve, and isredirected to transport the sample plug to the nebulizer in the inject position.382.2.5.2 Nebulizer OptimizationGas flows and acquisition parameters were recently optimized by a VGtechnician and were not re-optimized for the nebulizer work in this study. The torchposition and the ion lens were tuned by aspirating a solution of 10 ppb indiumcontinuously through the nebulizer. From the SIM mode, the signal from the 300flow injection loop was determined to be approximately 60 s long. Analytesignal was acquired for the complete duration of this signal.2.3 Results and Discussion2.3.1 Electrothermal Vaporization Optimization2.3.1.1 Modifications To Obtain Better PrecisionInitially, extremely poor signal precision was found for the ETV as suppliedby the manufacturer. This lack of reproducibility may have been due to severalfactors; splattering during the drying step, poor contact between the furnace andthe electrodes or the transient nature of the signal. Upon inspection of the sampleduring the drying stage, no splattering was observed. Poor contact between thefurnace and the electrodes was ultimately found to be the largest contributor to thelack of precision. The addition of a much stronger spring to push the moveable gasinlet housing, which houses one of the electrodes, against the furnace reduced therelative standard deviation of replicate sample injections from 25 - 50% down to3 - 8%.It was thought that precision could again be improved further if the extremelyshort signal produced by the ETV could be extended to allow for more scansacross the signal pulse. Elongation of the ETV signal may be achieved by rampingthe vaporization step. Three types of ramping were used (Figure 2.4); a slow ramp(8 s in duration), a fast ramp (1 s in duration) and a step function to vaporizationtemperature were compared. Initially, an indium signal could be observed at1200°C, thus this was the vaporization temperature selected for the step function39program. To elongate the ETV signal, the profiles ramped "through" thevaporization temperature. Ramp profiles started with a step function to 1000°Cfrom the ash temperature, a ramp portion starting at 1000°C and ended at 1400°Cfollowed by a cleaning step.1500 —213002 1100is900.0 Ashing Stage700.0500.0 I^1.1^I I155 160^165 170^175150time (s)Figure 2.4 Expanded vaporization stage for the various ramping profiles. (1) step function tovaporization temperature (2) 1 s ramp through vaporization temperature (3) 8 s ramp throughvaporization temperatureThe step function gave best sensitivity and precision. The long temperature rampafforded signals approximately 1/4 to 1/3 the magnitude of the signals obtainedwith the step function profile. No improvement in precision was observed with the 8s ramped vaporization step. Fast ramping afforded sensitivity similar to thoseobserved using the step function profile. However, the level of precision obtainedusing the fast ramp was still somewhat inferior to those obtained by the stepfunction profile with average RSD's of 9.5% and 5.5% respectively. This differencein precision may not be a result of the difference in the time to reach thevaporization step since a step function still requires a finite amount of time to reachvaporization temperature. Instead, it is possible that the difference in the way theinstrument electronics handle the step function and the sharp ramp gives rise to thedifference in precision.40Since no advantage was gained by ramping the vaporization stage, the stepfunction was used for the majority of the work done. Effects of Ash TemperatureAshing temperatures that were investigated were 250°C, 600°C, 800°C, and1000°C. It was found that ashing temperatures of 250°C and 600°C gavesensitivities and precision that were quite similar (3-8%). Ash temperature of800°C resulted in unacceptably high RSD's (as high as 59%). Low levels ofprecision obtained at this temperature may be attributed to the temperatureregulation switching from the thermister to the optical pyrometer. Ramping of thetemperature profile through 650°C has been observed to confuse the electronics,leading to uneven drying or sometimes, no drying at all. Ash temperatures at thelow range of the optical pyrometer resulted in poor reproducibility. An improvementin precision at ash temperatures of 1000°C suggests that higher optical emissionfrom the furnace results in better precision than those found at 800°C. However, theprecision is still poor at 1000°C (-17%) and analyte losses will increase withincreasing ash temperatures. For the moment, it would appear that the limit of theash temperature is 600°C for this system unless improvements in the hardware andsoftware can be made. Effects of Vaporization TemperatureVaporization temperature was optimized by varying the temperature andinjecting replicate samples of 1 ppb solutions (Figure 2.5). For high boiling orrefractory carbide forming elements in which freon was used, the optimaltemperatures were identical at -1950°C.41This lead to the conclusion that the limiting factor was the dissociation of freon toform fluorine radicals. Once formed, these radicals react with the element to formvolatile metal fluorides which typically have boiling points that are much lower than1950°C.12000 NamCECc.)a)cf)10000 -08000 —0 6000 —TtC)'CT)a)4000 —To 2000 -00 I^l^I^l}^i^l^1^1^Iii !i^i^i^I1000 1200 1400 1600 1800 2000 2200 2400Vaporization Temperature (°C)Figure 2.5 Vaporization temperature effects on Hf signal.For elements that do not require freon, the optimum vaporizationtemperatures varied from 1950°C for In and Ga to 2200°C for Al. A representativeheating profile is represented in Figure 2.6. The drying step is simply a slow rampto the ashing temperature of 600°C. The ash step is held for 60 s followed by astep function to the vaporization temperature. The vaporization step is held for 5 sfollowed by a clean step, usually at 2400°C, which is also held for 5 s to expel anyresidual analyte that may have remained.42025002000112 1500Et.s.Ec) 1000500.00.000 111111.1 I I i^i 1^i I I^I^.1^120^40 60 80 100^120Time (s)Figure 2.6 Representative of heating profile used in this study. Effects of Nebulizer Gas FlowOptimization of the carrier flow was accomplished by replicate injections of1 ppb solutions of Hf and In at varying nebulizer gas flows. Optimum carrier gasflow was found to be the same for both elements. After each change in the gasflow, the extraction and collector voltages were re-tuned. As expected, theextraction voltage decreased as the nebulizer flow rates increased since theanalyte region is pushed forward with increasing nebulizer gas flow.As can be seen from the graph below (Figure 2.7), the optimum gas flow is at0.8 Urnin. At flow rates higher than this, the residence time of the analyte vapor inthe cooling block of the ETV may be too short for the efficient formation ofmicroparticulates. This would greatly affect the transport efficiency and thus oneobserves a decrease in signal. The optimization of hafnium was performed on twodifferent occasions. The results were therefore normalized to the nebulizer flow at0.8 L/min to account for the differences in sensitivities.431.11.0a)co 0.90• 0.80cc• 0.701■1E• 0.60O0.500.400.30 a7.-`0.5^0.6^0.7^0.8^0.9^1^1.1Nebulizer Gas Flow (1../min)Figure 2.7 Nebulizer Gas optimization on a 1 ppb solution of Hf. Acquisition Parameters - Effects of Dwell TimeDwell times between 1 ms and 10 ms were evaluated. Triplicate injectionsof a 0.7 ppb solution containing 50:50 235 U: 238 U isotopes were performed todetermine isotopic accuracy, sensitivity and precision (Table 2.2).Similar isotope ratio accuracies were obtained for the different dwell timesinvestigated. However, a dwell time of 5 ms was found to give the best isotoperatio precision.44Dwell Time(ms)Average IsotopeRatioIsotopeRatioPrecisionAbsoluteSignalPrecisionAverage CPSat m/z 2381.000 0.975 3.06% 4.17% 7.72 x 1045.000 0.980 0.40% 7.17% 8.02 x 10410.000 0.979 0.53% 14.1% 6.85x 104Table 2.2^Effects of dwell time on Isotopic ratio accuracy, precision and sensitivity and absoluteprecision for a 0.7 ppb solution of 50:50 235 U/238U. The expected isotope ratio for235U:238U is 1.The effects of dwell time on precision may be explained when the method ofacquisition is considered. In the peak jumping mode, which is the mode used here,the time needed per sweep is given by the following:time = number of peaks [quadrupole settle time + (points per peak)(dwell^)]timesweepThe quadrupole settle time is the amount of time the quadrupole waits after'arriving' at a peak prior to acquiring. This helps to clear the signal from theprevious peak and prevent carry-over or memory effects from the previous isotope.The 'points per peak' is the number of points to be sampled in a peak. Dwell timeis the time that the quadrupole will spend acquiring at the desired mass. So, thenumber of sweeps across the signal is not linearly dependent on the dwell time.That is, decreasing the dwell time by 1/2 will not increase the number of sweeps bya factor of 2. So, when the dwell time is decreased, there is an overall decrease inthe amount of time the quadrupole actually spends acquiring even when there isan increase in the number of sweeps.On the other hand, increasing the dwell time would decrease the number ofsweeps during the signal pulse. Since the ETV signal is very short and intense, thenumber of scans over this transient signal is very important. As shown in Figure4500-"ci 4000-0 3500-c.' 3000 -Z1-)ci- 2500-0,2000-=8 1500 -1000 -500 -0 ^0 1 2 3452.8, the majority of the signal intensity from an EN acquisition is very short, -200-300 ms.Time (s)Figure 2.8 Representative signal obtained from the ETV for Th.If only a few scans are obtained over this very short period of time where a signal ofhigh intensity exists, this would result in lowered sensitivity (since the signal isaveraged over the 3 s integration period) and decreased precision. In using theETV as a sample introduction method, care should be taken to optimize the dwelltimes to give optimum sensitivity and precision. Acquisition Parameters - Effects of Acquisition TimeIn the SIM mode, the transient signal observed from the ETV wasapproximately 1.0-1.5 seconds in duration depending on element andconcentration. The sensitivity that could be obtained using the peak jump modewould be optimal if a minimal acquisition time could be used. Unfortunately, thesignal was found to migrate over a window ranging from 0.5-1.0 s from the time ofvaporization. Thus an acquisition time of 2.0 s starting from 0.5 s after the stepfunction was required. Because the integration time was limited by the software tostart and acquire in whole second intervals from the point of vaporization, 3.0 s wasthe minimal acquisition time possible to encompass the transient signal.462.3.1.7 Optimization of Torch Position and Ion Lens TuningTuning the ion lens and the torch position for the ICP-MS in the ETV moderequires a continuous strong and steady signal. Since the signals obtained fromthe ETV are transient, it would be very cumbersome to tune the ion lens to theparticular mass of interest. In order to optimize the mass of interest, multiple ETVruns would be required while changing the lens settings each time until amaximum was obtained. Since the length of time required to complete thisoperation would be so time consuming as to be impractical (estimated to be 6 - 8hours), it was thought that the ion lenses and the torch position could be optimizedusing a species already present in the plasma or using an element which has ahigh vapor pressure, and could be continuously introduced to the system.To avoid contamination of the ICP-MS system, the minor isotope of theargon dimer, Are - 76 was initially chosen 1 , 2 . It was found that tuning to this dimerdid not give consistent results. The lens tuning, especially the extraction voltage,changed quite drastically over short periods of time. In addition, the sensitivityvaried by a factor of 2 to 3 from day to day when using Are - 76 to optimize thesystem. It is thought that by tuning to a molecular species such as Are we may betuning to the "wrong" part of the plasma. The site of maximum molecular ionformation may be different from that of atomic ions. Tuning to Are - 76 wasultimately abandoned and alternate species for tuning were sought.The ideal species needed to tune the ICP-MS ion lens in the ETV mode hadto possess several properties:1. has volatility for ease of introduction2. exhibits no memory effects3. does not interfere with elements of interest4. forms only elemental ions5.^has m/z close to the mass of interest47Using this criterion, iodine was thought to be a good candidate for use as atune element. Iodine is a volatile species, monoisotopic, and does not isobaricallyinterfere with any elements. An aqueous solution of iodine was prepared fromwhich 25 was injected into the ETV and dried at 200°C. The ion lens and thetorch position were then tuned to the 1-127 signal. After the lenses were tuned, theETV was fired up several times to drive off any residual iodine. Unfortunately,although iodine made it possible to tune to a non-interfering volatile species, thechemical interaction of the iodine with the analyte in subsequent sample runs wasdetrimental. Precision of the In signals obtained following the use of iodine to tunewere poor. Precision degraded to 61% RSD for a 10 ppb solution of In. After 4-5hours of operation, when most of the iodine had dissipated, the precision of thesame solution dropped to 3.0% RSD. Similar degradation in precision has beenobserved when using Freon with indium. The use of iodine for tuning the ion lenswhen running hafnium was more successful. Precision of three replicatemeasurements were similar to the normal levels obtained, 3 to 8% RSD, forsolutions of 0.1 ppb to 10 ppb.One possible explanation for these observations is that hafnium does notform iodides easily by simply reacting with elemental iodine. Hf14 can be formed byreacting with elemental iodine in a sealed tube at temperatures higher than 700°C- conditions somewhat more drastic than those present in the ETV during dryingand ashing stages 3 . However, indium monohalides may be prepared attemperatures as low as 350°C4 . These halides have relatively low boiling points of711-715°C. Thus ashing temperatures would be limited to lower than 350°C.Ashing at higher temperatures would lead to loss of analyte. Since the amount ofiodine available to form halides was not constant, the amount lost should vary fromsample to sample leading to the degradation of precision for In.48Other elements that were considered as tune elements include Xe and Kr.These elements are present in trace amounts in the argon gas and have highionization potentials, 12.13 eV and 13.999 eV for Xe and Kr respectively 4 . It wasthought that due to their high ionization potential, they would be ionized mostefficiently at the hottest part of the plasma. One drawback to using Xe and Kr wasthat signals derived from these contaminant species were quite weak. Tuning theion lens to Xe and Kr did not result in very good sensitivities and it was thought that,again, the wrong regions of the plasma were probed. Similarly, 29Si+ signalscould be used if the source of Si in the plasma originated only from the injectortube. If this was true, tuning to 29 Si+ should represent the analyte region.Unfortunately, as with Xe and Kr, the Si tune gave sensitivities lower than thoseobtained from tuning to Ar2+ - 76 thus indicating that there are other sources of Si+in the plasma.Finally, a volatile metal was chosen as a tune species. Several microlitres ofa 1000 ppm mercury solution were injected into the ETV furnace and dried at-200°C. The resulting mercury signal was used to tune the torch and ion lenssettings. No effect on precision was found for the metals of interest and tuning wasfound to be stable for the duration of the run. Unfortunately, analysis of trace levelsof mercury could not be performed on this instrument for a long (approximately 1-2weeks) period of time owing to system contamination. Effects of Freon AdditionFreon was used for refractory metals such as Ti, Hf, Zr, Th and U. Shownbelow, in Figure 2.9, are mass spectra of 1 ppb injections of Th with and withoutfreon gas. The erratic and poorly defined signals were probably a result of carbideformation due to the refractory nature of the metal. Freon introduction gave rise towell defined, discrete signals, with little to no memory effects and a higher transportA27-242118-15-12-9 -6 -3 -4500 -4000-0'pc   3500 -0.,3000 -(1.,ci- 2500-o2000-=(.9 1500 -1000 -500 -0 ^0 1 2 349efficiency. Freon flow was optimized by varying the flow rate and measuring theresponse of triplicate measurements.1^2^3Time (s)o0Time (s)Figure 2.9 Thorium signals obtained by the ETV with and without Freon. (A) without Freon (B) withFreonIn the case of Hf, Th and U, the optimum freon flows were found to be thesame (^0.7 mUmin). Analyte signal levels off at higher flow rates. No interferenceswere observed with the introduction of freon.Unfortunately, in the case of Ti and Zr, an increase in background was foundwith freon use. The interference for Ti at masses 47, 48, 49 and 50 was found to bethe SiF+ species (discussed later). Thus, the analysis of Ti was performed usingthe lowest freon flow that could be obtained with the mass flow controller.Unfortunately, even with the introduction of small amounts of freon, the peaks2.5aTorna)ea2.11.7§(1e73 e1.3CaEIIS0.900.50 "^I^i^i^I l^i^i^10^0.2^0.4^0.6^0.8^1^1.2Freon Flow Rate (mL/min)Figure 2.10 Freon effects on Hf Signal for a 1 ppb solution. Signals were normalized to a freon flow ratecommon to both days. In this case, the common freon flow was 0.4 mL/min. Error bars areone standard deviation of the measurements.200160p16 -120‘1)(t) —7680.0 —a)73 40.0 —0 . 00 ^.1^Ia^I a a^I,^a 0^0.2^0.4^0.6^0.8^1^1.2Freon Flow Rate (mUmin)Figure 2.11 Freon effects on Hf background. Error bars are one standard deviation of themeasurements.obtained with Ti are still quite broad and ill-defined. The origin of the backgroundinterferences for Zr is unclear. The magnitude of the interference is not as severe50A*NM,51as that for Ti and a slightly higher freon flow rate (0.25 mt./min) was possiblewithout much of an increase in the background (Figure 2.12).6001 500g3aia 400cr)c38 300715cin• 200k)Ws 100c.10 ••0g0.00^I I I I I I I I I I I I II 1 I I I II I I 1 I II I I I l IL I 1 I I 1 1 1 1 1 1 1 10^0.1^0.2^0.3^0.4^0.5^0.6^0.7^0.8Freon Flow Rate (mi./min)Figure 2.12 Zirconium background as a result of Freon Flow(s) Zr - 91, (m) Zr - 90.Optimum sensitivity for Zr was also found at high flows (>0.5 mi.../min) of freon.Freon was required for Zr analysis because of the refractory nature of Zr. However,the negative effects of an increased background on detection limits, the flow rate offreon was confined to ^0.25 mi./min.2.3.2 Analytical Figures of Merit2.3.2.1 Evaluation of Sensitivity and PrecisionThe analytical figures of merit for this study are presented in Table 2.3. Thedetection limits were evaluated based on 3 awl of five replicate injections of a2.3 N HNO3 blank solution.Sensitivity for the ETV mode of sample introduction was found to beconsistently higher than that of the nebulizer mode. The sensitivities obtained have52been shown in terms of absolute concentration (signal/pg and signal/fmole) ratherthan in relative concentration (signal/ppb). It is more valid to compare thesensitivities at the absolute level since relative sensitivity is volume dependent. Forexample, relative sensitivity for the ETV mode can easily be multiplied by injectinga larger volume or by multiple injections. Nebulizer sensitivity can be increased byincreasing the volume of the injection loop. However, there is an upper limit fornebulizer sensitivity. This upper limit is obtained through continuous sampleintroduction.ElementSensitivity Detection^Limits Detection^LimitsNeb(CPS/pg)ETV(CPS/pg)Neb(pg)ETV(pg)Neb(fmol)ETV(fmol)In 4.7 5.5 x 103 5.7 0.016 49 0.14TI 4.9 1.4 x 103 14.7 0.0054 72 0.027Al .4.2 1.4 x 103 36.2 1.9 1340 72Hf 3.1 2.5 x 103 2.90 0.020 16.5 0.117Zr 3.4 1.5 x 103 2.72 0.019 30.2 0.21Ca 1.7 1.8 x 103 5.2 0.0041 73 0.057Fe 4.3 2.6 x 102 1370 2.1 24500 36Th 11.1 8.8 x 103 0.102 0.0035 0.440 0.015Ti 3.4 4.8 x 102 6.8 0.98 141 20U 10.2 4.5 x 103 0.559 0.0045 2.35 0.019Table 2.3^Comparison of absolute sensitivity and detection limits for pneumatic nebulizer and ETVmodes.Upon examination of the values obtained, it is clear that the sensitivity inmost cases was increased by two to three orders of magnitude. This is due to theincreased transport efficiency afforded by the ETV over the nebulizer, where over99% of the sample is lost as large droplets in the spray chamber as described inthe introduction. Some enhancement effects are suspected for the very large(greater than 2 orders of magnitude) improvements in sensitivity. These effectsmay stem from matrix effects as a result of using a dry plasma or to chemical53contributions from the graphite furnace. The rationale for these observations is stillunclear.As a result of the large increase in the sensitivity, detection limits wereimproved. For elements which do not suffer from high background or isobaricinterferences, the detection limits were lowered by over two orders of magnitude.However, for the elements which have very high backgrounds in the ETV mode,such as Ti and Al, absolute detection limits were typically lowered by only oneorder of magnitude. Although the detection limits were improved, Ti was found tosuffer from memory effects. Lengthy and repeated tube firings to remove residualTi makes this element very cumbersome to analyse by ETV.Precision obtained using the ETV were found to be comparable to that foundwith the nebulizer mode at low concentrations (<1 ppb) though inferior for higherconcentrations. Typical RSD values obtained for In by both methods of sampleintroduction are shown in Table 2.4. Similar results were observed for most of theelements studied.In Concentration(ppb)ETV Precision(% RSD)NebulizerPrecision (% RSD)0.010 9.58 7.820.100 5.82 4.800.500 1.90 6.641.00 5.37 3.785.00 4.37 0.80Table 2.4^Precision obtained for In by ETV and Nebulizer modes.Day-to-day variation in instrument performance may cause some of the aboveprecision estimates to deviate from the overall trend. At low levels, the signals fromthe nebulizer mode are close to the detection limit. Any changes in backgroundnoise level will have a large effect on the signal. The signals obtained using theETV mode are well above the detection limits at these concentrations. A portion of54the %RSD obtained may be traced to the method of sample injection. Theprecision found using manual injection of 25 1.11_ of sample by pipet has been foundto be -4% (i.e. ±1 44 Instrumental factors, pipeting error and imprecision due tothe short transient signal may combine to give the observed 2 - 10% RSD values.At higher levels, the precision improves for the nebulizer mode since the S/N ishigh. RSD values for the ETV remain similar to those obtained for low levels for thesame reasons mentioned above. Evaluation of LinearityLinear dynamic range for most of the elements (Table 2.5), spanned over 3orders of magnitude with the exception of Al and Ti. Again, high backgroundcounts for these elements resulted in a reduction of linear dynamic range. Anexample of this linear range is shown in Figure 2.13 for Hf.Element Nebulizer ETV r2In 103 103 1.0000TI 103 103 0.9980Al 104 101 0.9958Hf 104 103 0.9973Zr 104 103 0.9998Goa 104 103 0.9997Fe <101 NE NETh 103 103 1.0000Ti 103 >101 0.9982U 104 103 0.9997Table 2.5^Linear dynamic range obtained with pneumatic nebulizer and ETV. NE - not evaluated2.3.2.3 Comparison of Sample ThroughputAs with graphite furnace atomic absorption spectroscopy, the ETV mode ofsample introduction is slower than the nebulizer mode. The approximate time persample, including sample injection; drying, ashing and vaporization stages; and556.05.55.0i7(.1.5 4.50 4.03.5^)0.ar3.02.50-2^-1.5^-1^-0.5^0^0.5Log concentrationFigure 2.13 Optimized Log-log calibration curve for Hf using ETV sample introduction. Error bars are onestandard deviationcool down, was three minutes per sample (20/hr). This rate of sample throughputwas one-half of the rate of the nebulizer mode, which was about 1.5 minutes persample (40/hr). Evaluation of Multielement CapabilitiesMultielement analysis on the ETV is more difficult than that of the nebulizerfor several reasons. Primarily, the optimum heating parameters often vary fromelement to element. For example, very low ashing temperatures would be requiredif Hg analyses were desired. It may be necessary to compromise freon flows ifsimultaneous determinations were desired. For example, if Hf was analysedsimultaneously with Zr, freon flows would have to be adjusted so that the isobaricinterferences for Zr are at a reasonably low level without losing too much sensitivityfor Hf.Indium could not be measured simultaneously with refractory elements sincethe introduction of freon would form halides with In at low ashing temperatures56resulting in the loss of In. Gallium also could not be measured simultaneously withthe refractory metals since freon also contributes to isobaric interferences at m/z 69from the formation of 12C 19 F3+. Great care and some compromise must be madewhen trying to perform multielement analyses using the ETV.Elements chosen for evaluating the multielement capabilities of the ETVwere Zr, Th, and Hf. A freon flow of 0.7 mL/min was used, which resulted in anincrease in the 90Zr background. This increase in the background was reflected inthe increased detection limits for 90Zr.A slight decrease in the sensitivity and more significant decreases inprecision was observed in the multielement mode. Relative standard deviations forfive replicate injections of a 1 ppb solution ranged from 14-18%. However, theisotope ratio precision obtained for the five isotopes monitored were comparable toprecision obtained with monitoring fewer isotopes. The RSD value obtained for theZr (90:91) isotope ratio was attributed to high backgrounds from freon use whichsignificantly affected the minor isotope. Results from this study are summarized inTable 2.6.Sensitivity DetectionLimitsAbsolutePrecision (%)Isotope^Ratio *Precision (%)Element Neb ETV Neb ETV Neb ETV Neb ETV(cps/PP) (cps/PA) (Pa) (139) (%) (%) (%) (%)Zr 3.4 4.93 x 102 4.75 0.696 2.34 13.2 1.97 18.0Hf 4.2 6.77 x 102 1.94 0.0205 2.02 18.2 1.58 1.50Th 11.7 3.15 x 10 3 1.15 0.0157 1.91 15.8 1.30 2.93Table 2.6^Results of multielement analysis for five isotopes by ETV and nebulizer modes.Isotope Ratios for Zr - 90:91, Hf - 177:178, Th - 232:178The loss of precision and sensitivity was attributed to the short signal pulse which,with five isotopes monitored, allowed only 48 scans across the desired masses in57the 3 s acquisition. Since the signal pulse is only 1s in duration from baseline tobaseline, only 16 sweeps were made across the peak.A similar evaluation for the nebulizer mode was performed on a 10 ppbsolution of the same elements as those introduced to the ETV. The changes in thesensitivity, detection limits and precision found in the ETV were not matched in thenebulizer mode. This may be explained by the extended signal pulse of 60 s whichallows 600 sweeps for the same five isotopes monitored. The larger number ofsweeps should average any random noise spikes and give a better representationof the analyte signal than the 16 sweeps could for the ETV mode.Even with the above mentioned degradation in absolute precision, theabsolute sensitivity and detection limits obtained using the ETV were still, with theexception of Zr, two orders of magnitude better than those obtained using thenebulizer mode. The increase in Zr background, due to the increased freon flow,leads to the reduction in detection limits by one order of magnitude, still somewhatbetter than the detection limits obtained using nebulizer mode. In order to maintaingood absolute precision, it would appear that only 2 - 3 isotopes may be analysedsimultaneously. Since good isotope ratio precision could be obtained, it wouldappear that isotope ratio techniques, such as the method of internal standard orisotope dilution, should be incorporated for multielement analysis. Isotope Ratio Accuracy and PrecisionIsotope ratio precision using the ETV were evaluated by triplicate injectionsof each solution (Table 2.7). Natural isotopic abundance's were used for Ga, Zrand Hf. Precision at concentrations higher than 1 ppb are comparable to thoseobtained using the nebulizer mode. However, due to the increased sensitivity ofthe ETV mode, isotope ratio precision was much improved over those of thenebulizer mode for lower concentrations. Typically, good precision (3%) can be58expected for levels - 0.1 ppb in the ETV mode while similar precision in thenebulizer mode would require concentrations greater than 1 ppb for most metals.The difference between the expected isotope ratio and the measuredisotope ratio (accuracy) obtained using the ETV were better than values obtainedusing the nebulizer for concentrations below 1 ppb. This was thought to be due tothe increased sensitivity of the ETV over the nebulizer mode. At concentrationsbelow 1 ppb, the S/N ratios in the nebulizer mode are low. Therefore anyvariations in the noise would have a large effect on the minor isotope, and thus, theisotope ratio. In the ETV mode, the S/N ratios are significantly higher for the sameconcentrations, thus any variation in noise would affect the isotope ratio to a muchlesser extent. Typically for concentrations at the 0.05 ppb levels, ETV isotope ratioaccuracy of 4% or lower may be achieved. At concentrations over 1 ppb, the S/Nincreases for the nebulizer mode and the minor isotopes are less affected byvariations in the background noise. Accuracy of isotope ratios between the ETVand nebulizer modes is comparable at higher concentrations.In general, good isotope ratio precision, 3% RSD, was obtained for the ETVwith concentrations -0.1 ppb. Similar levels of precision using the nebulizer wouldrequire concentrations of an order of magnitude higher. Again, this observation isdue to the increased sensitivity in the ETV.The utilization of an internal standard also looks very promising for the ETV(Table 2.8). The precision of the Ga/In isotope ratio (-44 amu apart) ranged from0.17% to 4.49% for three replicate injections of solutions with concentrationsbetween 0.05 - 0.5 ppb. This is superior to the 15-159% precision obtained for thesame concentrations in the nebulizer mode. It should be noted that 0.05 ppb is atabout the detection limit of Ga in the nebulizer mode.59Hf (ppb) Nebulizer ETVmeasured177:178IsotopeRatio RSDmeasured/naturalmeasured177:178IsotopeRatio RSDmeasured/natural0.010 0.78 29% 1.15 0.774 7.74% 1.140.050 0.59 20% 0.86 0.712 8.36% 1.050.100 0.54 5.7% 0.80 0.710 3.33% 1.041.00 0.6791 0.75% 0.9967 0.6791 0.75% 0.99675.00 0.674 1.91% 0.989 0.6802 0.52% 0.9983Zr (ppb) Nebulizer ETVmeasured90:91IsotopeRatio RSDmeasured/naturalmeasured90:91IsotopeRatio RSDmeasured/natural0.020 -2 608% -0.38 4.77 10.5% 1.040.100 5-0 36.2% 1.1 4.81 3.60% 1.040.459 4.48 8.27% 0.977 4.272 0.33% 0.93081.001 4.727 1.42% 1.030 4.486 0.50% 0.97764.667 4.577 0.31% 0.9974 4.505 0.57% 0.98169.8 4.518 0.78% 0.9845Ga (ppb) Nebulizer ETVmeasured69:71isotopeRatio RSDmeasured/naturalmeasured69:71IsotopeRatio RSDmeasured/natural0.010 0.3 745% 0.2 1.90 22.7% 1.260.050 0.4 698% 0.3 1.539 1.95% 1.0220.100 1.77 29.8% 1.18 1.551 3.09% 1.0300.500 1.50 9.05% 0.993 1.462 1.94% 0.97071.00 1.457 5.73% 0.9673 1.4613 0.62% 0.97025.00 1.404 1.80% 0.9322 1.4527 0.57% 0.964410.00 1.465 1.44% 0.9723 1.453 1.11% 0.9645Table 2.7^Precision and accuracy for Hf, Zr and Ga for nebulizer and ETV modes. Natural isotope ratiosare Zr (90/91) - 4.589, Hf (177/178) - 0.681, Ga (69/71) - 1.506.60Isotoperatio^(Ga:ln)Concentration(ppb)ETV RSD(%)Neb RSD(e/e)71:115 0.050 4.50 1600.100 1.57 39.70.500 0.17 14.9Table 2.8^Ga/In Isotope ratio precision obtained using the EN and nebulizer modes.2.3.3 Isobaric Interference - Problematic Elements2.3.3.1 TitaniumThe purpose of introducing freon into the nebulizer gas stream was topromote the formation of more volatile halide species, such as TiF4 6 , duringvaporization. Upon adding freon, an increase in background was observed (Table2.9 and Figure 2.14) at m/z 47-49. Isotopic abundance's for masses 47, 48 and 49were found to be 90.2 ± 0.5%, 5.4 ± 0.5% and 4.4 ± 0.1%, which strongly suggestthat the interference originated mostly from the SiF+ species. Isotopic ratiosexpected for SiF+ were 92.2%, 4.7% and 3.1%, stemming from the naturalabundance of silicon.Freon Flow Rate CPS 47 CPS 48 CPS 49(mL/min)0.15 79973 5567 30720.25 115682 7667 43090.50 1009326 56806 37579Table 2.9 SiF+ interfering species with increasing Freon.In an effort to avoid freon (and the resulting high background at mass 48)while still gaining the advantages of a more volatile titanium halide, treatment withHF was studied. It was hoped excess HF would be removed during the drying andashing stages leaving the only source of fluoride contained in the TiF n compounds. 612800-2400-• 2000-1600-1200-C)• 800-4000A..,.I^I^ I^l^I45 46 47^48 49 50m/z3500 --Dg 3000 -21°,3 2500 -'13. 2000 -ws 1500o 1000 -500 -0 ^BI^I^I^1^I^—F.—45 46 47 48 49 50m/zFigure 2.14 Blank spectra for m/z 45 - 50 in ETV mode. (A) Without Freon (B) With Freon.This would greatly reduce the fluoride content in the plasma. The halide formedmay be titanium tetrafluoride 7 , TiF4, titanium oxide difluoride 8 , TiOF2, or the titaniumoxide hydroxyfluoride, [TiO(OH)F] 8 . During the drying and ashing stage, anincrease in the signal for mass 48 was detected. It was thought that TiFn may belost at this stage. However, it was found that the isotopic ratios during this stagewere 92 ± 1%, 5.0 ± 0.2% and 3±1% for m/z 47, 48 and 49 respectively, indicatingthat the signals were due to SiF+ formation and not loss of the analyte.Unfortunately, it appeared that the Ti compound formed with HF in the graphite62furnace was the TiO(OH)F compound that loses HF during the drying and ashingstages to form T102. The dioxide form is undesirable as it is refractory, and formstitanium carbides with the graphite furnace at elevated temperatures, thus leadingto high memory effects. Little enhancement in sensitivity over the nebulizer modeof operation was observed and severe memory effects were evident under theseconditions.Since fluorine is required to efficiently volatilize Ti and prevent carbideformation, perhaps the source of silicon could be reduced or removed, lowering theSiF+ interference. One possible source of silicon is the quartz transport tubeleading from the furnace. A Teflon® insert was made to fit inside the quartz tubewith approximately a centimeter allowance on the end closest to the furnace. Thisallowance was a precaution taken to avoid the melting of the Teflon during thevaporization stage. The purpose of this insert was to minimize the fluoride contactwith the quartz. Unfortunately, SiF+ signals were not reduced with thismodification.Another source of SiF+ may be in the torch itself. For the freon to react withthe quartz, high temperatures are required (>1700°C). Thus the torch, specificallythe injector gas tube which is in close proximity to the plasma, could very easilycontribute to the isobaric interference. A corrosion resistant torch, which has aremovable alumina insert as the injector tube, may be used to reduce the amountof interference (Figure 2.15). However, as shown in Figure 2.16, SiF+ signalsincreased dramatically with increasing Freon flow even with the corrosion resistanttorch. Again, isotopic abundance's strongly support that the interference was dueto SiF+. The source of this Si may be the outer concentric tubes of the torch wherescorching have been observed. Similar increases in SiF+ signals were observedin the nebulizer mode when HF is used with corrosion resistant components (torch,spray chamber, nebulizer).D EFigure 2.15 Corrosion resistant torch. ; (A) Teflon® ball and socket joint (B) Aluminosilicate injector tube(C) Normal quartz tube on torch (D) & (E) Cool and Auxiliary gas input.oUN1.21.0106106 •aa.coc2a)_co8.06.04.0105105105-Li_05 2.0 105 •0.0-1^I$I^I^I 111 III II III III^I I^I II III 11^I0^0.1^0.2^0.3^0.4^0.5^0.6Freon Flow Rate (mUmin)Figure 2.16 Effect of Freon flow rates on SiF+ signal at m/z 48 with a corrosion resistent torch.This is evidence that the source of SiF+ in the ETV mode originated from the torchsince it is the only common component in both modes of sample introduction.Replacement of the normal torch with a corrosion resistant torch did reducethe background from that observed with a normal torch by approximately 60006364CPS on a blank solution at the lowest freon flow possible. However, this is stillsignificantly higher than that observed using the nebulizer. The gain in sensitivityby a factor more than 200 over the nebulizer mode compensated for the increasedbackground giving detection limits which were an order of magnitude lower (basedon signals three times the standard deviation of the blank solution). This set upwas used for all experiments reported unless other specified. AluminumHigh background signals at m/z 27 were observed for 25 JAL injections of2.3 N HNO3 blank solution for aluminum. These high counts were much higherthan those observed in the nebulizer mode. Similar signals were found when noliquid was introduced into the graphite furnace. Typical signals obtained formasses 23 to 27 in the peak jumping mode are shown Table 2.10.Description CPS 24 CPS 25 CPS 26 CPS 27no freon (HNO3 blank) 71273 10146 26281 51020with freon (HNO3 blank) 29589 6756 63178 208406with freon (dry) 2688 1312 14964 224689Table 2.10 Signals observed for rn/z 24 - 27 in ETV mode with and without Freon.It is thought that the vaporization of the carbon from the graphite furnacecontributed to the increased signal. If some carbon from the furnace was vaporizedduring the vaporization stage, strong candidates for the interferences observed arecarbon containing species. Some of the possibilities are listed in Table 2.11.It was first thought that the increased interference for m/z 27 originated fromCN+ species. The origin of the carbon could be from the graphite tube and thenitrogen from air entrained into the plasma.65m/z Possible^Interference27 12c15N+13c14N+1X2 1 H3+12c14N1H+26 12c14N+12c2 1H2+25 12c21H+24 i2c2+Table 2.11 Possible carbon-containing interferences for aluminum. Natural isotopic abundances ofminor isotopes: 13C - 1.1%, 15N - 0.36%.When 0.15 mL/min of freon was added, significant increases in signals wereobserved, which agrees with the theory that the isobaric interference containscarbon. If CND was the dominant interferent, then the isotope ratio for m/z 26: m/z27 that would be expected is 66:1. However, the observed m/z 26: m/z 27 isotoperatio was 0.23 ± 0.03. Thus CN+ was not the major interferent for Al.It was found that the background interference at m/z 27 remained the samewhen comparing the response obtained from a 25 injection of 2.3 N HNO3 acidblank and a dry tube. However, for masses 24-26, the background was lower forthe dry tube firing. Thus, the interference for m/z was not contained within the acidblank.Thompson and Houk proposed that C2Hn+ species may mask the Al+signals 9 . The signals observed for m/z 24 and 25 show a decrease when freonwas introduced. This would not be expected to occur if C2H n+ species were themajor interferent since the introduction of carbon from freon should increase thesesignals. In addition, Lamoureux has reported low backgrounds for Al in the ETV-ICP-MS systemic'. It was not clear at this point, whether or not the interference atm/z 27 originates from carbon containing species. Other possible interferences aredescribed below.66The increases in background for masses 26 and 27 may be a result of anincrease in the doubly charged species. Some possible doubly chargedinterferents are listed below (Table 2.12).m/z Interference272654cr++54 Fe++40Ar14N++38Ar160++40Ar12c++Table 2.12 Possible doubly charged interferences for rn/z 26 and 27.The atomic doubly charged interferents, Cr++ and Fe++, were easilyeliminated as possible interferents. Although the signal at m/z 54 was high, thesignal at m/z 53 remained at background levels even with the introduction of freon.This indicated that little chromium was present in the system. When 56 Fe, themajor isotope of iron, was monitored, it was found that it was present at trace levelsand so, from natural abundance, any contribution to 54Fe would be very small andcould not account for the increases observed at m/z 27. The molecular ion,38Ar160+ may also be eliminated. Again, the increase in signal as a result of freonaddition was small for 40Ar160+ is small so that any contribution it may have on38Ar160-4-, and subsequently 38Ar160++ (m/z 27) , is negligible. Thus, the majority ofthe signal intensity at mass 54 may be attributed to the molecular ion, 40Ar14 N+.The ArN+ signal did not increase with the addition of freon. Therefore, the doublycharged species 38Ar14 N++ was not expected to increase to give in an enhancedisobaric interference for m/z 27.. At this point, the origin of the Al interference is not conclusive. Somepossible interferent such as CN+, double charged molecular and atomic species,were not found to be the source of high isobaric interference in this system.Aluminum contamination from the graphite tube, mobilized by the introduction of67freon, may be the origin of high backgrounds. Further studies are required toelucidate the identity of the Al interference. ZirconiumAn increase in background levels was observed as the flow of freon allowedto increase (Table 2.13). It was soon evident that the species affecting m/z 91 wasnot the same as those which affected m/z 90 since the ratios of 90:91 varied withfreon flow rates. This observation is also evidence that the contamination is notdue to Zr contamination in the graphite tube.Freon Flow Rate (mL/min) CPS m/z 90 CPS m/z 91 CPS m/z 92 CPS m/z 930.070.7026813792586361205629128922Table 2.13 Effect of Freon on Zr rniz 90 - 93. Zr 90 - 51.4%, Zr 91 - 11.2%, Zr 92 - 17.1%Tube contamination would result in constant isotope ratios with values similar tothose of natural isotopic abundances. Possible interferences may be various NISH -and CuSi+ species (Table 2.14).Mass Possible^Interferences90 62Ni28si+61m29si+6oNi3Osi+91 62Ni29Si+61 Ni3°Si+63CU28Si+60N112C19F+9 2 63CU29S1+61Ni12C19F+60Ni13c19F+93 63CU30Si+65Cu 28Si+62Ni12c19F+Table 2.14 Possible NiSi and CuSi interferences. Natural isotopic abundances for minor isotopes:60Ni _ 26.1%, 61Ni _ 1.13%, 62Ni _ 3 . 59% , 29si 4 . 7% , 3osi _ 3 . 1%68The source of Ni and Cu may be from the erosion of the nickel extraction cones andthe copper plated cooling blocks respectively. The isotope ratios expected whenNiSi+ species are the only interferences are shown in Table 2.15.DominantInterferentIsotope Ratio ExpectedValueIsotope RatioValue MeasuredNISI+ 90:91 20 0.18 ± 0.0290:92 20 2.6 ± 0.1CuSi+ 91:92 19.6 14 ± 391:93 2.09 7.7 ± 0.5NiCF+ 91:92 18.75 14± 391:93 7.27 7.7 ± 0.5Table 2.15 Comparison of isotope ratios expected for various interferences and the isotope ratiosobserved for masses 90-93.Clearly, the isotope ratios obtained using the ETV did not agree with the valuesexpected for NiSi+. Therefore, we can rule out NiSi+ as a major source ofinterference in Zr determination.The measured 91:92 isotope ratio suggests that CuSi+ may be a majorinterferent. However, the isotope ratio for 91:93 is somewhat higher than the valueexpected for CuSi+. Instead, it is possible that the interference at m/z 91 could bethe molecular ion 60Ni12c19F+. The isotope ratio obtained for 91:93 was 7.7 ± 0.5which is in good agreement with the expected value of 7.27 calculated from isotopeabundances. However, the isotope ratio measured for 91:92 was 14 ± 3. Thelarge error was to due, in part, to the low signal counts obtained for m/z 92. Thisratio was slightly lower than the expected 91:92 ratio of 18.75 if NiCF+ was the onlyinterferent present at mass 92.Clearly, the addition of freon results in increases in isobaric interferences forZr. The interference at mass 91 may be due to NiCF+, CuSi+ or a combination ofthese two species. It is unclear what species are responsible for the increase inbackground for m/z 90. Some sources of interference, such as tube contamination69and NiSi+, have been eliminated but further studies will be needed to determineconclusively the interfering species at masses 91 and IronThe most predominant isobaric interference for Fe is the Ar0+ ion whichoverlaps the most abundant isotope for Fe at mass 56. The primary source of thisinterferent originates from the aqueous sample matrix, which is the major supply ofoxygen. As can be seen from Figure 2.17 and Table 2.16, the background in thenebulizer mode is quite high at over 300000 counts per second for a blanksolution. Thus separating the sample matrix from the analyte can greatly reducethe background.Analysis Mode CPS 56 Standard Deviation (CPS)Neb 328618 1959ETV 2948 193ETV-dry 2563 119Table 2.16 Comparison of ArO signals obtained by nebulizer and EN modesSeparation of the matrix from the analyte was accomplished by thermally drying thesample in the ETV. As can be seen in Table 2.16 the background at mass 56 canbe reduced by two orders of magnitudes. Unfortunately, the background could notbe reduced to the near zero levels found by Evan& 1 and Huttonl. The levelsobtained in this work were approximately 3000 counts per second. Thus, only asmall portion of the background signal coul d be attributed to any Fe contaminationin the acid blank. The Ar0+ signal was also monitored for a dry plasma during theashing stage. The signal obtained was similar in magnitude as that during thevaporization stage with a dry tube. Thus, little, if any, contamination was attributed70056000—48000—40000—32000—24000—16000—80000 ^2800 —2400 —2000 —1600 —1200 —800 —400 —0 —A53^54 55^56^57^58^59B153^54^55^56^57^58^592100 —18001500 —0cn0_ 1200900 —600 —300 —0 ^C153^54^55^56^57^58^59m/zFigure 2.17 Representations of Fe blank solutions obtained with the nebulizer and ETV modes;(A) Nebulizer mode (B) ETV with 25 pl. blank solution (C) ETV with no blank solution71to contamination in the graphite tube. So it appears that the most probably sourceof interference was due to oxygen containing species in the argon supply gas ordue to air entrainment into the plasma.In the initial stages of this work, it was found that although the drying andashing stages were very lengthy, higher backgrounds at mass 56 were still foundduring vaporization. This was later thought to originate from any moisture whichmay have condensed on the cooler edges of the furnace or on the quartz tube.This particular problem was perhaps a result of the furnace design (Figure 2.18).2to 1CP -01^ Figure 2.18 Drawing of furnace and quartz tube.(1) Quartz tube (2) Cooling block (3) Graphite furnace(4) proposed site of condensationThe cooling block is designed such that, during the vaporization step, the analytevapor was cooled to form microparticulates and prevent analyte condensation onthe transport tube 11 . However, this system was found to also condense water onthe quartz tube during drying and ashing stages. When the vaporization stage is72reached, the sudden rush of hot air carries some of the water vapor with it and inthis way, contributes to the Ar0+ signal.To circumvent this problem, the drying and ashing stages were completedwith the carbon pencil removed and the nebulizer gas reduced to 0.4 mL/min. Theomittance of the carbon pencil insertion allowed the water vapor to escape throughthe hole thereby preventing the vapor from entering to the cooling block andcondensing on the transport tube. Back pressure from the plasma would "push"back and force the nebulizer gas to exit through the hole. The nebulizer gas wasreduced to insure that the nebulizer gas would not push the vapor forward and tomaintain a positive pressure to prevent condensation on the "gas inlet" side of thesystem. By taking these precautions, the Ar0+ background could be reducedsignificantly. However, the low backgrounds obtained by Evans 11 could not bereproduced in the present work.2.3.4 Analysis of Seawater SamplesSeawater samples were processed in a similar procedure as that describedin chapter 3. The eluents were analysed by ETV and by nebulizer. Using anexternal calibration curve gave results that clearly did not agree (Table 2.17).Method Calibration Curve(PPb)Standard Additions(PPb)ETVNebulizer1.27 ± 0.040.14 ± 0.010.194 ± .0080.23 ± 0.03Table 2.17 Comparison of Ga concentrations obtained by calibration curve and standard additions forboth methods of sample introduction.There are two possibilities that can explain these observations; isobaricinterference or matrix suppression/enhancement. Isotope ratios obtained from bothmethods produced 69:71 isotope ratios expected from the natural abundance ofGa. It may be concluded that the discrepancy was not due to isobaric interference.73For example, if chloride interference was present, the 69:71 isotope ratio woulddeviate significantly from the natural Ga isotopic abundance.Matrix effects can be corrected by the method of standard additions. Asdemonstrated (Table 2.17), the method of standard additions gave results whichwere in much better agreement between the two methods. The matrix effects maybe a result of organics in the eluent from the resin, the reagents used or somethingspecies the seawater sample. Further investigations into the origin of the matrixeffect in the nebulizer mode will be discussed in chapter 3.2.4 ConclusionsThe ETV is a useful alternative to the Meinhard nebulizer as a method ofsample introduction. Its increased transport efficiency has demonstrated increasesin sensitivities of over two orders of magnitude for most of the elements studiedwhen compared to the nebulizer mode. Significant improvements in detectionlimits were found for many of the elements investigated. Detection limits obtainedin this study ranged from 0.019 - 72 fmole. Elements with high atomic masses werefound to be quite interferent free and had detection limits of 0.019 - 0.14 fmole.Elements at the middle and lower end of the mass scale, such as Ti, Al and Zr,suffered isobaric interferences, especially with the introduction of freon. Detectionlimits for these elements were 20 - 72 fmole. Freon was found to be very effectivefor forming volatile fluorides with high boiling or refractory carbide forming metalssuch as Hf, Th and U. The linear dynamic range was -3 orders of magnitude formost elements. However, the maximum concentrations in the linear dynamic rangeare normally found to be less than 10 ppb, thus making the ETV a complementarymethod of low level detection method for the pneumatic nebulization system.Precision obtained using the ETV is 2-10%, somewhat inferior to the precisionobtained using the nebulizer. Significant reduction in the RSD may be obtained74using an autosampler, since manual injections were found to have approximately4% RSD.It is concluded that the most efficient way to utilize the ETV would be byemploying isotope dilution techniques. The ETV shows comparable precision inisotope ratio measurements to the nebulizer method. Analysis by externalcalibration or by standard additions (required since matrix enhancement was foundfrom the analysis of seawater eluents) is lengthy and cumbersome.752.5 References1^Hulmston, P.; Hutton, R. C. Spectroscopy 1991,6 (1), 35 - 38.2^Carey, J. M.; Evans, E. H.; Caruso, J. A.; Shen, W. L. Spectrochim. Acta1991, 46B(13), 1711-1721.3^Canterford, J. H.; Colton R. Halides of the Second and Third Row TransitionMetals ; John Wiley & Sons Ltd: New York, 1968.4^N. N. Greenwood; A Earnshaw Chemistry of the Elements; Pergamon PressLtd: Great Britain, 1984.5^Handbood of Chemistry & Physics, 63rd ed. CRC Press: Boca Raton, 1983.6^Codell, M. Analytical Chemistry of Titanium Metals and Compounds;Interscience Publishers Inc: New York, 1959.7^Barksdale, J. Titanium, Its Occurrence, Chemistry, and Technology, 2nd ed;Ronald Press Co.: New York, 1966.8^Canterford, J. H.; Colton R. Halides of the First Row Transition Metals ; JohnWiley & Sons Ltd: New York, 1968.9^Thompson, J. J.; Houk, R. S. Anal. Chem. 1986,58(12), 2541-2548.10^Lamoureux, M., Carleton University, personal communications, 1992.11^Evans, E. H.; Carey, ; Caruso, J. A. Spectrochim. Acta 1991, 46B(13),1711-1721.76Chapter 3Determination of Trace Levels of Titanium, Gallium and Indiumin the Central Pacific Gyre3.1 IntroductionTrace metals exist in very low concentrations in seawater due their rapidremoval from the oceans. Removal mechanisms, such as particle scavenging andbiological uptake, result in large temporal and spatial variability in the oceans. Bydetermining such variabilities in the ocean, we can elucidate the biogeochemicalcontrols on these elements and evaluate their potential to act as tracers of oceanprocesses. Titanium, gallium and indium are the elements that will be focused onin this study.There have been few published papers on titanium in natural waters. Orianset al. measured Ti in the northeastern Pacific and the north Atlantic using ICP-MSafter seawater preconcentration with a chelating resin 1 . They found Ticoncentrations of 5-300 pM, with a distribution similar to Cu in the subarctic regionsof the north Pacific. Titanium exhibits surface depletion and concentrationsincreasing linearly with depth. This similarity suggests that Ti, analogous to Cu, iscontrolled by a complex combination of input and removal mechanisms. Titaniumconcentrations in the North Atlantic are higher than the Pacific, but with similardepth dependence which supports a scavenging-type distribution for Ti. Titaniumin estuarine waters was recently studied by Uehara et al using HPLC afterevaporative preconcentration 2 and Skrabal et al using cathodic strippingvoltammetry (CSV) 3 . They discovered Ti concentrations were higher in thesewaters (0.2-67 nM) than those of the open ocean and that Ti removal was duemainly by particle or colloidal scavenging.Early studies by Culkin and Riley estimated that the average galliumconcentration in seawater near the English Channel was -430 pM4 . More recent77studies using clean methods and advanced instrumentation has shown averageseawater gallium concentrations to be much lower13 . Gallium has recently beenstudied as a potential analog of A1 5,6 . Shiller found gallium concentrations in thenorthwest Atlantic ocean to be 10 - 45 pM 6 while Orians and Bruland determinedopen north Pacific ocean Ga concentrations to be 2-30 0.4 5 . The lowerconcentration seen in the older waters of the deep north Pacific, relative to thenewly formed deep north Atlantic waters, shows a scavenged-type interoceanfractionation. Comparisons of dissolved Ga to Al showed that Ga is enriched withrespect to Al in seawater relative to their crustal abundance 5,6 . This enrichmentwas found for both the Atlantic and Pacific oceans. Several proposed theories to tryto account for this enrichment include preferential removal of Al, preferentialdissolution of Ga from the atmospheric dust or the presence of a Ga source not yetidentified. An increasing gradient of dissolved Ga concentration with longitudetowards the Asian dust in the North Pacific and towards Saharan dust in the NorthAtlantic supports an aeolian input for dissolved Gas. No fluvial input of dissolvedGa to the open ocean was found to either the eastern north Pacific or the westernnorth Atlantic. Generally, vertical distributions of dissolved Ga in the Pacific Oceanshow low surface values, a subsurface maximum, an intermediate minimum and abottom maximum which suggests a variety of addition input and removalmechanisms. These mechanisms include intermediate and bottom water sourcesas well as scavenging removal and possibly nutrient type cycling in the upperwaters. Further studies into gallium distributions in other oceanic regimes mayhelp elucidate its chemistry in the ocean.There are no recent publications regarding the oceanic behavior of indium.Early studies using very expensive and time consuming neutron activation analysisfound average indium concentrations in seawater to be 1-35 pM20,21. Clearly,insufficient data is available to accurately deduce the oceanic behavior of indium.783.2 Experimental3.2.1 Study SiteSeawater samples were collected on board the Russian research vessel,"Aleksandr Vinogradov" in April 1991. Samples used for this study were from thePacific central gyre at the station labeled AV 10 (27° 46.5'N, 174° 59.4'E) in Figure3.1. Also included on this map are the sites from previous studies that will beutilized later for comparison.3.2.2 Seawater Sample CollectionSeawater samples were collected at AV 10 (174 59.4' E, 27 46.5' N) usingTeflon® lined General Oceanics "Go-Flo" bottles. These 30 L samplers weresuspended on a Kevlar line. The "Go-Flo" bottles, sealed prior to entering theocean, were opened by a pressure sensor at 5-10 m depth, once they have left thesurface region where potential contaminants are higher. Bottles were then loweredto the desired depths, and termination of sampling was triggered by a Teflon®messenger. Once retrieved, all seawater samples were filtered through 0.45 p.mPoretics® polycarbonate membrane filters on board the ship in a make-shift cleanarea. This was followed by acidification to pH 1.5 - 2.0 with 6 M double distilledhydrochloric acid (Seastar Chemicals) and storage in acid-cleaned polyethylenestorage bottles.3.2.3 Reagents and SolutionsAll reagents were prepared using distilled deionized (DDI) water (18 Mf2,Nanopure, Barnstead). The acids used were environmental grade nitric acid(HNO3) (Anachemia), quartz double distilled hydrochloric acid (HCI), ammoniumhydroxide (NH4OH), and acetic acid (HOAc) (Seastar Chemicals). Concentrated(0.2 M) reagent grade potassium acid phthalate (KHP) (Baker) was cleaned prior touse by passing the solution through 1 mL gravity packed vinyl polymer (TSK)79Figure 3.1 Map indicating study site for this study and sites that will be used for comparison. Exactlocations for each site are listed below. Each location will be referred to their abbreviationin the discussion that follows. Aleksandr Vinogradov station 10 (AV10) 174 59.4' E 2746.5' N, Vertex IV (IV) 155° 07'W 28° 15'N, Vertex VA (VA) 139° 34'W 33° 06'N, VertexVC (VC) 122° 38'W 36° 06'N, Vertex V7,T7 (V7,T7) 145° 00'W 50° 00'N.80immobilized 8-hydroxyquinoline resin, synthesized in house following publishedmethods7 . Modifications to the synthetic route will be discussed in Chapter 4.Standards of Ti, Ga, and In were prepared by serial dilution of 1000 ppmatomic absorption standards (Johnson Matthey Inc.) with the appropriate acidmatrix. An isotopically enriched 49Ti standard solution (49Ti - 96.25%, 48Ti -2.71 %) was prepared previously from an enriched solid standard of 49TiO2 (OakRidge National Laboratories). The TiO2 was dissolved with a 4 mL acid mixture of25% HF, 25% HCI and 50% HNO3 followed by digestion by microwave 8 . After100-fold dilution with water, the concentration of this solution was determined byreverse isotope dilution with a known solution of 48Ti of natural abundance.3.2.4 Column PreparationTeflon® columns and fittings were acid washed prior to use. Approximately1 mL of gravity packed TSK 8-hydroxyquinoline resin was placed in each of thesecolumns. The resin was cleaned by gravity eluting (-0.2 mi./min) 20 mL of 2.3 NHNO3 through each column. At this point, column blanks are collected from eachcolumn. Water adjusted to pH -3.8 was passed through the columns to pH adjustthe resin prior to pumping seawater.3.2.5 Seawater Sample ProcessingAll sample processing was performed in a Class-100 filtered air clean roomto avoid air borne contamination.Acidified seawater samples (4 L) were used for these analyses. A 100 mLaliquot was removed from a number of the samples, prior to any further processing,for the purposes of a spike recovery test. The preconcentration procedure usedhere is similar to one previously developed 8 . Modifications to the procedure willbe discussed further in Section 3.3.1. The final procedure is presented in thissection. Each 4 L sample was spiked with an amount of 49Ti which would result in81isotope ratios of 49Ti to 48Ti approximately equal to unity. Estimates were basedon preliminary studies performed without isotopic dilution and assuming similar Tidistributions with those previously reported'. After allowing the samples toequilibrate for a minimum of three hours, the seawater samples were pH adjustedto 3.8±0.2 with NH4OAc and NH4OH. Samples were then pumped using avariable speed MasterFlex US pump (Cole Parmer) at a rate of 2.5-3.5 mUminthrough the 8-hydroxyquinoline resin. Pumping rates were closely monitoredduring the first 8 hours to insure that pumping rates were within the desired limits.Columns were then resuspended with 1 mL of DDI water, and washed sequentiallywith 2 x 500 ;IL of 0.1 M KHP (to remove sulphate interferences), 8 x 1.0 mL 5%NH4OAc (pH -6) (to remove Ca interference), and 5 x 3 mL of DDI water. Thedesired metals were eluted from the resin into 4 mL acid clean polyethylene bottlesby passing 7 x 500 aliquots of 2.3 M HNO3.3.2.6 Spike Recovery TestsSpike recovery experiments were performed to determine the efficiency ofthe concentration technique. Two resins were used: the TSK 8-hydroxyquinolineresin and the Chelex-100 (Bio-Rad). For the conditions described above, therecoveries of 50 ng spikes of Ga, In and Ti were evaluated to be 95±9 % ( 69Ga),98±3 °/0( 116 1n) and 100±3 % (47TO respectively with the TSK 8-hydroxyquinolineresin. Chelex recoveries were similar to those obtained for TSK 8-hydroxyquinoline resin at 100±7 % for Ga and 98±4 % for In. Titanium results werenot obtained for Chelex due to high Ca interferences. It should be noted that Cainterference for 48 Ti was also found to be relatively high for the TSK8-hydroxyquinoline resin. Due to the slow exchange kinetics, Chelex required over20 mL of 2.3 M acid to attain the above recoveries while the TSK8-hydroxyquinoline resin can be eluted completely with less than 4 mL of acid.Although the volume of eluent can be reduced by evaporation with subsequent re-82dissolution for Ga and In, Ti cannot be treated by this method. Titanium requiresthe use of HF for dissolution. Not only is HF corrosive, but as discussed previously(Section, the use of fluoride-containing eluent results in SiF+ interferencewith Ti isotopes of interest. Thus, the analysis of seawater samples was carried outusing the TSK 8-hydroxyquinoline resin.3.2.7 ICP-MS Operating ConditionsColumn eluents were analysed by solution nebulization ICP-MS in the peakjumping mode. Sample solution was introduced using a 300 pi flow injectionsample loop with uptake rates of 0.6 mL/min. Signal duration, determined withsingle ion monitoring (SIM) mode, under these conditions was approximately 60seconds. Data acquisition began three seconds after the sample had reached theICP and continued for 45 seconds to truncate the tailing of the signal resulting inbetter sensitivity without sacrificing reproducibility. From the SIM mode, it wasestimated that over 90% of the signal was measured with the 45 second acquisitiontime.3.3 Results and Discussion3.3.1 Attempts to Remove Isobaric InterferencesIt was discovered that the source of calcium interference was not only fromthe sea water, but also from the KHP solution used in this study. This interferencepersisted even though the KHP solution had been cleaned by passing it throughthe 8-hydroxyquinoline resin prior to use. To remedy this problem, the eluents fromthe columns could be collected, diluted with 750 mL distilled deionized water andreprocessed according to the procedure described in Section 3.2.5. With no KHPrinse used during the second elution sequence, this led to significant removal ofthe calcium interference. The level of Ca removal was inconsistent, however, and8348oa+ interference was evident periodically. These additional procedures also ledto extended processing time.In an attempt to reduce the Ca interference without additional processing,the 5% NH4OAc solution wash was added to the processing sequence after theKHP wash. Some removal of Ca was observed. Again, Ca removal was notconsistent, thereby affecting the signal of 48Ti. Complete removal of the Cainterference has, as of yet, not been achieved. Further work is needed to developbetter methods of Ca removal. Reagents, such as the KHP solution, could becleaned more efficiently by passing them through a higher capacity resin, such asChelex, or by using a larger amount of the TSK 8-hydroxyquinoline resin.Alternately, by monitoring the "Ca isotope a correction could be made for theinterference from calcium at m/z 48. By using this method of correction, satisfactoryresults for Ti were obtained. Thus the final seawater processing procedure used isthis study is as described in Section The Method of Isotope DilutionIsotope dilution techniques were used for the analysis of Ti. The principle ofisotope dilution analysis (ID) is that, by altering the natural ratio between twoisotopes in the sample with an amount of an accurately known quantity of anisotopic spike (usually a minor isotope), the concentration of the analyte present inthe original sample may be determined from the measured isotope ratio. A majoradvantage of ID analysis is that ratios, rather than absolute sensitivities aremeasured. Thus, once the isotopic spike has been added and equilibrated, loss ofanalyte or incomplete recovery is not important since the isotope ratio will notchange. Also, ID analysis corrects for instrumental drift and matrix effects sinceboth isotopes are affected equally. It should be noted that ID techniques do notaccount for isobaric interferences. The only instrumental requirement of IDanalysis is that mass bias between isotopes must remain constant.84The amount (moles) of the minor isotope in the sample, bX s , can becalculated from the formula 9 :bXs bX t (R rn - R t ) (R - R )s^m( 3.1)where bXt is the number of moles of the minor isotope in the spike added, R m is theisotopic ratio for the mixture, Rt is the isotopic ratio of the spike solution, R s is theisotopic ratio of the sample. By incorporating the isotopic abundance of the minorisotope, f, the total amount of analyte, X s , may be determined.Xs bX t (R m - Rt) As = f (Rs - Rm)(3.2)It has been shown that, in theory, the most accurate results for ID analysesare obtained when the measured ratio, R m , equals the square root of the ratios ofthe spike and natural isotope product .' 0 . For Ti, with abundances in the spikesolution of 49T1 - 96.25%, 48-ri - 2.71%, this optimum spike ratio is 49/48 = 1.6.However, in practice, counting statistics dictate that maximum precision is obtainedfor isotope ratios of unity. In these studies, isotope ratios between 1 and 1.3 wereused.3.3.3 Method of Standard AdditionsThe method of standard additions as well as the calibration method wereused to determine Ga and In concentrations. Standard additions provide a monitorand a correction for signal suppression/enhancement caused by the sample matrixwhich would not be detectable by the method of external calibration. Due tosample limitations, one point standard additions were used in this study oncelinearity in this concentration range had been verified. To minimize extrapolation85error, the added standard resulted in concentrations approximately five times thatof the original sample 11 .3.3.4 Matrix EffectsIt was found previously that both indium and gallium suffered some form ofmatrix suppression at low concentrations (Section There are threepossible origins of the matrix effects; the seawater, the resin, or the reagents. Toevaluate the latter two possibilities a standard solution of 0.2 ppb Ga, Ti and In in2.3 M HNO3 was passed through the resin. Since this acid matrix is used to elutethe columns in the seawater studies, no metal retention is expected. Any deviationin instrumental response from the standard solution that was not passed throughthe column should therefore be a result of the resin or resin-reagent effects.Nine columns were used. Each column was cleaned with 20 mL of 2.3 MHNO3, and pH adjusted with 25 mL pH adjusted DDI water. The columns werethen divided into three groups and the various reagents were passed through asshown in Table 3.1.Column # KHP 5% NH40Ac DDI water(2x500 A) (8x1 mL) (5x3 mL)1 - 3 - - 44 - 6 4 - 47 - 9 4 4 4Table 3.1^Column washes for matrix effect experiment.All columns were washed with 5 x 3 mL of DDI water followed by elution with 4 x 1mL the standard solution. The eluent was collected and then analysed bynebulizer ICP-MS and compared with the signal obtained from the originalstandard solution.All eluents displayed similar degrees of signal suppression (-20 - 30%) forIn. Thus, it would appear that no signal suppression originated from the reagents,86KHP and NH4OAc. Matrix suppression is thought to be a result of the residue fromthe 8-hydroxyquinoline resin. Little or no signal suppression was observed forgallium in this experiment. Thus, perhaps part of the matrix suppression for Gafound for the nebulizer mode in section 2.3.4 originated from the seawater obtainedfrom the Halifax Harbour. It should be noted that the matrix effects for In were notobserved with high concentrations, such as the levels used for recovery studies. Itis not clear why matrix effects should only be observed only for low concentrations,however, it is obvious that care must be taken to correct for these effects duringanalysis. Only the nebulizer mode was used in these experiments. The use ofETV-ICP-MS, especially when employing the method of standard addition, was toocumbersome It was not necessary to use the ETV in these experiments sincesufficient signal was obtained in the nebulizer mode of operation for all elements.3.3.5 Titanium DistributionThe distribution of titanium found in the central gyre (Figure 3.2) shows asurface maximum (-100 pM), a sub-surface minimum (-50 pM) and increasingconcentrations with depth to a bottom maximum (-230 pM). Slightly elevateddissolved Ti levels are observed at -400-1000 m. This type of behavior suggeststhat titanium distribution in the central Pacific gyre is governed by a combination ofexternal inputs and removal mechanisms.The observed surface maximum with a subsurface minimum suggests anatmospheric input (aeolian source) with removal at mid-depth by scavenging ontoparticles. Results of this study are somewhat different from those found previously 1where no surface maximum was observed. This discrepancy is likely to be due tothe difference in atmospheric fluxes to the high latitude eastern North Pacific vs thecentral gyre. AV 10 is significantly further west and south of the previously studiedsites (VA and VII). Since atmospheric input is carried by the westerlies, the aeolian111KhloaliriliiIIIIIIIIIIie-i^ "01-0-1MN=el-^ .Iiilli illiiii 111111^if i 1—87Titanium (pM)50 100^150 200 2500.000 01000E, 2000_caa)0 300040005000Figure 3.2. Titanium distribution at AV 10. No error bar for the Ti concentration at 400 m since onlytwo determinations were performed.signature would be expected to be much more pronounced in western regions,closer to the Asian dust sources. The vertical distribution of dissolved Ti in theupper waters resembles that of dissolved Al where there is a rapid decrease inconcentration in the top 300m. Dissolved Al has been found to have a surfacemaximum from aeolian input with rapidly decreasing concentrations at mid-depthdue to rapid scavenging removal.The Ti distribution also exhibits slightly elevated levels at mid-depth (400-1000 m). This observed increase may be correlated to the oxygen minimumobserved at this station (Figure 3.3). Although Ti itself will not be reduced undersuboxic conditions, Ti may be released by association with Mn oxide particles.Manganese is known to be reduced and transported in suboxic environments 12 .Dissolved Ti concentrations below 1500 m were found to increase almostlinearly with depth to -230 pM. This type of distribution would suggest a bottom40005000Oxygen (gM)0.000 0 50 100 150 200 2501000200030000ii:4■11.-4■444.•••••••••■11-4■1•4■11-41-4114■•■••••■11:688water source such as a pore water flux or sediment surface remineralization. Atthis point, the exact nature of the bottom source of dissolved Ti is not known.Further studies of pore waters are needed to elucidate the dissolved Ti bottomsource.Titanium (pM)0 50 100 150 200 2500.0001000E 2000a)p 300040005000Figure 3.3 Titanium and 02 distribution at AV 10. No error bar for the Ti concentration at 400 m sinceonly two determinations were performed. Collection of the oxygen data was performed bythe Institute of Ocean Science (Sidney, BC). No error estimates were supplied.3.3.6 Gallium DistributionThe distribution of gallium (Figure 3.4) shows intermediate concentrations atthe surface (-17 pM) with a slight subsurface maximum at 200 - 300m (-20 pM), aminimum at -1000 m (5-10 pM) followed by increasing concentration with depth toa maximum at bottom waters (30 pM). The values found here are much lower thanthose reported in an earlier study by Culkin and Riley 4 (-500 pM), but similar tothose found more recently by Orians and Bruland 13 and Shiller6 . This profileshows subsurface and bottom sources of dissolved gallium as well as scavengingremoval in the water column.The source of dissolved gallium to the surface has been proposed to be fromdissolution of atmospheric dust6,6 . This data supports aeolian input, in that highervolivisiverivi Tills wiz IIIa-.11111 sisal a ii 145111111. Itm'IIIIIIIIiIiiiiirwoiNIIIIIII11111117el'1-Q11-e-1leileilei^i..IIIIIIIIIIIHI hiHIIIIIIIIIIIIIIL:89surface values (-17 pM) are found in this study than those found further east 5(<10 pM). Aeolian input from the Asian dust sources would predict that dissolvedgallium concentrations increase towards the west.01000_ 2000E"ELa)fn 300040005000Gallium (pM)0^5^10 15 20 25 30 35Figure 3.4 Gallium distribution at AV 10. The data points are a combination of the values obtainedusing the external calibration method and the method of standard additions. No matrixeffects were observed for Ga in these samples.The subsurface maximum and the mid-depth minimum may be acombination of two processes, horizontal advection and/or a vertical processinvolving exchange with sinking particles. Gallium taken up at the surface may bereleased in this region to give the subsurface maximum, which might be expected ifbiological uptake into soft tissues is occurring in surface waters. After release,gallium may then be scavenged by sinking particles resulting in the minimumobserved at 1000 m as was suggested previously5.• .•• •• S.E 400600I 180°W^170°W^160°W^150°W^140°W^130°W^120°W••••••90An alternate subsurface source may be due to lateral advection. Contoursgenerated by combining this data with previous values help to visualize thepossible horizontal processes (Figure 3.5).LongitudeFigure 3.5 Contours generated by combining data obtained in this study with those previouslyobtained5 . Numbers on the contour lines have units of pM.The subsurface maximum may be due to horizontal advection from the west wherethe isopycnal surfaces outcrop. This far western region of the Pacific is a region ofhigh dust input and thus a high Ga signal is expected. At this point, the source ofthe subsurface maxima is not clear. Additional data for sites further west arerequired to further investigate this theory.Dissolved gallium concentrations below 1000 m increase almost linearlywith depth to a bottom maximum. This is indicative of a bottom water source suchas diffusion from sediments or a surface sediment remineralization.11 1^1^1^1^1 1 1^1 1^I^I^I^lei e 1 l^III•••••••913.3.7 Indium DistributionThere are no other reliable data that may be used for comparison with this Indata. The behaviour of indium (Figure 3.6) is similar to that of gallium, withintermediate surface values (-0.3 pM), a subsurface maximum at -500 m (0.45pM), a mid-depth minimum at -1500 m (0.16 pM) and some increase inconcentration with depth to 0.3 pM in deep waters. The subsurface maximum issomewhat deeper than that observed for Ga, and the deep water values are lowerrelative to surface waters. A more detailed discussion of In will follow in Section3.3.9.1.Indium (pM)0.0^0.10^0.20^0.30^0.40^0.50010002000E_ca)p 300040005000Figure 3.6 Indium distribution at AV 10. Due to sample limitations, only two replicate determinationswere performed. The average values of the duplicates are presented here.3.3.8 Vertical Advection-Diffusion ModelA simple vertical advection-diffusion model may be used to approximate thedeep-water scavenging times for non-conservative elements with bottom sources.The model may be expressed as 14 :ra2[Ch^ra[Ch0 = Dz az2 - vz^+ Jwhere Dz is the turbulent mixing coefficient, [C] is the concentration of the element,z is the depth, v is the rate of seawater advection and J is the term that accounts forscavenging removal. It can be shown that, by rearranging the solution to the abovedifferential equation, the advection-diffusion model may be expressed by 14 :[C] = a + — z + 138vz(3.4)where a and 13 are constants, and 8 is the potential temperature. Alternately,potential temperature may be replaced by a conservative tracer such as salinity.By using multiple linear regression, a, J/v z and [3 can be determined, and thus, J,the scavenging term may also be determined assuming an upwelling rate of 4 m/yr.Temperature - salinity plots (Figure 3.7) at depthes below 1000 m at thestudy site show a linear relationship which allows application of a vertical, one-dimensional, advection-diffusion model 14 .Dissolved titanium data from below 1000 m, plotted with respect to salinitydemonstrates a concave contour indicative of scavenging removal (Figure 3.8).Residence times for dissolved titanium in this region are estimated at 500 - 600years, somewhat longer than the 100-200 years previously reported for dissolvedTi in the Subarctic currents. Longer residence times may be due to the reducedproduction in this central gyre location, leading to decreased particle flux in thewater column, and thus, decreased scavenging.92(3.3)C ^AO/••••••••••r 4.03.5—0; 3.02 2.5a) (%.)34.2^34.4^34.6^34.893Figure 3.7 Temperature vs. salinity plot for depths below 1000 m at AV 10. Data was obtained fromother scientists. Data provided did not contain uncertainties.Titanium (pM)100^150^200^25034.534.7534.5534.634.6534.7Figure 3.8 Titanium vs. Salinity in the deep Pacific ocean.94A plot of dissolved Ga in deep water with respect to a conservative tracer,salinity, also demonstrates evidence of scavenging behaviour (Figure 3.9). Usingthe simple advection-diffusion model, the scavenging residence times areestimated to be approximately 600±100 years. This is in agreement with previousvalues of 750±100 years 13 reported for station IV, also in the central gyre.Again, a plot of salinity with respect to In at depths greater than 1500 mindicate that indium is, similar to Ga and most other hydrolysis dominated species,possibly removed by particle scavenging. Unfortunately, the sparsity of dataavailable in the deep waters for In does not allow for an estimate of residence timesat this point.^Gallium (pM)^Indium (pM)0^5^10^15^20^25^30^35^0.19^0.23^0.26^0.30-81a34.5034.6034.7034.80Figure 3.9 Gallium - Salinity and Indium - Salinity Plots for the deep Pacific Ocean. Note: Indium -Salinity plot was obtained by curve fitting.34.1034.2034.3034.403.3.9 Interelement ComparisonsFrom the distribution of dissolved Ga and In (Figures 3.2 and 3.3), it wouldappear that both elements are governed by similar inputs and removalmechanisms. To be able to rigorously compare the oceanic chemistry anduptake/removal rates of the two elements, one must know their initial relative111111111111111111 14 ,y1• /• /Ff .0.977• /O tt 1111 lilt^ !III I'[Ga]/[In]10^20^30^40^50^6095sources and concentrations. To date, such information is not available for theactual Ga and In source. However if we consider only atmospheric dust sourcesand assume average crustal abundance for these elements in the atmosphericinput, then the expected Gail!, ratio would be approximately 250. Variations fromthis ratio would indicate, if our source estimate is correct, that removal rates forthese two elements vary.The dissolved Ga/i n ratios plotted vs depth (Figure 3.10) demonstrate arelatively linear decrease in the Ga/i n ratio with depth down to 1000 m. This changein ratio suggests that the controls for dissolved In and Ga are different since similarinput and removal processes and rates would result in constant Ga/in ratios. Thischange in ratio may be a result of Ga depletion by preferential scavenging relativeto In. In surface waters, closer to the atmospheric source, ratios are closer tocrustal values. As the effects of scavenging increase with depth, the Ga/i n ratiodecreases. Another possible explanation is that In is more involved in nutrient[Gal/[In]0 20 40 60 80 10012014071111^'II^IIII 111117• ••••5000  ^1200Figure 3.10 Plots of Ga/In ratios vs. depth at AV 10. The expected Ga/in ratio for crustal abundance is250.0.0001000— 2000_c300040000.000200.0400.0E600.00800.0100096cycling than Ga. Nutrients are depleted in surface waters and enriched at mid-depth. If In was controlled to a significant degree by nutrient cycling, it wouldexperience depletion at surface waters (increased Ga/in ratio) and In enrichment atmid-depth (decreased Ga/i n ratio). It is interesting that this decrease is linear.In deep waters, using the limited data available, the Gail !, ratio is closer to thecrustal than the upper water column. This increase in Ga/i n ratio may be a result ofthe closer proximity to the sediment source and would have a closer value to thecrustal abundance. At this point, based on the only available In profile, it is difficultto draw many conclusions about the factors controlling In distributions in the ocean.3.3.10 Comparison of Ti, Ga and In with AluminumIt is often useful to compare the concentrations of hydrolysis-dominated tracemetals against those of aluminum. Aluminum is the most abundant metal in theearth's crust and has been shown have a high atmospheric input in the surfaceoceans. Aluminum is also one of the most reactive species in the oceans.Dissolved Al is rapidly removed from the surface waters to give a minimum at 1000m 15 . Slightly higher dissolved Al concentrations found in deep waters suggest thepossibility of a bottom source.Since Al values were not available for this study site, values obtained in aneighboring site (IV) were used. Since dissolved Al levels have been previouslyfound to be relatively uniform in the central Pacific gyre 16 , it is thought that Al levelsand distributions at AV 10 and (IV) should be similar, It is likely, however, that Allevels at AV 10 would be slightly higher in the surface waters, since it is situatedwest of (IV) (elevated surface concentrations for Ga are found at AV 10 vs IV). If thiswas the case, then the calculated element/Ai ratios in the surface waters may beslightly over estimated, but still useful for comparison. A comparison of Ga/A1,97and Ti/Al ratios in seawater with those expected from crustal abundance (Table 3.2)is useful since much more is known about Al behavior in seawater.ElementRatio17Expected Ratio(mmoiimonSurfaceEnrichmentDeep WaterEnrichmentGa/AiIn/AITi/Al.0672.67x10-436.5904500.9750150011Table 3.2^Table of estimated enrichment factors of Ti, Ga and In with respect to Al for surface(<500 m) and deep (1500-4500 m) Pacific waters. Enrichment is defined asobserved/expected. Surface WatersIn surface waters of the central north Pacific, the Ti/Al ratio was found to beclose to that expected in the crust. If the Ti and Al sources were similar to thatexpected from crustal abundance and both elements encounter similar degrees ofdissolution, then this ratio suggests that the residence times of Ti and Al are similar(1-4 years) 15 . This short residence time is also consistent with the observedhorizontal concentration gradient between the present study site and the highlatitude North Pacific (V7,T7) 1 . Element surface residence times much longer thanthe surface water residence time (>50 years) would result in mixing of input andoutput signals and thus a conservative surface distribution would be expected.Gallium-aluminum ratios show an enrichment factor of -90. This enrichment in thesurface waters could suggest that Ga would have a residence time ofapproximately 100-350 years (90 times that of Al). This estimate is not reasonablesince a surface concentration gradient has been observed for Ga in the easterncentral Pacific gyre. Several possible explanations may be invoked in an attemptto account for the additional Ga. These include preferential scavenging of Al insurface waters, an additional Ga source not yet identified, a greater abundance ofGa in the atmospheric source relative to the average crust or an enhanced98dissolution of Ga from aeolian materials. The vertical advection-diffusion model fordeep ocean waters at this station has predicted that Ga has a longer residence(-600 years) time than Al (-100-200 years at IV) which suggests that Al may bepreferentially scavenged in comparison to Ga. A possible surface source, inaddition to crustal dust, is atmospheric Ga emissions related to coal burning 8 thatcould enter into the surface waters. Ga has been reported to show nutrient typecycling in the upper water column which could also contribute to increased Garesidence time. The fraction of Al that dissolves from atmospheric sources wasevaluated to be 5-10% 18 . It is possible that a larger fraction of Ga dissolves inseawater8 A combination of the above mentioned factors may contribute to the Gaenrichment over Al in surface waters.Many of the arguments used to explain Ga enrichment in the upper waters,such as increased dissolution, may be applied to try to unravel the large Inenrichment. The In/A1 ratio demonstrates an enrichment over the Ga/A1 ratios.Perhaps In is less particle reactive than Ga and thus would have a longerresidence time. Other factors, such as an additional source, may account for the Inenrichment as well. At this point, not enough data is available to accuratelypostulate many theories on what processes control In in the surface waters. Deep WatersThe enrichment of Ti/A1 in deep waters (11 times) may be a result ofpreferential scavenging of Al over Ti. Scavenging times of Ti in the deep watersare estimated at -600±100 years, considerably longer than the -50-150 yearspreviously estimated for A1 15 . This 4-10 fold difference in residence times,combined with the prominent bottom input observed for Ti, is likely to explain the11-fold enrichment of Ti in deep waters.Gallium-aluminum enrichment is -750 times the expected ratio from crustalabundance. Again, the presence of a significant Ga bottom source combined with99reduced particle reactivity compared to Al may partly contribute to such anobservation. However, since deep water residence times for Ga and Ti are similarand Ti is only enriched by one order of magnitude, preferential scavenging is notexpected to be the dominant cause of Ga enrichment. An additional source of Gato the whole ocean appears to be needed to explain the observed Ga enrichmentthroughout the water column. It is not thought that the Ga emissions from coal use,which might be invoked to explain surface enrichment, could have penetrated thedeep Pacific waters. The Pacific deep ocean waters are roughly 500-1000 yearsold. The use of large amounts of coal, enough to emit significant amounts of Ga,has only occurred within the last 300 years or less. Thus deep Pacific waters pre-date this era. At this point, the origin of the large Ga/Ai enrichment is still uncertain.The enrichment of In/A1 ratio in deep waters determined to be over threeorders of magnitude higher than those predicted by natural crustal abundances. Inorder for variations in scavenging intensity to explain this observation, a residencetime of a hundred thousand years for In would be required. This residence time isclearly unreasonable since oceanic mixing times are on the order of thousands ofyears, thus, allowing the oceans to completely mix the external input and outputsignals. Spatial distribution would not be observed with depth for such longresidence times unless internal cycles dominate the In distribution. Though someinternal cycling may be involved, the distribution of In does not follow any of theknown nutrients. This large enrichment thus suggests that our estimate of theatmospheric input is incorrect. That is, the abundance of In in the atmosphericsource is not of crustal abundance.It has previously 13 been argued that Al was preferentially scavenged relativeto Ga due to its speciation in seawater. From a thermodynamics point of view, itcan be shown that Al, Ga, In and Ti primarily exist in seawater as various hydroxidespecies (Table 3.3)19.100Element^Species Abundance (%)Al(OH)3 39Al(01-1)4- 61Ga(OH)3 1Ga(OH)4- 99TiO(OH)2 100In(OH)3 97In(OH)4- 3Table 3.3^Speciation of elements in seawater at pH 8.2.It had been hypothesized that Ga enrichment is due to the dominance of thenegatively charged Ga(OH)4 - species. Since oceanic particles have a net negativecharge 13 , Ga(OH)4 - is not as likely to be scavenged as a neutral specie, such asAl(OH)3. Indium is also predicted to exist predominantly as the neutral specieIn(OH)3. Extension of the above theory predicts that the Gail, ' ratio should beenriched due to increased scavenging of In. However, this was not observed inthis study. In fact, Ga/i n ratios were found to be lower than the ratio that would beexpected from crustal abundance. Thus, preferential scavenging does not appearto depend on molecular charge.It must be stated, however, that until better measures of Ga and In sourcesare available, it will be difficult to determine whether enhanced input or decreasedremoval is responsible for their enrichments in seawater.3.4 ConclusionsThe concentrations of titanium, gallium, and indium have been determinedin seawater by first preconcentrating with a TSK 8-hydroxyquinoline resin followedby analysis with ICP-MS.The range of dissolved Ti concentrations found was 50 - 230 pM consistentwith previous observations. Dissolved Ti showed elevated surface values (-100101pM) decreasing to a minimum at -250 m (-50 pM) and subsequently increasingwith depth to a bottom maximum (-230 pM). The distribution of Ti in surface watersstrongly supports an atmospheric source for titanium. Previous studies have foundTi surface concentrations to be much lower (< 10 pM) at a site (V7,T7) that wassituated much further from the Asian dust sources. The subsurface minimumobserved here suggest particle scavenging removal. A slight increase in Ticoncentration between -400-1000 m may be correlated to the oxygen minimumobserved at this site (AV 10). This suggests that Ti may be associated with Mnoxide cycling. The increasing concentration with depth strongly suggests a bottomsource, details of which are still unknown. A vertical advection-diffusion modelpredicts that residence times for Ti in deep waters to be approximately 500 - 600years.Gallium distributions found at AV 10 showed intermediate surfaceconcentrations (-17 pM) with a slight subsurface maximum at 200-300 m (-20pM), a minimum at -1000 m (5 - 10 pM) and an increase with depth to a maximumat bottom waters (30 pM). This type of distribution suggests a subsurface and abottom water source with scavenging removal throughout. Correlating this data toprevious Ga data obtained in the eastern North Pacific shows an increase indissolved gallium surface values towards the west, and strongly indicates anatmospheric source of gallium. Latitudinal contours generated with the combineddata set may indicate that the subsurface maximum is due to horizontal advectionfrom surface sources further west. Additional profiles obtained at study sites furtherwest would more clearly demonstrate the feasibility of the theory. Application of avertical advection-diffusion model to the data is this study predicts that dissolvedGa has a residence time of 600±100 years and is in agreement with previousstudies.Dissolved indium distribution at AV 10 was found to be similar to that of Ga,suggesting that the two elements may be governed by similar processes.102Dissolved In concentrations are -0.3 pM at the surface with a subsurface maximumat 500 m of 0.45 pM, decreasing to a minimum at 1000 - 1500 m of 0.12 pM andincreasing slightly at bottom waters to 0.28 pM. The Ga/in ratios are lower thanexpected from crustal sources and decrease with depth in the upper water column.One explanation may be that Ga is preferentially scavenged with respect to In inthe upper 1000 m. Alternately, if In undergoes more internal cycling than Ga, theGa/I n ratio would be expected to be depleted at mid-depth.Comparison of Ti, Ga and In concentrations with Al concentrations at (IV)showed enrichment for all three elements. Titanium enrichment over Al was foundto be 11-fold by comparison with crustal abundances. This enrichment may beexplained by preferential scavenging of Al and an increased bottom source for Ti.Comparing Ga/Ai and In/AI ratios show that both elements are significantly enriched(750- and 1500-fold respectively) over the crustal abundance. These largeenrichment factors may be attributed to a variety of sources. These may includepreferential dissolution of Ga and In; Ga and In sources not yet identified fromatmospheric aerosols; additional nutrient type cycling causing a regenerative input;and significant bottom water inputs. Variable scavenging removal rates do notappear to be able to explain these enrichments. Further studies are needed toadvance our understanding of behavior exhibited by these trace elements.1033.5 References1^Orians, K. J.; Boyle, E. A.; Bruland, K. W. Nature 1990, 348, 322-325.2^Uehara, N.; Morimoto, K.; Shijo, Y. Analyst 1991,116, 27-29.3^Skrabal, S. A.; Ullman, W. J.; Luther, G. W. Mar. Chem. 1992,37, 83-103.4^Culkin, F.; Riley, J. P. Nature 1958,181, 179-180.5^Orians, K. J.; Bruland, K. W. Geochim. Cosmochim. Acta 1988, 52,2955 2962.6^Shiller, A. M. Geochim. Cosmochim. Acta 1988, 52, 1879-1882.7^Landing, W. M.; Haraldsson, C.; Paxeus, N. Anal. Chem. 1986, 58,3031 3035.8^Orians, K. J.; Boyle, E. A. 1992 submitted for publication to Anal. Chem..9^Holland, G.; Eaton, A. N. Applications of Plasma Source MassSpectrometry; Royal Society of Chemistry, Cambridge: UK, 1991.10^Fassett, J. D.; Paulsen, P. J. Anal. Chem. 1989, 61(10), 643A-649A.11^Marshall, H.; Presented at the 75th Canadian Chemical Conference andExhibition, Edmonton, AB, June, 1992; paper EA 183.12^Martin, J. H.; Knauer, G. A. Nature 1985, 314, 524-526.13^Orians, K. J.; Bruland, K. W. Nature 1988, 332 , 717-719.14^Boyle, E. A.; Sclater, F. R.; Edmond, J. M. Earth Planet. Sci. Lett. 1977, 37,38-54.15^Orians, K. J.; Bruland, K. W. Earth Planet. Sci. Lett. 1986, 78, 397-410.16^Orians,K. J.; Bruland. K. W. Nature 1985, 316, 427-429.17^Greenwood, N. N.; Earnshaw, A. Chemistry of the Elements, PergamonPress: Oxford, 1984.18^Maring, H. B.; Duce, R. A. Earth Planet. Sci. Lett. 1987,84, 381-392.19^Turner D. R.; Whitfield, M, Dickson, A. G. Geochim. Cosmochim. Acta1981, 45, 855-881.20^Chow, T. J.; Snyder, C. B. Earth Planet. ScL Lett. 1969, 7, 221-223.21^Matthews, A. D.; Riley, J. P. Anal. Chim. Acta 1970, 51, 287-294.104Chapter 4Resin Development4.1 IntroductionIon exchange resins have a wide range of applications in analyticalchemistry, including the separation of ions from solution matrix andpreconcentration of trace metals from aqueous samples. Several resins have beendescribed earlier for this type of work; Chelex 1 , silica gel immobilized8-hydroxyquinoline 2 and TSK immobilized 8-hydroxyquinoline 3 . Chelex is acommercially available imide diacetate chelating resin. Although it has a highchelating capacity, Chelex swells and shrinks with changing pH, and is thereforecumbersome to work with. The functional chelating group in the latter two resins is8-hydroxyquinoline. 8-Hydroxyquinoline is a widely used chelating agent in liquid-liquid extraction for trace metal preconcentration and has a high affinity for bothtransition group and heavy metals. This property may be transferred to a solidphase by immobilizing 8-hydroxyquinoline on a solid support. Two of the solidsupports used are silica gel and TSK resin. The silica gel solid support exhibitsgood mechanical strength, and resistance to swelling and shrinking; however,silica gel is not stable and dissolves at high pH. Dissolution exposes fresh silicagel surfaces from which trace metals may be leached. Clearly, this potential sourceof contamination greatly inhibits the use of this material for trace metal analysis.Fractogel TSK is a highly cross-linked polymer containing methacrylic, ether andsecondary alcohol groups. The physical properties of the TSK resin are quitesimilar to those of silica gel. Unlike silica gel, this resin is stable under both inacidic and basic conditions, making the TSK gel resin an excellent support for ionexchange applications.Fractogel®, the original TSK resin is reported to have good capacities (-300gmol Cu(II)/g of resin) 3 . However, using a resin manufactured by Toyopearl®105(reported to be identical to Fractogel® which is no longer available), a reduced andvariable capacity was obtained (<1 p.mol Cu(II)/g of resin). Extensive researchefforts by our group over the last few years has not been fruitful in elucidating thecause of this reduced capacity and problematic synthesis. A portion of this thesishas therefore been devoted to investigating this problem, since access to a reliableion exchange resin is of fundamental importance to this type of trace analysis. Inaddition, the potential of a new ion exchange resin, using a new solid support toimmobilize a derivative of 8-hydroxyquinoline, has also been investigated.4.2 Experimental4.2.1 ReagentsAll reagents were analytical grade and were used as received unlessotherwise indicated. All water used was analytical grade type I distilled deionizedwhich was made using a Barnstead Nanopure water purification system.4.2.2 Synthesis of TSK 8-Hydroxyquinoline Resin 3Approximately 30 mL of the Toyopearl® TSK-Gel HW-75 (Supelco) resinslurry was washed sequentially with the following: 2 x 100 mL aliquots of 0.5 MNaOH; 3 x 100 mL aliquots of distilled deionized water; 2 x 100 mL aliquots of 1.0M HCI; 3 x 100 mL aliquots of distilled deionized water; 2 x 100 mL aliquots of 95%ethanol; 2 x 100 mL aliquots of acetone; and 2 x 100 mL of chloroform. All filteringtook place under vacuum conditions using a sintered glass funnel. The resin wasfiltered, then dried under vacuum (-0.1 torr) at -60-70°C overnight. The dried resin(3.08 g) was then added to a mixture of 6.18 g (0.033 mol) p-nitrobenzoyl chloride(Aldrich) and 15 mL (0.11 mol) of triethylamine (BDH) in 195 mL of chloroform andrefluxed under nitrogen for 48 hours. The reaction mixture106was filtered and washed with chloroform. The resulting faint yellow product wasdried under vacuum at ambient temperatures overnight.A solution of 2.13 g (0.027 mol) of Na2S (Fisher) in 20 mL 0.1 M acetic acidat pH -7, was added to the esterified resin to reduce the nitro moiety. The reactionmixture was allowed to stir for several hours to give the yellow amine product. Theproduct was filtered and washed with 4 x 10 mL of acetone to remove anyelemental sulfur which may have formed, followed by 3 x 20 mL of distilled water toremove the acetone.The resin was transferred to a clean reaction vessel with 5-8 mL of distilledwater. A cold solution of 5.0 g (0.072 mol) sodium nitrite (Eastman) in 1.0 M aceticacid was added to the resin. The reaction proceeded at 0°C for 45 minutes, thenthe diazotized resin was filtered and washed with cold water. The cold resin wastransferred to the reaction flask with 5 mL of cold water. A cold solution of 4.0 g(0.027 mol) 8-hydroxyquinoline (BDH) in 200 mL of 95% ethanol was added to theresin. If the coupling was successful, the resin turns dark orange to dark red at thispoint. The reaction was allowed to stir for two hours. The product was then filteredand rinsed with: 2 x 75 mL 0.5 M NaOH; 3 x 100 mL water; 2 x 75 mL 1.0 N HCI;and 3 x 100 mL water. The resulting resin was stored in water. A summary of thereaction sequence described is shown on the following page in Figure Affi-Prep® CouplingApproximately 25 mL of the cold (-20°C) Affi-Prep® (Bio-Rad) resin wasslurried and filtered through a sintered glass filter. Care was taken such that theresin did not filter dry. After washing the resin with 8 x 100 mL of cold (3°C) 10 mMsodium acetate at pH -4.5, the moist resin was transferred to a cold (0°C) 500 mLflask with 250 mL of 20 mM sodium acetate (BDH) at pH 7.1070^RefluxNO2CH2Cl2Et3NHCINO2CINH2OAcNENNaNO2, HOAcpH -70°COHN:NEtOHpH - 7NENOHReaction sequence for the coupling of 8-hydroxyquinoline to the TSK-GelResin.Figure 4.125°CNa2SNO2^0.1M HOAcpH - 7 NH2A 100 mL solution of 0.03 M 5-amino-8-hydroxyquinoline dihydrochloride (Aldrich)at pH -6 was added to the reaction vessel. The reaction sequence for the Affi-Prep® coupling is shown in Figure 4.2. After 2 hours of stirring, the reactionmixture was allowed to warm to room temperature. The coupled resin was stirredovernight before filtering through a sintered glass funnel. The resin was thenwashed with 5 x 200 mL distilled deionized water and stored in 60 mL distilleddeionized water.4.2.4 Capacity Test: Affi-Prep® ResinApproximately 0.5 mL of the coupled Affi-Prep resin was gravity packed in a0.8 x 4 cm Poly-Prep Chromatography column (Bio-Rad, Richmond CA) shown inAffi-Prep® Resin^5-amino-8-hydroxyquinoline011Coupled product^OH0 OH• 2HCI1 08Figure 4.3. Due to the instability of the 5-amino-8-hydroxyquinoline linkage to acid,acidic solutions were not used to clean the columns prior to loading the coppersolution. Column blanks were evaluated by eluting a column which had not beenexposed to the copper solution. After washing with 10 mL of water, the resin wasloaded with 3 x 10 mL aliquots of 20 ggig copper solution at pH -6. Interstitialcopper was removed with 10 mL of distilled deionized water. Chelated copper waseluted with the appropriate solution."Extensor Arm"0 H /Oso,).L N'-• ri lr. .)Fl^0Figure 4.2 Affi-Prep® coupling reaction sequence.109123Figure 4.3 Poly-Prep Chromatography column. 1) End cap 2) 10 mL reservoir3) 20 p.m frit to prevent the resin from escapingTwo neutral solutions, ethylenediaminitetraacetic acid (EDTA, Mallinckrodt)and 8-hydroxyquinoline, were evaluated in addition to the acid eluents. Becausethese solutions are neutral, they were passed through the resin for the purposes ofcleaning the resin, and to obtain a column blank. Copious amounts of water werepassed through the resin to remove interstitial 8-hydroxyquinoline solution prior toCu loading. Washing and eluting procedures for these neutral solutions weresimilar to those used for the acid eluent.Analysis of copper content, using matrix matched standards, was performedby flame atomic absorption spectroscopy (Perkin Elmer 560).4.2.5 Capacity Tests: TSK 8-Hydroxyquinoline ResinCapacity tests performed on the TSK immobilized 8-hydroxyquinoline resinwere very similar to those of the Affi-Prep resin. Approximately 0.5 mL of resin wasgravity packed in the Poly-Prep Chromatography column. Each column waswashed with 20 mL of 2.3 N HNO3 to displace any previously chelated metals.Resin loading was accomplished by passing 3 x 10 mL aliquots of 20 gg/g copper1 1 0solution at pH-6 through each column. Interstitial copper was removed with 10 mLof distilled deionized water. Chelated copper was eluted with 2.3 N HNO3.4.3 Results and Discussion4.3.1 Studies on 8-Hydroxyquinoline Immobilization onto TSKThe synthetic sequence for the coupling of 8-hydroxyquinoline to the solidsupport consisted of four steps (Figure 4.1). Elucidation of the step responsible forthe decreased chelating capacity (compared to the immobilization onto Fractogel®as reported by Landing et al. 3 ) was desired. It was not possible to isolate thecoupling step since the intermediate diazonium salt is unstable and maydecompose explosively. The study was accomplished by the diazotization andcoupling of a model compound, ethyl p-amino benzoate (1), with 8-hydroxyquinoline under the same conditions as those used to immobilize 8-hydroxyquinoline onto the Toyopearl TSK resin. These simple compounds wereused as models to simplify the spectral analysis of the resulting product.Compound 1 was treated with NaNO2 and HOAc followed by 8-hydroxyquinoline(1.1 equivalent) in a similar manner to the resin coupling.COOEtCOOEt NaNO2, HOAcpH -7, 0°CNH21OH3Figure 4.4 Model compound (ethyl p-amino benzoate) reaction sequence.111A strong color change from pale yellow to bright red occurred upon the addition of8-hydroxyquinoline to the diazonium salt, which suggests that this coupling wasquite facile and that failure at these two steps was not the source of reducedcapacity. Typically, in an unsuccessful resin coupling, little or no color change wasobserved denoting little 8-hydroxyquinoline coupling. The color change in themodel compound is due to the formation of the azo product and not other by-products. A 1 H NMR of the crude reaction mixture showed that starting materialswere present. However, the appearance of new peaks at 7.3-7.4, 8.6-8.7, and 9.0-9.1 ppm, which were expected for coupled product, were also observed. Theproduct was separated by flash column chromatography (methylene chloridefollowed by ethyl acetate), and preliminary characterization was performed. Gaschromatography (GC) of the isolated material showed that- the materialdecomposes on heating to give starting materials. The GC integrations showedthat approximately 99% of the product decomposed within the GC, but theremaining 1% gave retention times consistent with those expected for the desiredproduct. 1 H NMR spectra of the purified material had new peaks consistent withthe predicted structure of the product.The work with the model compound indicates that the diazo couplingreaction was successful (albeit in low yield -30%). The low yield may be due to theuse of only 1.1 equivalents of 8-hydroxyquinoline in the model compound couplingreaction sequence. This is not expected to result in a low yield in the resincoupling since the 8-hydroxyquinoline is present in 3-5 fold excess. Thus, theproblem was due to one of the earlier two steps in the resin reaction sequence.The second step was not likely to be the limiting step since Na2S is a strongreducing agent, and should easily reduce the nitro moiety into an amine. Theesterification of the secondary alcohol on the resin must therefore be the limitingstep. Esterification efficiency could be reduced by the presence of water reacting112with the acyl chloride to convert it to the unreactive carboxylic acid. Althoughmethylene chloride is not hygroscopic, precautions were taken to perform thereaction under nitrogen with freshly-distilled solvent, in addition to using excessacid chloride. These precautions did not overcome the low-capacity problem.Since the reagents were not the cause of failure, perhaps it was the resinitself. The procedure required that the resin be dried overnight in an oven prior touse, however, any water hydrogen-bonded to the secondary alcohol groups wouldnot be removed under these conditions. Hydrogen bonding with water wouldprevent any reaction of the hydroxyl moiety with the acid chloride since the latterreagent would be hydrolyzed before esterification could occur. Drying the resinunder vacuum gave inconsistent results, as the resulting capacities obtained werelow and variable. Finally, the resin was dried under heat (60°C) and vacuumovernight prior to use. The resulting capacities were significantly higher than thosepreviously found (-30 gmol Cu(II)/g of resin). The capacity achieved with theToyopearl® resin was still lower than that reported with the Fractogel® resin, eventhough both resins should have the same structure and functional groups. Clearly,further work is needed to optimize the capacity of this resin. However, the capacityobtained through these studies was sufficient for use in the preconcentration oftrace metals from seawater.The capacity of this new 8-hydroxyquinoline immobilized resin (30±2 gmolCu(II)/g of resin) is 30 times greater than the previously synthesized resin. Thisincrease was probably due to the increased esterification efficiency resulting fromthorough drying of the resin under vacuum oven conditions prior to use. From thework on the model compound, it was evident that the 8-hydroxyquinoline couplingefficiency in the last step needs to be increased. The yield in the coupling reactionmay be improved by adjusting the pH more carefully during the coupling step, or bysubstituting hydrochloric acid, a more commonly used acid media for this type ofreaction 4 , for acetic acid in the diazotization step.1134.3.2 Studies on Affi-Prep® ResinCapacity tests on the Affi-Prep® resin were performed using several differenteluents. The linkage of either the extensor arm or the 5-amino-8-hydroxyquinolinegroup were unstable and as a result, some of the 5-amino-8-hydroxyquinolinegroups were cleaved each time the resin was eluted with 2.3N HNO3. Thecolumns were not washed with acid prior to Cu(II) loading. Instead, a columnpacked with the same amount of resin was eluted without Cu(II) loading and usedas the blank. A more dilute acid eluent, such as 1% (0.14N) HNO3 was used.Similar degradation of the 5-amino-8-hydroxyquinoline immobilization wasobserved. The capacity obtained with acid eluents are shown in Table 4.1.Eluent Calculated Capacity2.3 N HNO3 34 + 31% (0.14 N) HNO3 34 ± 3EDTA 22 ± 18-hydroxyquinoline 5 ± 1Table 4.1^Resin capacities obtained from Affi-Prep® resin.The resin capacity was indicative of quantitative coupling of the 5-amino-8-hydroxyquinoline to the solid support. The active ester content of the Affi-Prep®quoted by the manufacturer was -30 gmole/g or resin. It was found that theunreacted resin possessed some chelating properties (-3 gmole / g or resin). Thusa capacity of 34 ± 3 is indicative of the combined resin and coupled 5-amino-8-hydroxyquinoline capacity. Acid blanks were passed through the resin exhibitedsignals below detection limits for the FAA.Neutral eluents, such as aqueous solutions of 8-hydroxyquinoline or EDTA,were used in an attempt competitively remove the Cu from the resin. Unfortunately,even with the large volumes used (50 times the resin volume), only a fraction of thecopper may be removed by this method when compared to the acid elution method114(Table 4.1). Competitive ion exchange was not found to be effective with EDTAand 8-hydroxyquinoline for reasonable eluent volumes.4.4 ConclusionsResin capacity of the TSK immobilized 8-hydroxyquinoline has beenimproved over 30-fold by the thorough drying of the resin under vacuum ovenconditions prior to use.A new resin has also been developed by the coupling of 5-amino-8-hydroxyquinoline to a solid support called Affi-Prep®. Although good resincapacity was obtained, the weak linkages of either the extensor arm or thechelating agent resulted in decoupling of the 5-amino 8-hydroxyquinoline underacidic conditions from the Affi-Prep® resin. Some eluting abilities were observedwith 8-hydroxyquinoline and EDTA solutions. However, the solutions were not aseffective at elution as acids. In light of the matrix effects observed in the ICPcaused by the presence of organic compounds, the Affi-Prep® resin is not viablefor trace metal preconcentration without subjecting the eluent to evaporation anddigestion to remove the organic residue eluted from the resin after thepreconcentration step.1154.5 References1. Paulson, A. J. Anal. Chem. 1986, 58, 183-187.2. McLaren, J. W.; Mykytiuk, A. P.; Willie, S. N.; Berman, S. S. Anal. Chem.1985, 7, 2907-2911.3. Landing, W. M.; Haraldson, C.; Paxeus, N. Anal. Chem. 1986, 58,3031-3035.4. Streitwieser, A. Jr.; Heathcock, C. H. Introduction to Organic Chemistry,2nd ed.; Macmillan: New York, 1981; Chapter 25.116Chapter 5SummaryThe method of sample introduction has been the weak link in elementalanalysis. Solution nebulization suffers from low transport efficiency and molecularinterferences. The ETV has been shown to be a useful alternative to thenebulization as a method of sample introduction. Its increased transport efficiencyhas demonstrated sensitivities more than two orders of magnitude higher than thenebulizer mode. Subsequently, significant improvements in detection limits werefound for many of the elements investigated. Signals from high mass elements werefound to be interferent free and possessed detection limits in the range of 0.019 -0.14 fmole. Elements at the middle and lower end of the mass scale, such as Ti, Aland Zr, suffered isobaric interferences in the ETV mode. These interferences wereenhanced with the introduction of freon into the nebulizer gas. Detection limits forthese elements were in the range of 20 - 72 frnole. When it did not contribute toisobaric interferences, freon was found to be very effective for forming volatilefluorides with high boiling or refractory carbide forming metals such as Hf, Th and U.The linear dynamic range was -10 3 for most elements. The maximumconcentrations in the linear dynamic range are normally found to be less than 10ppb, thus making the ETV a complementary method of ultra-low level detectionmethod for the pneumatic nebulization system. Precision obtained using the ETV is2-10%, somewhat inferior to the precision obtained using the nebulizer. Significantreduction in the RSD may be obtained using an autosampler, since manualinjections were found to have approximately 4% RSD. Matrix effects were found forthe ETV mode for the analysis of processed seawater samples.It is concluded that the most efficient way to utilize the ETV would be byemploying isotope dilution techniques. The ETV shows comparable precision inisotope ratio measurements to the nebulizer method. Analysis by external117calibration or by standard additions is lengthy and cumbersome but is required sincematrix enhancement was found from the analysis of seawater eluents.The concentrations of titanium, gallium, and indium were determined inseawater collected in the central Pacific Gyre (174° 59.4'E, 27° 46.5'N). Theanalysis procedure involved a solid-liquid extraction using a TSK immobilized 8-hydroxyquinoline resin followed by analysis using ICP-MS in nebulizer mode usingboth the method of external calibration as well as standard additions. The range ofdissolved Ti concentrations found was 50 - 230 pM consistent with previousobservations. Dissolved Ti showed elevated surface values, a minimum at -250 m,and a bottom maximum. Combined with previous studies, the distribution of Ti insurface waters strongly supports an atmospheric source for titanium. The observedsubsurface minimum suggests scavenging removal in the water column. A slightincrease in Ti concentration at depths between -400-1000 m may be correlated tothe oxygen minimum observed at this site. The increase in concentration with depthstrongly suggests a bottom source. A vertical advection-diffusion model predicts thatresidence time for Ti in deep waters to be approximately 500 - 600 years.Gallium distributions found at this site showed intermediate surfaceconcentrations, a slight subsurface maximum at 200-300 m, a minimum at -1000 mand a bottom maximum. This type of distribution suggests a subsurface and abottom water source with scavenging removal throughout. Correlating this data toprevious Ga data obtained in the eastern North Pacific shows an increase indissolved gallium surface values towards the west, and strongly indicates anatmospheric source of gallium. Latitudinal contours generated with the combineddata set may indicate that the subsurface maximum is due to horizontal advectionfrom surface sources further west. Additional profiles obtained at study sites furtherwest would more clearly demonstrate the feasibility of the theory. Application of a118vertical advection-diffusion model to the data is this study predicts that dissolved Gahas a residence time of 600 ± 100 years.Dissolved indium distribution at this station was found to be similar to that ofGa, suggesting that the two elements may be governed by similar processes.Dissolved In distributions show intermediate surface values, a slight subsurfacemaximum at 500 m, a minimum at 1000 - 1500 m and a slight increase at bottomwaters. The Ga/in ratios are lower than expected from crustal sources and decreasewith depth in the upper water column which suggests that Ga and In are controlledby different processes or by similar processes at different rates.Comparison of Ti, Ga and In concentrations with Al concentrations at (IV)showed enrichment for all three elements. Titanium enrichment over Al may beexplained by preferential scavenging of Al and a Ti source for bottom waters.Comparing Ga/,6,1 and In/Al ratios show that both elements are significantly enriched(750- and 1500-fold respectively) over the crustal abundance. These largeenrichment factors may be attributed to preferential dissolution of Ga and In; Ga andIn sources not yet identified from atmospheric aerosols; additional nutrient typecycling causing a regenerative input; and significant bottom water inputs. Variablescavenging removal rates do not appear to be able to explain these enrichments.Further studies are needed to advance our understanding of behavior exhibited bythese trace elements.In recent years, the synthesis of the TSK immobilized 8-hydroxyquinoline hasproduced resins with low chelating capacities. Thus, efforts to improve the couplingsequence or to find a suitable new resin were merited. The failure of the couplingsynthesis was discovered to be due to water hydrogen-bonded to the resin. Resincapacity of the TSK immobilized 8-hydroxyquinoline has been improved over 30-foldby the thorough drying of the resin under vacuum oven conditions prior to use.119Further improvements to the chelate capacity may be achieved by changingthe acid media from acetic acid to hydrochloric acid, a more commonly used acid inthe diazotization step.A new resin has also been developed by the coupling of 5-amino-8-hydroxyquinoline to a solid support called Affi-Prep®. Although good resin capacitywas obtained, the weak linkages of either the extensor arm or the chelating agentresulted in decoupling under acidic conditions. 8-hydroxyquinoline and EDTAsolutions had some eluting abilities, but these were not as effective as the acideluents. In light of the matrix effects observed in the ICP caused by the presence oforganic compounds, the Affi-Prep® resin is not viable for trace metalpreconcentration without subjecting the eluent to lengthy evaporation and digestionprocedures to remove the organic residue eluted from the resin after thepreconcentration step.


Citation Scheme:


Citations by CSL (citeproc-js)

Usage Statistics



Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            async >
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:


Related Items