Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Development and characterization of furnace atomization plasma excitation spectrometry Hettipathirana, Terrance Dayakantha 1993

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

Item Metadata

Download

Media
831-ubc_1993_fall_phd_hettipathirana_terrance.pdf [ 7.23MB ]
Metadata
JSON: 831-1.0059353.json
JSON-LD: 831-1.0059353-ld.json
RDF/XML (Pretty): 831-1.0059353-rdf.xml
RDF/JSON: 831-1.0059353-rdf.json
Turtle: 831-1.0059353-turtle.txt
N-Triples: 831-1.0059353-rdf-ntriples.txt
Original Record: 831-1.0059353-source.json
Full Text
831-1.0059353-fulltext.txt
Citation
831-1.0059353.ris

Full Text

DEVELOPMENT AND CHARACTERIZATION OFFURNACE ATOMIZATION PLASMA EXCITATION SPECTROMETRYbyTERRANCE DAYAKANTHA HETTIPATHIRANAB.Sc., University of Colombo, Sri Lanka, 1985M.Sc., University of British Columbia, 1989A THESIS SUBMITTED IN PARTIAL FULFILLMENT OFTHE REQUIREMENTS FOR THE DEGREE OFDOCTOR OF PHILOSOPHYinTHE FACULTY OF GRADUATE STUDIESDEPARTMENT OF CHEMISTRYWe accept this thesis as conformingto the required standardUNIVERSITY OF BRITISH COLUMBIAJuly 1993© Terrance Dayakantha Hettipathirana, 1993In 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.(Signature) Department of ^The University of British ColumbiaVancouver, CanadaDate  kf-A-n 1-31^1 3DE-6 (2/88)iiABSTRACTFurnace Atomization Plasma Excitation Spectrometry ( FAPES ) is a newemission spectrochemical method which employs a graphite furnace foranalyte atomization and an atmospheric pressure plasma sustained insidethe furnace for analyte excitation. The primary objective of this work was tocharacterize the fundamental processes that are occurring within the plasmaduring the analyte atomization, vaporization, and excitation.Background spectra are dominated by emission features from COP, N2+,OH, NH, and N2. The plasma background emission is most intense near theradio frequency ( RF ) electrode and less intense near the graphite furnacewall. The Fe and Pb-excitation temperatures are in the range of3000 - 5000 K at RF powers between 10 and 100 W. The Fe-excitationtemperature also exhibits a spatial dependence. The emission features ofCO+ and N2+ indicate that this plasma source is capable of exciting energylevels as high as 20 eV.Both atomic absorption and emission experiments show a non-uniformtemperature distribution along the length of the RF electrode. Thetemperature lag of the RF electrode relative to the furnace wall causescondensation of analytes on the RF electrode and subsequent second-surfacevaporization resulting in two peaks in the temporal response of the emissionsignal. Analyte condensation on the RF electrode is severe at low RF powersand can be observed when high amounts of analyte are deposited. Similartemporal responses are observed for simultaneously measured atomicabsorption and emission signals for Ag and Pb. The time-resolvedPb-excitation temperature also suggests that the temporal emission profilesof these analytes in FAPES are determined by atomization and vaporizationcharacteristics of the analyte rather than by excitation characteristics.Results obtained for Pb also show an early shift in the appearance andpeak temperatures in FAPES compared with those in Graphite FurnaceAtomic Absorption Spectrometry ( GFAAS ), probably because of a shift in thedissociation equilibria of Pb species in the gas-phase. Experimental resultsshow the presence of high levels of CO in the FAPES source due to theenhanced oxidation of graphite on the RF electrode and on the graphitefurnace wall in the presence of the plasma. At RF powers higher than 50 W,the Pb emission intensity decreases. The highest signal-to-noise andsignal-to-background ratios are observed at relatively low RF powers ( about20 W).For both Pb and Ag, the major cause of interference effects from NaC1 isthe formation of volatile molecular species which are lost prior toatomization, and consequently, leads to a decrease in the atomic emissionintensity in FAPES. The interference effect for both Pb and Ag in a NaNO3matrix is complex and exhibits both condensed and gas-phase effects.The work presented in this thesis demonstrates that the FAPES sourcehas the potential to be a potent excitation source for atomic emissionspectrometry. This work also identifies the limitations of FAPES andsuggests further improvements to the instrumentation.ivTABLE OF CONTENTSAbstract^ iiTable of Contents^  ivList of Tables  viiiList of Figures^  ixList of Abbreviations xvAcknowledgements^ xviiChapter 1Introduction^ 11.1 FAPES: Historical Development^ 21.2 Furnace Atomization Plasma Excitation Spectrometry^ 81.3 Analyte Atomization: The Graphite Furnace^ 111.3.1 L'vov Furnace^ 121.3.2 Massmann Furnace 131.3.3 Limitations of Massmann-type Furnaces^ 151.3.3.1 L'vov Platform^ 161.3.3.2 Integrated Contact Cuvette 171.3.4 Atomization Mechanisms 181.4 Analyte Excitation: The Plasma^  201.4.1 The RF Discharge at Atmospheric Pressure^ 211.4.2 The RF Discharge in FAPES 251.4.3 Excitation Mechanisms^ 261.4.4 Plasma Temperature 291.4.5 Spectroscopic Temperature Measurements^ 311.5 Overview of the Thesis^ 34Chapter 2The Experimental System^  362.1 Instrumentation  362.1.1 The Plasma Source Work-Head^ 38V2.1.2 The Atmospheric Pressure RF Discharge^ 412.1.3 Spatially Resolved Intensity Measurements 432.1.4 Spectral Isolation and Detection^ 432.1.5 Atomic Absorption Measurements 442.1.6 Graphite Furnace Temperature Measurements^452.2 Data Acquisition^  462.3 Data Processing 472.4 The FAPES Method 47Chapter 3Spectral and Excitation Characteristics of the AtmosphericPressure Helium Plasma Source in FAPES^ 493.1 Introduction^ 493.2 Experimental 513.2.1 Plasma Background and Iron Emission Spectra^ .513.2.2 Spatially and Temporally Resolved Spectra  .533.3 Results and Discussion^ 543.3.1 Background Emission Spectra^ 543.3.2 Spatially Resolved Background Emission Intensities^ 613.3.3 Temporally Resolved Background Emission Intensities. . 643.3.4 Fe-Excitation Temperatures^ 693.3.5 Atmospheric Pressure Ar Plasma  743.4 Summary^ 76Chapter 4Temporal Emission and Absorption Characteristics of Silver, Lead,and Manganese in FAPES^ 784.1 Introduction^  784.2 Experimental  804.2.1 Signal Presentation^ 804.2.2 Appearance and Peak Temperatures^ 814.2.3 Reagents^ 814.2.4 Procedure  814.3 Results and Discussion 82vi4.3.1 Silver^  834.3.2 Lead 934.3.3 Manganese 1074.4 Summary^  116Chapter 5Emission Characteristics and Figures of Merit for Lead inFAPES^ 1185.1 Introduction^  .1185.2 Experimental 1215.2.1 Signal Presentation^ 1225.2.2 Reagents^  1225.2.3 Procedure 1225.3 Results and Discussion 1235.3.1 Time-resolved Pb-excitation temperature^ 1235.3.2 Effect of RF power^ 1285.3.3 Signal-to-Noise ratio  1335.3.4 Figures of merit 1435.4 Summary^  149Chapter 6Effects of Sodium Chloride and Sodium Nitrate on the Lead andSilver Emission in FAPES^ 1516.1 Introduction^  .1516.2 Experimental 1546.2.1 Signal Presentation^ 1546.2.2 Reagents^  1546.2.3 Procedure 1556.3 Results and Discussion 1566.3.1 Reflected Power^ 1566.3.2 Temporal Response of Na ( as NaCl and NaNO3 ) ^ 1586.3.3 Effects of NaC1 and NaNO3 on the Pb Emission Intensity .1616.3.4 Effects of NaC1 and NaNO3 on the Ag Emission Intensity .1726.3.5 Effects of ascorbic acid and phosphoric acid 178vii6.4 Summary^  185Chapter 7Conclusions^  188Bibliography 195viiiLIST OF TABLESTABLE^ PAGE3.1. Transitions of species observed in helium plasma atatmospheric pressure in 215 - 515 nm range.^ 583.2. Fe ( I ) lines used for the temperature measurement andrelevant spectral data [83]. ^ 705.1. Wavelengths,^Excitation^energies,^and^Spectralcharacteristics for Pb lines used for the Pb-excitationtemperature determination [122].  124ixLIST OF FIGURESFIGURES^ PAGE1.1. Schematic diagram of the FANES source. ( Adapted fromH. Falk, E. Hoffmann, and Ch. Ludke, Spectrochim. Acta,283, 39B ( 1984 ), with permission of Pergamon JournalInc. ) 51.2. Schematic diagram of the HA-FANES source. ( Adaptedfrom N. E. Ballou, D. L. Styris, and J. M. Harnly,J. Anal. At. Spectrom., 1141, 3 ( 1988 ), with permission ofThe Royal Society of Chemistry. ) 71.3. Schematic diagram of the FAPES source. ( Adapted fromD. C. Liang and M. W. Blades, Spectrochim. Acta, 1049,44B ( 1989 ), with permission of Pergamon Journal Inc. ) 91.4. Schematic diagram of the L'vov graphite furnace andelectrode assembly. ( Adapted from B. V. L'vov,Spectrochim. Acta, 53, 24B ( 1969 ), with permission ofPergamon Journal Inc. ) 131.5. Schematic diagram of the Massmann graphite furnace.( Adapted from H. Massmann, Spectrochim. Acta, 215, 23B( 1968 ), with permission of Pergamon Journal Inc. )  141.6. Schematic representation of the L'vov platform inside thegraphite furnace.^  171.7. Schematic diagram of an RF discharge system ^ 242.1. Schematic diagram of the experimental system  372.2. Schematic diagram of the plasma source work-head^ 392.3. End view of the (a) vertical mount and (b) horizontal mount.The dash line represents the RF connector. 413.1. Background Spectra of helium plasma. (a) 215 - 300 nm,(b) 300 - 400 nm, and (c) 400 - 515 nm.^ 57FIGURES^ PAGE3.2. Emission spatial profile for COP  623.3. Temporal emission behavior for (a) He ( I ), (b) COP,(c) N2+, and (d) OH at RF powers of 14, 18, 22, 26, and30 W (e) furnace temperature corresponding to thesediagrams   683.4. Emission spectrum for 50 ng Fe at an RF power of 20 W  713.5. Iron excitation temperature as a function of RF power. 733.6. Background spectrum of the Ar plasma for the spectralregion, 220 - 260 nm  754.1. Temporal response of the Ag atomic absorption signal at anRF power of 0 W for 0.5 ng of Ag deposited on the furnacewall with ( - - - ) and without ( ) the graphite co-axialrod; background ( • • ); and the temperature profile of thefurnace wall ( — — ). 844.2. Temporal response of the Ag atomic absorption signal at anRF power of 0 W for 0.5 ng of Ag deposited on the graphiteco-axial rod ( ) and tungsten co-axial rod ( - - - ); andbackground ( • • ).   864.3. Temporal response of the Ag atomic emission signal at anRF power of 20 W for 0.5 ng of Ag deposited on the furnacewall ( ) and RF electrode ( - - - ); and background( • • ).  884.4. Temporal response of the Ag atomic emission signal at anRF power of 20 W for Ag deposited on the furnace wall.Analyte amounts: 0.05 ( — ), 0.25 ( - - - ), 0.50 ( - • - • - ),and 5.0 ng ( — — — ); and background ( • • • ) at 20 W  904.5. Temporal response of the Ag atomic emission signal for0.5 ng of Ag deposited on the furnace wall at an RF powerof 14 ( — ), 20 ( - - - ), 26 ( - • - • - ), 32 ( — — — ), and38 W ( - ••• - ••• ); and background at 20 W ( • ).   91xiFIGURES PAGE4.6. Temporal response of the Ag atomic absorption signal for0.5 ng of Ag deposited on the furnace wall at an RF powerof 20( • • • ) and 40 W ( - - - ).  924.7. Temporal response of the Pb atomic absorption signal at anRF power of 0 W for 5 ng of Pb deposited on the furnacewall with ( - - -) and without ( ) the co-axial rod; andbackground ( • • • ).   944.8. Temporal response of the Pb atomic absorption signal at anRF power of 0 W for 5 ng of Pb deposited on the graphiteco-axial rod ( ) and tungsten co-axial rod ( - - - ); andbackground ( • • • ).   974.9. Temporal response of the Pb atomic emission signal at anRF power of 20 W for 5 ng of Pb deposited on the furnacewall ( — ) and RF electrode ( - - - ), 0.5 ng of Pbdeposited on the furnace wall ( -•-•-, x 10); and backgroundat 20W(•••).   1004.10. Atomic absorbance for 5 ng of Pb deposited on the furnacewall as a function of RF power: for 1 - 5 ( • ), and 1 - 3 s( • ).   1014.11. Temporal response of the Pb atomic emission signal for 5 ngof Pb deposited on the furnace wall at an RF power of 14^ ), 20 ( - - - ), 30 ( - • - • - ), and 40 W ( — — — ); andbackground at 20 W ( ••• ). ^ 1054.12. Temporal response of the Pb signal for 5 ng of Pb depositedon the furnace wall at an RF power of 40 W: emission) and absorption ( - - -).  1064.13. Temporal response of the Mn atomic absorption signal at anRF power of 0 W for 1.25 ng of Mn deposited on the furnacewall with ( - - - ), and without ( ) the co-axial rod; andbackground ( • • • ).  1084.14. Temporal response of the Mn atomic absorption signal at anRF power of 0 W for 1.25 rig of Mn deposited on thegraphite co-axial rod ( — ), tungsten co-axial rod ( - - - );and background ( • • ).  109xiiFIGURES^ PAGE4.15. Temporal response of the Mn atomic emission signal at anRF power of 20 W for 1.25 ng of Mn deposited on thefurnace wall ( — ) and RF electrode ( - - - ); andbackground at 20 W ( - - • ).  1114.16. Temporal response of the Mn atomic emission signal for1.25 ng of Mn deposited on the furnace wall at anRF power of 14 ( — ), 20 C- - - ), 30 ( - • - • - ), and 40 W( — — — ); and background at 20 W ( ). 1134.17. Atomic absorbance for 1.25 ng of Mn deposited on thefurnace wall as a function of RF power: for 2.8 - 5.5 ( • ),and 3.5 - 5.5 s(•).  1144.18. Temporal response of the Mn signal for 1.25 ng of Mndeposited on the furnace wall at an RF power of 40 W:emission ( ) and absorption ( - - - ).   1155.1. Time-resolved Pb-excitation temperature (^) at anRF power of 50 W and the corresponding graphite furnacetemperature ( - - - ) during the atomization step of Pb.^ 1255.2. Effect of the RF power on the net emission intensity ofPb ( • ) and on the Boltzmann factor ( - - - ). ^ 1295.3. Temporal response of the Pb emission signal at anRF power of 20 W for 0.5 ng of Pb ( ^ ); background( - - - ). ^ 1345.4. Lead signal-to-noise ratio ( SNR ) as a function of signalintensity. 1385.5. Effect of RF power on the signal-to-noise ratio ( SNR: • )and the signal-to-background ratio ( SBR: • ) for Pb ^ 1395.6. Effect of RF power on the net background emission ( • ) andthe^relative^standard^deviation^of^thebackground ( RSDB: 0).  1405.7. Effect of the spectral bandwidth on the backgroundemission intensity at 283.3 nm; 20 W ( • ) , 40 W ( • ). ^ 141FIGURES^ PAGE5.8. The analyte calibration graph for Pb   1445.9. Effect of integration time on the sensitivity of the Pbemission intensity. The values in the parentheses give thecorrelation coefficient of the least square fit for eachcalibration graph  1455.10. Effect of integration time on the detection limit ( DL ) forPb.  1466.1. Temporal response of the reflected power at an RF power of20 W for the deposition of distilled water ( — ), and160 ng of Na as NaC1 in distilled water ( • • • ); and thetemperature profile of the graphite furnace ( - - - ).  1576.2. Temporal response of the Na emission signal at anRF power of 20 W for the deposition of 160 ng of Na asNaC1 ( — ), and 160 ng of Na as NaNO3 ( - - - ); andwater blank ( • • • ).   1596.3. Temporal response of the Pb emission signal at anRF power of 20 W for the deposition of 0.5 ng ofPb ( — ) , 0.5 ng of Pb and 160 ng of Na as NaC1 ( - - - ),and 0.5 ng of Pb and 160 ng of Na as NaNO3 ( — — ). 1636.4. Interference effect of 160 ng of Na as NaC1 ( • ) and 160 ngof Na as NaNO3 ( • ) on the Pb emission intensity as afunction of RF power. 1696.5. Temporal response of the Pb emission signal at anRF power of 14W for 0.5 ng of Pb (^) and 0.5 ng ofPb and 160 ng of Na as NaC1 ( - - - ). .1706.6. Interference effect on the Pb emission intensity at an RFpower of 20 W as a function of amount of Na as NaCl.^ 1716.7. Temporal response of the Ag emission signal at anRF power of 20 W for 0.25 ng of Ag ( — ) , 0.25 ng of Agand 160 ng of Na as NaC1 ( - - - ), and 0.25 ng of Ag and160 ng of Na as NaNO3 ( • • • ). ^  ,173xivFIGURES PAGE6.8. Interference effect of 160 ng of Na as NaC1 ( • ) and 160 ngof Na as NaNO3 ( • ) on the Ag emission intensity as afunction of RF power. 1756.9. Interference effect on the Ag emission intensity at an RFpower of 20 W as a function of amount of Na as NaCl.  1766.10. Interference effect on the Ag emission intensity at anRF power of 20 W as a function of amount of Na as NaC1 in1 % HNO3 ( 0 ), and Na as NaNO3 ( • ). 1776.11. Temporal response of the Pb emission signal at anRF power of 20 W for 0.5 ng of Pb ( — ) , 0.5 ng of Pb in0.25 % ( w/v ) ascorbic acid ( - - - ), and 0.5 ng of Pb in1.5 % ( w/v ) ascorbic acid ( — — ). 1806.12. Interference effect on the Pb emission intensity at anRF power of 20 W as a function of concentration of ascorbicacid   1816.13. Temporal response of the Agdeposition of 0.25 ng of Ag (0.25 % ( w/v ) ascorbic acid ( - -1.5 % ( w/v ) ascorbic acid ( —20 Wemission signal for the— ) , 0.25 ng of Ag in- ), and 0.25 ng of Pb in—) at an RF power of^ 1836.14. Temporal response of the Pb emission signal for 0.5 ng ofPb ( — ) , 0.5 ng of Pb in 2.5 % ( v/v ) phosphoricacid ( — — — ), 2.5 % ( v/v ) phosphoric acid blank ( - - - ),and water blank ( • • • ) at an RF power of 20 W  184LIST OF ABBREVIATIONSAAS^atomic absorption spectrometryADC analog-to-digital converterAES^atomic emission spectrometryAPF-CCP atmospheric pressure furnace capacitively coupledplasmaCFAES^carbon furnace atomic emission spectrometryCRA carbon rod atomizerd. c.^direct currentDL detection limitEIE^easily ionizable elementFANES furnace atomization non-thermal excitationspectrometryFAPES^furnace atomization plasma excitationspectrometryGF graphite furnaceGFAAS^graphite furnace atomic absorption spectrometryHA hollow-anodeHC^hollow cathodeHCL hollow cathode lampHGA^heated graphite atomizerICC integrated contact cuvetteICP^inductively coupled plasmaICP-OES inductively coupled plasma optical emissionspectrometryIP^ionization potentialISM industrial, scientific, and medicalxviLTE^local thermal equilibriumLPDA linear photodiode arrayMS^mass spectrometricNRC national research councilPMT^photomultiplier tubeRF radio frequencyRSD^relative standard deviationRSDB relative standard deviation of the backgroundS^sensitivitySBR signal-to-background ratioSDB^standard deviation of the backgroundSNR signal-to-noise ratioSQRT^square rootSTPF stabilized temperature platformTE^thermal equilibriumv/v volume-to-volume ratiow/v^weight-to-volume ratioobi standard deviation of the backgroundACKNOWLEDGEMENTSI would like to express my sincere appreciation to my research supervisor,Dr. M. W. Blades, for his guidance and encouragement throughout the courseof this project.I wish to thank members of my guidance committee, Dr. W. R. Cullen,Dr. L. Burtnick and Dr. D. Dolphin for their valuable suggestions.I thank Dr. K. Orians for lending a sodium hollow cathode lamp, whichwas used in the interference studies. I would like to extend my thanks toMr. M. Vagg in the Mechanical Services Shop of the Chemistry Departmentfor constructing the FAPES source and for his prompt help. Thanks also goto Mr. M. Carlisle of the Electronics Shop for technical assistance.Finally, I would like to thank my wife for her patience and encouragementduring the past few years.1CHAPTER IINTRODUCTIONAdvancements in many areas of analytical chemistry are driven byenvironmental needs, made possible by technological development, and basedon progress in fundamental scientific understanding. Analytical atomicspectroscopy is one branch of analytical chemistry, wherein the analyticalspectroscopist develops, improves, and characterizes spectrochemical sources.The "characterization studies" in analytical atomic spectroscopy reflect theneed for solving problems in chemical measurement in atomic spectrometry.The primary objective of the work described in this thesis is to characterizethe radio frequency ( RF ) helium plasma source at 13.56 MHz in FURNACEATOMIZATION PLASMA EMISSION SPECTROMETRY ( FAPES ) as a spectrochemicalsource for elemental analysis. As compared with "The Seven Ages of anAnalytical Method" described in H. A. Laitinen's editorial in AnalyticalChemistry [1], this new spectrochemical method, FAPES, has undergone thefirst two ages, i.e., "conception" and "experimental measurement". The thirdand fourth stages described in this thesis are the "development ofinstrumentation" and "characterization". Yet to come are the fifth stage,"applications", the sixth stage, "establishment of procedures" and the seventhstage, "recognition as an accepted method".2The discussion that follows in this chapter is devoted to an introduction toFAPES and some special topics related to the work described in this thesis.The last section of this chapter presents the scope of the thesis.1.1 FAPES: HISTORICAL DEVELOPMENTAustralian physicist Alan Walsh's classic publication on Atomic AbsorptionSpectrometry ( AAS ), describing flame AAS [2], revolutionized atomicspectrometric methods. During the 1970s, flame AAS became the mostwidely employed spectrometric method for the determination of metallicelements. Even today, flame AA,S is a useful method for a variety ofanalyses. Although flame AAS has interferences, these interferences can beeasily controlled. But, refractory oxides, for example, those of B, V, Ta, andW, are only partially dissociated in the flame, and therefore, are difficult toanalyze by employing flame AAS. Since the introduction of the nitrousoxide-acetylene flame in 1966 [3], there have been no major advances in theflame method, which appears to have reached a plateau of development.Graphite Furnace Atomic Absorption Spectrometry ( GFAAS ), firstdescribed three decades ago by Boris L'vov [4], appears to be the mostsensitive atomic spectrometric method for the determination of trace metals.Unlike the flame AAS method, wherein Walsh employed continuousnebulization of the sample into the flame to provide a steady state absorptionsignal, L'vov converted a very small sample volume to an atomic vapor inside3a resistance-heated graphite furnace. This graphite furnace method yieldsthe best detection limits for absolute amounts and very few techniques cansuccessfully compete with those detection levels [5]. However, AAS is wellrecognized as a "single element at a time" technique(').In GFAAS, 1 to 100 III, of aqueous sample is deposited onto the innerfurnace wall, dried, and then vaporized into a furnace volume of less than1 cm3, to create a highly concentrated atom cloud. The analyte transportefficiency into the observation volume is close to 100 %, compared with1 to 5 % for other conventional sources ( flame or ICP ). Furthermore, theresidence times of analyte atoms in the observation volume are about twoorders of magnitude higher than those found in other sources. Consequently,analyte densities in the observation volume are several orders of magnitudehigher in the furnace methods, resulting in higher sensitivities, and hence,lower detection limits. The graphite furnace also facilitates matrixmodification in situ, because chemical and thermal pre-treatment of thesample can be incorporated into the heating program.Several kinds of plasma spectrometric methods have been employed inemission spectrometry [6], but the most prevalent is the Inductively CoupledPlasma Optical Emission Spectrometry ( ICP-OES ) described by(1) Some commercial instruments are now available with limited multi-elemental capability.Smith-Hieftje 8000 flame furnace AA ( Thermo Jarrel Ash, Franklin,Massachusetts, USA) is configured with eight hollow cathode lamps that permit theoperator to determine eight elements without changing hollow cathode lamps, and toperform determinations up to four at a time.4Greenfield et al. in 1964 [7], and Wendt and Fassel in 1965 [8]. Unlike theAAS method, emission spectrometric methods such as ICPs are inherentlymulti-elemental techniques. Inductively coupled plasma is also relativelyfree from interferences, and those refractory oxides only partially dissociatedin the flame are completely dissociated at high ICP temperatures. However,on a concentration basis, ICP detection limits are 10 to 100 times higher andon an absolute basis, 1000 times higher than GFAAS [5].Because of the low detection limits in GFAAS, the graphite furnace hasbeen employed as an atomization source for emission spectrometric methodswhich are capable of simultaneous, multi-elemental analysis( 2). Ottaway andShaw employed the high temperature produced during the atomization stepof the analyte in a graphite furnace to thermally excite the analyteatoms [10]. This method, Carbon Furnace Atomic Emission Spectrometry( CFAES ), is limited by the energy available for thermal excitation. Forelements having resonance wavelengths below 300 nm, the detection limitsare much poorer than those for GFAAS [10]. Furthermore, at temperaturesabove 2500 K, the intense emission from the furnace wall is a major spectralinterference in the visible region [11].(2) The graphite furnace is an atomization source, not only for emission spectrometricmethods but also for mass spectrometric ( MS) methods, such as GF-ICP-MS, withfemto gram level detection limits [9]. However, a discussion about such methods isbeyond the scope of this thesis.5graphite furnacehollow cathode anode —I> >- a na lyteemissiongraphite furnacepower supplyFigure 1.1. Schematic diagram of the FANES source.( Adapted from H. Falk, E. Hoffmann, andCh. Luclke, Spectrochim. Acta, 283, 398 ( 1984 ),with permission of Pergamon Journal Inc. )Falk et al. reported a low pressure, direct current ( d. c. ) glow dischargesustained inside a graphite furnace, with the furnace employed as a HollowCathode ( HC ), and a point or ring external to the furnace as theanode [12,13]. This spectrometric method has been designated FurnaceAtomization Non-thermal Excitation Spectrometry ( FANES ), orHC-FANES. Figure 1.1 depicts a schematic diagram of the FANES source.The furnace was a Massmann-type ( Section 1.3.2), and was graphite orpyrolytic-graphite coated. The sample was introduced by depositing a ALvolume onto the inner wall of the graphite furnace, and vaporized into thedischarge during the high temperature atomization step of the analyte. The6analyte atoms are excited while inside the discharge and emit characteristicline spectra that can be used for simultaneous multi-elemental analysis.Detection limits of the FANES determinations are comparable to those ofGFAAS. Because of the low pressure requirement of the glow dischargeoperation, sample introduction is somewhat laborious and time consuming.Ballou et al. also employed a d. c. glow discharge [14], which wasconceptually similar to the FANES source. Figure 1.2 illustrates a schematicdiagram of the HA-FANES source. In this source, the graphite furnace wasemployed as a Hollow Anode ( HA), and a graphite rod, which was orientedco-axially with the furnace and extended the entire furnace length, acted asthe cathode. The graphite furnace was an integrated contact cuvette( Section 1.3.3.2 ) and the axial electrode was a pyrolytic-graphite coated rod.Analytical figures of merit for HA-PANES are not much known. Theoperation of this source is similar to that of FANES and requires lowpressures.Liang and Blades first reported an atmospheric pressure plasma sourceinside the graphite furnace for analyte excitation [15]. This plasma sourcehad the same geometry as that of HA-FANES, but was operated with anRF power source. The spectrometric method with such an RF plasma sourcehas been designated Furnace Atomization Plasma Excitation Spectrometry( FAPES ) [16,17]. Compared with the low pressure operation, theatmospheric pressure operation was expected to offer convenient sampleintroduction and increased residence time of analyte atoms within theintegrated contactcuvette: hollow anodecathode7to d. c. power supplyFigure 1.2. Schematic diagram of the HA-FANES source.( Adapted from N. E. Ballou, D. L. Styris, andJ. M. Harnly, J. Anal. At. Spectrom., 1141, 3( 1988 ), with permission of The Royal Society ofChemistry. )graphite furnace [15]. A more detailed description of FAPES is presented inthe next section.As stated previously, the development of new spectrochemical methodsstem from the need for determining ultra-trace concentrations of analytes invarious samples. During the last two to three decades, improvements indetection limits, use of small sample volumes, and fast analysis have beenachieved [18]. Detection limits have gone from microgram to sub-picogramlevels or even single atom detection, and sample size has gone from millilitersto microliter volumes [19]. With the variety of analytical methods available8today, no method is really rendered obsolete by another. There is no "best"method; the choice depends on one's skills and on the nature of the chemicalmeasurement.The historical development [20,21] and the general status [5,22-24] ofGFAAS has been reviewed elsewhere. Comprehensive reviews of thehistorical development of various plasma sources [6], development ofICP [25], graphite furnaces for sample introduction in plasma sourcespectrometry [26], GF-ICP-OES methods [27], and ICP-mass spectrometricmethods [28-30] are available. Recently, a graphite furnace has also beenemployed as a sample introduction method for d. c. arc plasmas [31-33].1.2 FURNACE ATOMIZATION PLASMA EXCITATION SPECTROMETRYFigure 1.3 illustrates a schematic diagram of the RF plasma source describedby Liang and Blades [15]. This plasma source consists of a conventionalMassmann-type graphite furnace work-head ( Instrumentation Laboratory,model IL 455) and a co-axial thoriated-tungsten RF electrode. The rearoptical window ( through which light from the hollow cathode lamp wasnormally directed ) of the furnace work-head was removed and replaced withthe RF electrode. The operating frequency was 27 MHz and the RF powerdelivered to the plasma was about 20 W. Helium was used as the plasma gasof this atmospheric pressure RF plasma. Liang and Blades suggested thatthe mode of power coupling to the plasma was primarily capacitive in(a) (a)=^to RF powersupply<I-- RF electrode9graphite furnaceto furnace powersupply (a)Figure 1.3. Schematic diagram of the FAPES source.( Adapted from D. C. Liang and M. W. Blades,Spectrochim. Acta, 1049, 44B ( 1989 ), withpermission of Pergamon Journal Inc. )nature [15]. Liang and Blades designated this plasma source as AtmosphericPressure Furnace Capacitively Coupled Plasma ( APF-CCP ). However, thisplasma source is now recognized as a FAPES source in the literature [34].The analyte sample was a small brass chip about 5 mg. Some emissionspectra of Cu and Zn were recorded when the furnace was heated toabout 800 °C. Liang and Blades suggested that at low RF powers, thedominant sample introduction mechanism to the plasma wasRF sputtering [15]. A relatively high gas temperature and, as a consequence,10a reduction in gas-phase chemical interferences compared with GFAAS, wereexpected within the graphite furnace with the plasma.In a follow up study, Smith et al. determined the analytical figures ofmerit for Ag [17]. For Ag, the absolute detection limit was 0.3 pg and thelinear dynamic range was 2-orders of magnitude. Smith et al. also reported adecrease of 20 % and 50 % in the Ag emission intensity when 2.3 j.ig of Na ispresent as NaC1 and NaNO3, respectively.Sturgeon et al. employed a Perkin-Elmer furnace work-head ( modelHGA-2200 ) with a co-axial graphite electrode [16]. The power delivered tothe helium plasma was about 50 to 70 W. A 10 AL aliquot of test solution, Cdor Mn, was deposited on the furnace wall and subjected to the atomizationstep as in GFAAS. The plasma background emission and transient emissionsignals for Cd and Mn were given. The detection limits for Cd and Mn were36 pg and 52 pg, respectively.In the next section, the analyte atomization within the graphite furnace isdiscussed. The various graphite furnace designs along with the role of thegraphite furnace for analyte atomization and vaporization are presented.111.3 ANALYTE ATOMIZATION: THE GRAPHITE FURNACEThe graphite furnace is a resistance-heated cylindrical device. Liquidsamples can be deposited onto the inner furnace wall by means of amicro-pipette through a hole in the furnace. To prevent the oxidation ofgraphite at high temperatures, the furnace is coated with a thin layer ofpyrolytic-graphite and purged with an inert gas.The sample is placed on the furnace wall ( wall atomization ) or on aseparate device inserted into the furnace ( platform atomization,Section 1.3.3.1). In the case of platform atomization, the platform isprimarily heated by the radiation from the furnace wall. The sample on thefurnace wall ( or on the platform ) can be subjected to thermal treatment.Generally, there are two thermal treatment stages: drying and ashing.During the drying stage, solvent is evaporated. During the ashing stage, anyorganic residues in the sample are removed by decomposition andvaporization. The next stage is the vaporization and atomization ( or simply:atomization ) stage, during which the analyte and the remainder of thesample form a vapor cloud inside the furnace. During the atomization step,the temperature of the furnace ( atomization temperature ) can reach amaximum as high as 3200 K.In the next section, some graphite furnace designs are presented. A moredetailed discussion about graphite furnaces in AAS is availableelsewhere [35-39].121.3.1 L'vov FurnaceThe first successful non-flame atomizer in the form of a carbon-rod electrodeand a graphite furnace was described by L'vov [4,40]. The L'vov furnace hasnow been widely described in the GFAAS literature, and is depicted inFigure 1.4. The graphite furnace was 30 to 50 mm long, with an innerdiameter of 2.5 to 5 mm The furnace was heated up to about 2700 K.The sample was introduced into the furnace on a carbon-rod electrode of6 mm diameter. The head of the electrode was shaped to fit the orifice in thewall of the graphite furnace. The graphite furnace was heated for 20 to 30 sand the electrode was moved into the orifice of the graphite furnace. Theauxiliary electrical heating of the electrode was turned on for 2 to 3 s, and theatomic absorption was measured. The electrode was then lowered away fromthe graphite furnace, and the system was ready for the introduction ofanother electrode carrying a sample. Because the atomic vapor can penetratethe walls of the furnace, the inner wall was lined with a thin layer ofpyrolytic-graphite, Ta or W foil.In the original form of the L'vov furnace, vaporization of the sample wasachieved by a d. c. arc formed between the carbon-rod electrode and anauxiliary electrode mounted under the furnace [4]. To eliminate the furnacebackground, light from the hollow cathode lamp was modulated. L'vov alsoemployed two-channels that permit simultaneous recording of absorption fortwo elements, one of which can be an internal standard [40].13graphite furnace/ \light pathI (a) ?water cooled ends (a)carbon—rod electrodeFigure 1.4. Schematic diagram of the L'vov graphitefurnace and electrode assembly. ( Adapted fromB. V. L'vov, Spectrochim. Acta, 53, 2413 ( 1969 ),with permission of Pergamon Journal Inc. )1.3.2 Massmann FurnaceThe Massmann furnace [41], a simplified version of the L'vov graphitefurnace, is depicted in Figure 1.5. The Massmann furnace was 55 mm long,6.5 mm in internal diameter, and had a wall thickness of 1.5 mm. Theelectrical contacts were made at both ends of the furnace. Unlike the L'vovmethod, liquid samples were deposited onto the inner furnace wall through asmall hole in the center of the furnace by means of a micro-pipette, whilesolid samples were inserted from one side of the furnace. Sample volumes of5 to 200 I.LL were used. The temperature of the furnace reached 2900 K14Figure 1.5. Schematic diagram of the Massmann graphitefurnace. ( Adapted from H. Massmann,Spectrochim. Acta, 215, 23B ( 1968 ), withpermission of Pergamon Journal Inc. )within a few seconds. The furnace was enclosed in a chamber and purgedwith Ar.The Massmann graphite furnace has been further developed forcommercial AAS instruments. Almost all the research work carried out inGFAAS during the last three decades employed two commercial graphitefurnaces, namely, Perkin-Elmer model, Heated Graphite Atomizer ( HGA )and Varian Techtron model, Carbon Rod Atomizer ( CRA ). The HGA issimilar to the Massmann-type furnace. The CRA is a 1.5 mm diameter rodwith a transverse hole of 1.5 mm in diameter. The CRA is supported by two15graphite electrodes, which are mounted on water cooled terminal blocks andconnected to the furnace power supply. Recent models of the Varianspectrometers contain Massmann-type furnaces. A more detailed descriptionof these commercial graphite furnaces is available in reference [39].1.3.3 Limitations of Massmann -type FurnacesSome fundamental limitations are inherent in the Massmann-type furnaces.Temporal non-isothermality and spatial non-isothermality are two of themajor limitations [42]. Temporal non-isothermality occurs when analyteatoms appear in the observation volume during a time in which thetemperature of the gas-phase is low and is rapidly changing to theatomization temperature. The temperature range within which analyteatoms persist in the furnace, and hence, the residence time of analyte atoms,is dependent on the nature of the analyte and the accompanying matrix [43].As a result, the degree of atomization is often low and matrixdependent [43,44]. In addition, temperature dependent kinetic effects on theatomization of the analyte cannot be eliminated [43].Spatial non-isothermality is the non-uniform temperature distributionalong the furnace length and is caused by the heating characteristics of thewater cooled ends of the furnace. This non-uniformity results in vaporcondensation on the cooler end-regions of the furnace, causing memoryeffects [43]. In addition, recombination of the sample vapor leaving thefurnace through the cooler end-regions is a major contribution to spectral16interferences: molecular absorption and light scattering, in AAS [40]. Spatialnon-isothermality of the furnace also has a severe effect in CFAES due toself-absorption [45].1.3.3.1 L'vov PlatformL'vov first employed a graphite platform installed inside the furnace, fromwhich the sample was vaporized rather than from the furnace wall [43]. Thegraphite platform is sometimes called the "L'vov platform" or the "StabilizedTemperature Platform" ( STPF ) [44,46]. A schematic diagram of the L'vovplatform is provided in Figure 1.6.The platform slides into the furnace operated in the usual manner, exceptlonger thermal treatment times are necessary. The platform is heatedprimarily by furnace radiation during the heating cycle of the furnace, andhence, the temperature of the platform lags compared with that of thefurnace wall. Therefore, analyte vaporization and atomization are delayeduntil the gas-phase temperature within the furnace reaches the atomizationtemperature. In addition, due to the temperature difference between thefurnace wall and the platform, a higher heating rate for the platform isachieved in the initial stages of the atomization step [43].A more detailed discussion on theoretical and experimental studies on theplatforms for GFAAS has been published by Wu et al. [47].17L'vov platformgraphite furnaceFigure 1.6. Schematic representation of the L'vov platforminside the graphite furnace.1.3.3.2 Integrated Contact CuvetteWhen a L'vov platform is used with conventional end-heated Massmann-typefurnaces, temporal isothermality can be achieved. However, spatialnon-isothermality of the furnace is not overcome. Spatial non-isothermalitycan be eliminated by employing a side-heated Integrated Contact Cuvette( ICC ) [48]. In the ICC, the full length of the furnace starts to heat at thesame time ( transverse heating ), and therefore, achieves spatialisothermality. It should be noted here that both the ICC and a platformshould be employed to achieve spatial and temporal isothermality. The ICC18is the furnace type used for most of the work described in this thesis. Aschematic diagram of the ICC is available in Figure 1.2.1.3.4 Atomization MechanismsAtomization mechanisms are dependent on the nature of the analyte and theaccompanying matrix, and are very complex in nature. Many studies onatomization mechanisms have been reported in the literature and variousproposed mechanisms have been subjected to considerable debate over theyears. Reference to these studies will be given, when appropriate, inChapter 4 of this thesis. For an introduction to atomization mechanisms, thefollowing discussion is warranted.When the analyte sample is in a nitric acid solution, oxide of the analyte isformed from the nitrate during the ashing step or the atomization step.Mechanism I. Reduction of Oxide by CarbonMO(8)Carbon^>ReductionM(B,1)CO(g)Vaporization>^M(g) In this process, analyte oxide is reduced by carbon in the furnace wall toform analyte atoms either in solid or liquid form, which are then vaporized tothe gas-phase.19Mechanism II. Thermal DissociationMO(s)ThermalM(s,1)DissociationVaporizationM(g) 1/202In this mechanism, analyte oxide thermally dissociates on the furnacewall. The thermal dissociation follows the gas-phase appearance of theanalyte atoms.Mechanism III. Dissociation of Oxide VaporMO(s)Vaporization^> MO(g)Thermal+ 1/202DissociationAccording to this mechanism, analyte oxide vaporizes from the furnacewall before dissociating in the gas-phase. The gas-phase dissociationequilibrium of the analyte may be affected by the amount of oxygen which, inturn, is determined by the amount of CO in the gas-phase ( Sections 4.3.2 and6.3.5).Mechanism W. Dissociation of Halide VaporVaporization^ Thermal^>^MX(g)^ M(g)^X(g)Dissociation20Dissociation ( or formation ) of analyte halides in the gas-phase isimportant, especially when a chloride matrix is present in the sample.The next section describes the analyte excitation in the plasma. Somecharacteristics of RF discharges at atmospheric pressure along with theexcitation mechanisms of the analyte will be presented.1.4 ANALYTE EXCITATION: THE PLASMAUnlike low pressure RF gas discharges, atmospheric pressure capacitivelycoupled gas discharges appear to find no applications in plasma processingtechniques. Therefore, these RF plasmas are rarely reported in theliterature. However, the high frequency discharges at atmospheric pressurehave been studied in a few occasions as discharges in the circuit breakers inhigh frequency power generators [49,50]. Some fundamental characteristicsof RF ( mainly 1 - 25 MHz) gas discharges at atmospheric pressure havebeen reported [51-54].Recently, RF discharges at low pressure have been employed in massspectrometry [55], and emission spectrometry [56]. Those interested in adetailed discussion on low pressure discharges ( both d. c. and RF ) may referto Chapman [57]. Only a brief description leading to the nature of theRF discharge at atmospheric pressure is given in the next section.211.4.1 The RF Discharge at Atmospheric PressureFor the simplicity of the discussion, first consider two electrodes of equalarea, at a certain distance apart at low pressure. When a sufficient d. c.voltage is applied, a discharge strikes between the electrodes. In thisdischarge, a cathode dark space and a glow space can be seen. A cathode fallof potential develops across the dark space, leaving the glow space nearlyfield free. For self-sustainment ( current continuity ) of the discharge, asteady state electron concentration must be maintained. In a d. c. glowdischarge, this electron concentration is mainly caused by secondary emissionthrough positive ion bombardment on the cathode.If, instead of a d. c. voltage, a low frequency alternating voltage is applied,it can be observed that the discharge behaves as though it has twoalternating cathodes. This system is a succession of short-lived d. c.discharges, because at low frequencies, there is ample time for the dischargeto become fully extinguished. The discharge is extinguished when thecathode potential drops below the discharge sustain value because of thebuild up of a self-bias d. c. potential on the cathode.The nature of the d. c. potential on the cathode of d. c. and alternatingcurrent discharges should be noted. In the case of a d. c. discharge, thepotential at the cathode is equal to the applied potential difference betweenthe two electrodes. In contrast, in an alternating current discharge, the d. c.potential at the electrodes is a self-bias voltage. This self-bias voltage arisesbecause the electrons have much higher mobility than the heavier ions and so22are easily collected on an electrode whenever it becomes positive with respectto the glow space [58].If the frequency of the applied voltage is increased, it is observed that theminimum pressure at which the discharge operates is reduced [59]. Thisreduction indicates that there is an additional source of ionization other thanthe secondary electron emission from the electrodes. This additional sourceof electrons arises when electrons, oscillating in the time dependent electricfield, undergo collisions with the plasma gas atoms to cause ionization.Therefore, the high voltage electrode that is necessary in a d. c. glowdischarge for the secondary electron emission is not required to sustain theRF discharge [60]. Furthermore, the cathode glow attached to each electrodeis the same as in the d. c. case. The human eye cannot follow quick intensitychanges and "sees" that the cathode glow is attached to both the electrodes.However, the cathode glow changes its position and is attached only to theelectrode that acts as the cathode in any particular half-cycle [52].In addition to the frequency of the discharge, the pressure is also animportant parameter that affects the discharge characteristics. The pressureaffects the discharge characteristics in two ways. Firstly, in a low pressuredischarge, the mean free path of the electrons and ions is high. The electricfield in the cathode dark space causes the acceleration of positive ionsthrough the dark space toward the cathode. These accelerated positive ionsimpinge on the cathode, and cause the sputtering of the electrode materialand the emission of secondary electrons. However, at atmospheric pressure,23the mean free path of the ions is short, and therefore, ions can be consideredas stationary. Secondly, the high pressure discharge is essentially a lowcurrent glow discharge. Transition from glow to arc discharge can beproduced by an increase in current under constant pressure ( or by anincrease in pressure at constant current ), and a considerable fall in thedischarge voltage, which is associated with a change in the electron emissionmechanism from the cathode [61]. In the case of arc, electron emission fromthe cathode is mainly thermionic and field emission.In atmospheric pressure discharges, a positive space-charge region is builtup in front of the appropriate electrode in each negative half-cycle. Atfrequencies above 1 MHz, the length of this space-charge region does notexceed 5 x 10-3 cm ( in air ), neither does the mean free path of an ion [51].Voltage of the RF discharge depends on the nature and the distance betweenthe electrodes, and is 300 V or higher, and carries less than 1 A ofcurrent [54]. The time scale of the application of the RF voltage is such thatfrequencies in the order of 1 MHz and above resulting in a pseudo-continuousplasma. The reignition voltage in each half cycle is dependent on theelectrode distance. It has also been shown that reignition voltage drops at acertain electrode distance [53]. This observed drop in the reignition voltagewas attributed to the residual charge carriers at higher electrode distancesand was not due to the space charge effect [53].If the current density increases, an RF arc discharge with a brighterappearance than a glow can be observed [54]. An arc does not usuallyplasma source24matching--C>networkRF powersupply-Figure 1.7. Schematic diagram of an RF discharge system.maintain a stable position, but moves around on the electrode surface. TheRF glow can turn temporary into an arc at any time. The transition to thearc is also favoured by conditions that facilitate electron emission, such asrough electrode surface and salt deposits [54].Figure 1.7 depicts the main components of an RF discharge system. Itconsists of an RF power supply, a matching network, and the discharge. TheRF power supply can be operated at the frequencies permitted by the UnitedStates Federal Communications Commission for industrial, scientific, andmedical ( ISM ) equipments [62]. The purpose of the matching network is to25maximize the RF power coupled to the plasma and to protect the RF powersupply.1.4.2 The RF Discharge in FAPESThe atmospheric pressure RF plasma source within the graphite furnace hasa bright region surrounding the RF electrode and a less intense plasma fillsthe remainder of the graphite furnace volume It is also observed that whenthe RF power coupled to the plasma is increased, the bright regionsurrounding the electrode extends along the RF electrode beyond the lengthof graphite furnace. The appearance of the extended RF glow along theRF electrode marks the onset of arcing between the RF electrode and thefurnace wall. This RF discharge inside the graphite furnace is not yetwell-understood. The voltage-current characteristics are not known and amathematical model describing the discharge has not been developed.The atomic spectrometric utility of the RF discharge at atmosphericpressure should be noted. Firstly, the gas discharge inside the graphitefurnace has been used only for analyte excitation process. Analyteintroduction into the excitation volume is achieved during the hightemperature atomization step of the analyte. Secondly, the dischargecontains unequal electrode areas causing different current densities andelectric field strengths on each electrode during each half-cycle. Therefore,this plasma source is not radially symmetrical along the furnace length.Finally, the discharge contains hot electrodes with varying temperatures up26to about 3000 K depending on the experimental conditions. However,depending on the RE' power and the furnace temperature, the RF plasmaappears to change into an RF arc with the evolution of the thermionicelectrons.1.4.3 Excitation MechanismsAlthough it is beyond the scope of this thesis to go into details of variousexcitation mechanisms in plasmas, a brief discussion is warranted. A moredetailed discussion about excitation mechanisms is availableelsewhere [63,64].In the previous section, the term "gas discharge" is widely used. Gasdischarge is a general term used to describe both glow discharges andplasmas. The glow discharge is the generic name for an extensive class of gasdischarges with a cathode fall and a space charge. On the other hand, aplasma can be simply described as a partially ionized gas with an equalnumber of positive and negative charges. In such a plasma, the principalspecies present are neutral atoms, ions, and unbound or free electrons.The collisional processes are assumed to be the most importantmechanisms through which analyte excitation and ionization occur.Collisional processes can be of the elastic or inelastic type depending onwhether the internal energies of the colliding particles are conserved.Particles usually have two types of energy: kinetic and potential energy. In27an elastic collision, there is interchange of kinetic energy only. In aninelastic collision, there is no such restriction and potential energies can alsochange. Therefore, kinetic energy of one particle may be lost to excite orionize the second particle.The following discussion is limited to the inelastic collisions ( excitation )and radiative processes ( de-excitation: characteristic line spectra ) involvingthe analyte atomic species. The relevant species considered here are, M;ground state analyte atom ( or ion ), M*; excited state analyte atom ( or ion ),He; ground state plasma gas, He*; excited state plasma gas, He*,m, excitedmetastable state, and e; free electron. The following are the importantcollisional radiative processes.Mechanism I. Electron Impact ExcitationM + e ^>^M* + eIn the electron impact excitation, a fast moving electron collides with ananalyte atom ( or ion ) in a lower energy state and the kinetic energy of theelectron is taken up to excite the analyte atom ( or ion ). Excess kineticenergy is carried away by the primary electron. This process is assumed tobe the most important mechanism for the excitation of analyte in theplasmas. The reverse process is called collisional de-excitation.28Mechanism II. Collisional ionization+ e  ^M+^+ e + eIn this process, an atom is ionized by impact with a fast moving electron.The kinetic energy of the electron must be at least as large as the ionizationpotential of the atom ( or greater than the binding energy of the ionizingelectron in a particular excited state ). The reverse process is calledthree-body recombination.Mechanism III. Penning ExcitationM + He*,ni M* + HePenning excitation ( or ionization ) of the analyte is caused by collisionwith metastable helium atoms. For such an energy transfer, resonancecondition applies. In addition, in an atmospheric pressure plasma, the lifetime of the metastable species is very short. Therefore, this mechanism isassumed to be insignificant in atmospheric pressure plasmas.Mechanism IV. Radiative RecombinationAti*^+^hucont,29In this case, M is the analyte ion and M* is an analyte atom in the excitedstate. The energy of recombination ( hi,) produced in this reaction has acontinuous range, because the colliding electrons posses a range of energies.The recombination of helium ions with electrons produces the continuumbackground emission.Mechanism V. Radiative De-excitationM*In the radiative de-excitation, the excited analyte atom ( or ion ) relaxes toa lower state by emitting a quantum of radiation energy. Radiativede-excitation is the most important mechanism for emission spectrometry,because it creates characteristic line spectra. The reverse process is calledabsorption.1.4.4 Plasma TemperatureThe temperature of the plasma is an important consideration in atomicspectrometry. Relatively high temperature plasma sources such as ICPs giverise to intense line emission for the analyte and low matrix interferenceeffects in analytical determinations. As such, high temperature plasmasources are considered to possess superior analytical merits compared withthose of low temperature plasmas.30A unique temperature for a system can be specified only if the system is ina state of thermal equilibrium However, for laboratory plasmas,temperature described by Planck distribution is not fulfilled. Furthermore,partially ionized gases like plasmas consist of heavy particles such as atomsand ions, as well as electrons. For such a system, one temperature can beconsidered when the heavy particles and electrons have the same kineticenergy. For most laboratory plasmas, existence of such an energyequilibrium is rare.If thermal equilibrium is not established, a single temperature can not beassigned to the plasma. This non-thermal equilibrium leads to severaldifferent definitions of temperatures in the plasma depending on the speciesconsidered. These are electron kinetic temperature ( Te, from the Maxwellvelocity distribution ), gas kinetic temperature ( Tg, from the Maxwellvelocity distribution ), excitation temperature ( Texe, from the Boltzmannenergy state distribution ) and ionization temperature ( Tim, from the Sahaequation ). However, when the source is in a state of local thermalequilibrium ( LTE ), a unique temperature can be defined for each point inthe source but allowing for the possibility of different temperatures atdifferent points. A more comprehensive description of plasma temperaturesand the relevant distribution functions are available elsewhere [65,66].311.4.5 Spectroscopic Temperature MeasurementsSpectroscopic methods are popular for the measurement of plasmatemperatures, especially in high temperature plasmas, because they arenon-invasive. Measured temperatures can be used to compare differentplasma sources with respect to their ability to atomize, ionize, and exciteanalyte species. The measurement of the excitation temperature using theBoltzmann distribution is described below.The emission intensity of a spectral transition from a higher energylevel ( p ) to a lower energy level ( q) can be expressed by:Pq^1147C ) X Apq X Np x ( hupq )^(1.1),where IPq ( erg cm-2 sr-1 r1 ) is the absolute intensity of the emission line,APq ( -1) is the Einstein transition probability, Np ( cm-3 ) is the numberdensity of atoms in the upper energy level p, L ( cm ) is the source length,h ( 6.6262 x 10-27 erg s) is the Planck constant, and voq ( s-1 ) is thefrequency ( Xpq is the wavelength ) of the spectral line emitted [66].The number density of particles in excited level p ( No ) is given by theBoltzmann distribution:=^( Ngp/Q(T) ) exp( -Ep/kTexe )^(1.2),32where N ( cm-3 ) is the total atom ( or ion ) density, gp is the degeneracy ofthe upper level p, Q(T) is the partition function, Ep ( erg ) is the energy of theupper energy level p, k is the Boltzmann constant ( 1.3805 x 10-16 erg K-1 ),and Texe ( K ) is the excitation energy. One can combine equations (1.1) and(1.2) to obtain:ln( ipqXpq/gpApq )^=^ln( kN)-Ep1( kTexe )^(1.3),where k is a constant. Therefore, the excitation temperature, Texe, can bedetermined by measuring the relative intensity of each spectral line andplotting ln( IpqApq/gpApq ) as a function of E. The slope of the line( - 1/kTexe ) then provides the Texe. Alternatively, Apq may be replacedby fqp ( the oscillator strength ), and ln( IpqXpq3/gqfqp ) is plotted as afunction of E. In this case, gq is the degeneracy of the lower level q of thetransition. This spectroscopic method of temperature determination isnormally called the slope method.For spectroscopic temperature measurements, the relative emissionintensity measurements must not suffer from self-absorption and should beproportional to the absolute emission intensity. It should be noted that thevalue of ln( k N) will vary during the atomization step of the analyte,because the total atom density in the furnace varies, but will be a constantfor all lines of an element at any specific instance.33Alternatively, for two upper energy levels, p and q, one can combineequations (1.1) and (1.2), and rearrange to:ln( Ip/lq )^=^ln[( gpAp ) / ( gqAq )] - ( Ep-Eq )/kTexe^(1.4).1p^'togTherefore, by measuring the relative intensities of two lines at differentenergies, it is possible to calculate the excitation temperature as a function oftime during the atomization step. This method of temperature determinationis called the two line method.The following considerations should be noted for spectroscopic temperaturemeasurements. As stated previously, a unique temperature can only be usedto describe the plasma if the plasma is in thermal equilibrium. With anon-thermal equilibrium, a variety of temperatures can be used to describethe plasma, depending on the statistical distribution and the thermometricspecies used for the measurement [67,68]. The accuracy of spectroscopicallymeasured temperatures is dependent on the accuracy of the transitionprobabilities [67]. Furthermore, if the plasma is not in LTE, theexperimental temperature is applicable only to the energy levels of thethermometric species used to drive it [69].341.5 OVERVIEW OF THE THESISFurnace Atomization plasma excitation spectrometry ( FAPES ) is a newspectrometric method, and hence, the work described in this thesis is focusedon the characterization of the RF plasma in FAPES as a spectrochemicalsource for elemental analysis. An effort has been made to distinguish theeffects of the plasma on the analyte from those of graphite furnace withoutthe plasma. The scope of the thesis is presented below.A new experimental system, developed for the measurement of temporallyand spatially resolved transient signals, is described in Chapter 2. Thisexperimental system can be employed to acquire two data channelssimultaneously. A complete software package is written to acquire andprocess the transient signals obtained from the FAPES source.Spectral, spatial, and temporal emission characteristics of the RF plasmasource are studied. To this end, time averaged Fe-excitation temperatures ofthis new plasma source are measured. Both atomic absorption and atomicemission are measured to study the effect of plasma on the temporal responseof the analyte emission signal. Analyte atomization, vaporization andexcitation characteristics are discussed. Moreover, the temporal behavior ofthe analyte emission signal is discussed with respect to physical andchemical phenomena. This work is described in Chapters 3 and 4 of thisthesis.35Chapter 5 deals with the effect of RF power on the Pb emissioncharacteristics and the plasma background emission. The time-resolvedPb-excitation temperature is measured to study the effect of the Pb-excitationprocess on the observed signal. Factors that affect the analytical figures ofmerit for a FAPES determination are also discussed.Sodium chloride and NaNO3 matrix interference effects in FAPES areinvestigated with the aim of understanding interference mechanisms and theeffect of the RF plasma on the interference effects. The temporal behavior ofthe analyte and the interferent in the gas-phase is studied. Similarities anddifferences between NaC1 and NaNO3 interference effects on Ag and Pbemission intensities are also presented in Chapter 6.36CHAPTER 2THE EXPERIMENTAL SYSTEMAs stated in the introduction, one of the objectives of this thesis was to setupan experimental system to characterize the atmospheric pressure heliumplasma source at 13.56 MHz in FAPES. This chapter describes theexperimental system developed for the characterization of this RF plasmasource in FAPES. A description of this experimental system is also availableelsewhere [70,71].2.1 INSTRUMENTATIONFigure 2.1 presents a schematic block diagram of the experimental systememployed for the acquisition of spatial and time-resolved signals. The maincomponents of this experimental system were: an RF power supply, anRF matching network and the plasma source work-head for igniting andsustaining the plasma, and a furnace power supply for resistance-heating thegraphite furnace. In addition, two lenses (Li and L2 ), a monochromator,and a photomultiplier tube ( PMT) were employed for the spectral isolationand detection, along with a current amplifier, an analog-to-digitalconverter ( ADC ), and a computer ( PC/AT ) for the data acquisition.ADCPC / ATCurrent_AmpLock_in_AmpHCLPower SupplyImpedanceMatching NetworkPlasmaWork—HFurnacePower SupplyL1RF Power supplyourceadL2MonochromatorPMTTRG^11_ ^CH1 7* VCH2Figure 2.1 Schematic diagram of the experimental system.38The ADC data channels are labelled as CH1 and CH2 in Figure 2.1. Whenthe atomic absorption signal was measured simultaneously with the atomicemission signal, a lock-in-amplifier was employed to detect the hollowcathode lamp ( HCL ) signal from the PMT output as depicted in Figure 2.1.For other experiments, CH2 was directly connected to the output from thecorresponding signal transducer ( for example, optical pyrometer ) to acquirethe data. The trigger channel for the ADC is labelled as TRG in Figure 2.1.The sections that follow in this chapter, describe each of these componentsalong with data acquisition, data processing, and the FAPES method.2.1.1 The Plasma Source Work-HeadFigure 2.2. provides a schematic diagram of the plasma source work-head.The main components of this plasma source work-head were: apyrolytic-graphite coated integrated contact cuvette ( ICC ) and apyrolytic-graphite coated co-axial rod ( Ringsdoff-Werke, BadGodesberg,Germany ); and a high current furnace support structure, machined fromcopper and Macon, contained in a five-way hollow cube machined fromaluminium The ICC was 19 mm long, 57 mm in internal diameter, and7.1 ram in outer diameter. The co-axial rod was 09 mm in diameter, andextended the full length of the graphite furnace. The hollow five-way cubewas 6 x 6 x 6 inch with 5 inch diameter ports in five of the six sides. Each ofthese ports was fitted with a 5.5 inch diameter "0"-ring sealed aluminiumflange. This five-way cube could be evacuated to about 1 mtorr, if necessary.39Figure 2.2. Schematic diagram of the plasma sourcework-head.The furnace support structure and the water cooling system of thegraphite furnace were designed similar to the method described byBallou et al. [14]. The graphite furnace was resistance-heated by employing afurnace power supply ( Model IL-555, formerly Instrumentation Laboratory;now Thermo-Jarrell Ash, Waltham, MA, USA), unless otherwise noted. Theplasma gas entered the plasma source work-head from a small inlet on theflange with the furnace support structure. The plasma gas outlet was locatedon the flange opposite to the flange with the furnace support structure.40Thus, the plasma gas had no directed flow through the graphite furnace. Forall experiments described in this thesis, a gas flow rate of 6 SCFH ( the gasflow rate reading of the furnace power supply in units of standard cubic feetper hour of air ) was used. Silver atomic absorption signals measured as afunction of gas flow rate, from 4 to 20 SCFH, showed no significant effect oneither peak height or peak area measurements.The co-axial graphite rod was powered through an RF connector and twovertical metal blades, in the manner depicted in Figure 2.2, unless otherwisenoted. This arrangement for the co-axial rod ( as depicted in Figure 2.2 ), forthe simultaneous measurement of absorption and emission signals, is calledthe "vertical-mount ". The alternative arrangement, when the co-axial rod isdirectly connected to the RF connector, is called the "horizontal-mount ".The end view of these two arrangements is depicted in Figure 2.3. In thisthesis, the RF powered co-axial rod is called the " RF electrode ", unlessotherwise noted.The plasma was viewed through a 1 inch diameter quartz window on thefront-side flange. A similar quartz window on the opposing backside alloweda light source to be directed through the graphite furnace so that an atomicabsorption experiment can be carried out or the horizontal-mount can beinstalled instead of the quartz window. The analyte sample was depositedonto the inner furnace wall ( or onto the RF electrode ) through a smallscrew-top port mounted on the top flange. The plasma source work-head was41RF connector --{>- r'imetal blades --->RF electrode(a )^(b)Figure 2.3. End view of the (a) vertical mount and(b) horizontal mount. The dash line representsthe RF connector.constructed in the mechanical services shop of the chemistry department atUniversity of British Columbia.2.1.2 The Atmospheric Pressure RF DischargeA helium gas discharge was sustained inside the graphite furnace byemploying a 13.56 MHz RF generator ( Model RFX-600, Advanced Energy,Fort Collins, CO, USA), an automatic power tuner ( Model ATX-600,42Advanced Energy, Fort Collins, CO, USA), and an impedance matchingnetwork ( Model 5017-000-G, Advanced Energy, Fort Collins, CO, USA). Theoutput of the impedance matching network was coupled to the RF electrodethrough a variable 1 - 101.ill inductor. This arrangement could be used toignite and sustain a helium discharge over the RF power range between12 and 25 W. At RF powers between 25 and 50 W, occasional arcing wasobserved between the RF electrode and the furnace wall. Althoughconsistent arcing was observed, a gas discharge could be maintained atRF powers between 50 and 150 W.When operating at RF powers below 75 W with an atomizationtemperature of about 2100 K, the reflected power was stable below 1 - 2 Wthroughout the atomization step. At RF powers higher than 75 W, thereflected power could be maintained below 1 - 2 W, except during theatomization step. At an RF power of 100 W, after about 3.5 s into theatomization step, the reflected power started to increase to a value as high asthe applied RF power. Furthermore, under these conditions, the reflectedpower would stay at these high values, even after the completion of theatomization step. Thus, temperature of 2100 K sets the upper limit of theatomization temperature that can be used with this experimental setup.Moreover, reflected power increased frequently during the atomization stepwhen the horizontal-mount was employed compared with the vertical-mount,probably due to arcing between the furnace wall and the base of theRF connector.43In addition, sufficient cooling of the furnace support structure made ofcopper during the atomization step and removal of the hot gases from theplasma source work-head after the atomization step were two importantconsiderations for the usable atomization temperature with thisexperimental system. Once the atomization temperature was set, theheating rate was limited by the IL-555 furnace power supply.2.1.3 Spatially Resolved Intensity MeasurementsThe plasma source work-head was mounted on a post which in turn wasmounted on a crank-driven linear translation stage which allowed thework-head to be moved laterally relative to the detection system. Forspatially resolved emission intensity measurements, the plasma sourcework-head was translated in increments of 0.1 mm. The movement of theplatform was monitored using a precision displacement indicator gauge( Model 2047-11, Mitutoyo, Japan ). For the spatially resolved experiments,emission intensities were recorded on a chart recorder. The plasma wasviewed, 1.2 mm from the RF electrode ( unless otherwise noted ), on theopposite side of where the sample was deposited.2.1.4 Spectral Isolation and DetectionA 0.35 m Czerny-Turner monochromator ( Model 270, Schoeffel-MacPherson,MA, USA) with a holographic grating with 2400 lines/mm was employed,with an entrance slit of 70 gm wide and 2 mm high, unless otherwise noted.44For atomic absorption experiments, a 1: 1 image of the hollow cathode lightsource ( Hamamatsu, Japan ) was formed at the center of the furnace byemploying a 20 mm diameter, 75 mm focal length fused silica lens: Li( Melles Griot, Irvine, CA, USA). The 1: 1 image of the plasma was formedat the entrance slit of the monochromator by employing a 35 mm diameter,100 mm focal length fused silica lens: L2 ( Melles Griot, Irvine, CA, USA).Signals were detected by employing a PMT ( Model R955, Hamamatsu,Middlesex, NJ, USA). The PMT was operated at - 600 V. The output of thePMT was amplified by employing a current amplifier ( Model 427, Keithley,Middlesex, NJ, USA). The rise time of the current amplifier was 0.3 ms,unless otherwise noted. The image of the plasma source was aligned with theentrance slit of the monochromator by employing a HeNe laser ( Miles Griot,Irvine, CA, USA).2.1.5 Atomic Absorption MeasurementsFor atomic absorption measurements, output of the current amplifier wasdivided and one of the channels ( CH2 ) was selectively detected by alock-in-amplifier ( Model 121, PAR, Princeton, NJ, USA) with a 30 ms timeconstant. The HCL source used for the atomic absorption experiments wasmodulated at 200 Hz by employing a pulsed power supply ( ElectricalServices Shop, Department of Chemistry, UBC ). This pulsed power supplycan modulate up to 500 Hz and can deliver up to 100 mA of current. TheHCL output was saved, and absorbance was calculated, when necessary.452.1.6 Graphite Furnace Temperature MeasurementsThe radiation emitted from the graphite furnace during the atomization stepwas monitored using an optical pyrometer ( Ircon Series 1100, Model 11 x 30,IL, USA) which viewed the sample introduction hole ( without theRF electrode ) through a 1 inch diameter quartz window on the top flange.The optical pyrometer was mounted on a support-arm which in turn wasfixed to the plasma source work-head by a vertical post. The opticalpyrometer could be slid along the support-arm which in turn could berotated 3600 around the vertical post. This arrangement allowed the opticalpyrometer to be focused onto a selected position on the graphite furnace.The output from the pyrometer was amplified by an amplifier ( ElectricalServices Shop, Department of Chemistry, UBC ), and then digitized. Thedigitized data were converted into absolute temperature by using thecalibration data provided by the pyrometer manufacturer and fitted to aSt order polynomial least square fit. The graphite furnace was assumed as agrey body radiator with an emissivity of 0.7. The temperature below therange of the optical pyrometer was determined by extrapolating furnacetemperature back to the ash temperature. The ash temperature wasmeasured using a thermocouple ( Model 80TK, John Fluke, Everett, WA,USA). The atomization temperature was the maximum temperaturereached by the graphite furnace during the atomization step of the analyte.The appearance temperatures of analytes were calculated based on theirappearance times and the temperature-time profile of the graphite furnace.46The appearance time was defined as the time taken for the absorbance or theemission signal to reach the average base line plus two standard deviations ofthe base line noise. The peak temperature was defined as the temperature atwhich maximum absorption or emission occurs.2.2 DATA ACQUISITIONThe data acquisition was started when the ADC was triggered by a triggersignal from the graphite furnace power supply at the start of the atomizationstep. The input from single channel or two channels was digitized with a12-bit resolution by employing a sixteen channel ADC ( Model ADM12-10,Quatech, Akron, OH, USA) capable of operating at a maximum samplingrate of 30 KHz, and stored by using a 12 MHz IBM PC/AT compatiblecomputer.The data acquisition software facilitated two signals: analyte andbackground, to be acquired, and along with the background corrected signal,to be stored in the computer. The rate of data acquisition ( maximum of250 Hz) and the number of data points per channel ( maximum of2000 points ) were software selectable, and were limited by the freeconventional memory allocated by the Ver. 4.0 of MS-DOS. The ADC wasaddressed by incorporating the software provided by Quatech Co.472.3 DATA PROCESSINGThe data processing software was written in Turbo Pascal ( BorlandInternational, CA, USA), and facilitated the calculation of diagnosticinformation, such as peak height, peak area, and the peak width of thetemporal response of the emission signal. The other options included werebackground correction, signal averaging, smoothing, displaying, andgenerating plots for HP Tm plotters.All graphs and curve fittings shown in this thesis were completed with theaid of Axum: Technical Graphics and Data Analysis program ( TriMetrix,WA, USA ).2.4 THE FAPES METHODThe analyte sample was deposited onto the inner furnace wall ( or onto theRF electrode ), and was subjected to the thermal treatment. The duration ofthese heating steps could be selected and furnace temperature program wasin auto mode. Within the next 10 s lag time, RF power was applied to theRF electrode and the plasma was ignited. This lag time allowed sufficienttime to ignite and to stabilize the plasma before the start of the atomizationstep. After this lag time, the atomization step was started. Data acquisitionwas automatically triggered at the start of the atomization step. Theduration of the atomization step was limited to a minimum of 5 s by the48furnace power supply. All signals were acquired for 8 s. The next step wasthe cooling step of the graphite furnace. After the cooling step, the systemwas ready for the next sample. The sample throughput was 4 to 6 samplesper hour.It should be noted that this chapter describes only the experimentalsystem developed. The specific experimental methods along with theexperimental parameters will be presented with each study. The nextchapter will present some basic characteristics of the RF plasma.49CHAPTER 3SPECTRAL AND EXCITATION CHARACTERISTICS OFTHE ATMOSPHERIC PRESSURE HELIUM PLASMA SOURCE IN FAPES3.1 INTRODUCTIONSpectral characteristics of a plasma are among the most importantconsiderations for the successful utilization of a plasma source in emissionspectrometry. Spectral characteristics may affect the background and noisein the background in a spectrochemical determination. Signal-to-backgroundratio ( SBR ) and noise in the background are two important characteristicsthat affect the detection limit of an analytical method [72,73]. Spectralcharacteristics of plasma sources are also useful when studying thefundamental properties of plasmas.In addition to spectral characteristics, it is well known that the excitationtemperature is an important fundamental property which is closelyassociated with the analyte excitation and ionization. In high temperatureplasmas, analytes produce intense atomic ( or ionic ) line spectra which canbe used for analytical determinations. The emission intensity of an analytedepends on the total number of atoms ( or ions ) in the volume from which thesignal is collected and the fraction of atoms ( or ions ) that are in the excited50state. The population of a given excited state is proportional to theBoltzmann factor: e- "TE, exe ( Section 1.4.5). For example, for an excitedstate with an energy of 4.3 eV, the Boltzmann factor is 25 times higher at4500 K than at 3000 K. If the number of atoms per unit volume remainsconstant, emission intensity should increase by a factor of 25 when thetemperature changes from 3000 to 4500 K. The more intense analyteemission provides a higher sensitivity in analytical determinations.Furthermore, many excited states are populated in a high temperatureplasma, so that an alternative line can be used when a spectral interferenceoccurs by an emission feature in the background or another interfering linein a multi-elemental analysis. Moreover, high temperature plasmas providean excitation environment relatively free from matrix interferences.In this study, some of the basic characteristics of the helium plasma sourceat 13.56 MHz in FAPES were investigated. Spectral, spatial, and temporalprofiles of the plasma emission features, and the excitation temperaturemeasured by using iron as a thermometric species are presented in thischapter. Furthermore, the importance of these plasma properties to FAPESwill be discussed. The work described in this chapter has been previouslypublished by Hettipathirana and Blades [70].513.2 EXPERIMENIALThe experimental system described in Chapter 2 and a horizontal-mount forthe RF electrode were employed in the present study. The followingexperimental methods and parameters should also be noted.The plasma source work-head, employed for the collection of backgroundspectra depicted in Figure 3.1, was an IL-455 ( formerly InstrumentationLaboratory, now Thermo-Jarrel Ash, Waltham, MA, USA) Massmann-typefurnace work-head fitted with a pyrolytic-graphite coated RF electrode [15].This furnace work-head could be quite effectively purged and sealed, allowingspectra to be collected without ingress of atmospheric pressure gases. Thishelium plasma source was operated at 27 MHz and at an RF power of about20W.3.2.1 Plasma Background and Iron Emission SpectraA 1 m Czerny-Turner monochromator ( Model 2061, Schoeffel-McPherson,MA, USA ) was used for the measurement of plasma background spectra, andiron spectra for excitation temperature measurements. This monochromatorwas equipped with a holographic grating with 1200 lines/mm( Model AH-3264, Schoeffel-McPherson, MA, USA). The image of the plasmawas formed at the entrance slit of the monochromator by employing a150 mm focal length, 50 mm diameter piano-convex fused silica lens ( Oriel,Stratford, CT, USA). For the Fe-excitation temperature measurements, the52image of the plasma was taken from a distance of 1 mm from theRF electrode using a slit height of 2 mm and a slit width of 70 pm such thatthe area of the discharge sampled had the same dimensions ( 1: 1 imaging ).The source of Fe for the temperature measurement was from amicro-pipette deposition of 51.LL of a 10 ppm Fe solution. Iron solutions in1 % ( v/v ) HNO3 were prepared by dissolving analytical grade FeSO4.7H20( BDH, Toronto, Canada ). Five replicate depositions were made at eachRF power setting. The sample was dried at about 390 K for 1 min, with a1 min thermal pre-treatment step at about 770 K ( during which the plasmawas ignited ). The temperature was then ramped to 2100 K. Emissionspectra were recorded by integrating the signal for 7.5 s. This integrationtime is longer than is actually required because the signal only lasts for aperiod of about 1 - 2 s. However, difficulties in synchronizing the timing ofthe graphite furnace atomization step and the beginning of an arrayintegration period precluded the use of a shorter integration time.The detector employed was a linear photodiode array ( LPDA: ModelRL-2048S, Reticon, Sunnyvale, CA, USA). The array integration period andreadout were controlled by employing a satellite controller board( Model RC1021, Reticon, Sunnyvale, CA, USA). Using an entrance slit of70 gm, the LPDA spectrometer provided a resolution of 0.06 nm whileallowing the simultaneous measurement of a spectral window about 40 nmwide. The wavelength vs diode calibration was carried out by measuringemission lines of Cd, Pb, and Fe. A thermoelectric cooler53( Model CP14-71-10L, Melcor, Trenton, NJ, USA ) mounted on the backside ofthe LPDA allowed it to be cooled to -20 °C.Data acquisition was carried out by interfacing the LPDA to a 12-bit ADC( Model ISC-16, RC Electronics, Santa Barbara, CA, USA) which wasinterfaced to a 12 MHz IBM PC/AT compatible computer which also was usedto set the integration time for the LPDA. In-house software for acquisition ofthe LPDA spectra was available.3.2.2 Spatially and Temporally Resolved SpectraA 0.35 m Czemy-Turner monochromator ( Model 270, Schoeffel-McPherson,MA, USA) equipped with a holographic grating with 1200 lines/mm wasemployed for spatial and temporal emission measurements. The 1: 1 imageof the plasma was formed at the entrance slit of the monochromator byemploying a 100 mm focal length, 35 mm diameter fused silica lens ( MellesGriot, Irvine, CA, USA). The entrance slit was 50 gm wide and 1 mm high.The detector employed was a PMT ( Model R955, Hamamatsu, Middlesex,NJ, USA). The PMT output was amplified by employing a currentamplifier ( Model 427, Keithley, Middlesex, NJ, USA). Output from thecurrent amplifier was converted to digital form by employing an ADC( RC Electronics, Santa Barbara, CA, USA) and, stored in a 12 MHzIBM PC/AT compatible computer for further processing. The data acquisition54was accomplished by the software provided by RC Electronics Co. Fortemporally resolved measurements, 4000 points were taken at 500 Hz.3.3 REsuurs AND DLSCUSSION3.3.1 Background Emission SpectraTypical background emission spectra for a 20 W helium plasma source areprovided in Figure 3.1; for the spectral regions, 220 - 300 run ( Figure 3.1.a),300 - 400 nm ( Figure 3.1.b ), and 400 - 520 nm ( Figure 3.1.c ). In general,the wavelength range, 200 - 500 nm, is considered as the analytically usefulwavelength range in atomic emission spectrometry. The most prominentspectral features in the background emission spectrum are: CO + ( Firstnegative system of CO), OH, NH, N2 ( second positive system of N2 ), andN2+ ( first negative system of N2 ). These spectral features are listed inTable 3.1. Reference data on molecular spectra are available [741These spectra were collected when the furnace was at room temperatureand the plasma source work-head was continuously purged with helium. Thecomposite spectra were constructed by integrating 12 different photodiodearray spectral windows, each 40 nm wide, into a single background emissionspectrum. Integration times ranging from 2 to 10 s were used. Figure 3.1allows a comparison of the relative intensities of the spectral features.However, it should be noted that the spectra have not been corrected for thespectral response of the measurement system.•-•0 ,---1/ '-CMI 0^CNI 0 -C.l wi et 0^csie-1.01.tO+ (^x21+ )1250 -1000 -750 -500 -250 -0 N.L14,661466,tymikAmAiliweAsiusimm220 230 240 250 260 270 280 290 300Wavelength ( nmFigure 3.1.a. ( continued on page 56)1 1 1 1 1 1 r 1200 - X 1 02160 -120 -80 -40 -0O.dioF^I-300 310 320 330 340 350 360 370 380 390 400Wavelength ( nm )Figure 3.1.b. ( continued on page 57)1 111 11200 - x 1 021 60 -1 20 -80 -40 -0400 410 420 430 440 450 460 470 480 490 500 510 520Wavelength ( nm )Figure 3.1. Background Spectra of helium plasma. (a) 215 -300 nm, (b) 300 - 400 nm, and (c) 400 - 515 nm.58Table 3.1. Transitions of species observed in heliumplasma at atmospheric pressure in 215 - 515 nmrange.Emitting species^Transition^Line wavelength or(0,0) Band-Head( observed range )C (I)^1131 - 3-S0^ 247.8He(I)^1131 - 10 501.5He ( I ) To - 3S1^ 388.8First negative^B2Z+ - vz+ 219system of CO ( 215 - 280)OHSecond positivesystem of N2NHFirst negativesystem of N2A2E+ - X2II^306( 280 - 320)C311u - B311g^337( 295 - 400 )A3II - X3E-^336( 336 - 337)B2Y,u+ - x2rig+^391( 330 - 470)59These prominent emission features in the 27 MHz helium plasma ( seeSection 3.2) are also observed in the 13.56 MHz helium plasma operating atan RF power of 20 W. For the RF power range between 20 and 50 W, nofrequency dependence on the background emission features is noted.Furthermore, during an atomization step of 1800 K, all prominentbackground emission features observed in the 27 MHz plasma are observedin the 13.56 MHz plasma At both frequencies, an increasing continuumbackground beyond 450 nm is also observed because of the black-bodyemission from the furnace wall.In the 220 to 270 nm region, the spectra are dominated by emission bandsof CO+ ( Figure 3.1.a ). However, these CO + bands are not observed when Aris used as the plasma gas ( Section 3.3.4). It is well known that CO + isreadily excited in helium discharges as a result of selective excitation of theB2Z+ state of COP, according to the following reactions [75,76];^He2++ CO ^> CO( B2Z+ ) + 2He^(3.1),and He (21S, 23S ) + CO ^> C0+( B2Z+ ) + He ( 11S ) + e^(3.2).The variable potential energy of He2+ ( 18.3 - 20.3 eV) primarily accountsfor the excitation of CO + ( B2) state by the resonant charge-transfermechanism [76]. The presence of these CO + bands in FAPES suggests thatthe helium plasma source could be quite effective for the selective excitationof analyte atoms and molecular species in cases where the energetics are60favorable. The other dominant feature is the emission from the differentvibrational bands of N2+. The excitation of the N2+ system is also bycharge-transfer from He2+, which selectively populates the B2Zu+ electroniclevel leading to intense ( B-X ) band emission [76].The occurrence of these molecular bands presents a relatively intenseplasma background upon which analytical signals must be measured. Mostof these background spectral features arise from traces of impurities in theplasma gas and molecular species desorbed from the furnacewall ( Section 3.3.3). The approximate concentration of impurities in thebottled helium used in these experiments were: 112< 1 ppm, 02 < 3 ppm, N2 .-.5 -25 ppm, CO2 < 1 -2 ppm, H20 <5 ppm, and total hydrocarbons <5 ppm.However, in graphite furnace methods, the sample matrix is usually the mainsource of contaminants in the gas-phase. In an analytical determination, thegas-phase within the graphite furnace may be dominated by various samplematrix decomposition products as well as water vapor and acid decompositionproducts. These species can be readily adsorbed on the furnace wall anddesorbed during the atomization step of the analyte [77]. Therefore, in a realanalysis, the plasma background is likely determined by the composition ofthe sample matrix. The intensities of background emission features are alsodependent on the RF power.Sturgeon et al. reported emission features for the spectral region190 - 430 nm: CO, OH, CN, NH, and He ( I ) at 388.9 nm for a 50 W plasma( frequency was not given ) [16]; and for the spectral region 200 - 510 nm: NO,61CO, N2, OH, and He ( I ) lines at 388.8, 447.1, 471.3, 492.2, and 501.6 run fora 50 W plasma at 13.56 MHz [78]. In a subsequent publication, the presenceof CO+ was reported for a 50 W helium plasma at 40 MHz [79]. Some He ( I )lines ( 447.1, 471.3, and 492 2 nm ) observed at 13.56 MHz heliumplasma [78], were also absent in the helium plasma at 40 MHz [79]. Theabsence of CO+ emission features in the 13.56 MHz helium plasma sourceemployed by the Sturgeon group is probably due to a spatial difference orRE' power coupling efficiency difference between the plasma sourcesemployed by two groups, or both. The CO + emission intensity is low in theintermediate region between the RF electrode and the furnacewall ( Section 3.3.2).3.3.2 Spatially Resolved Background Emission IntensitiesFigure 3.2 is a plot of the spatial distribution of emission from CO + at 20 W,observed by translating an image of the plasma source laterally inincrements of 0.1 mm There are some features of the CO + spatialdistribution which should be noted. The spatial profile peaks at the center ofthe plasma source adjacent to the RF electrode and also near the wall of thegraphite furnace. Most intense CO+ emission is observed adjacent to theRF electrode than near the furnace wall. In addition, a more luminous regionsurrounding the RF electrode and a less luminous plasma filling theremainder of the graphite furnace volume can also be seen in thisatmospheric pressure helium RF plasma in FAPES.IiI 1IiII1I1I1111111II1II1IIII1II1IIIII1111••,•- -,..•^ss,50 -40 -30 -20 -10-IIIIfII0 _60 -620^1^2^3^4^5^6Distance ( mm )Figure 3.2. Emission spatial profile for COP.The intense CO+ emission near the RF electrode is likely due to adifference in RF field strengths at the RF electrode and at the furnace wall.This difference in the RF field strengths arises as a result of the differencesin the surface area of electrodes: RF electrode and furnace wall. The higherRF field strength at the RF electrode leads to a high energetic excitationregion adjacent to the RF electrode compared with that adjacent to thefurnace wall. The low excitation region between the RF electrode and thefurnace wall can be evidenced by very low emission intensity of CO + observedbetween the RF electrode and the furnace wall. Sturgeon et al. also reported63a He-excitation temperature difference of 400 K or more between the sheathadjacent to the RF electrode and the rest of the plasma [80]. Furthermore, inlow pressure RF discharges, a higher field strength appeared at the smallerelectrode [60].At this time, it is not known whether the major source of CO is from theoxidation of the carbon surface on RF electrode and furnace wall, orsputtering of carbon from the RF electrode ( and from the furnace wall ) tooxidize in the gas-phase. However, in an atmospheric pressure heliumplasma, it is most likely that CO + is mainly formed from the precursor CO,formed from the oxidation of carbon on the RF electrode and the furnace wall.The high reactivity at the RF electrode can also be evidenced by the gradualdegradation of the RF electrode. This CO formation, in the presence of theplasma, is an additional source of CO which is absent in GFAAS.Similar spatial behavior is observed for the He ( I )1. and N2+, although theprofiles for OH and NH show significant emission in the intermediate regionbetween the RF electrode and the furnace wall. Significant emissionobserved between the RF electrode and the furnace wall for OH and NHcompared with that of CO+ is likely due to differences in excitationmechanisms, high radiative life times, or both. The exciting species for botht Spatial distribution of the Ar ( I ) emission line at 415.85 nm, when plasma sourcework-head described in Chapter 2 is employed as a low pressure hollow anode d. c.discharge, is similar to that of He ( I ) from the atmospheric pressure plasma in FAPES.For the low pressure plasma source, a cathode voltage of 460 V and pressure of 1.2 torrwere used.64OH and NH is He [75]. The radiative life times for OH and NH are 0.69 and0.43 gs, respectively [81]. The radiative life times for CO+ and N2+ are 0.05and 0.06 gs, respectively. Regardless, spatial behavior of the molecularspecies is indicative of a strong excitation region adjacent to the RF electrodein this plasma source.The spatial distribution of analyte emission is not measured because thetransient nature of the signal makes this a time consuming measurementprone to error. The use of a slitless spectrograph combined with an imagingdetector, similar to the system used by Gilmutdinov et al. for the collection ofabsorption shadow-grams [82], is one possible method for acquiring thisinformation.3.3.3 Temporally Resolved Background Emission IntensitiesThe variation of emission intensity for He ( I ), COP, N2+, and OH wasmeasured during a dry ( no aqueous sample was deposited ) atomization stepin order to examine the effect of furnace wall temperature on the emissionintensity for these species. The results are given in Figure 3.3.a for He ( I ),Figure 3.3.b for CO, Figure 3.3.c for N2+, and Figure 3.3.d for OH, atRF powers of 14, 18, 22, 26, and 30 W, as marked on each figure. Thetime-resolved emission intensity profiles given in Figure 3.3 ( a - d ) show thevariation of the intensity of a band-head during the atomization step. Theatomization temperature profile corresponding to these figures is given inFigure 3.3.e.65For He ( I ) ( Figure 3.3.a ), there is an increase in the emission intensityat the beginning of the atomization step with increasing RF power and theemission intensity is relatively stable throughout the atomization step. Theincrease in the helium emission intensity, at the beginning of the atomizationstep, is due to the increase in excitation when the RF power is increased.The relatively stable helium emission intensity shows that the RF powercoupling efficiency to the plasma is stable during the atomization step andnot affected by the thermionic electrons at these furnace temperatures.Moreover, the helium emission signal shows a small increase during the peakatomization step. This small increase in emission intensity may be due to achange in the He-excitation temperature. However, Sturgeon et al. reportedthat, for a 100 W plasma, the He-excitation temperature was =affected bythe furnace wall temperatures between 500 and 2500 K [80].For CO+ ( Figure 3.3.b ), there is a dramatic increase in the emissionintensity during the atomization step, followed by a depression at the peakatomization temperature, and a plateau as the furnace cools. This increaseduring the atomization step is most likely due to an increase in the amount ofCO in the plasma as a result of increased desorption of CO from theRF electrode and the furnace wall. The depression at the peak atomizationtemperature could be attributed to recombination of CO + with the therrnionicelectrons liberated from the heated RF electrode to form CO. The thermionicelectron emission from the RF electrode can also account for the time lagbetween the depression and the peak atomization temperature, as well as the300 -250 -200 -150 -100 -50'^I^I^F^-•^I^•^I^-•^I 1.100 0 0.5 1.0 1 .5 2.0 2.5 3.0 3.5 4.0 4.5 5.0 5.5 6 0Time ( s )66•3.3.b30 W^ .••"---------,^/ /^\^/ ,-.„^•-s,...-_,26 W /^,^-''`'' --._/^/^(-- -.2.\^ 'l22 W^ ,,18 W/i „./ .--- .-'^/ r.....---.•,_./ ..^\^i. •••. \^, ''......../.^-^•- -•:‘^•14 W800700600500 -400 -300200100^------ —0'•^00 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0 5.5 6 0Time ( s )Figure 3.3.a and 3.3.b. ( continued on page 67)670 0 0.5 1 .0 1 .5 2.0 2.5 3.0 3.5 4.0 4.5 5.0 5.5 6 0Time ( s ),0.0 0.5 1 .0 1 .5 2.0 2.5 3.0 3.5 4.0 4.5 5.0 5.5 6.0Time ( s )Figure 3.3.c and 3.3.d. ( continued on page 68)681 2000 0 0.5 1.0 1 .5 2.0 2.5 3.0 3.5 4.0 4.5 5.0 5.5 6 0Time ( s )Figure 3.3. Temporal emission behavior for (a) He ( I ),(h) CO', (c) N2+, and (d) OH at RF powers of 14,18, 22, 26, and 30 W; (e) furnace temperaturecorresponding to these diagrams*.shift in the depression to earlier times with an increase in RF power. Thisobserved depression may also be due to the recombination of He2+ withthermionic electrons, and hence, change in the excitation characteristicsfor COP.* The X-axis in Figure 3 can be compared with similar figures in the rest of this thesis byadding 1.4 s.69If recombination of the thermionic electrons with molecular ions is thecase, these molecular ions could act as a buffer, minimizing the effect ofthermionic electrons on the total electron density in the discharge. The datafor N2+ ( Figure 3.3.c ) support this suggestion because the temporal behaviorat the atomization step for N2+ is similar to that observed for COP, and thedepression appears near the temperature maximum In addition, thetemporal behavior for CO+ and N2+ are dependent on both the incidentRF power and the furnace temperature. In general, as the RF power isincreased, at a given point in the atomization step, the emission intensity isincreased as a result of the increase in excitation.For OH ( Figure 3.3.d ), there is an initial increase, then a decrease inemission during the atomization step. Most probably, the temporal profile isthe result of initial desorption of H2O and OH, and subsequent dissociation,or increase in electron recombination with exciting species He.3.3.4 Fe-Excitation TemperaturesThe time-averaged Fe-excitation temperatures were evaluated by using sixFe ( I ) emission lines in the 370 - 377 nm spectral region, which wererecorded by employing the photodiode array spectrometer described in theexperimental system. The background subtracted emission intensities weresubstituted into the Boltzmann distribution and the Fe-excitationtemperatures were evaluated from the slope method ( Section 1.4.5).Because the lines used for the Fe-excitation measurement are all in close70Table 3.2. Fe ( I ) lines used for the temperaturemeasurement and relevant spectral data [83].Wavelength ( nm ) E ( cm-1 ) log gqfqp372.256 27560 -1.28373.486 33695 0.31373.713 27167 -0.57375.823 34329 0.00376.379 34547 -0.19376.719 34692 -0.34wavelength proximity, no spectral response correction was applied for themeasurement system. The Fe lines that were used along with their spectraldata are listed in Table 3.2. These spectral data were taken from Bridgesand Komblith [83]. The error in these gf values is about 10 %.There are several advantages to the use of a photodiode arrayspectrometer. One is that the blank background can be recorded andspectrally stripped from the analyte spectrum, ensuring the removal ofinterfering background lines. This procedure was used for the measurementof temperatures reported in this thesis. A second advantage is that the71400 -350:300:250:200 -150 -100 -0)0)NYNtr) Ntf) 0)0) ll,^CDal cl^N^I ' )^TZ t tr 3 2 g ; 2^I:, Tzr.: cd 6 ci .- N N ei Ki ei NI- 4 4 6 6 6ID CO N N NNN-rshr, N. N. NNNNP.1 NI M PI 1.1^NI^P.')^CV 01^Pn Pe) rn P.) MI^P )P1P■— —1...k_________A............A..../A111----r50 -^ 10I^•^I366^368^370^372^374^376^378^380Wavelength ( nm )Figure 3.4. Emission spectrum for 50 ng of Fe at anRI' power of 20 W.emission from all lines is recorded simultaneously, for a single sampledeposition, which significantly improves the precision, particularly for atransient signal obtained from graphite furnace methods. A typical emissionspectrum of Fe at an RF power of 20 W is provided in Figure 3.4.The variation of the Fe ( I ) excitation temperature with RF power ispresented in Figure 3.5, which covers the RF power range 15 to 50 W. Thetemperatures measured, from a distance of 1 mm from the RF electrode,range from 3100 to 4200 K. The data in Figure 3.5 also demonstrate that the72temperature increases linearly with an increase in RF power over theRF power range studied. For comparison, a temperature of about 4500 Kcorresponds to a low flow Ar-ICP operating at a power of about 600 W [68]. Itis clear that either the radiative and convective heat losses from the plasmasource in FAPES are much lower than those from a ICP or the RF fieldstrength is much higher in this RF plasma source in FAPES, or both. Anexcitation temperature of 2600 K was reported for thermal excitation inCFAES [84]. At an RF power of 50 W, excitation temperature of FAPES is1600 K higher than that of CFAES. For FAPES, the Fe-excitationtemperature was also measured at a different position, closer to the furnacewall. The measured temperature is 3100 ± 90 K at an RF power of 45 W. Asmentioned earlier, the difference in temperature near the RF electroderelative to the furnace wall also shows that the plasma is hotter adjacent tothe RF electrode. Based on the Fe-excitation temperatures, this plasmasource in FAPES clearly has the potential to be a potent excitation source foratomic spectrometry.Sturgeon et al. also reported several different temperatures for a13.56 MHz helium plasma in FAPES, including Fe and He-excitationtemperatures. The Fe-excitation temperature was measured using Fe(C0)5,which was continuously introduced into the furnace work-head and thefurnace was maintained at 970 K in order to dissociate the ironpentacarbonyl. The Fe-excitation temperature was measured only at anRF power of 100 W. Measurement at lower RF powers was not possible as aresult of poor signal-to-background ratio, attributed to partial quenching of73 450042003900360033003000^1^I^1^1^1^t^I^i^i10 15 20 25 30 35 40 45 50 55 60Power ( W)Figure 3.5. Iron excitation temperature as a function ofRF power.the plasma due to Fe(C0)5 and its decomposition products. However, theHe-excitation temperatures were measured at RF powers between 25and 100 W.The Fe-excitation temperature reported by Sturgeon et al. was 2920 K fora 100 W plasma [80]. For the RF plasma source described in this thesis, thistemperature would correspond to an RF power of 10 W. It is unclear at thistime whether the difference in Fe ( I ) temperature is due to variability insources, spatial differences, or differences in the RF power coupling74efficiency. It should also be noted that the energy levels used bySturgeon et al. for the Fe-excitation temperature measurements weredifferent from those used in this thesis. In addition, the effect of the presenceof excessive amounts of Fe(C0)5 or CO on the excitation characteristics of theplasma source in FAPES is not known. Nevertheless, considering thesedifferences in the temperature evaluation, the Fe-excitation temperaturereported by Sturgeon et al. is in reasonable agreement with those reported inthis thesis.3.3.5 Atmospheric Pressure Ar PlasmaA discussion on some experimental observations of the Ar plasma source inFAPES is presented in this section.An Ar discharge at 13.56 MHz could be ignited at powers exceeding about30 -40 W but would not operate in a stable manner at higher powers. If,after ignition, the power was rapidly reduced to 8 - 10 W, a diffuse Ar plasmacould be maintained at 13.56 MHz. However, it has been found that it ismuch easier to ignite and operate an Ar plasma at 27 MHz. The ease ofoperation of the helium plasma can be attributed to the higher electronionization rate of helium compared with that of Ar.The difference in background spectral features in helium and Ar should benoted. Figure 3.6 presents the 40 nm wide wavelength window around236 nm for the Ar plasma at 27 MHz. The intense emission from CO + in the1250 -x io1 000 -750 -500 -250 -075I^ I^ I^I^I^I220 225 230 235 240 245 250 255 260Wavelength ( nm )Figure 3.6. Background spectrum of the Ar plasma for thespectral region, 220 - 260 nm.helium plasma is absent in the Ar plasma at 27 MHz. Furthermore, intenseemission from C ( I ) at 247.89 xim can be observed in the Ar plasmacompared with that in the helium plasma. This intense emission from C ( I )in Ar plasma compared with that of helium plasma is likely due to the heavyAr ion bombardment on the RF electrode, higher temperature of the Arplasma, or both. Sturgeon et al. also reported the presence of C ( I ) line at247.85 nm in the Ar plasma, but not in the helium plasma [78,791763.4 SUMMARYSome basic characteristics of the atmospheric pressure helium plasma sourceat 13.56 MHz in FAPES are presented and discussed in this chapter.Background emission spectra for the wavelength region, 200 - 500 rim, aredominated by emission features from COP, OH, NH, N2+, and N2. Theemission features of CO+ and N2+ indicate that this plasma source is capableof exciting energy levels as high as 20 eV, and the energy transfer mechanismin this case is near-resonant charge transfer.Background emission is most intense near the RF electrode and lessintense near the furnace wall. The intense emission near the RF electrode islikely due to a difference in RF field strengths at the RF electrode and at thefurnace wall. The higher RF field strength at the RF electrode leads to ahigh energetic excitation region adjacent to the RF electrode compared withthat adjacent to the furnace wall.Time-resolved studies show complex emission characteristics for plasmaemission features during the atomization step. This complex emissionbehavior of the molecular species is indicative of the evolution of thermionicelectrons from the RF electrode during the high temperature atomizationstep. The time-averaged Fe ( I ) excitation temperature for the 13.56 MHzhelium plasma source is in the range, 3100 - 4200 K, for RF powersbetween 15 and 50 W. The Fe-excitation temperatures also show a spatialdependence and a higher temperature is observed adjacent to the77RF electrode. These Fe-excitation temperatures show that the FAPES sourcehas the potential to be a potent excitation source for atomic spectrometry.As is the case with most spectrochemical plasma sources, this RF plasmasource is spatially non-homogeneous and shows a strong excitation regionnear the RF electrode. Because the plasma source is spatiallynon-homogeneous, care must be exercised in the measurement andcomparison of fundamental and analytical characteristics as they may bespatially dependent.78CHAPTER 4TEMPORAL EMISSION AND ABSORPTION CHARACTERISTICS OFSILVER, LEAD, AND MANGANESE IN FAPES4.1 INTRODUCTIONThe temporal response of the analyte absorption signal is an importantdiagnostic tool in GFAAS, and has often been combined with kinetic andthermodynamic calculations to determine the analyte atomizationmechanisms during the high temperature atomization step of theanalyte [85-88]. Under temporally non-isothermal conditions, temporalresponse of the analyte signal is also an important factor that affects thefigures of merit and matrix effects for analytical determinations byGFAAS [44,89].The temporal response of the analyte signal in FAPES is likely differentfrom that in GFAAS due to three primary reasons. Firstly, the atomizationcharacteristics of analytes in FAPES may be different from those in GFAASbecause of the plasma inside the graphite furnace. The plasma may affectatom-formation processes through participation in the equilibria forcondensed and gas-phase species. Gas-phase concentrations of differentspecies that affect the atomization characteristics of analytes may be79different with the plasma inside the graphite furnace compared with thosewithout the plasma. Secondly, the vaporization characteristics of analytes inFAPES may be different from those in GFAAS because of the RF electrodeinside the graphite furnace. The RF electrode can act as a second-surface,where analytes can condense and then vaporize to form a second peak.Thirdly, if the excitation characteristics change during the high temperatureatomization step of the analyte, then the temporal response of the analytesignal in FAPES would be different from that in GFAAS. In this case, thetemporal response of the analyte signal is determined not only byatomization and vaporization characteristics of the analyte but also byexcitation characteristics of the analyte in the plasma.Smith et al. reported double peaks for Ag at high analyte amounts and lowRF powers, and speculated that the appearance of double peaks was due tothe atomization characteristics of Ag in FAPES [17]. Sturgeon et al. reportedthe appearance of double peaks for Cd at low RF powers in FAPES [16]. Inaddition, an early shift in the position of peak maximum for Cd, Cu, Ni, andBe, and no shift for Fe, Pb, P, and Bi were observed with increasingRF power [78]. Sturgeon et al. suggested that these observed shifts were dueto the increased plasma volume and density which occur with increasingRF power, more efficient excitation via electron collision, possible changes inthe actual observation zone, and reduced self-absorption in the larger, hotterplasma [78]. For HA-FANES, Riby et al. reported two unresolved peaks forCr, and suggested that Cr condensation on the electrode was the origin ofthese two peaks [90].80In an effort to understand the temporal response of the analyte signal inFAPES, a study of time-resolved atomic absorption and atomic emission wasundertaken. This chapter presents the results of this study for Ag, Pb, andMn. Effects of the plasma and the RF electrode on analyte atomization andvaporization characteristics are discussed. In addition, effects of the plasmaexcitation process on the temporal response of analyte signal during theanalyte atomization and vaporization are presented. The work described inthis chapter has been previously published by Hettipathirana andBlades [71].4.2. EXPERIMENTALA complete description of the experimental system employed to acquiresimultaneous atomic absorption and atomic emission signals is given inChapter 2 of this thesis. This experimental system and a vertical-mount forthe RF electrode were employed. The following experimental methods andparameters should be noted.4.2.1 Signal PresentationFour replicate measurements were averaged and subjected to a 25-pointSavitzky-Golay smoothing procedure [91]. Analytical parameters, such asabsorbance, peak area, and peak width, were calculated for each sampledeposition.814.2.2 Appearance and Peak TemperaturesThe appearance temperatures of analytes were calculated based on theirappearance times and the temperature-time profile of the graphitefurnace ( also see Section 2.1.6). The appearance time was defined as thetime taken for the absorbance or the emission signal to reach the averagebase line plus two standard deviations of the base line noise. Therepeatability in determining the appearance temperature was less than1.5 %. The peak temperature was defined as the temperature at whichmaximum absorption or emission occurs.4.2.3 ReagentsAll analyte solutions in 1 % ( v/v ) HNO3 were prepared from serial dilutionof 1000 mg L-1 stock solutions prior to analysis. Silver, Pb, and Mn solutionswere prepared by dissolving analytical grade AgNO3 and Pb(NO3)2 ( both,BDH, Toronto, Canada ), and reagent grade MnSO4.H20 ( MCB, Ohio, USA).Nitric acid solutions were prepared by using the analytical gradereagent ( BDH, Toronto, Canada ).4.2.4 ProcedureA 5 I.LL aliquot of analyte solution was deposited onto the furnace side wall oron the RF electrode by employing an Eppendorf 0.5 - 10 gla UltraMicropipette. The plasma source work-head was purged with helium ( UnionCarbide, Toronto, Canada ). The furnace temperature was increased to 470 K82in 45 s to dry, and was maintained at 470 K for another 45 s to ash thesample. This total time interval was sufficient to exclude the water vaporinside the plasma source work-head before plasma ignition. Within the next10 s lag time, the plasma was ignited. After this lag time, the furnacetemperature was ramped to 2050 K in 5 s for all determinations. After eachsample atomization step, a dry ( without deposition ) atomization step wascarried out to clean the graphite furnace. Four replicate measurements werecarried out for each determination. The blank determinations were carriedout by depositing the same amount of 1 % ( v/v ) HNO3 solution.All determinations were carried out by using atomic resonance lines ofAg ( 328.07 nm ), Pb ( 283.30 nm ), and Mn ( 279.28 rim ).4.3 RESULTS AND DISCUSSIONExperiments where atomic absorption is measured without the plasma, arereferred to as " AA at an RF power of 0 W" in this chapter. For thoseexperiments, the RF electrode is referred to as the "co-axial rod ". Whenatomic absorption is measured with the plasma, the corresponding RF powerwill be specified. The RF electrode is a graphite rod unless otherwise noted.It should also be noted that all specified masses of analytes are from theircorresponding salts dissolved in 1 % HNO3.834.3.1 SilverThe effect of a graphite co-axial graphite rod on the atomic absorption signalwas studied by depositing the analyte sample on the furnace wall and thetemporal response was recorded with and without the co-axial graphite rodinstalled ( AA at an RF power of 0 W). The results of this experiment, when0.5 ng of Ag is deposited on the furnace wall, are given in Figure 4.1.When the sample is deposited on the furnace wall without the co-axialgraphite rod, a peak with an appearance temperature of 1130 K and a peaktemperature of 1430 K is observed ( solid line ). When the sample isdeposited on the furnace wall with the co-axial graphite rod installed ( dashline ), two peaks are observed. The first is a very small peak with anappearance temperature of 1240 K and a peak temperature of 1400 K. Thesecond more intense peak has an apparent peak temperature of 1680 K.Different amounts of sample deposited on the wall result in a constant peakratio for these two peaks. These observations suggest that the co-axial rodacts as a condensation site for species vaporized from the furnace wall withsubsequent " second-surface " vaporization from the co-axial rod. The smallsize of the initial peak as compared with the second peak shows that thecondensation is essentially instantaneous. The delay in the onset of thelarger absorbance signal when the co-axial rod is installed is the result ofdelayed heating of the co-axial rod relative to the furnace wall, somewhatsimilar to the situation encountered during the radiant heating of aL'vov platform.1.21.00.80.60.40.20.0 • ••22001 900CD1 600CD1 300 ar-1-1000700844000.0^1.0^2.0^3.0^4.0^5.0^6.0^7.0^8.0Time ( s )Figure 4.1. Temporal response of the Ag atomic absorptionsignal at an RF power of OW for 0.5 ng of Agdeposited on the furnace wall with ( -) andwithout ( — ) the graphite co-axial rod;background ( • • • ); and the temperature profileof the furnace wall ( — — — ).The process in which the cooler surfaces inside the graphite furnace act ascondensation sites for analytes vaporized from the furnace wall is consistentwith recent studies reported for GFAAS [92,93]. L'vov et al. observed doublepeaks when the sample was deposited on the furnace wall with a platforminside the furnace [92]. Hocquellet employed a "second-surface trap " insidethe furnace to condense the analyte vaporizing from the furnace wall and85then to vaporize later in the atomization step to achieve the temporalisothermality [931.The temperature difference between the furnace wall and the co-axial rod,and the effect of graphite on the atomization process for Ag were studied bydepositing 0.5 ng of Ag on co-axial rods of graphite and tungsten ( AA at anRF power of 0 W). The results of this experiment for deposition on thegraphite rod ( solid line ) and tungsten rod ( dash line ) are given inFigure 4.2. Except for the small difference in the appearance temperatures( and the peak temperatures ), identical peak shapes are observed forsamples deposited on graphite and tungsten rods. These differences intemperatures are probably due to the difference in radiant heating rates forthese two types of co-axial rods.The time difference of 1.2 s in the appearance of the atomic absorptionsignal for the sample deposited on the furnace wall without the graphiteco-axial rod ( solid line, Figures 4.1 ) and that for the sample deposited on thegraphite co-axial rod ( solid line, Figure 4.2) demonstrates that there is anappreciable temperature difference in the temperature-time profiles for thefurnace wall and the co-axial rod. In addition, there is a difference in thepeak time for the Ag signal from the second-surface vaporization ( dash line,Figure 4.1 ) and the direct deposition of Ag on the graphite co-axial rod ( solidline, Figure 4.2). Silver deposited directly on the co-axial rod appears laterin time. This time difference is likely due to a difference in thetemperature-time characteristics between second surface vaporization sites1 .00.80.60.40.20.086 --0.0^1 .0^2.0^3.0^4.0^5.0^6.0^7.0^8.0Time ( s )Figure 4.2. Temporal response of the Ag atomic absorptionsignal at an RF power of 0 W for 0.5 ng of Agdeposited on the graphite co-axial rod ( — )and tungsten co-axial rod ( - - - ); andbackground ( • • • ).on the co-axial rod and the sample deposition site on the co-axial rod. Thereis no doubt that the co-axial electrode does not heat uniformly and it is likelythat the RF connector end of the co-axial rod is cooler than the rest.Unfortunately, getting an accurate spatially resolved measurement of theco-axial rod temperature by itself without collecting scattered radiation fromthe furnace wall is very difficult.87The identical peak shape of signals, from the deposition on both graphiteand tungsten rods ( Figure 4.2 ), is consistent with the atom-formationmechanism for Ag in GFAAS. The accepted mechanism of Ag atom-formationin GFAAS is the direct thermal dissociation of the oxide and volatilization ofthe metal [94]. The calculated distribution of species as a function oftemperature shows that, Ag exists primarily as metallic Ag on the furnacewall at temperatures below 1075 K, and sublimes to form gas-phase Ag above1075 K [94]. Thermogravimetric studies indicate that AgNO3 decomposes at880 K to form the oxide [95]. Silver oxide decomposes at about 615 K to formmetallic Ag [95].Figure 4.3 provides the results of a FAPES study in which 0.5 ng of Ag isdeposited both on the furnace wall ( solid line ) and on the RF electrode( dash line ) at an RF power of 20 W. The emission peak for the sampledeposited on the furnace wall ( peak width of 500 ms ) has an appearancetemperature of 1170 K and a peak temperature of 1410 K. The sampledeposited on the RF electrode results in a broader peak ( peak widthof 750 ms ) with an apparent appearance temperature of 1160 K and a peaktemperature of 1460 K. The early appearance of the emission signal whenthe sample is deposited on the RF electrode indicates that the temperature ofthe sample deposition site of the RF electrode may be slightly hotter than thefurnace wall. The emission profile for the sample deposited on theRE' electrode has more signal after the initial appearance, which could resultfrom a difference in the temperature-time characteristics of the RF electrodeor the initial vaporization of Ag into the more energetic excitation region88 20001 600120080040000.0^1.0^2.0^3.0^4.0^5.0^6.0^7.0^8.0Time ( s )Figure 4.3. Temporal response of the Ag atomic emissionsignal at an RF power of 20 W for 0.5 ng of Agdeposited on the furnace wall ( — ) andRF electrode ( - - - ); and background ( • • • ).adjacent to the RF electrode. The tailing edges of the two peaks in Figure 4.3are identical, indicating similar atom dissipation processes.The difference in appearance time between the atomic absorption signalwhen the sample is deposited on the co-axial rod ( solid line, Figure 4.2) andthe atomic emission signal when the sample is deposited on theRF electrode ( dash line, Figure 4.3) is about 1.4 s, indicating that thetemperature of the RF electrode is higher when compared with that without89the plasma. The difference in appearance time between the atomicabsorption signal ( solid line, Figure 4.1) and the atomic emissionsignal ( solid line, Figure 4.3) when the sample is deposited on the furnacewall, is 100 ms and indicates that there is probably no appreciable differencein the temperature of the inside furnace wall with and without a plasma at20W.The effect of the amount of Ag on the emission signal for FAPES isprovided in Figure 4.4. As the amount of Ag increases from 0.05 to 0.5 ng,the peak temperature increases by about 120 K, the appearance temperaturedecreases by about 60 K, and the peak width increases from 400 to 610 ms.However, the peak for 5 ng ( long-dash line ) is shifted to higher peak andappearance temperatures compared with those for 0.05 ng, and its peakwidth is 360 ms. With the exception of the signal for 5 ng Ag, theseobservations are consistent with those observed for Ag in GFAAS [96,971The most significant difference in temporal behavior as a function of theamount of Ag is that the deposition of 5 ng of Ag resulted in a second peak ata later time. However, this peak disappeared when the RF power coupled tothe plasma was increased from 20 to 30 W. The formation of a second peakat a higher deposited amount can be attributed to the second-surfacevaporization of Ag which condenses on the cooler areas of the RF electrode.As pointed out earlier, the RF electrode may not heat uniformly and it islikely that the RF connector end of the electrode is cooler than the rest.-I^,^,^,^,^,0.0^1 .0^2.0^3.0^4.0^5.0Time ( s ),6.0^7.0 8.01 .20.090Figure 4.4. Temporal response of the Ag atomic emissionsignal at an RF power of 20 W for Ag depositedon the furnace wall. Analyte amounts:0.05 ( — ), 0.25 ( - - - ), 0.50 ( - • - • - ), and 5.0 ng( — — — ); and background ( • • • ) at 20 W.The results of a study on the effect of changes in RF power on the emissionsignal for 0.5 ng of Ag deposited on the furnace wall is provided in Figure 4.5.The most significant observations are: (a) the apparent appearancetemperature decreases from 1220 K at 14 W to 940 K at 38 W, (b) the peaktemperature decreases by about 130 K as the RF power is increased, (c) thepeak area increases as the RF power is increased, and (d) the peak widthincreases from 480 ms at 14 W to 670 ms at 38 W.80060040020091 -_-0.0^1.0^2.0^3.0^4.0^5.0^6.0^7.0^8.0Time ( s )Figure 4.5. Temporal response of the Ag atomic emissionsignal for 0.5 ng of Ag deposited on the furnacewall at an RF power of 14 (  ), 20 ( - - - ), 26( - • - • - ), 32 ( — — — ), and 38W ( - ••• - ••• ); andbackground at 20 W ( • • • ).These emission characteristics shown in Figure 4.5 were investigated bysimultaneous measurement of atomic absorption and atomic emissionsignals. The temporal response of the atomic absorption signal for 0.5 ng ofAg, measured at an RF power of 20 ( solid line ) and 40 W ( dash line ), isgiven in Figure 4.6. For clarity, only absorption signals are given. The mostsignificant observation is that not only the emission signal but also theatomic absorption signal is shifted to earlier times when the RF power is920.50.40.30.20.10.00.0^1 .0^2.0^3.0^4.0^5.0^6.0^7.0^8.0Time ( s )Figure 4.6. Temporal response of the Ag atomic absorptionsignal for 0.5 ng of Ag deposited on the furnacewall at an RF power of 20 ( — ) and 40 W ( - - ).increased. In contrast to peak area of the atomic emissionsignal ( Figure 4.5 ), that of the atomic absorption signal ( Figure 4.6)decreases when the RF power is changed from 20 to 40 W. However, peakwidth of the atomic absorption signal increases only marginally ( about50 ms ).The observed shifts in both atomic emission ( Figure 4.5) and atomicabsorption ( Figure 4.6) signals as the RF power is increased may be due toan increased evaporation rate of Ag from the furnace wall when the plasma is93present [98]. It is also possible to have some plasma-assisted heating of thesurface of the furnace wall at high RF powers. However, it should be notedhere that Pb ( Section 4.3.2 ) and Mn ( Section 4.3.3 ) do not show asignificant change in the peak temperature as RF power is increased withinthe same range. The decrease in peak area of the Ag absorption signal isprobably due to an increase in evaporation loss of Ag. However, this effectcannot be seen from the emission signal because of the increase in excitationwhen the RF power is increased.4.3.2 LeadLead has been widely studied and, judging by reports in the literature, hasbeen the subject of some controversy. Double peaks and temporal shifts ofthe absorbance signal for Pb in GFAAS in the presence of samplecontaminants and different gaseous contaminants have been reported bymany authors. These double peaks and temporal shifts have been attributedto different chemical forms [99], condensation and vaporization from thefurnace wall [43,100], effects of chemisorbed oxygen on the furnacewall [101], gas-phase dissociation of the oxide [102], and formation of apyrolytic-graphite layer on non-pyrolytic graphite furnace wall [103].In order to study the effect of the graphite co-axial rod on the atomicabsorption signal for Pb, 5 ng of Pb was deposited on the furnace wall and theresponse was recorded with and without the co-axial rod installed, anexperiment analogous to Ag experiments ( AA at an RF power of 0 W). The1 .4  1 .2 -1 .0 -0.8 -0.6 -0.4 -0.2 -940.0 -0.0^1 .0^2.0^3.0^4.0^5.0^6.0^7.0^8.0Time ( s )Figure 4.7. Temporal response of the Pb atomic absorptionsignal at an RF power of OW for 5 ng of Pbdeposited on the furnace wall with ( ) andwithout ( — ) the co-axial rod; andbackground ( • • • ).results are shown in Figure 4.7. The atomic absorption signal of the sampledeposited on the furnace wall without the co-axial rod ( solid line ) is acomposite of two closely spaced peaks, a smaller peak that appears on theleading edge of the signal that has an appearance temperature of 1080 K, anda much larger peak with a peak temperature of 1400 K. The atomicabsorption signal of sample deposited on the furnace wall with the co-axialrod installed ( dash line ), shows two peaks with an appearance temperature95of 1220 K for the first peak. The first, less intense peak has a peaktemperature of 1400 K and the second, more intense peak has an apparentpea temperature of 1770 K. This, broad, second peak has a width of 500 ms.The response for Pb with the co-axial rod installed is consistent with theprocess involving initial vaporization from the furnace wall, condensation onthe co-axial rod, and subsequent second-surface vaporization from theco-axial rod as it heats with a time lag relative to the furnace wall. Therelative intensity of the early peak to that of the latter peak is quite differentfrom that observed for Ag. This is probably the result of different values ofsticking coefficients for Ag, Pb, and Pb0 on graphite.Currently it is believed that Pb can be produced through either reductionof Pb0 by carbon to form gas-phase Pb and CO, or through directvaporization of Pb0 and subsequent dissociation to form Pb and 02. ForGFAAS, Campbell and Ottaway calculated an appearance temperature of1000-1100 K based on the reduction of Pb0 by carbon, and observed anappearance temperature of 1000 K [104]. The combined thermodynamic andkinetic calculations carried out by Sturgeon et al. showed reduction of Pb0 bycarbon and dissociation of gas-phase Pb2 to form Pb [105]. The experimentalappearance temperature reported by these authors was 1040 K [105].Frech et al. calculated the distribution of Pb as a function of temperature andshowed the existence of both gaseous Pb and Pb0 at temperatures around900 K [94]. Suzuki et al. calculated the atomization energies for Pb in amolybdenum microtube atomizer and suggested gas-phase thermaldissociation of Pb0 to form Pb with an appearance temperature of961270 K [106]. The gas-phase dissociation of Pb0 was also reported byGilchrist et al. [102,107]. The presence of Pb0 and Pb during the atomizationof GFAAS was established by mass spectral studies [108,109]. In a massspectrometric study to determine the dissociation energy of Pb0, gas-phasePb0 was observed at 1000 - 1150 K from the vaporization of Pb0 [110].Regardless of the diverse atom-formation mechanisms suggested, the averageappearance temperature reported by five authors was 1060 K [43].It is possible that the initial peak shown in Figure 4.7, when the sample isdeposited on the furnace wall without the RF electrode ( solid line ), is theresult of Pb formed from Pb0 reduction by carbon and the latter, moreintense peak, results from direct vaporization and subsequent dissociation ofPb0 in the gas-phase. Some supporting evidence for this hypothesis isprovided in Figure 4.8 which provides the results of an atomic absorptionexperiment ( AA at an RF power of 0 W) in which 5 ng of Pb is deposited onboth graphite ( solid line ) and tungsten ( dash line ) co-axial rods. Theapparent appearance time was shifted later in time by about 400 ms ( about115 ms for Ag ) when the sample was deposited on the metal rod relative tothe graphite rod. However, the difference in peak times is about 70 ms( about 60 ms for Ag ). The peak width of the sample deposited on the metalrod ( 360 ms ) is less than that on the graphite rod ( 480 ms ). It should benoted that it is quite probable that the heating rates for the two co-axial rodsare different and the differing peak times are a reflection of this effect.Nevertheless, the signal for deposition on the graphite co-axial rod appears tobe a composite of two peaks, similar to the shape seen in Figure 4.7 when the971 .41 .21 .00.80.60.40.20.00.0^1 .0^2.0^3.0^4.0^5.0^6.0^7.0^8.0Time ( s )Figure 4.8. Temporal response of the Pb atomic absorptionsignal at an RF power of OW for 5 ng of Pbdeposited on the graphite co-axial rod ( — )and tungsten co-axial rod ( - - - ); andbackground ( • • • ).sample is deposited on the furnace wall ( solid line ), although much less welldefined. In contrast, the signal from the metal rod appears to be a singlewell-defined peak. The falling edges of the two peaks are identical.The appearance of a shoulder on the rising edge of the Pb signal from thegraphite surface ( solid line, Figures 4.8) also shows that atomization fromthe graphite surface is due to more than one mechanism. However, themechanism of early Pb release is absent for the tungsten rod ( dash line,98Figure 4.8 ). This observation is consistent with an initial release of Pbthrough reduction of Pb0 by carbon to form the gas-phase Pb and CO. Thelarger peak, obtained from the tungsten rod, corresponds to directvaporization and subsequent dissociation of Pb0 in the gas-phase.The second peak in Figure 4.7 ( with the co-axial rod installed, dash line )has a slowly rising edge compared with the peak from the furnace wallwithout the co-axial rod ( solid line ). This difference in the rising edge showsa difference between initial vaporization from the first-surface ( furnacewall ) and subsequent vaporization from the second-surface ( co-axial rod ). Ifthe initial slow reduction of Pb0 by carbon on the furnace wall at theappearance temperature ( shoulder in the rising edge of the signal ) and thevaporization of Pb0 with subsequent dissociation in the gas-phase ( the peakwith the maximum absorbance ) take place, then second-surface vaporizationfrom the co-axial rod is most probably in the form of Pb0 rather than Pb.Lead oxide vaporized from the co-axial rod should undergo rapid dissociationin the gas-phase because the temperature of the gas-phase at this time ishigh compared with initial vaporization from the furnace wall. However,condensation of Pb0 on the co-axial rod should produce an increase in thesurface coverage compared with initial deposition of the sample on thefurnace wall and a concurrent increase in the graphite-Pb0 interaction. Thisincreased surface coverage and the temperature non-uniformity of theco-axial rod may be the reason for the broad peak width and slowly rising andfalling edges of the Pb signal for the second-surface vaporization from theco-axial rod. Furthermore, the similar shape of the rising and falling edges of99the second peak when the sample is deposited on the furnace wall with theco-axial rod indicates the continued generation of Pb even after the peaktemperature is reached.Figure 4.9 provides the emission signal for 5 ng of Pb deposited on thefurnace wall ( solid line ) and on the RF electrode ( dash line ), and 0.5 ng ofPb deposited on the furnace wall ( dash-dot line ). The signal for 0.5 ng of Pbis plotted at 10 times the actual peak height. Compared with Ag, Pb showsquite complex temporal behavior in FAPES. With respect to the emissionsignal for 0.5 ng of Pb deposited on the furnace wall, the appearance time ismuch earlier than the atomic absorption signal without the co-axialrod ( solid line, Figure 4.7). This is in contrast to the results obtained for Ag.In addition, the emission signals in Figure 4.9 show a close similarity of peaktimes for samples deposited on the furnace wall and on the RF electrode.Another feature of the data shown in Figure 4.9 is that, when 5 ng of Pb isdeposited, a second broad peak is observed between 2.5 and 4 s and this peakappears whether Pb is deposited on the furnace wall or on the RF electrode.It seems that the temperature of the sample deposition site of theRF electrode and that of the furnace wall are similar at this instance, andhence, Pb species can condense on the cooler areas of the RF electroderegardless of the initial sample deposition site. As mentionedearlier ( Section 4.3.1), the RF connector end of the electrode can be coolerthan the rest, even in the presence of the plasma.1 7501 4001 05070035001000.0^1.0^2.0^3.0^4.0^5.0^6.0^7.0^8.0Time ( s )Figure 4.9. Temporal response of the Pb atomic emissionsignal at an RF power of 20 W for 5 ng of Pbdeposited on the furnace wall ( — ) andRF electrode ( - - - ), 0.5 ng of Pb deposited onthe furnace wall ( -•-•-, x 10); and background at20 W(•••).Some supporting evidence for the condensation of Pb species on theRF electrode is given in Figure 4.10, where absorbance is plotted as afunction of RF power. The total absorbance for both peaks ( 1 - 5 s ) does notchange significantly, but the absorbance for the first peak ( 1 - 3 s ) increasesas the RF power increases. The increase in absorbance for the first peak, andhence, a decrease in absorbance for the second peak are due to reducedcondensation Pb0 on the RF electrode when the RF power is increased. A0.50.40.30.20.10.010 15^20^25^30^35Power ( W)40^45101Figure 4.10. Atomic absorbance for 5 ng of Pb deposited onthe furnace wall as a function of RF power: for1-5(•)tand1-3s(•).similar situation cannot be seen from the emission experiment ( seeFigure 4.11 ) because of the increased excitation with increasing RF power.The lowering or increasing appearance temperature of Pb has beenwell-documented in GFAAS literature. McLaren and Wheeler [99] and Imaiand Hayashi [103] reported peak doubling in the presence of 1% ( w/v )ascorbic acid and the appearance of an early peak. In the latter study, nosignificant difference in appearance temperatures was observed whenpyrolytic-graphite-coated graphite was used, but the appearance temperature102decreases from 1130 to 1030 K when 1% ( w/v ) ascorbic acid was present( corresponding peak temperatures are 1510 and 1250 K, respectively ).Salmon et al. [101], and Sturgeon and Berman [111] observed a late timeshift of Pb signal when 02 was added to the sheathing gas. Cendergren et al.reported an early time shift and doubling of Pb signal when CO was used asthe purge gas [112]. A mass spectral and atomic absorption study by Bassand Holcombe showed that the early appearance of Pb from oxygenatedsurface was due to increased CO concentrations in the gas-phase [113]. Bassand Holcombe attributed the effect of CO to the heterogeneous equilibrium ofPb0 on the surface and the gas-phase Pb, CO and CO2 [113]. A study byGilchrist et al. showed appreciable amounts of CO when ascorbic acid wasused with slow heating rates [114]. In another study, Gilchrist et al. reportedan early time shift of signal by adding H2 and CO to the purge gas [102].They reported a decrease in appearance temperature of Pb from 1110 to840 K when 2.1 % ( v/v ) CO was added to the purge gas. This was attributedto the gas-phase equilibrium involving dissociation of Pb0 to form Pb and 02.In summary, these GFAAS studies show that the atomization of Pb isstrongly affected by the gas-phase composition, especially by the amountof CO in the gas-phase.It is quite possible that the presence of higher amounts of CO in theplasma shifts the appearance temperature of Pb in FAPES ( Figure 4.9)compared with that without the plasma ( Figure 4.7). At present, calculatedor experimentally determined partial pressures of CO and 02 have not beenreported for FAPES. However, Sturgeon et al. reported the presence of CO in103the RF plasma source at 13.56 MHz [78]. In addition, CO + emission isobserved for the 13.56 MHz plasma in FAPES ( Section 3.3.1).Time-resolved studies during a dry atomization step showed an increase inemission intensity for CO + in FAPES ( Section 3.3.3). This increase in CO+intensity during the atomization step is due to increased desorption of COfrom the furnace wall and from the RF electrode. In FAPES, loss of carbonfrom the RF electrode is observed and this loss is an additional source ofcarbon and CO which is absent in GFAAS. Furthermore, presence of NaNO3( effect of 02 concentration, Section 6.3.3 ) and ascorbic acid ( effect of CO andCO2 concentrations, Section 6.3.5 ) shows pronounced effects on the emissionsignal for Pb in FAPES.Carbon monoxide has been shown to affect the atomization of Pb because itis involved in a variety of homogeneous and heterogeneous equilibriaincluding;2Pb0(0 ■^ 2Pb(0^+ 02^(4.1),02 +2C0 ^ 2CO2 (4.2),Pb0 (5) + C(s) Pb(g) + CO(g) (4.3),and Pb0(s) + CO(g) ., Pko + C 02^(4.4).The early release of Pb ( Figure 4.9) can be due to the combined effects ofequilibria (4.1) and (4.2). The presence of CO in the plasma could act as asink for 02 ( equilibrium 4 2) such that the equilibrium (4.1) is shifted to the104right leading to an earlier appearance temperature for Pb when the plasma ispresent. The equilibrium (4.3) indicates that the initial generation of Pb bythe reduction of Pb0 by carbon ( in atomic absorption experiments withoutthe plasma ) should be shifted to a later appearance temperature as CO inthe gas-phase increases.Sturgeon and Willie reported a coincident appearance of Pb0 emission ( at286.62 nm ) and Pb ( at 217.0 rim), when 1 gg of Pb ( as nitrate in 1 % v/vHNO3 ) was deposited on the furnace wall at an RF power of 100 W [115].The atomization temperature was 1600 K with a heating rate of about0.2 K ms-1. This simultaneous appearance of Pb0 and Pb is in contrast tothe graphite furnace - mass spectral studies where Pb0+ signal precedes thePb+ signal [108,109]. However, for FAPES, the simultaneous appearance ofPb and Pb0 is consistent with the experimental observations provided in thisthesis.Figure 4.11 gives the effect of the RF power on the emission signal from5 ng of Pb deposited on the furnace wall. The appearance temperature of Pbwhen the sample is deposited on the furnace wall at 14 W is 1010 K and at40 W is 770 K, although an additional early peak can be seen at RF powersabove 30 W. The cause of this early peak is not known at this time; however,it could arise from the reduction of Pb0 by CO according to theequilibrium (4.4). The temporal response of the emission signal of Pb inFAPES is complex because at least four peaks are observed at highRF powers, indicating a number of atomization or vaporization mechanisms.1051 000 800 -600 -400 -200 -0.0^1.0^2.0^3.0^4.0^5.0^6.0^7.0^8.0Time ( s )Figure 4.11. Temporal response of the Pb atomic emissionsignal for 5 ng of Pb deposited on the furnacewall at an RF power of 14 (^), 20 ( - - - ), 30( - • - • - ), and 40 W (^); and backgroundat2OW(•••).Compared with Ag, there is no significant shift in peak temperature of themajor peak as the RF power is increased from 20 to 40 W. This offers supportfor the notion that, for FAPES, the atomization of Pb is controlled bygas-phase chemistry rather than direct thermal desorption.The effect of the excitation process on the temporal response of theemission signal is studied by simultaneously measuring atomic emission andatomic absorption signals. Figure 4.12 gives the results of this experiment at- 0.4>- 0.3^c",)'-0,a-a0- 0.2 0CD1061 0008006004000.5- 0.1i\• ^a''''s .^is\^i^.^1 0.03.0^4.0^5.0^6.0^7.0^8.0Time ( s )Figure 4.12. Temporal response of the Pb signal for 5 ng ofPb deposited on the furnace wall at anRF power of 40W: emission ( ) and ( . . . )absorbance.an RF power of 40 W. Similar to Ag, the temporal response of the emissionsignal is similar to that of the absorbance signal. This similarity in temporalresponse of both absorption and emission is most likely due to the dominanceof atomization and vaporization characteristics of Pb in the plasma. Thesimilar temporal responses also show no significant effect of plasmaexcitation process on the emission signal for Pb in FAPES.1074.3.3 ManganeseThe results of an atomic absorption ( AA at an RF power of 0 W) study inwhich 1.25 ng of Mn was deposited on the furnace wall with ( dash line ) andwithout ( solid line ) the co-axial rod installed are provided in Figure 4.13.The appearance temperature of the signal for the sample deposited on thefurnace wall without the co-axial rod is 1470 K and the peak temperature is1700 K. This absorption signal has a sharp rising edge compared with thefalling edge, and has a peak width of 800 ms. The appearance temperature ofthe signal with the co-axial rod installed is 1380 K with the two peak maximaat 1560 and 1950 K. These results are similar to those for Ag and Pb; i.e.,with the co-axial rod, condensation and subsequent second-surfacevaporization from the co-axial rod are observed.The results of an atomic absorption ( AA at an RF power of 0 W)experiment in which 1.25 ng of Mn was deposited on graphite ( solid line )and tungsten ( dash line ) rods are given in Figure 4.14. There is a noticeabledifference between appearance times and peak absorbance for Mn depositedon the two different materials. The signal from the sample deposited on thegraphite rod is an overlap of two closely spaced peaks. When the sample isdeposited on the tungsten rod, the apparent appearance temperature isobserved at a higher value. Although these results suggest that thepredominant mechanism for the formation of Mn in the graphite furnace isMnO reduction by carbon rather than MnO vaporization and thermaldissociation, some differences between the two traces should be expected as aresult of the difference in heating rates and thermal conductivity of the1081 .21 .00.80.60.40.20.00.0^1 .0^2.0^3.0^4.0^5.0^6.0^7.0^8.0Time ( s )Figure 4.13. Temporal response of the Mn atomicabsorption signal at an RF power of 0 W for1.25 ng of Mn deposited on the furnace wallwith ( - - - ), and without ( ) the co-axialrod; and background ( • • • ).tungsten vs the graphite rods. It should also be noted that the bond energy ofMnO is 4.1 eV ( that of Pb0 is 3.8 eV ) [81].The atomization mechanisms for Mn in GFAAS have been studied in a fewoccasions. A study by Smets has suggested that carbon reduction of MnO asthe mechanism of atom-formation in GFAAS [116]. In contrast, Aggett andSprott studied Mn atom-formation on a tungsten strip and suggested thatthermal dissociation of the oxide is the atom-formation mechanism [117].1091 .81 .51 .20.90.60.30.00.0^1 .0^2.0^3.0^4.0^5.0^6.0^7.0^8.0Time ( s )Figure 4.14. Temporal response of the Mn atomicabsorption signal at an RF power of 0 W for1.25 ng of Mn deposited on the graphite co-axialrod ( - ), tungsten co-axial rod ( - - - ); andbackground ( • • • ).Frech et al. proposed the thermal dissociation of MnO as the mechanism ofatom-formation [94]. McNally and Holcombe suggested that Mn was formedvia desorption of molecular aggregates of MnO from the surface of thegraphite furnace [118].The most interesting observation in the data provided in Figure 4.14 is,when the sample is deposited on the tungsten rod, the absorption peakmaximum appears 200 ms after the furnace wall reached its maximum110temperature. This observation may be attributed to a continued increase inthe temperature of the tungsten rod even after the furnace wall reached itsmaximum temperature, or a time lag between initial vaporization of MnOfrom the tungsten rod and thermal dissociation in the gas-phase.The results of a FAPES study in which 1.25 ng of Mn was deposited on thefurnace wall ( solid line ) and on the RF electrode ( dash line ) at 20 W isgiven in Figure 4.15. The appearance temperature of the emission signalfrom the sample deposited on the furnace wall is 1340 K and on theRF electrode ( apparent appearance temperature ) is 1420 K. The emissionsignal for the sample deposited on the furnace wall is a composite of twopeaks. The peak temperature of the sample deposited on the furnace wall is1920 K and that of sample deposited on the RF electrode is 1870 K. Thedelayed appearance of Mn signal when the sample is deposited on theRF electrode, indicates a temperature lag in the sample deposition site of theRF electrode compared with the furnace wall. Furthermore, compared withthe atomic absorption signal for the sample deposited on the furnacewall ( solid line, Figure 4.13), atomic emission signals have a sharp fallingedge and a narrower peak width of about 650 ms ( Figure 4.15).11120001 6001 20080040000.0^1.0^2.0^3.0^4.0^5.0^6.0^7.0^8.0Time ( s )Figure 4.15. Temporal response of the Mn atomic emissionsignal at an RF power of 20 W for 1.25 ng of Mndeposited on the furnace wall ( — ) andRF electrode ( - - - ); and background at 20 W( • • • ).It is apparent from the comparison of the data in Figures 4.13 and 4.15that Mn condenses on the RF electrode when deposited on the furnace walland then vaporizes as the electrode heats radiatively. Because of thiscondensation, major peaks shown in Figure 4.15 ( solid line and dash line )originate from the RF electrode regardless of the sample deposition site.These results suggest that after 3 s into the atomization step, thetemperature of the RF electrode lags compared with that of the furnace wall.112However, this "time limit" most likely depends on the heating rate of thegraphite furnace. The narrow peak width of the emission signals is likelydue to Mn vaporization from the RF electrode at a instance when thegas-phase and furnace wall temperatures are high compared with that of theRF electrode. In atomic absorption experiments, when the sample isdeposited on the furnace wall, analyte vaporizes into a gas-phase where thetemperature lags compared with that of the furnace wall. The difference inthe rising edge of the atomic absorption signal for the sample deposited onthe graphite rod ( solid line, Figure 4.14) compared with that for the sampledeposited on the RF electrode at 20 W ( dash line, Figure 4.15) is likely dueto the difference in the heating characteristics of the graphite rod with andwithout the plasma. The falling edge of the emission signals has similarshapes ( Figure 4.15) indicating similar atom dissipation processes from thegraphite furnace because both peaks are from Mn vaporized from theRF electrode regardless of the sample deposition site.The results of a FAPES study in which the RF power is increased from14 to 40 W for 1.25 ng of Mn deposited on the furnace wall are provided inFigure 4.16. The effect of RF power on the sample deposited on the furnacewall shows that the maximum peak intensity ( and peak area ) of the mainpeak increases about five times when power changes from 14 to 40 W and asmall pre-peak is observed that also increases in intensity in proportion tothe larger peak. There is a small shift in the peak temperature to a lowervalue as the RF power is increased. The increase in intensity for both first1 2001 0008006004002001130.0^1.0^2.0^3.0^4.0^5.0^6.0^7.0^8.0Time ( s )Figure 4.16. Temporal response of the Mn atomic emissionsignal for 1.25 ng of Mn deposited on thefurnace wall at an RF power of 14 ( ), 20( - - ), 30 - • - • - ), and 40W ( — — — ); andbackground at 20 W ( . . . ).and second peaks also indicates no significant effect of RF power on thetemperature of the RF electrode after 3 s into the atomization step.Some supporting evidence for the effect of RF power on the temperature ofthe RF electrode can be seen from the atomic absorption measured as afunction of RF power. The results of this experiment are provided inFigure 4.17. The integrated areas for both the composite peak ( 2.8 to 5.5 s)and the second peak ( 3.5 to 5.5 s ) decrease with increasing RF power. If the1.00.80.60.40.20.010 15^20^25^30^35Power ( W)40^45114Figure 4.17. Atomic absorbance for 1.25 ng of Mn depositedon the furnace wall as a function of RF power:for 2.8 - 5.5 ( • ), and 3.5 • 5.5 s ( • ).RF power affects the temperature of the RF electrode after 3 s into theatomization step, then the intensity of the first peak should increase and theintensity of the second peak should decrease ( a similar situation to Pb ).Figure 4.18 depicts simultaneously measured atomic emission andabsorption signals for Mn deposited on the furnace wall at an RF power of40 W. As observed for Ag and Pb, the temporal response of the emissionsignal is similar to that of the absorption signal. This similarity in theemission and absorption signals can probably be attributed to the dominance1 .00.8_1 200  1 000 -800 -600 -400 -200 -o i-1150.0^1.0^2.0^3.0^4.0^5.0^b.0^7.0^8.0Time ( s )Figure 4.18. Temporal response of the Mn signal for 1.25 ngof Mn deposited on the furnace wall at anRF power of 40 W: emission (  ) andabsorbance ( - - - ).of atomization and vaporization characteristics of Mn on the temporalresponse of emission signal rather than excitation characteristics. However,the peak time of the second more intense peak of the emission signal showsan early shift compared with that of simultaneously measured atomicabsorption signal ( both peaks originate from the RF electrode ). This shift inthe peak time of Mn emission signal may be due to the early appearance ofMn atoms in the more energetic excitation ( temperature ) volume adjacent tothe RF electrode relative to the absorption volume ( also see Section 5.3.1).1164.4 SUMMARYThe temporal response of the analyte emission signal in FAPES has beenstudied for Ag, Pb, and Mn by using both atomic absorption and atomicemission spectroscopy. The analyte sample was deposited on the furnacewall and on the RF electrode, and atomic emission and absorption profiles arecompared with and without the RF plasma.Without the plasma inside the graphite furnace, the co-axial rod is heatedby radiation emitted from the furnace wall during the high temperatureatomization step. Because of this radiative-heating, the temperature of theco-axial rod lags compared with that of the furnace wall. Therefore, withoutthe plasma inside the graphite furnace, the co-axial rod acts as acondensation site for atoms and oxides that subsequently vaporize( second-surface vaporization ) at higher furnace temperatures. With theplasma inside the furnace, the co-axial rod ( RF electrode ) is additionallyheated by contact with the plasma. The temperature-time characteristics ofthe RF electrode with the plasma are different from those without theplasma. The RF electrode does not heat uniformly and it is likely that theRF connector acts as a heat-sink.With the plasma inside the graphite furnace, the RF electrode heats suchthat for Ag and Pb, condensation is not observed except at high analyteamounts and low RF powers. For Ag and Mn, the appearance temperaturesand peak temperatures are not significantly affected by the plasma relative117to the response obtained in atomic absorption experiments without theplasma. However, for Pb, the appearance and peak temperatures are shiftedsignificantly earlier. This early appearance of Pb in the RF plasma is likelydue to the presence of CO in the gas-phase, generated from oxidation ofgraphite on the furnace wall and RF electrode.During the later stages into the atomization step, it appears that thetemperature of most parts of the RF electrode lags compared with that of thefurnace wall. It is during this time that Mn, which has a relatively highappearance temperature, vaporizes from the furnace wall. This temperaturelag results in Mn condensation on the RF electrode and second-surfacevaporization. The results obtained for Ag, Pb, and Mn suggest analytecondensation on the RF electrode which results in the appearance of twopeaks, the more intense peak determined by the RF power. At present, theeffect of the heating rate of the graphite furnace on the temporal response ofthe emission signal in FAPES is not known. However, it is likely that theheating rate of the graphite furnace affects the temperature-timecharacteristics of the RF electrode, and hence, analyte condensation andsecond-surface vaporization characteristics from the RF electrode.The temporal response of the simultaneously measured atomic absorptionand atomic emission signals shows similar leading and falling edges for bothAg and Pb. These results suggest that the temporal response of theseelements in FAPES is determined by the atomization and vaporizationcharacteristics of the analytes rather than by the excitation characteristics.118CHAPTER 5EMISSION CHARACTERISTICS AND FIGURES OF MERITFOR LEAD IN FAPES5.1 INTRODUCTIONAnalytical figures of merit such as (1) the sensitivity of the method, whichcorresponds to the slope of the calibration graph, (2) precision ( relativestandard deviation: RSD ), which is the reciprocal signal-to-noiseratio ( SNR ), and (3) the detection limit ( DL), which is the lowest amount( or concentration ) of analyte which gives a signal ( Xdi ) significantlydifferent from the background signal ( Xbi ), determine the success of ananalytical method. The DL can be expressed by:DL =kablS(5.1),S which relates the detection limit to the standard deviation of thebackground ( a) and sensitivity ( 8). A value of k = 3 is recommendedfor 99.86 % confidence level. Noise in the background is normally taken to bethe standard deviation of the background. Therefore, both the intensity ofthe analyte signal and noise in the measurement ( i.e., in both analyte and119background signals ) affect the sensitivity, precision, and the DL for ananalyte in a spectrochemical determination. The DL is sometimes stated asthe analyte concentration which provides a SNR of 3. This definition for theDL should be interpreted with caution because it assumes that the noise inthe background and analyte signal are the same at low analyte signal levels.Furthermore, DLs are not measured directly, but rather are extrapolatedfrom measurements made at higher concentrations, where the highersignal-to-background ratio ( SBR ) permits easier measurement of theanalyte signal.The intensity of the analyte emission signal depends on the concentrationof analyte, particular electronic transition, the excitation temperature, andhence, on the RF power applied to the plasma. On the other hand, noise inthe measurement is due to the (1) shot-noise, which is proportional to thesquare root of the signal intensity (2) flicker noise, which is proportional tothe signal intensity, and (3) detector noise, which is independent of the signalintensity [119]. Shot-noise is due to the random arrival of photons at thedetector whereas flicker noise is the low frequency fluctuations and drift.The major sources of flicker noise involve random drift of light sources,analyte atom-formation, and detection. However, the cause of flicker noise isnot known [119]. Detector noise is due to the random thermal motion ofcharge carriers. Both shot and detector noise are distributed at allfrequencies, and are referred to as "white noise ". For a shot-noise limitedsignal, SNR is proportional to the square root of the signal intensity, whereasfor a flicker noise limited signal, SNR is independent of the signal intensity.120A more detailed review of SNR in analytical spectrometry is availableelsewhere [120].The dominant noise in an atomic absorption signal, in the analytical_ _range ( absorbance between 0.2 and 1) in GFAAS, comes from the analyteabsorption fluctuations ( flicker or drift ) [120]. However, for graphitefurnace - plasma methods, such as FAPES and PANES, additional sources ofnoise may be present in the signal. These additional sources of noise in thesignal are the plasma background flicker and shot-noise that arise during theatomization step and the analyte emission flicker and shot-noise that ariseduring the excitation. At the analytical wavelength, there may be astructured background by emission from molecular species. In theatmospheric pressure plasma source in FAPES, about 80 % of theanalytically useful wavelength region ( 200 - 500 nm ) is dominated byemission from molecular species such as CO+, N2, and OH. Variations of theexcitation temperature during the analyte excitation and condensation of theanalyte on the RF electrode may also contribute to a poor precision in themeasurement.For FAPES, effects of the RF power on the emission characteristics of theEmalyte or on the background have not been studied. Smith et al. employed aplasma at an RF power of 20 -30 W [17], whereas Sturgeon et al. employedone at 50 W [78], during the determination of figures of merit for FAPES.Sturgeon et al. reported a highly structured background in the vicinity of theanalytical lines: NO band near lines for Cd, Fe, and Be; CO bands near Ni;121and OH bands near Bi and Pb [78]. Sturgeon et al. also reported an optimumSBR for Cd ( 228.2 nm ) at 50 W and for Pb ( 217.0 am) at 75 W [121], but adetailed study has not been reported. For HC-FANES, Falk et al. reportedthat the limiting noise, and consequently, the DLs, were determined not bythe recombination continua but by extraneous background molecularemission bands [13]. For HA-FANES, Riby et al. reported emission from CNand NH in the plasma background, and employed wavelength modulation tocorrect the background at 357.9 nm for the determination of Cr [90]. Hollowcathode-FANES and HA-FANES are normally operated at forward powersbetween 20 and 60 W.In an effort to understand the emission characteristics of analytes inFAPES, the effect of RF power on the analyte emission intensity and noise inthe measurement was investigated. Lead was examined as the test analytein this study. Lead is one of the few elements which requires detection ofultra trace levels for the analysis of biological and environmental samples,and appears to be the most widely determined element by GFAAS. Thetime-resolved Pb-excitation temperature was also measured as a function ofRF power, to study the effect of analyte excitation process on the temporalresponse of the Pb emission signal.5.2 EXPERIMENTALA complete description of the experimental system employed to acquireatomic emission signals is given in Chapter 2 of this thesis. This122experimental system and a vertical-mount for the RF electrode wereemployed. The following experimental methods and parameters should benoted.5.2.1 Signal PresentationFour replicate measurements were averaged and subjected to a 25-pointSavitzky-Golay smoothing procedure [91]. Analytical parameters, such aspeak height, peak area, and peak width, were evaluated for each sampledeposition.5.2.2 ReagentsAll Pb solutions in 1 % ( v/v ) HNO3 were prepared from serial dilution of a1000 mg L-1 stock solution prior to analysis. Lead and HNO3 solutions wereprepared by dissolving analytical grade Pb(NO3)2 and HNO3 ( both, BDHToronto, Canada ).5.2.3 ProcedureA 5 gl, aliquot of solution was deposited onto the furnace side wall with anEppendorf 0.5 - 10 ilL UltraMicro pipette. The plasma source work-head waspurged with helium ( Union Carbide, Toronto, Canada ). The furnacetemperature was increased to 400 K for 60 s to dry, and was maintained at650 K for another 20 s to ash the sample. This total time interval is123sufficient to exclude the water vapor inside the plasma source work-headbefore the plasma ignition. Within the next 10 s lag time, the plasma wasignited. After this lag time, the furnace temperature was ramped to 2000 Kin 5 s for all determinations. The blank determinations were carried out bydepositing the same volume of 1 % ( v/v ) HNO3 solution. After each sampleatomization step, a dry atomization step ( without deposition ) was carriedout to clean the graphite furnace.Four replicate measurements were carried out for each determination andintegrated for 3 s: from 0.5 to 3.5 s into the atomization step ( unlessotherwise noted ).All determinations were carried out by using an atomic resonance line ofPb ( 283.30 rim) and a spectral bandwidth of 0.06 nm ( unless otherwisenoted ).5.3 RESULTS AND DISCUSSION5.3.1 Time-resolved Pb-excitation temperatureThe time-resolved Pb-excitation temperature was evaluated by using the twoline method ( Section 1.4.5). The two lines of Pb ( I ) used in the excitationtemperature measurement were 280.1 and 283.3 rim. Spectralcharacteristics of these two Pb lines are given in Table 5.1. Transition124Table 5.1. Wavelengths, Excitation energies, and Spectralcharacteristics for the two Pb lines used for thePb-excitation temperature determination [122].Wavelength ( nm )^E ( eV)^gpAp^)^280.19^5.7439^4.3 x 109^283.30 4.3749 1.8 x 108probabilities were obtained from Corliss and Bozman [122]. At eachwavelength, four replicates of 0.5 ng of Pb in 1 % HNO3 ( four replicates of1 % HNO3 for blank ) were measured and the averaged temporal emissionprofile was calculated. From this averaged temporal emission profile, thetime-resolved Pb-excitation temperature was evaluated as a function ofRF power.Figure 5.1 depicts the time-resolved Pb-excitation temperature ( solid line )during the atomization of Pb at an RF power of 50 W. The variation of thewall temperature of the graphite furnace ( dash line ) is also given. Thepeak-to-peak averaged Pb-excitation temperature for a 50 Wplasma ( Figure 5.1) is 4560 ± 100 K ( from 1 - 3.5 s). At 20 and 100 W, thePb-excitation temperatures are 3990 ± 80 K ( from 1 - 2 s), and 4840 ± 110 K( from 1 - 2.5 s), respectively. The calculated Pb-excitation temperature was1 8001 600a=—101 400 D-00a'1 200 CCD-I...--■1000 ■_.....1.5 3.08003560005400480042003600300010 2.0^2.5Time ( s )125Figure 5.1. Time-resolved Pb-excitation temperature( — ) at an FtF power of 50W and thecorresponding graphite furnace temperature( - - - ) during the atomization step of Pb.limited to a specific time duration due to the transient nature of Pb emissionintensity and the background correction error.The time-resolved Pb-excitation temperature provided in Figure 5.1 showssome random rather than systematic fluctuations within the precision of themeasurement. This time-resolved temperature measurement also shows thatthe Pb-excitation temperatures are not affected by the thermionic electronswhen the furnace temperature is below 1800 K at 50 W, because the126thermionic electron concentration is too low to be significant for the furnacetemperature range in which Pb is present in the plasma. However, at anRF power of 100 W, the reflected power starts to increase just after theemission signal of Pb reaches that of the background; after 3.5 s into theatomization step. This increase in the reflected power is probably due to asignificant increase in the evolution of thermionic electrons from theRF electrode at higher RF powers.Sturgeon et al. also reported that the He-excitation temperature was notinfluenced by the furnace wall temperatures up to about 2700 K [80]. Themeasured He-excitation temperature was about 3450 K at 50 W and 3650 Kat 100 W. For the measurement of He-excitation temperatures, the higherfurnace temperature range, 1670 - 2650 K, was accessed by recording thetransient response of helium ( I ) lines with a preset upper temperature, andthe lower furnace temperature range, 300 - 1330 K, was accessed understeady state conditions. For FAPES, Sturgeon et al. used an atomizationtemperature of 2700 K [16,78,79,123]. They also reported an increase in thereflected power with increasing furnace temperatures above1800 K [16,79,123].To date, for FANES sources, neither the time-resolved nor the peakexcitation temperatures for analytes have been reported. For HC-FANES,Falk et al. also reported no significant change in the helium or Ar excitationtemperatures although the electron density decreases significantly when thefurnace temperature exceeds 1700 K limit [124]. However, Falk et al.127reported a significant decrease in the cathode voltage ( about 85 % of theinitial voltage of 300 V) at a constant current when the cathode ( graphitefurnace ) temperature was increased above 1700 K [124]. This change in theelectrical characteristics has been attributed to the thermionic electrons fromthe graphite furnace wall. For HA-FANES, the maximum atomizationtemperature reported is 2200 K. At furnace temperatures above 1800 K,changes in the current-voltage characteristics due to the evolution ofthermionic electrons have been reported [125].With reference to the previously observed shifts in the absorption andemission maxima ( Section 4.3 ), a brief discussion is warranted. For anatomic absorption measurement, absorbance is proportional to the number offree atoms: N, in the gas-phase at a given instance [43,126], hence, themaximum absorption ( Amax ) at Nmax ( maximum number of free atoms inthe gas-phase ). In the case of simultaneous atomic absorption and atomicemission measurements, the maximum emission ( Imax ) appears earlycompared with Amax. This early shift indicates that Imax occurs at a timewhen the number of free atoms in the gas-phase is increasing. However, ifthe excitation temperature ( Tex e) is decreasing at this time, then theemission intensity can decrease because the emission intensity isproportional to Np: the number of excited atoms in the gas-phase, andhence, the Boltzmann factor: CE/kTexe ( Section 1.4.5).The early shift observed for the Mn emission maximum is 200 ms relativeto the absorption maximum ( Figure 4.18). This observation is in contrast to128the pronounced late shifts in 'max from Amax reported for CFAES [10,127],where thermal excitation of the analyte is solely dependent on the graphitefurnace temperature. Although the observed shift for Mn may be explainedby a decrease in Mn-excitation temperature while the furnace temperature isincreasing, both emission and absorption profiles show similar falling edges.Therefore, possible changes in Mn-excitation temperature do not explain theearly shift of peak emission relative to peak absorption. It should be notedhere that Mn atomizes at a medium graphite furnace temperature.However, at higher furnace temperatures, the effect of thermionicelectrons may be significant. The excitation characteristics may change dueto the therrnionic electrons emitted from the RF electrode ( and from thefurnace wall ). However, it is difficult to evaluate the effect of thermionicelectrons at higher furnace temperatures on the excitation temperature,because the thermionic electrons can also affect the plasma impedance, andhence, the net energy coupled to the plasma.5.3.2 Effect of RF powerFigure 5.2 depicts the net emission intensity of 0.5 ng of Pb as a function ofRF power. Also plotted in Figure 5.2 is the Boltzmann factor: e-FikTexe ( dashline ), for Pb as a function of RF power. The Pb-excitation temperatures at20, 50, and 100 W were used to calculate the Boltzmann factor. AlthoughBoltzmann distribution predicts an exponential increase in emission250020001 5001 00050001290^20^40^60^80^100Power ( W)5E-54E-5co0,=--,3E-5 0m=-1-12E-5 0°,0--,1E-5Figure 5.2. Effect of the RF power on the net emissionintensity of Pb ( • ) and on the Boltzmannfactor ( - - - ).intensity with increasing RF power, the emission intensity for Pb increasesup to about 50 W and then decreases. This decrease in emission intensity forPb may be due to losses during the pre-atomization or atomization steps,changes in the excitation characteristics, or an increase in ionization of Pbwith increasing RF power.The pre-atomization and atomization losses in the graphite furnacemethods are due to the high volatility of the analyte molecular species, and aslow heating rate of the graphite furnace coupled with a lower atomization130temperature. For GFAAS, it has been shown that Pb can be ashed up toabout 800 K without loss of analyte [128,129]. However, during the slowatomization step employed in the experiments described in this chapter, somePb may be lost as undissociated Pb0 before the gas-phase temperature ishigh enough. The atomization losses may also be significant due to theincreased diffusion with increasing temperature at high RF powers.RF sputtering may also cause a loss of analyte from the furnace wall prior toatomization. However, Sturgeon et al. reported no measurable decrease inthe Ag emission signal even during an extended 2 min period of plasmaoperation prior to the atomization step [79].The Pb-excitation temperature shows an increase with increasingRF power ( Section 5.3.1), and therefore, the decrease in the Pb emissionsignal intensity at high RF powers cannot be due to a change in theexcitation temperature. However, it is possible that the excitationcharacteristics are affected by the presence of thermionic electrons at highRF powers. The possible changes in the excitation characteristics aredecreased collisional excitation rate ( as a consequence of changes in theelectron number density for excitation in the plasma ) and increasedcollisional de-excitation rate ( due to the low energy thermionic electrons ).Although no changes in the reflected power are observed before 3.5 s into theatomization step, some temporal characteristics of ionic species such as CO+and N2+ are attributed to the evolution of thermionic electrons( Section 3.3.3). Furthermore, the He-excitation temperature measurementfor FAPES by Sturgeon et al. [80], and Ar and He-excitation temperature131measurement for HC-FANES by Falk et al. [124], show that the thermionicelectrons do not affect the excitation temperature at higher furnacetemperatures; however, they can influence the electron density of theplasma [124]. Sturgeon et al. have not reported the variation of electrondensity with furnace wall temperature for FAPES, because of the increase inreflected powers above 1800 K [80].For plasma spectrometric methods, analyte ionization may also besignificant. When the RF power coupled to the plasma is increased, it notonly increases the analyte excitation, and hence, increases the emission butalso increases the analyte ionization. Compared with high temperatureplasma sources like ICPs ( with the excitation temperature ranging from4000 - 7000 K), the low pressure d. c. plasma sources in FANES ( with theexcitation temperature ranging from 1000 - 3000 K) are considered to be lowionization plasma sources [13]. However, for the RF plasma source inFAPES, the excitation temperature ranges from 3000 - 5000 K, depending onthe RF power and the thermometric species used for the measurement( Sections 3.3.4 and 5.3.1). Evidence for ionization of Ag in FAPES isdiscussed in Section 6.3.4. Silver with an ionization potential ( IP) of7.5 eV ( that of Pb is 7.4 eV ), shows an ionization suppression when a highamount of Na ( IP is 5.1 eV) is present during the atomization. Analyteionization may be significant at high RF powers, and hence, a source for thesuppression of atomic emission signal intensity in FAPES.132The Pb-excitation temperature for the helium RF plasma source in FAPESis similar to that reported for Ar-ICP operating at an RF power of1.5 kW [130]. Similar excitation temperatures for FAPES and ICP can beattributed to the fact that radiative and convective heat losses from theFAPES source are much lower than those from an ICP, or to the fact that theelectric field strength is much higher in the FAPES source ( Section 3.3.4).For the ICP, Blades and Horlick reported norm temperatures ( temperatureat which the maximum in emission intensity occurs ) between 4000 and5000 K for the emission lines with excitation potentials between 4 and 5 eV,respectively [131]. At temperatures higher than the norm temperature,emission intensity decreases as the ionization become significant.Nevertheless, the exact cause of the decrease in the Pb emission signal athigh RF powers ( Figure 5.2) is not known at present, and needs furtherstudy.Sturgeon et al. have compared the relative intensities for a number of ioniclines at an RF power of 50 W with those of ICP, and suggested that thedegree of analyte ionization was significantly lower in the FAPES source [78].The ionization temperature reported by Sturgeon et al. for FAPES rangedfrom 4800 - 5900 K at an RF power of 100 W, depending on the thermometricspecies [80], compared with the ionization temperature for ICP which rangedfrom 6700 - 8000 K for RF powers between 1 and 1.5 kW [68]. However,Sturgeon et al. acknowledged that about 85 % of the applied RF power wasdissipated in the tuning network for their experimental system, with the133result that the plasma was apparently operating on only a few watts of realpower [79].For HC-FANES, Falk et al. reported an increase in the emission intensityfor Ni, Cr, Cu, Fe, Co, and Al when the discharge current is increased from20 to 60 mA [124]. However, the power coupled to the HC-FANES source isonly 18 W at a discharge current of 60 mA. For HA-FANES, Hamly et al.reported a constant emission intensity for Cu and Cd above a thresholdcurrent ( 50 and 20 mA for Cu and Cd, respectively ) when the dischargecurrent was changed from 10 to 80 mA, but the reason for these emissioncharacteristics was not known [125].5.3.3 Signal-to -Noise ratioFigure 5.3 depicts the temporal response of the Pb emission signal for 0.5 ngof Pb ( solid line ) and that of the background ( dot line ) at an RF power of20 W. For the study of SNR and SBR, the standard deviation of fourreplicates of analyte emission signal is assumed to be equal to the noise inthe " net " signal. The propagation of error method shows that, for theaddition of two independent quantities, the total variance is the sum of thetwo variances. However, the analyte flicker noise and the blank flicker noiseare not an exact addition because analyte flicker noise occurs during thesample and not during the blank [119]. But, at high SBRs, that the totalnoise in the measurement is equal to the noise in the signal, is a goodapproximation. It should be noted that these SNRs are overestimates of the1 5001 20090060030001340.0^1.0^2.0^3.0^4.0^5.0^6.0^7.0^8.0Time (s)Figure 5.3. Temporal response of the Pb emission signal atan RF power of 20 W for 0.5 ng of Pb (  );and background ( true SNRs, and the overestimation depends on the number of measurementsbecause the standard deviation is a negatively biased estimate of thevariance [132].It was noted that the best SNR was obtained by peak area measurementcompared with peak height measurement because the magnitude of the peakarea was high compared with that of the peak height. Furthermore, thestandard deviation of the peak height measurement was low compared with135that of peak area measurement. These observations are most likely due tothe slow heating rate of the graphite furnace.For GFAAS, the effect of heating rate on the peak height or the peak areahas been reported on a number of occasions. Torsi and Tessari reported anincrease in peak height with increasing heating rate [133]. Gregoire et al.reported an increase in peak height and an exponential decrease in peak areafor relatively volatile elements, but an exponential increase in peak area forrelatively non-volatile elements when the heating rate was increased [134].Zhou et al. reported, for Ag and Pb, a decrease in peak area and an increasein peak height when the heating rate was increased under isothermalconditions [135]. However, a constant peak area and an increased peakheight were observed when the heating rate was increased undernon-isothermal conditions [135]. Falk suggested an increase in peak heightand a decease in peak area when the heating rate was increased [136]. Insummary, it appears that the effect of increasing heating rate is to increasethe peak height at the expense of the peak area. The effect of heating rate onpeak height or peak area may also depend on the nature of the analyte.Although it has been widely believed that improved precision and linearityof the calibration graph can be obtained by peak integration, experimental ortheoretical studies on the merits of peak height vs peak area measurementshave been limited. The main reason for this limited number of studies havebeen the complexities in dealing with the low frequency flicker noise intransient signals. Schramel evaluated the peak height and the peak area136measurements for GFAAS, and suggested that the peak area gives betterprecision and increased linearity in the calibration graph [137].Sturgeon et al. reported similar DLs and precision, but lower sensitivity andwider linear range for the measurement of peak area compared with that ofpeak height for the CRA furnace [138]. However, for the HGA furnace,Sturgeon et al. reported increased sensitivity for the peak areas comparedwith peak height measurement [139]. The difference between analyticalcharacteristics of CRA and HGA furnaces was attributed to the increaseddiffusion losses from the CRA furnace [139]. For Cu and Pb in GFAAS,Harnly reported that the precision for peak areas integrated from theappearance time to tije2 was 30 % better than the precision for areasintegrated over the whole peak and were 10 % better than those for peakheights [140]. However, for V, the precision for peak height measurementswas better than that for peak area measurements, regardless of theintegration interval [140].Furthermore, it has been recognized that integration of the entire peakdoes not necessarily yield the maximum SNR. Piepmeier predictedimprovements in precision by up to a factor of six when transient absorbancepeaks with long tails were only partially integrated; however, there was noloss of precision when the Gaussian absorbance peaks were integrated over99 % of the total peak [141]. Laeven and Smit estimated the optimum peakarea for a shot-noise limited Gaussian peak when the peak was integratedfrom the peak maximum minus 1.4a to the peak maximum plus1.4a ( where a is the variance ) [142]. However, the estimation of theP^P137optimum peak area for a flicker noise limited peak was complicated [142].Voigtman also estimated the optimum peak areas for shot-noise limitedsignals with different peak shapes [143]. In a recent publication, Kale andVoigtman predicted not well defined boundaries for the integration of theflicker noise limited peaks [144].In summary, it appears that the determination of the optimum integrationtime interval for the maximum SNR for a flicker noise limited transientsignal is difficult, especially for the transient signals in graphite furnaceswhere atomization and residence times play an important role in determiningthe duration of the peak, and hence, noise in the signal. The intent here is tostudy how noise in the analyte emission signal and that in the backgroundvary with the RF power rather than to determine optimum values for theSNR. All emission signals ( both the analyte and background ) wereintegrated for 3 s, starting from 0.5 s into the atomization step, which coversabout 95 % of Pb signal at 20 W.Figure 5.4 is a plot of SNR as a function of Pb emission intensity at anRF power of 20 W. These data illustrate that SNR is constant at highersignal intensities. In addition, both SBR and signal-to-background noiseratio increase linearly with the signal intensity. Thus, at higher SBRs,limiting noise in the Pb signal is due to the analyte flicker noise which has acharacteristic linear dependence with signal intensity. Furthermore, themaximum SNR obtained for the Pb signal is about 12, which corresponds toan RSD of about 8 %. • •_•- i 1^1'120151 0138500^1000 1500 2000 2500 3000Emission Intensity ( orb. units )Figure 5.4. Lead signal-to-noise ratio ( SNR ) as a functionof signal intensity.Figure 5.5 depicts plots of SNR and SBR as a function of RF power. Theeffect of RF power on the net Pb emission intensity is previously provided inFigure 5.2. The SNR decreases up to about 50 W and then shows a marginalincrease as the RF power increases. However, the SBR steadily decreases asthe RF power increases. The data presented in Figure 5.5 also show that thehighest SNR and SBR are at 20 W. The marginal increase in the SNR athigh RF powers is likely due to the changes in the background noisecharacteristics.13920^40^60^80^100Power ( W)Figure 5.5. Effect of RF power on the signal-to-noiseratio ( SNR: • ) and the signal-to-backgroundratio ( SBR: • ) for Pb.Figure 5.6 depicts the net background emission at 283.3 nm and therelative standard deviation of the background ( RSDB ) as a function ofRF power applied. These data illustrate that the background emissionsteadily increases up to about 50 W and then shows a marginal increase. Incontrast, the RSDB decreases with increasing RF power. This decrease inthe noise in the background signal at high RF powers can account for themarginal increase in SNR at low SBRs ( Figure 5.5).2000^15^30^45^60^75Power (090^105 1205040302010140Figure 5.6. Effect of RF power on the net backgroundemission ( • ) and the relative standarddeviation of the background ( RSDB: 0).The plasma background emission is an important diagnostic tool in atomicspectroscopy. If the plasma background emission results from continuumemission as a consequence of radiative recombination ( Section 1.4.3), theplasma background can be related to the electron number density and theelectron temperature in the plasma [145,146]. The nature of the backgroundemission signal at the analytical wavelength of Pb ( 283.30 nun) was studiedby varying the spectral bandwidth of the monochromator. Figure 5.7 depictsthe effect of spectral bandwidth on the square root of the net background141201510500.00 0.02 0.04 0.06 0.08 0.10 0.12 0.14 0.16Spectral Bandwidth ( nm )Figure 5.7. Effect of the spectral bandwidth on the squareroot of the background emission intensity at283.3 nm;20 W(• ),40W(•).emission intensity. The amount of background emission collected increaseslinearly with the spectral bandwidth up to about 0.08 nm and then increasesexponentially with increasing spectral bandwidth. In addition, an increase inRF power increases the background emission intensity.If the continuum emission is imaged on an entrance slit, the measuredintensity is proportional to the square of the spectral bandwidth. Theexponential increase of the background emission with the increase of spectralbandwidth above 0.08 nm ( Figure 5.7) indicates a partial spectral overlap.142It was also observed that, an increase in the spectral bandwidth increases thePb emission intensity linearly; however, increasing the spectral bandwidthabove 0.08 nm causes a decrease in SNR, probably due to the higher level ofnoise in the background. The temporal response of the background at awider spectral bandwidth is similar to that of the OH at306 nm ( Section 3.3.3). The background emission feature around 283 nmwavelength is the ( 1,0 ) OH vibrational band at 281 nm ( Section 3.3.1).At narrow spectral bandwidths and low RF powers, the dominant source ofbackground noise may come from the scattered blackbody radiation from thegraphite furnace, which generates both shot-noise and a drift noise due to thevarying temperature of the graphite furnace during the atomization step. Atwide slit widths, the flicker noise in the background at 283.30 nm is likelydue to the significant drift in the characteristic temporal emission of OHduring the atomization step and the variation of the amount of OH present inthe plasma. The variation of the amount of OH present in the plasma is dueto the desorption of OH from the furnace wall and the dissociation of 11 20 inthe gas-phase. The background transients collected without the ash step orthe high temperature atomization step show a large variation in thebackground emission intensity between replicates with increasing RF power.This observed variation of background emission between replicates may beattributed to a change in the amount of OH present in the plasma gasbetween each data acquisition cycle.143For HA-FANES, Harnly et al. reported a maximum SNR for Cu and Cd at50 and 20 mA, respectively, when the discharge current was varied from 10to 80 mA [125]. For the calculation of SNR, Hamly et al. assumed that thenoise in the signal was independent of the atomization process: black-body,line, and molecular emission were negligible [125]. In a subsequentpublication, Riby et al. reported a maximum SNR for Cr between 60 and70 mA at 160 Torr [90]. The low SNR at low and high currents wasattributed to the plasma instabilities. Riby et al. reported that, at lowcurrents, the discharge fluctuated between partial and total coverage of thecathode, and at high currents, sparking occurred between the base of thecathode and the graphite furnace [90]. It should be noted that the SNR wascalculated as the average peak area for triplicate determinations divided bythe standard deviation of seven blank atomizations [90].5.3.4 Figures of meritFigure 5.8 provides the calibration graph for the determination of Pb inFAPES. Four replicate measurements were carried out for eachconcentration studied. The RF power used for the collection of data providedin Figure 5.8 and the rest of this study was 20 W.In order to study the effect of analyte condensation ( on the RF electrode )on figures of merit, calibration graphs were obtained at different integrationtimes. The variation in sensitivity ( slope of the calibration graph ) as afunction of integration time is plotted in Figure 5.9. The integration interval1444000300020001 00000.00^0.50^1.00^1.50^2.00^2.50^3.00Amount of Lead ( ng )Figure 5.8. The analyte calibration graph for Pb.was varied up to 3.5 s ( from 0.5 to 4 s into the atomization step ) for theblank and different analyte amounts because of the unsymmetrical temporalresponse of Pb with respect to the peak maximum as a result of analytecondensation on the RI? electrode. Figure 5.9 shows a linear increase insensitivity when the integration time is increased. Data provided inFigure 5.9 also show that analyte condensation on the RF electrode lowersthe sensitivity of the calibration graph at shorter integration intervals as canbe seen by the continued increase in sensitivity even after the first peakmaximum. The values in the parentheses ( Figure 5.9) give the correlation14520001 6001 2008004000 .0^0.5^1.0^1.5^2.0^2.5^3.0^3.5^4 0Integration Time ( s )Figure 5.9. Effect of integration time on the sensitivity ofthe Pb emission intensity. The values in theparentheses give the correlation coefficient ofthe least square fit for each calibration graph.coefficient of the least square fit for each calibration graph as a function ofintegration time. Loss of linearity of the calibration graph is not observedeven after integration into the peak maximum of the second peak with higheramounts of Pb.Figure 5.10 is a plot of the DL as a function of integration time. The DL isdefined as the concentration of the analyte which gives a signal equal to theblank signal plus three standard deviations of the blank, and was calculated150.0120.090.060.030.0__• ^ 4,----41--.-_1460.00 0^0.5^1 .0^1.5^2.0^2.5^3.0^3.5^4 0Integration Time ( s )Figure 5.10. Effect of integration time on the detection limit( DL ) for Pb.using equation 5.1. The standard deviation of the blank was determined byfifteen replicates. These data illustrate that, as the integration timeincreases, the DL decreases and then levels off. The advantage of gain insensitivity of having a longer integration time is counterbalanced by the highintegrated flicker noise in the background. This effect of the integrated noisecan be seen from the level off in the DL at longer integration times. Atshorter integration times, the integrated flicker noise in the backgroundis low.147The absolute DL calculated by using an integration time of 1 s ( from 1.5 to2.5 s) yields 65 pg and the concentration DL is 13 gg L -1 . The previouslyreported DL for Pb is 46 pg ( peak area without the use of a chemicalmodifier ) for FAPES [78], and 15 pg ( peak area with the use of a chemicalmodifier ) for GFAAS [147]. Sturgeon et al. predicted improvements of10 to 100-fold for the DLs in FAPES by analyte atomization from a platform,use of chemical modifiers and use of an ICC, more efficient transfer opticsand image coupling to a high resolution spectrometer, and backgroundcorrection [78].Precision of a Pb determination is about 7 - 10 %. This low precision islikely due to the manual pipetting of the analyte sample onto the graphitefurnace side-wall, tendency of the sample droplet to spread over or run alongthe furnace side-wall, and non-repeatability of the analyte condensation onthe RF electrode. For GFAAS with manual pipetting, precision is about3 - 5 % (120).It was noted that the partial integration of the Pb emission peaks yieldseven poorer precision. If analyte condensation on the RF electrode is notrepeatable for each atomization step, then the analyte emission intensitymay not be repeatable even with the integration of the whole peak. Thisnon-repeatability of the analyte emission intensity likely arises from thetemporal non-isothermality of the graphite furnace during the atomizationstep. The diffusion, and hence, the residence time of different portions of theatomic population in the gas-phase are dependent on the gas-phase148temperature. Therefore, the analyte emission intensity can change betweeneach atomization step due to the non-repeatability of the analytecondensation in temporal non-isothermal conditions. Therefore, undertemporal non-isothermal conditions, non-repeatability of the analytecondensation on the RF electrode is a likely contributing factor for the poorprecision of the determination of Pb in FAPES.The calibration graph given in Figure 5.8 shows a linear range of about1.4-orders of magnitude for Pb determination by FAPES. The linear dynamicrange reported by Sturgeon et al. for 9 analytes was about 3-orders ofmagnitude [78], and that reported by Smith et al. for Ag was 2-orders ofmagnitude [17]. For these studies [17,78], end-heated Massmann-typegraphite furnaces were employed. It is well known that self-absorption limitsthe linear dynamic range when this type of end-heated Massmann-typefurnaces are used in CFAES [45]. A similar situation is very likely to exist inFAPES, if one uses a short RF electrode leaving a cooler region at theopen-end of the graphite furnace. In addition, if one integrates only a portionof the rising edge of the emission peak, the linear dynamic range shouldincrease without sacrificing the DL, because self-absorption is less likely atearly stages of the atomization. For GFAAS, the linear dynamic range spansover 1.5-orders of magnitude [120].It is important to understand the limitations of these figures of merit.These figures of merit give only a measure of the excitation ability of theRF plasma source in FAPES and the detection power of the experimental149system employed for this study. These kinds of figures of merit are useful forcomparing different spectrochemical sources. However, most of these kindsof so called" analytical " figures of merit reported in the literature ( when theanalyte to be determined was present in an isolated form, i.e. aqueousstandard solutions ) reflect direct instrumental procedures. With complexreal samples, severe matrix interferences can occur, and as a result, not onlythe detection ability is degraded but also systematic errors can occur.Furthermore, each analyte should be studied separately because the analyteemission and the plasma background emission may depend on the nature ofthe analyte and the analytical wavelength.5.4 SUMMARYThe effect of RF power, between 20 and 100 W, on the analyte emissionsignal and the plasma background emission is studied for the 283.3 nmPb ( I ) resonance line. The time-resolved Pb-excitation temperature for 50 Whelium plasma is 4560 ± 100 K. At 20 and 100 W, the Pb-excitationtemperatures are 3990 ±80 and 4840 ± 110 K, respectively.The highest Pb emission is observed at an RF power of about 50 W. Above50 W, the Pb emission intensity decreases. This decrease in emissionintensity may be attributed to analyte losses during the pre-atomization oratomization step, changes in plasma characteristics, or an increase inionization of Pb at high RF powers. The highest SNR and SBR are observed150at relatively low RF powers. At these low RF powers ( about 20 W), analyteflicker noise is the dominant noise in the emission signal.The absolute detection limit for Pb is 65 pg and the concentration DL is13 Lig L-1 with a relatively slow heating rate for the graphite furnace.Precision is about 7- 10 % for the Pb analysis by FAPES. This low precisionis likely due to the manual pipetting of the analyte sample onto the graphitefurnace side-wall, tendency of the sample droplet to spread over or run alongthe furnace side-wall, and non-repeatability of the analyte condensation onthe RP electrode. Analyte condensation on the RF electrode reduces thesensitivity but shows no effect on the linearity of the analytical calibrationcurve.151CHAPTER 6EFFECTS OF SODIUM CHLORIDE AND SODIUM NITRATE ONTHE LEAD AND SILVER EMISSION IN FAPES6.1 INTRODUCTIONAn important characteristic of an analyte is its susceptibility to be affected bycontaminants in the sample matrix during an analytical determination. Inemission spectrochemical methods based on a graphite furnace for analyteatomization and vaporization and a plasma source for analyte excitation, theanalyte atomization, vaporization, and excitation characteristics may beaffected by matrix effects found in graphite furnace methods and those foundin plasma methods.Matrix effects in graphite furnace methods, which have been studied byusing AAS, can be classified into two categories: (1) condensed-phaseinterferences, which are volatile compound formation, incompletevaporization due to occlusion, refractory compound formation, and change inthe rate of analyte supply, and (2) gas-phase interferences, which are shiftsin the dissociation equilibria due to stable compound formation, shifts inionization equilibria, and changes in the rate of analyte removal [148].152Matrix effects in plasma methods are broadly classified as transport( viscosity and surface tension effects ), vaporization, dissociation, andionization interferences. Additional sources of interference can occur as aresult of the interferent affecting the plasma characteristics. Examples ofthese sources of interference are collisional excitation rate changes for theanalyte due to the changes in the electron energy distribution or density inthe plasma, and ambipolar diffusion due to shifts in the spatial distributionof plasma constituents [149,150].Plasma spectrometric methods, such as FAPES and FANES, which arebased on a graphite furnace for analyte atomization and vaporization, aredistinct from most other common emission spectrometric methods because ofthe transient nature of the analyte and the interferent populations in theplasma source. In these emission spectrometric methods, the analyte and theinterferent populations may have different temporal characteristics.Therefore, gas-phase interferences may not be present in some instancesbecause the analyte and the interferent do not co-exist in the gas-phase.However, in graphite furnace methods, not only the analyte vaporizationefficiency but also the interferent vaporization efficiency to the gas-phase ishigh. As a consequence, when the analyte and interferent co-exist in thegas-phase, interference effects in the gas-phase are expected to be severe.To date, only a few studies have been reported on the interference effectsin FAPES. Smith et al. reported a slight enhancement of peak area for Ag atlow concentrations and a suppression at higher concentrations of both153NaNO3 and NaC1 [17]. The enhancement and the suppression effects wereattributed to Na rather than to the counter ion. Smith et al. speculated thatat low concentrations of Na, the signal was enhanced by a shift in theionization equilibrium, and at higher concentrations, the free atompopulation of Ag was being depressed or the excitation characteristics of theplasma were modified by the presence of Na [17]. Sturgeon et al. determinedCd and Pb in three marine reference samples from the National ResearchCouncil ( NRC ) of Canada, and it was necessary to employ the method ofstandard additions for quantification because of matrix effects [121]. In asubsequent publication, Sturgeon et al. reported a large reduction in Pbemission when NaC1 was present in standard solutions [123]. While theresults were not conclusive about the nature of the effect of NaC1 on Pbemission, experimental observations were attributed to a gas-phase effect dueto the co-existence of Pb and interferents in the gas-phase, quenching of theplasma, or condensation of NaC1 on the RF electrode [123]. Although it hasbeen widely speculated that the major cause of the NaC1 interference effect inFANES or FAPES methods is the gas-phase co-existence of the analyte, Na,and Cl, no mechanistic study has been reported to date. Obviously, a greaterunderstanding at a fundamental level will be needed before these effects canbe fully controlled.This chapter presents the results of the work undertaken to study themechanism of NaC1 and NaNO3 interference effects in FAPES [151]. Sodiumsalts constitute one family of interferents in both GFAAS and emissionspectrometric methods. In this study, Pb and Ag were examined as the test154analytes because of the expected severe gas-phase interference effects foranalytes with similar volatilities to Na. The effect of the addition of ascorbicand phosphoric acid, commonly used matrix modifiers for Pb in GFAAS, arealso presented.6.2 EXPERIMENTALA complete description of the experimental system employed to acquireatomic emission signals is given in Chapter 2 of this thesis. Thisexperimental system and a vertical-mount for the RF electrode wereemployed. The following experimental methods and parameters should benoted.6.2.1 Signal PresentationFour replicate measurements were averaged and subjected to a 25-pointSavitzky-Golay smoothing procedure [91]. Analytical parameters, such aspeak area and peak width, were calculated for each sample deposition.6.2.2 ReagentsAll analyte solutions in distilled water were prepared from serial dilution of1000 mg L-1 in distilled water stock solution prior to analysis. Lead and Agsolutions were prepared by dissolving analytical grade AgNO3 and Pb(NO3)2155( both BDH, Toronto, Canada ). Sodium solutions were prepared bydissolving analytical grade NaCl ( BDH, Toronto ) and NaNO3 ( Anachem,Portland, USA). Nitric acid solutions were prepared by using the analyticalgrade reagent ( BDH, Toronto ).6.2.3 ProcedureA 5 A aliquot of analyte solution was deposited onto the furnace side wallwith an Eppendorf 0.5 - 10 ill, UltraMicro pipette. The plasma sourcework-head was purged with helium ( Union Carbide, Toronto, Canada ). Thefurnace temperature was increased to 470 K in 45 s to dry, and wasmaintained at 470 K for another 45 s to ash the sample. This total timeinterval was sufficient to exclude the water vapor inside the plasma sourcework-head before plasma ignition. Within the next 10 s time lag, the plasmawas ignited. After this time lag, the furnace temperature was ramped to2050 K in 5 s for all determinations. The blank determinations were carriedout by depositing the same volume of distilled water, or the respectiveinterferent in distilled water ( unless otherwise noted ). After the analyteatomization step, a dry ( without deposition ) atomization step was carriedout to clean the graphite furnace. Four replicate measurements were carriedout for each determination.All determinations were carried out by using atomic resonance lines ofPb ( 283.30 rim) and Ag ( 328.07 rim). Sodium emission was measuredat 330.23 nm.1566.3 RESULTS AND DISCUSSION6.3.1 Reflected PowerReflected power is an important diagnostic tool in plasma spectrochemicalmethods because the changes in reflected power indicate changes in theplasma characteristics. In FAPES, as the graphite furnace heats during theatomization step of the analyte, the plasma impedance changes, because ofthe evolution of thermionic electrons from the hot graphite surfaces. Thischange in the plasma impedance necessitates the use of an impedancematching network, which acts to maintain the reflected power at a minimumand to protect the RF power supply.The change in the reflected power during the atomization step due to thepresence of NaC1, Na, and Cl, or ionization of Na in the plasma, was studiedat an RF power of 14, 20, 30, and 40 W. The results of this experiment at anRF power of 20 W for the deposition of distilled water ( solid line ), 160 ng ofNa as NaC1 in distilled water ( dot line ), and the temperature profile of thegraphite furnace ( dash line ), are given in Figure 6.1. There is no significantdifference between the reflected power with and without NaCl. At RF powersbetween 14 and 40 W, the average reflected power measured is about 1 Wwith some random rather than systematic fluctuations. In addition, noreflected power change is observed up to 3.214 of Na ( as NaC1 or NaNO3 )used in the experiments described in this chapter.15720001 7501 50012501 000750500--iCD3-0CD0-1C--s(D"--.■._,0.0^1.0^2.0^3.0^4.0^5.0^6.0^7.0^8.0Time ( s )Figure 6.1. Temporal response of the reflected power at anRF power of 20 W for the deposition of distilledwater ( - ), and 160 ng of Na as NaC1 indistilled water ( • • • ); and the temperatureprofile of the graphite furnace ( - - - ).Sturgeon et al. reported a reflected power loss when 2314 of Na ( as NaC1in 1 % HNO3 ) was atomized at an RF power of 100 W [123]. However, thereflected power loss was observed not during the atomization step of theanalyte but after the furnace reached a steady state temperature of 2100 K.Sturgeon et al. attributed this observed reflected power loss to the evolutionof thermionic electrons from the RF electrode because of the lag time whichoccurred between the establishment of maximum reflected power and steady158state furnace wall temperature. This lag time was attributed to the radiativeheating of the RF electrode by the furnace wall [123].It should be noted here that, for the experimental system described in thisthesis, a reflected power loss is observed even before the furnace reaches2100 K at RF powers higher than 75 W ( Sections 2.1.2 and 5.3.1). Thisobservation can be attributed to the evolution of thermionic electrons fromthe RF electrode because the onset of the reflected power loss is dependent onthe RF power. Furthermore, some changes in the temporal characteristics ofthe CO+ emission during the atomization step are also due to the evolution ofthermionic electrons from the RF electrode ( Section 3.3.3). The observedreflected power loss during the atomization step is most likely due to a highertemperature of the RF electrode for 75 W plasma employed in the workdescribed in this thesis. This higher temperature of the RF electrode can beattributed to an efficient RF power coupling to the plasma source. However,the extent of the RF power coupling efficiency is not known at present.Sturgeon et al. employed a manual impedance matching network in theirexperimental system, and reported 85 % internal loss of RF power within theRF tuner [79].6.3.2 Temporal Response of Na ( as NaC1 and NaNO3 )Figure 6.2 provides the temporal response of the Na emission signal for160 ng of Na as NaCl ( solid line ) and 160 ng of Na as NaNO3 ( dash line )deposited on the furnace wall at an RF power of 20 W. The temporal15920001 6001 20080040000.0^1.0^2.0^3.0^4.0^5.0^6.0^7.0^8.0Time ( s )Figure 6.2. Temporal response of the Na emission signal atan RF power of 20 W for the deposition of160 ng of Na as NaC1 ( — ), and 160 ng of Naas NaNO3 ( - - - ); and water blank ( • • • ).response of the emission signal of Na as NaC1 is different from that of Na asNaNO3. For NaC1, there are two peaks, and the more intense first peak hasan appearance temperature of 1080 K. For NaNO3, there is a single broadpeak and has a similar appearance temperature compared with Na as NaCl.When NaCl is deposited on the furnace wall, Na is likely formed from thegas-phase dissociation of NaCl:160NaCl(s) ^>^NaCl ^>^Na(g) +^Cl(g) (6.1).The existence of two peaks for the Na signal from NaC1 is indicative of thepresence of Na in the gas-phase at two different stages. The first peak isfrom initial vaporization from the furnace wall and gas-phase dissociation ofNaC1 to form Na ( equation 6.1) and the second peak is most likely fromNaC1 ( and Na ) which has condensed on the RF electrode and subsequentsecond-surface vaporization into the gas-phase ( similar to Pb0,Section 4.3.2).Campbell and Ottaway suggested that, in GFAAS, Na can be formed fromcarbon reduction of Na20 [104];4NaNO3(s) > 2Na20(s)^+ 4NO2^+ 02 (6.2),and Na2O() + C ^> 2Na(g)^+ CO^(6.3).For NaNO3 in FAPES, the broad Na peak shown in Figure 6.2 may be dueto an overlap of Na signals from the initial vaporization of Na from thefurnace wall ( equation 6.3 ), and second-surface vaporization subsequent tocondensation on the RF electrode. It is also possible that the broad Na peakfrom NaNO3 may be largely due to the second-surface vaporization of Nafrom the RF electrode rather than from the furnace wall because of aninstantaneous condensation of Na on the RF electrode.161The condensation of Na on the graphite surfaces has been also observed inGFAAS studies. In a study of spatial distribution of Na atoms within thegraphite furnace, Stafford and Holcombe observed lower Na atom densitynear the furnace wall than in the center of the furnace [152]. The observedNa atom gradients were attributed to a gas-phase dissociation mechanism,and strong interaction of atomic Na with graphite [152]. Huie and Curranreported adsorption of Na atoms on the upper-wall of the furnace at a ratehigher than the release of Na atoms during the early stages of theatomization [153].6.3.3 Effect of NaC1 and NaNO3 on the Pb Emission IntensityInterference effects of Na, an easily ionizable element ( EIE ), have beenextensively studied in flame AES and plasma spectrometry. In flame AES,an enhancement of the atomic emission intensity for alkali metals has beenobserved as a result of a shift in the ionization equilibrium toward the atomicform [154]. In plasma source emission spectrometry, both enhancement andsuppression of the analyte emission signal have been observed [155-163].These observations have been attributed to various modifications of plasmacharacteristics when Na is present. However, the root cause of theseobserved matrix effects is not well understood, and is still the subject ofconsiderable debate. In general, the effect of EIE interferences found in theplasmas was to enhance the atomic emission, and has been attributed toionization suppression or changes in the plasma characteristics rather than162to a chemical effect on the analyte because of the high temperatures ofplasma sources.Figure 6.3 depicts the temporal response of the Pb emission signal for0.5 ng of Pb ( solid line ), 0.5 ng of Pb and 160 ng of Na as NaC1 ( dash line ),and 0.5 ng of Pb and 160 ng of Na as NaNO3 ( long-dash line ) deposited onthe furnace wall at an RF power of 20 W. The effect of both NaNO3 and NaC1on the Pb emission signal is to reduce the emission signal relative to thatwithout the interferent.The appearance and peak temperatures for Pb in the presence of NaC1 arethe same as those for Pb alone. Moreover, the normalized temporalresponses of Pb, with and without NaC1, coincide within the error of themeasurement. These observations indicate that in the presence of NaCl, areduction in the excitation ability of the plasma, or a decrease in the analyteatom population in the plasma, or both, cause the reduction of the Pbemission intensity in FAPES.Gas-phase co-existence of the analyte and the interferents can be studiedby the comparison of temporal responses depicted in Figures 6.2 and 6.3. Thetemporal response of Na ( as NaC1 ) in Figure 6.2, and that of Pb inFigure 6.3 show that Na vaporizes later in time than Pb. Therefore, Pb, Na,and Cl co-exist in the gas-phase only during the later stages of Pbatomization while Pb is diffusing out of the plasma. The non-overlap oftemporal responses for Na as NaC1 ( Figures 6.2) and Pb ( Figures 6.3 ), and1631 .21 .00.80.60.40.20.00.0^1 .0^2.0^3.0^4.0^5.0^6.0^7.0^8.0Time ( s )Figure 6.3. Temporal response of the Pb emission signal atan RF power of 20 W for the deposition of 0.5 ngof Pb ( — ) , 0.5 ng of Pb and 160 ng of Na asNaC1 ( - - - ), and 0.5 ng of Pb and 160 ng of Na asNaNO3 ( — — ).the reduction of Pb emission intensity when Na ( as NaC1 ) is present,suggest that the major cause of NaC1 interference effect on Pb in FAPES isdue to a decrease in the Pb population in the plasma. It should be noted herethat an early appearance of Pb is observed in FAPES compared with theappearance of Pb in GFAAS ( Section 4.3.2).For GFAAS, Czobik and Matousek studied the effects of metal and alkalichlorides, including NaC1, on atomic absorption signals of Pb and Ni [164].164In the presence of 0.2 to 20 ng of NaC1, the signal for Pb showed a decrease inthe peak absorbance and a shift to lower appearance and peak temperatures.They attributed this signal modification to a depletion of analyte atompopulation in the gas-phase due to the formation of Pb chlorides and to acarrier effect of NaC1 on Pb. However, no NaC1 interference effect wasobserved for Ni because NaCl and its decomposition products were lost fromthe furnace before the Ni atomization temperature was reached [164]. Slavinand Manning reported a decrease in integrated absorbance for Pb as theamount of NaC1 was increased [128]. This effect was not observed to besignificant when a L'vov platform was used except at very high NaC1amounts ( about 0.1 %). The effect of NaC1 was attributed to the formationof volatile Pb chlorides in the condensed phase which are lost during theasiling stage [128], and the formation Pb chlorides in the gas-phase [165].Welz et al. used a dual-cavity platform to study the interference effects ofNaC1 on Pb [129]. Welz et al. suggested that while some Pb is lost as volatilePb chlorides during the ashing stage, Pb was also lost due to the analytevapor expulsion with the vaporization of the NaCl matrix [129].Therefore, it appears that the most significant effect of NaCl in GFAAS isnot due to Na, but rather to the counter ion: chloride. The interferingchlorides form stable molecular species with the analyte;i.e.," analyte-chloride ". These chloride interferences are most severe for thevolatile elements because the dissociation of analyte-chlorides in thegas-phase is less likely at lower temperatures. This is the reason that theuse of the L'vov platform is useful for these situations. In addition, at higher165heating rates of the graphite furnace, a higher gas-phase temperature isattained so rapidly that these molecular species can be dissociated beforedissipating out of the furnace. However, it is difficult to determine whetherthe formation of analyte-chlorides is in the condensed-phase prior tovaporization or in the gas-phase after the vaporization from the furnace wall.In an attempt to determine whether the effect of NaC1 originates in thecondensed-phase or gas-phase, Pb is deposited on the furnace wall with andwithout NaCl deposited on the RI? electrode. Even though the atomicpopulations of Pb, Na, and Cl do not co-exist in the plasma, it is possible thatundissociated NaC1 is present in the plasma at earlier times ( compared withthe appearance of Na ) because dissociation of NaC1 in the gas-phase isdependent on the gas-phase temperature. It should be noted here thatanalytes deposited on the RF electrode have a similar appearance time tothose deposited on the furnace wall when a 20 W plasma isused ( Chapter 4). Therefore, no significant difference in the temporalresponse of Na is expected when the NaCl is deposited on the RI? electrode.In fact, when the analyte deposited on the furnace wall has an appearancetime before 3 s into the atomization step, the appearance time can shift to aearlier time when deposited on the RF electrode.When 0.5 ng of Pb is deposited on the furnace wall and 160 ng of Na asNaC1 is deposited on the RF electrode at an RF power of 20 W, the sameintegrated emission intensity is obtained compared with that with Pb alone.This observation and the temporal non-overlap of atomic populations of Pb,166Na and Cl, suggest that the effect of Na ( as NaC1 ) is not due to theformation of Pb chloride in the gas-phase. Furthermore, similar integratedintensities with and without NaCl deposited on the RF electrode eliminatethe possibility of a reduction in excitation ability of the plasma when NaC1 ispresent ( also see the effect of RF power ). These observations indicate thatloss of Pb chloride ( formed in the condensed-phase ) during the ashing stepand the slow atomization step is the most likely cause for the reduction in Pbemission in FAPES.However, for GFAAS, it has been shown that Pb samples with NaC1 can beashed up to about 800 K without any significant loss of the absorbancesignal [128,129]. For the experiments described in this chapter, sample wasashed at 470 K. Therefore, the loss of Pb as chlorides during the slowatomization step may be the major cause of Pb signal suppression in thisstudy. When the analyte atomization process goes through ananalyte-chloride formation step and vaporization of analyte-chlorides occursat a temperature lower than the appearance temperature, it is essential thatthe graphite furnace reaches the final temperature quickly in order toprevent any analyte losses.The effect of NaNO3 on the Pb emission signal ( Figure 6.3) is morecomplex than that observed for NaCl. The Pb signal in the presence ofNaNO3 appears at a later time and has a narrower peak width. When 0.5 ngof Pb is deposited on the furnace wall and 160 ng of Na as NaNO3 isdeposited on the RF electrode at an RF power of 20 W, a reduction in167emission is observed but the peak shape, appearance temperature, and peaktemperature are the same as those obtained without NaNO3 deposited on theRF electrode. The identical peak shapes for Pb with and without NaNO3deposited on the RF electrode, and the difference in peak shapes with andwithout NaNO3 deposited on the furnace wall ( Figure 6.3) suggest that theeffect of NaNO3 on Pb partly takes place in the condensed-phase. The delayin the onset of the Pb emission signal in the presence of NaNO3 ( Figure 6.3)is probably due to the entrapment of Pb0 in the NaNO3 matrix. Therefore,Pb0 is prevented from vaporizing from the furnace wall and the narrow peakwidth of the Pb emission signal is due to the rapid expulsion of Pb0 from theNaNO3 matrix at higher temperatures. The reduction in the emissionintensity of Pb in the presence of NaNO3 can be attributed to a decrease infree atom population in the gas-phase, probably due to changes in thegas-phase dissociation equilibrium of Pb0. The gas-phase dissociationequilibrium of Pb0 can be affected by a change in the gas-phase 02concentration because of the decomposition products of NaNO3, according toequation (6.2).Effects of NaNO3 matrices on the atomic absorption signal, in particularon Pb, were studied in GFAAS. Eklund and Holcombe showed that largeamounts of nitrate can reduce the signal for Ag, Cu, and Ga in GFAAS as aresult of gas-phase oxidation of the metal by NO2 and 02 formed duringashing and atomization steps [166]. Cedergren et al. reported a depressiveeffect and narrowing of the absorbance peak width of Pb in GFAAS whenNaNO3 is present in the sample [112]. These observations were attributed to168an increase in partial pressure of 02 formed from the decomposition productsof NaNO3. Cedergren et al. also reported that in the presence of CO, the Pbpeak height increases without any apparent peak shape difference whenNaNO3 is present [112]. This effect of CO was attributed to a reduction in 02partial pressure. Holcombe et al. reported the formation of a double peak forPb when a 100-fold excess of NO3- ( as NaNO3 ) is present and delayedatomization in the presence of 1000-fold excess of NO3- [167]. This late shiftwas attributed to an increase in the sticking coefficient of Pb to theoxygenated surface. However, Sturgeon and Berman suggested that theobserved shift is due to the suppression of thermal dissociation of analyteoxides [111]. In a later study, Bass and Holcombe attributed the shifts inabsorbance signals to the presence of gas-phase CO and CO2 [113]. Insummary, these observations indicate both condensed and gas-phase effectson Pb in the presence of NaNO3 for GFAAS.Figure 6.4 depicts the effect of RF power on the interference effect of160 ng of Na ( as NaC1 and NaNO3 ) on the emission signal for 0.5 ng of Pb inFAPES. The emission intensity for the analyte alone is arbitrarily given avalue of 1. For Pb in NaCl and NaNO3, the magnitude of the interferenceeffect is essentially the same over the power range 20 to 40 W, although itbecomes a little less significant as RF power is increased over this range.This insignificance of RF power on the interference effect indicates a loss ofmolecular species ( for example, Pb chlorides and Pb0 ) without dissociatingin the plasma. For the thermal dissociation of these molecular species, andhence, to observe a reduced interference effect, plasma temperature should be_1.5 -1 .0 -0.5 -2.01690.0 ^10,^1^I^I^115^20^25^30^35^40^45Power ( W )Figure 6.4. Interference effect of 160 ng of Na as NaCI ( • )and 160 ng of Na as NaNO3 ( • ) on the Pbemission intensity as a function of RF power.increased significantly when the RF power is changed from 20 to 40 W. Thegas kinetic temperatures reported by Sturgeon et al. were independent of theRF power over the range 50 - 100 W [80]. The minimal effect of RF power onthe interference effect of NaNO3 on Pb also indicates that it is not theRF power but the gas-phase composition that limits the gas-phasedissociation of Pb0 in FAPES.An important feature of the plot in Figure 6.4 is that at 14 W, there is anenhancement in the emission for Pb in the presence of both NaC1 and15012090603001700.0^1.0^2.0^3.0^4.0^5.0^6.0^7.0^8.0Time ( s )Figure 6.5. Temporal response of the Pb emission signal atan RF power of 14 W for 0.5 ng of Pb ( — )and 0.5 ng of Pb and 160 ng of Na as NaC1 ( — ).NaNO3. The Pb emission signals with ( solid line ) and without ( dash line )160 ng of Na as NaCl at an RF power of 14 W are provided in Figure 6.5. Itis clear that in contrast to higher powers, the signal is enhanced at 14 W. Asimilar effect is seen with NaNO3 as the contaminant. Figure 6.5 also showsa slight late shift in the temporal profile. This enhancement of emission at14 W should be related to an increase in the analyte population in thegas-phase or to an enhancement in the excitation ability of the plasma.However, cause of this effect is not known at present, and needs furtherstudy.171 1 .21 .00.80.60.40.20.00.0E0^1 .0E3^2.0E3^3.0E3^4.0E3Amount of Na ( ng )Figure 6.6. Interference effect on the Pb emission intensityat an RF power of 20 W as a function of amountof Na as NaCI.Figure 6.6 depicts the interference effect from different amounts of Na ( asNaC1 ) on the Pb emission intensity from 0.5 ng of Pb at an RF power of20 W. It can be seen that even small amounts of Na cause a markeddepression in the Pb emission. As the amount of the interferent increases,the interference reaches a fairly stable level. Sturgeon et al. also observed alarge reduction of Pb intensity even with a small amount of NaC1 [123].172It should be noted that the temporal response for 0.5 ng of Pb and 160 ngof Na as NaC1 in 1% HNO3 ( v/v ) is similar to that of 0.5 ng Pb and 160 ng ofNa as NaNO3 in pure aqueous solution ( Figure 6.3). These similar peakshapes are because of the conversion of NaC1 to NaNO3 in the presence ofHNO3. A similar observation for Pb in GFAAS was reported by Frech andCedergren [168]. Nitric acid is commonly used in digestion procedures inbiological and environmental sample analyses, preceding the determinationby GFAAS.6.3.4 Effects of NaC1 and NaNO3 on the Ag Emission intensityFigure 6.7 provides the temporal response of the Ag emission signal for0.25 ng of Ag ( solid line ), 0.25 ng of Ag and 160 ng of Na as NaCl ( dashline ), and 0.25 ng of Ag and 160 ng of Na as NaNO3 ( dot line ) deposited onthe furnace wall at an RF power of 20 W. The effect of both NaC1 and NaNO3on Ag signal is to decrease the integrated emission intensity relative to thatwithout the interferent. The magnitude of the reduction in Ag emission ismuch greater for NaNO3 compared with NaCl, and hence, NO3- is a majorcause for Ag signal suppression. Both Na containing interferents also shiftthe Ag peak temperatures to lower values. An additional feature of the Agtemporal response is that the emission profile for Ag almost directly overlapswith the emission profile for Na ( Figure 6.2 ) when NaCl is present.The rising edge of the Ag emission signal from a pure aqueous solution ofAgNO3 ( Figure 6.7) is different from that observed from AgNO3 in1731 .21 .00.80.60.40.20.00.0^1 .0^2.0^3.0^4.0^5.0^6.0^7.0^8.0Time ( s )Figure 6.7. Temporal response of the Ag emission signal atan R.F power of 20 W for 0.25 ng of Ag ( — )0.25 ng of Ag and 160 ng of Na as NaC1 ( - ),and 0.25 ng of Ag and 160 ng of Na asNaNO3 ( • • • ).1 % ( v/v ) HNO3 ( Section 4.3.1). In a study on the mechanism ofvaporization of Ag in GFAAS, a change in Ag peak shapes was reported andwas attributed to the changes in the surface topology of Ag by chemisorbed02 [97]. Gold, with similar atomization characteristics to Ag, showed anunresolved double peak in Ar/02 for GFAAS compared with Ar alone, whichwas attributed to a surface effect of graphite [111]. However, for FAPES, thecause of this observed difference in peak shapes is not known at present, andneeds further study.174Figure 6.8 depicts the effect of RF power on the interference effect of160 ng of Na ( as NaC1 and NaNO3 ) on the Ag emission intensity for 0.25 ngof Ag in FAPES. Similar to Pb ( Figure 6.4 ), the effect of NaC1 is essentiallyindependent of the magnitude of the RF power, thus the interference effect isconstant and not proportional. This constant interference effect as a functionof RF power indicates a common interference effect for both Ag and Pb in thepresence of NaCl; i.e., formation of analyte-chlorides in the condensed-phase.It should be noted that both AgC1 and PbC1 have similar dissociationenergies ( 3.2 and 3.1 eV respectively ). It should also be noted that anidentical number of moles of Pb and Ag ( 0.0023 n mol ) were used in theseexperiments, and hence, the analyte to interferent molar ratio is the same forboth Pb and Ag. The effect of RF power also shows that the NaClinterference effect is not due to any changes in excitation characteristics or areduction in RF power coupled to the plasma.For Ag in NaNO3 ( Figure 6.8 ), the interference effect is severe atRF powers less than 20 W, and becomes less significant as the RF power isincreased to 40 W. This observation is in contrast to that for Pb in NaNO3.When Ag is present in NaNO3, a decrease in the gas-phase free Ag atompopulation is possible, due to the oxidation of gaseous Ag to form Ag20 byNO2 and 02 formed from decomposition of NaNO3 ( equation 6.2) asmentioned previously [166]. The reduction of interference as the RF powerincrease:, __ggests that gaseous Ag is lost as Ag20 at low RF powers, and thedissociation of relatively unstable Ag20 is increased at high RI? powers( i.e., ar increase in free Ag atom population in the gas-phase ). The bond175I ^1.0 ^0.8 -0.60.4 -0.2^0.0 ^10 15^20^25^30^35^40^45Power ( W)Figure 6.8. Interference effect of 160 ng of Na as NaCI ( • )and 160 ng of Na as NaNO3 ( • ) on the Agemission intensity as a function of RF power.dissociation energy of Ag0 is 2.3 eV, compared with that ofPb0 ( 3.8 eV ) [81].Figure 6.9 depicts the interference effect from different amounts of Na ( asNaC1 ) on the Ag emission intensity for 0.25 ng of Ag at an RF power of 20 W.Similar interference effects for both Pb ( Figure 6.6 ) and Ag ( Figure 6.9 ) arealso consistent with a common effect causing signal suppression when NaC1is present. In addition, except at 14 W, the interference effect of NaC1 onboth Ag ( Figure 6.8) and Pb ( Figure 6.4) as a function of RF power1761 .21 .00.80.60.40.20.00.0E0^1 .0E3^2.0E3^3.0E3^4.0E3Amount of Na ( ng )Figure 6.9. Interference effect on the Ag emission intensityat an RF power of 20 W as a function of amountof Na as NaCI.validates this notion of a common interference effect: loss of analyte-chlorideduring the slow atomization of the analyte. However, as the amount of NaC1increases, the interference effect shows a marginal decrease relative to lowamounts of the interferent ( Figure 6.9 ), probably due to the ionizationsuppression of Ag in the presence of large amount of Na as an EIE. It shouldbe noted that Ag and Na atomic populations co-exist in the gas-phase forAg-NaC1 system. An enhancement of Ag emission was also reported fornitrous oxide-acetylene flame when EIEs were present and was attributed toionization suppression of Ag in the flame [169]1771 .0E3^2.0E3^3.0E3^4.0E3Amount of Na ( ng )Figure 6.10. Interference effect on the Ag emission intensityat an RF power of 20 W as a function of amountof Na as NaC1 in 1 % HNO3 ( 0 ), and Na asNaNO3 ( • ).Figure 6.10 depicts the interference effect on Ag emission from differentamounts of NaC1 in 1 % ( v/v ) HNO3 and NaNO3 in water at an RF power of20 W. Even with very small amounts of interferent, the Ag emissionintensity is greatly depressed. Data provided in Figure 6.10 show that theeffect of NaC1 in HNO3 and NaNO3 is similar because of the conversion ofNaC1 to NaNO3 by HNO3 during the ashing stage or atomization stage.178In contrast to the study by Smith et al. [17], the work presented in thisthesis indicates two different interferent mechanisms from NaC1 and NaNO3on Ag. In the study by Smith et al., analyte and interferents ( both NaC1 andNaNO3 ) were prepared in dilute HNO3. In the presence of HNO3, NaC1 isconverted to NaNO3 during the ashing step of the sample as mentionedpreviously. This conversion is most likely the reason for similar effectsobserved for both NaC1 and NaNO3 [17]. However, Smith et al. reported onlya 50 % reduction of Ag emission ( peak area ) even when 2.3 gg of NaNO3 ( indilute HNO3 ) was present in the sample. It should be noted that theRF plasma source was operated at 27 MHz, and the RF power delivered tothe plasma was not measured [17]. Furthermore, 8 pg of Ag was depositedwith various amounts of interferents, and the blank determination wascarried out by depositing deionized water only.6.3.5 Effects of ascorbic acid and phosphoric acidAscorbic acid has been used as a matrix modifier for Pb determinations inGFAAS [99,170,171]. McLaren and Wheeler reported that the presence ofascorbic acid causes the early appearance of the Pb signal in GFAAS with adoubling of peak [99]. Tominaga and Umezaki reported the appearance of adouble peak in the presence of 0.05 % ( w/v ) ascorbic acid and a single peakwith a lower appearance temperature when 5 % ( w/v ) ascorbic acid waspresent [171]. Since these findings, a number of studies have been reportedon the mechanism of ascorbic acid effect on the Pb signal inGFAAS [103,111,114].179The interest on the effect of ascorbic acid on Pb emission in FAPES stemsfrom the observation of early appearance of Pb emission signal comparedwith the appearance of atomic absorption signal ( Section 4.3.2). This earlyappearance is attributed to the difference in the gas-phase composition inFAPES and GFAAS. This study was undertaken to determine the effect ofgaseous decomposition products of ascorbic acid on the Pb emission signal inFAPES.Figure 6.11 depicts the effect of different concentrations of ascorbic acid onthe temporal response of the Pb emission signal for 0.5 ng of Pb deposited onthe furnace wall at an RF power of 20 W. The appearance and peaktemperatures are the same with and without 0.25 % ( w/v ) ascorbic acid.However, a late shift in the peak temperature can be seen in the presence of1.5 % ( w/v ) ascorbic acid ( long-dash line ).Figure 6.12 presents the peak area of the Pb emission signal as a functionof ascorbic acid concentration. An enhancement of peak area at 0.25 % ( w/v )ascorbic acid followed by a gradual decrease as the ascorbic acidconcentration is increased, are observed. The effect of CO on the gas-phasedissociation of Pb0 is given in the following equilibria ( also seeSection 4.3.2);180 3000250020001 5001 00050000.0^1 .0^2.0^3.0^4.0^5.0^6.0^7.0^8.0Time (s)Figure 6.11. Temporal response of the Pb emission signal atan RF power of 20 W for 0.5 ng of Pb ( — ) ,0.5 ng of Pb in 0.25 % ( w/v ) ascorbic acid ( — ),and 0.5 ng of Pb in 1.5 % (w/v) ascorbicacid ( — — ).2Pb0 (g) 2Pb(g)^02^(6.4),and 02+ 2C0 ^ 2CO2^(6.5).The effect of CO and CO2 concentrations on the gas-phase dissociation ofPb0 is opposing. When CO concentration is increased, 02 concentrationdecreases, which leads to a shift in equilibrium (6.4) to right. Theenhancement of Pb emission in the presence of 0.25 % ( w/v ) ascorbic acid,1815.04.03.02.01 .00.00.00^0.25^0.50^0.75^1.00^1.25^1.50Concentration of Ascorbic Acid ( w/v )Figure 6.12. Interference effect on the Pb emission intensityat an RF power of 20W as a function ofconcentration of ascorbic acid.can be attributed to an increase in the CO concentration, and hence, adecrease in 02 concentration. However, at high CO 2 concentrations,equilibrium (6.5) may be shifted to left. The depression effect and the lateshift of the Pb temporal response can be attributed to an increase in the CO 2concentration as the ascorbic acid concentration is increased.The late shift of Pb in FAPES at higher concentrations of ascorbic aciddiffers from GFAAS observations where an early shift has been reported.This difference in appearance temperature can be attributed to the difference182in the gas-phase concentrations of 0 2 and CO, due to the plasma and to theeffect of ascorbic acid. It seems that the plasma has effectively increasedthe CO concentration ( similar to the effect of ascorbic acid in GFAAS, alsosee Sections 3.3.2 and 4.3.2), and the effect of ascorbic acid in FAPES is toincrease the CO2 concentration such that the net effect is a suppression of Pbemission and a late shift for the appearance of Pb. A late shift in theappearance of Pb atomic absorption signal was reported when CO2 wasadded to the purge gas [1021Figure 6.13 depicts the effect of increasing concentration of ascorbic acidon the temporal response of the Ag emission signal for 0.25 ng of Agdeposited on the furnace wall at an RF power of 20 W. When ascorbic acidconcentration is increased, there is a late shift in the peak temperature butno effect on the appearance temperature is observed. At higherconcentrations of ascorbic acid, Ag emission is an overlap of two peaks. Whenatomic absorption is measured without the plasma ( AA at an RF Power of0 W), no peak shift is observed but there is a decrease in the Ag absorbance.The effect of ascorbic acid on Ag emission in FAPES may be due to theformation of stable molecular species ( for example, Ag 2C2 ) when the plasmais present. The first peak is likely due to the residual Ag present on thegraphite furnace which gives the same appearance temperature. The laterpeak with a higher peak temperature can be attributed to the dissociationof these molecular species.1 5001 20090060030001830.0^1 .0^2.0^3.0^4.0^5.0^6.0^7.0^8.0Time (s)Figure 6.13. Temporal response of the Ag emission signalfor the deposition of 0.25 ng of Ag ( - ) ,0.25 ng of Ag in 0.25 % ( w/v ) ascorbicacid ( - - - ), and 0.25 ng of Pb in 1.5 % ( w/v )ascorbic acid ( - • - • ) at an RF power of 20 W.In addition to ascorbic acid, phosphate modifiers have been used inanalysis of Pb in GFAAS [172-174]. Matousek and Brodie used phosphoricacid to obtain higher ashing temperatures and to obtain more repeatablepeaks [172]. In a later study, Czobik and Matousek suggested aheterogeneous reaction to form metal-pyrophosphate which decomposes at ahigher temperature [173]. In a mass spectral study, Bass and Holcombe0.0^1.0^2.0^3.0^4.0^5.0^6.0^7.0^8.0Time (s)184(f)3002500200c150– 100Ccr)..E50LU0Figure 6.14. Temporal response of the Pb emission signalfor 0.5 ng of Pb (^) , 0.5 ng of Pb in2.5 % ( v/v ) phosphoric acid ( — — ),2.5 % ( v/v ) phosphoric acid blank ( - ), andwater blank ( • • • ) at an RF power of 20 W.proposed the formation of surface bound Pb2P207 which decomposes toform Pb0 at 1150 K and immediate reduction of Pb0 to form Pb [109).Figure 6.14 depicts the effect of 2.5 % ( v/v ) phosphoric acid on thetemporal response of the Pb emission signal for 0.5 ng of Pb at an RF powerof 20 W. The effect of increasing concentration of phosphoric acid on Pb andAg is to late shift the emission peaks for both elements. When Pb isdeposited on the furnace wall and phosphoric acid is deposited on the185RF electrode, no Pb emission signal is observed. This observation suggeststhat phosphoric acid or its decomposition products form stable molecularspecies with Pb in the gas-phase. These results shows that phosphoric acidcan be used as a matrix modifier for Pb when mixed in solution, or it can actas an interferent in the gas-phase.6.4 SUMIVIARYThe effect of NaC1 and NaNO3 on the emission intensity of Pb and Ag inFAPES is studied in an effort to understand the interference mechanism andto distinguish the plasma effects from graphite furnace effects. Theinterferent effect of both NaC1 and NaNO3 is to decrease the emissionintensity for both Pb and Ag. There are different features associated with thetemporal response for analyte with and without interferent which suggestthat the NaCl and NaNO3 interferences are due to two different mechanisms.For Pb in the presence of NaCl and NaNO3, the magnitude of theinterference effect is essentially the same over the power range, 20 to 40 W.However, the temporal response of the Pb signal in the presence of NaNO3shows a different effect from that obtained in the presence of NaCl. Inaddition, experiments carried out by depositing Pb on the furnace wall andinterferent on the RF electrode indicate that both NaC1 and NaNO3interferences are due to two different condensed-phase effects.186The interference effect of NaC1 on Pb is due to the formation ofPb chlorides in the condensed-phase which are lost during the slowatomization step of the analyte. The interference effect of NaNO3 on the Pbemission is due to both condensed-phase and gas-phase effects. Lead oxidemay be trapped in the NaNO3 matrix to yield a late shift in the Pb emissionpeak. The gas-phase dissociation of Pb0 may be hindered by theincreased 02 concentration due to the decomposition products of NaNO3.The heating rate of the graphite furnace likely plays an important role in thedissociation of Pb chlorides and Pb0 in the gas-phase because a highergas-phase temperature can be reached rapidly with a higher heating rate ofthe graphite furnace. Gas-phase reactions are also suggested for the effectsof ascorbic acid and phosphoric acid, widely used matrix modifiers, on the Pbemission intensity.The interference effect of NaCl on the Ag emission is similar to that ofNaC1 on the Pb emission. The interference mechanism is the loss of Ag aschloride during the slow atomization step of the analyte. For Ag in NaNO3,the interferent effect is caused by the loss of Ag as Ag20 in the gas-phase.Silver oxide is formed by the oxidation of gaseous-Ag due to thedecomposition products of NaNO3. The relatively unstable Ag20 isdissociated at high RF powers, and hence, show a reduced interference effectas the RF power is increased.Results obtained for both Pb and Ag show that the major cause ofinterference from NaCl and NaNO3 for these analytes is the chemical187interference effects found in GFAAS. A slight enhancement in atomicemission is observed at higher amounts of Na ( as NaC1 ) relative to lowamounts, probably due to the shift in the analyte equilibrium between atomicand ionic forms of Ag, caused by Na acting as an easily ionizableelement ( EIE ). The temporal overlap of atomic populations of analyte andNa in the gas-phase is only observed for the Ag-NaC1 system.188CHAPTER 7CONCLUSIONSThe basic characteristics of the helium plasma source at 13.56 MHz inFurnace Atomization Plasma Excitation Spectrometry ( FAPES ) have beeninvestigated as a spectrochemical source for elemental analysis. The primaryobjective of this study was to characterize the fundamental processesoccurring within the plasma source during the analyte atomization,vaporization, and excitation. An effort has been made to distinguish theeffect of the plasma on the analyte.Background spectra for the useful analytical wavelengthregion, 200 - 500 nm, are dominated by emission features from COP, N2+,OH, NH, and N2. The plasma background emission is most intense near theRF electrode and less intense near the graphite furnace wall. This spatialdependence of the background emission intensity arises because of thedifference in the RF field strengths at the co-axial RF electrode and at thefurnace wall due to the differences in the surface area of the two electrodes.The temporal response of these molecular emission features, during theatomization step of the analyte, shows complex emission characteristics,probably due to the recombination of ionic species with the thermionicelectrons and dissociation of molecular species. Experimental results alsosuggest the presence of higher levels of CO in the FAPES source comparedwith those in GFAAS. These higher levels of CO in FAPES are due to the189enhanced oxidation of graphite on the RF electrode and on the graphitefurnace wall in the presence of the plasma.The time-averaged Fe and Pb-excitation temperatures are in the range of3000 - 5000 K at RF powers between 10 and 100 W. The Fe-excitationtemperature at 50 W is similar to that of a low power Ar-ICP operating atabout 600 W. The measured Fe-excitation temperature is higher near theRF electrode compared with that near the furnace wall. These highexcitation temperatures in FAPES are likely caused by either lower radiativeand convective heat losses from the plasma in FAPES compared with thosefrom an ICP or higher RF field strength in the plasma in FAPES, or both.These excitation temperatures are also significantly higher than thosereported for CFAES. The emission features of CO+ and N2+ indicate that theFAPES source is capable of exciting energy levels as high as 20 eV. However,the energy transfer mechanism in this case is near-resonant charge transfer.The temporal response of Ag, Pb, and Mn emission signals during theatomization step was studied by simultaneous measurement of atomicabsorption and emission signals. Both atomic absorption and emissionexperiments show a non-uniform temperature distribution along the lengthof the RF electrode, and it is likely that the RE' connector-end of theRE' electrode acts as a heat-sink. The RF electrode heats significantly whenthe plasma is present inside the graphite furnace relative to that without theplasma, but no significant change in the furnace wall temperature isobserved. However, late into the atomization step, the temperature of most190parts of the RF electrode lags relative to that of the furnace wall. Thesedifferences in temperature-time characteristics between the graphite furnaceand the RF electrode cause condensation of analytes on the RF electrode andsubsequent second-surface vaporization resulting in two peaks in thetemporal emission profile. The relative intensifies of these two peaks areaffected by the RF power. Analyte condensation on the RF electrode is severeat low RF powers and can be observed when high amounts of analyte aredeposited. At present, the effect of the heating rate of the graphite furnaceon the temporal response of the emission signals in FAPES is not known.However, it is quite possible that the heating rate of the graphite furnaceaffects the temperature-time characteristics of the RF electrode relative tothe furnace wall, and hence, analyte condensation and second-surfacevaporization characteristics from the RF electrode. Furthermore, resultsobtained for Pb show an early shift in the appearance and peak temperaturesin FAPES compared with those in GFAAS. These early shifts are likely dueto the changes in the gas-phase composition in the FAPES source, and hence,to a shift in the dissociation equilibria of analyte oxide in the gas-phase.The similar peak shapes observed for simultaneously measured atomicabsorption and atomic emission signals for Ag and Pb show that the temporalemission response of these analytes in FAPES is determined by atomizationand vaporization characteristics of the analyte rather than by excitationcharacteristics. In addition, the time-resolved Pb-excitation temperatureshows that the excitation temperature is not affected by the thermionicelectrons during the atomization of Pb. At RF powers below 50 W and191furnace temperatures below 1800 K, conditions under which Ag and Pb arepresent in the plasma, the thermionic electron concentration in the plasma istoo low to significantly affect the excitation characteristics of these analytes.For Mn, an early shift in the emission signal is observed compared withthe simultaneously measured atomic absorption signal, probably due to ahigh energetic excitation region experienced by Mn when second-surfacevaporization occurs from the RF electrode. This observed shift for Mn cannotbe explained by a change in the Mn-excitation temperature. However, athigher furnace temperatures, thermal excitation of analyte and effects ofthermionic electrons may alter the plasma excitation characteristics duringthe atomization step which may result in complex temporal emissioncharacteristics for analytes with higher appearance temperatures.The effect of RF power, and hence, the excitation temperature on theanalyte emission signal was investigated for Pb. Results show a decrease inPb emission intensity at RF powers higher than 50 W. This decrease inemission intensity may be due to a premature loss of analyte, a change inexcitation characteristics or an increase in ionization of Pb at highRF powers. The highest signal-to-noise and signal-to-background ratios areobserved at relatively low RF powers ( about 20 W). Lead condensation onthe RF electrode at low RF powers does not affect the linearity of theanalytical calibration curve, but it does affect the sensitivity and precision forthe analytical determination.192The matrix effects of NaC1 and NaNO3 on Pb and Ag emission intensitieswere investigated to understand the interference mechanism and the effect ofthe plasma. For both Pb and Ag, the major cause of interference effects fromNaC1 and NaNO3 is chemical interference effects found in GFAAS which leadto a decrease in the atomic emission intensity in FAPES. The major cause ofthe signal suppression in a NaC1 matrix is the formation of relatively stablemolecular species and the subsequent losses during the slow atomizationstep. The interference effect for both Pb and Ag in a NaNO3 matrix iscomplex and exhibits both condensed and gas-phase effects. A slightenhancement in the Ag atomic emission intensity is observed at higheramounts of Na ( as NaC1 ) relative to low amounts, probably due to theionization suppression of Ag, caused by Na acting as an EIE. Theco-existence of the atomic populations of analyte and Na in the gas-phase isobserved for Ag-NaC1 system.Most of the experiments described in this thesis were limited to anatomization temperature of 2100 K and RF powers less than 50 W. Athigher furnace temperatures, thermionic electrons affect the plasmaimpedance, and hence, the RF power dissipated by the plasma. The majorcourse of these thermionic electrons is the RF electrode. Because of thereflected power changes late into the atomization step, application ofRF powers higher than 50 W is limited by the appearance temperature ofanalytes and the atomization temperature of 2100 K. Regardless, theseexperimental conditions should be extended to use higher atomizationtemperatures up to 3200 K and to use higher RF powers to achieve the full193analytical capability of graphite furnace methods. The RF power supply andmatching network employed for this study were designed for plasmaprocessing work. Therefore, the obvious remedy for the problem of variableRF power dissipation by the plasma is to modify the matching network todeliver a constant RF power during the high temperature ramp of theatomization step. Another option would be the use of a free-runningRF oscillator. Furthermore, the electrical characteristics of this atmosphericpressure plasma source are not yet understood. To this end, a comprehensivestudy of the current-voltage characteristics of the plasma source should beundertaken.Analyte condensation on the RF electrode at low RF powers can be avoidedby depositing the sample on the RF electrode. The experiments carried outby depositing the sample on the RF electrode show that the RF electrode canbe employed similar to a L'vov platform in GFAAS. When the RF electrode isemployed as a platform to vaporize the analytes, a combination of a high ashtemperature and a high heating rate for the furnace should be used toovercome the analyte condensation on the furnace wall. Alternatively, theplasma can be employed with a pre-heated furnace. When a pre-heatedgraphite furnace is employed, the analyte sample can be introduced into theplasma source via a secondary rod, similar to the L'vov method in the earlyyears of GFAAS. This method should overcome the temporalnon-isothermality and analyte condensation on the RF electrode because thefurnace and the RF electrode are pre-heated before the sample introduction.Furthermore, difficulties arising during a ramp atomization step to maintain194a constant RF power dissipation by the plasma can be avoided because theplasma can be ignited and allowed to stabilize before the sampleintroduction. Difficult and time consuming sample deposition onto thegraphite furnace side-wall can also be avoided by introducing the samplethrough a secondary rod.Moreover, specifically for the experimental system employed for this work,the plasma source work-head should be redesigned to reduce the void volume,to allow more efficient and uniform cooling of the graphite furnace and thefurnace mount, and to use higher furnace temperatures with a hold-step afterthe atomization step. In addition, use of a higher heating rate for thegraphite furnace during the atomization step is beneficial to improve theanalytical characteristics and to reduce the matrix effects in analyticaldeterminations.In summary, this thesis presents the first study of some important basiccharacteristics of the helium plasma source in FAPES as a spectrometricsource for chemical analysis. This study provides a greater understanding ofphysical and chemical phenomena occurring in the plasma at a fundamentallevel. Furthermore, this study identifies the limitations of FAPES andsuggests modifications to instrumentation to improve the analyticalcharacteristics for FAPES determinations. Moreover, this thesis suggestsfurther studies to improve the understanding of fundamental processesoccurring in the plasma.195BIBLIOGRAPHY[1] H. A. Laitinen, Anal. Chem., 2305, 45 ( 1973).[2]A. Walsh, Spectrochim. Acta, 108, 7 ( 1955).[3] M. D. Amos and J. B. Willis, Spectrochim. Acta, 1325, 22 ( 1966).[4]B. V. L'vov, Spectrochim. Acta, 761, 17 ( 1961).[5]W. Slavin, Anal. Chem., 589A, 58 ( 1986).[6]W. G. Schrenk, Appl. Spectrosc., 4,42 ( 1988).[7]S. L. Greenfield, I. L. Jones, and C. T. Berry, Analyst, 713, 89 ( 1964).[8]R. H. Wendt and V. A. Fassel, Anal. Chem., 920,37 ( 1965).[9]W. Shen, J. A. Caruso, F. L. Fricke, and R. D. Satzger, J. Anal. At.Spectrom., 451, 5 ( 1990).[10] J. M. Ottaway and F. Shaw, Analyst, 438, 100 ( 1975).[11] D. Littlejohn and J. M. Ottaway, Analyst, 553, 102 ( 1977).[12] H. Falk, E. Hoffmann, I. Jaeckel, and Ch. Ludke, Spectrochim. Acta,333, 34B ( 1979).[13] H. Falk, E. Hoffmarm, and Ch. Ludke, Spectrochim. Acta,283, 39B ( 1984 ).[14] N. E. Ballou, D. L. Styris, and J. M. Harnly, J. Anal. At. Spectrom.,1141, 3 ( 1988).196[15]D. C. Liang and M. W. Blades, Spectrochim. Acta, 1049, 44B ( 1989).[16]R. E. Sturgeon, S. N. Willie, V. Luong, and S. S. Berman,J. Anal. At. Spectrom., 669, 4 ( 1989).[17]D. L. Smith, D. C. Liang, D. Steel, and M. W. Blades, Spectrochim. Acta,493, 45B, ( 1990).[18]J. G. Grasselli, Anal. Chem., 677A, 64 ( 1992).[19]H. Falk, J. Anal. At. Spectrom., 255, 7 ( 1992).[20]S. R. Koirtyohann, Spectrochim. Acta, 663, 35B ( 1980).[21]S. R. Koirtyohann, Anal. Chem., 1024A, 63 ( 1991).[22]W. Slavin, Anal. Chem., 685A, 54 ( 1982).[23]G. M. Hieftje, J. Anal. At. Spectrom., 117, 4 ( 1988).[24]B. L'vov, Anal. Chem., 924A, 63 ( 1991).[25]G. A. Meyer, Anal. Chem., 1345, 59 ( 1987).[26]J. M. Carey and J. A. Caruso, CRC. Crit. Rev. in Anal. Chem.,397, 23 ( 1992).[27]H. Matusiewickz, J. Anal. At. Spectrom., 171, 1 ( 1986).[28]R. S. Houk, Anal. Chem., 97A, 58 ( 1986).[29]G. M. Hieftje and G. H. Vickers, Anal. Chim. Acta, 1, 216 ( 1989).[30]J. W. Olesik, Anal. Chem., 12A, 63 ( 1991).197[31] D. Slin.kman and R. Sacks, Anal. Chem., 1656, 62 ( 1990).[32]D. Slinkman and R. Sacks, Appl. Spectrosc., 155, 45 ( 1991).[33]D. Slinkman and R. Sacks, Anal. Chem., 343,63 ( 1991).[34]K. W. Jackson and H. Qiao, Ana/. Chem., 50R, 64 ( 1992).[35]G. F. Kirkbright, Analyst, 609,96 ( 1971).[36]A. Syty, CRC. Crit. Rev. in Anal. Chem., 155, 4 ( 1974).[37]R. Woodriff, Appl. Spectrosc., 413, 28 ( 1974).[38]R. E. Sturgeon and C. L. Chakrabarti, Prog. Analyt. Atom. Spectrosc.,5, 1 ( 1978).[39]S. R. Lawson, F. G. Dewalt, and R. Woodruff, Frog. Analyt. Atom.Spectrosc., 1, 6 ( 1983).[40]B. V. L'vov, Spectrochim. Acta, 53, 24B ( 1969).[41] H. Massmann, Spectrochim. Acta, 215, 23B ( 1968).[42]R. E. Sturgeon and C. L. Chakrabarti, Spectrochim. Acta,231, 32B ( 1977).[43]B. V. L'vov, Spectrochim. Acta, 153, 33B ( 1978).[44]W. Slavin and D. C. Manning, Spectrochim. Acta, 701, 35B ( 1980).[45] J. E. Marshall, D. Littlejohn, J. M. Ottaway, J. M. Harnly,N. J. Miller-lhli, and T. C. O'Haver, Analyst, 178, 108 ( 1983).[46] W. Slavin and G. R. Carnrick, Spectrochim. Acta, 271, 39B ( 1984).198[47]S. Wu and C. L. Chakrabarti, Prog. Analyt. Atom. Spectrosc.,111, 10 ( 1987).[48]W. Frech, D. C. Baxter, and B. Hutsch, Anal. Chem., 1973, 58 ( 1986).[49]J. R. Baired and P. D. Coleman, Proc. IRE, 1890, 49 ( 1961).[50]A. M. Howatson, An Introduction to Gas Discharges, Pergamon, London,1965.[51] H. A. Schwab and C. K. Man.ka, J. Appl. Phys., 696, 40 ( 1969).[52]R. F. Hotz, J. Appl. Phys., 1500, 41 ( 1970).[53]H. A. Schwab and R. F. Hotz, J. Appl. Phys., 1503, 41 ( 1970).[54]H. A. Schwab, Proc. IEEE, 613, 59 ( 1971).[55]D. C. Duckworth and R. K. Marcus, Anal. Chem., 1879, 61 ( 1989).[56]M. R. Winchester, C. Lazik, and R. K. Marcus, Spectrochim. Acta,483, 46B ( 1991 ).[57]B. Chapman, Glow Discharge Processes, John Wiley, New York, 1980.[58]S. M. Levitskii, Soy. Phys - Tech. Phys., 887, 2 ( 1957).[59]H. Norstrom, Vacuum, 341, 29 ( 1979).[60]H. R. Koenig and L. I. Maissel, IBM J. Res. Develop., 168, 14 ( 1970).[61] H. Y. Fan, Phy. Rev., 769, 55 ( 1939).[62]J. L. Vossen and J. J. O'Neill, Jr., J. Vac. Sci. Technol., 1052, 12 ( 1975).199[63]J. B. Hasted, Physics of Atomic Collisions, Butterworths, London, 1972.[64]A. von Engel, Ionized Gases, Oxford, London, 1965.[65]H. R. Griem, Plasma Spectroscopy, McGraw-Hill, New York, 1964.[66]P. W. J. M. Boumans, Theory of Spectrochemical Excitation, Hilger andWatts, London, 1966.[67] J. F. Alder, R. M. Bombelka, and G. F. Kirkbright, Spectrochim. Acta,163, 35B ( 1980).[68]M. W. Blades, B. L. Caughlin, Z. H. Walker, and L. L. Burton, Frog.Analyt. Atom. Spectrosc., 57, 10 ( 1987).[6911. Reif, V. A. Fassel, and R. N. Kniseley, Spectrochim. Acta,105, 28B ( 1973).[70]T. D. Hettipathirana and M. W. Blades, Spectrochim. Acta,493, 47B ( 1992).[71] T. D. Hettipathirana and M. W. Blades, J. Anal. At. Spectrom.,1039, 7 ( 1992).[72]P. W. J. M. Boumans, Spectrochim. Acta, 799, 45B ( 1990).[73]P. W. J. M. Boumans, Spectrochim. Acta, 431, 46B ( 1991).[74]R. W. B. Pearse and A. G. Gaydon, The Identification of MoleculerSpectra, Chapman and Hall, London, 1976.[75] C. B. Collins and W .W. Robertson, J. Chem. Phys., 701, 40 ( 1964).[76] M. Endoh, M. Tsuji, and Y. Nishimura, J. Chem. Phys., 5368, 79 ( 1983).[77]R. E. Sturgeon and H. Falk, J. Anal. At. Spectrom., 27, 3 ( 1988).200[78]R. E. Sturgeon, S. N. Willie, V. T. Luong, and S. S. Berman, Anal. Chem.,2370, 62 ( 1990).[79]R. E. Sturgeon, S. N. Willie, V. T. Luong, and J. G. Dunn, Appl.Spectrosc., 1413, 45 ( 1991).[80] R. E. Sturgeon, S. N. Willie, and V. T. Luong, Spectrochim. Acta,1021, 46B ( 1991).[81] A. A. Radzig and B. M. Smimov, Reference Data on Atoms, Molecules,and Ions, Springer-Verlag, Berlin Heidelberg, 1985.[82]A. Kh. Gilmutdinov, Yu. A. Zakharov, and V. P. Ivanov andA. V. Voloshin, J. Anal. At. Spectrom., 505,6 ( 1991).[83]J. M. Bridges and R. L. Komblith, Astrophys. J., 793, 192 ( 1974).[84]D. Littlejohn and J. M. Ottaway, Analyst, 208, 104 ( 1979).[85] S. L. Paveri-Fontana, G. Tessari, and G. Torsi, Anal. Chem.,1032, 46, ( 1974).[86] G. Torsi and G. Tessari, Anal. Chem., 839, 47 ( 1975).[87]D. A. Bass and J. A. Holcombe, Spectrochim. Acta, 1473, 43B ( 1988).[88] S. J. Cathum, C. L. Chakrabarti, and J. C. Hutton, Spectrochim. Acta,35, 46B ( 1991).[89]W. Slavin and D. C. Manning, Spectrochim. Acta, 955, 37B ( 1982).[90]P. G. Riby, J. M. Harnly, D. L. Styris, and N. E. Ballou, Spectrochim.Acta, 203, 46B ( 1991).[91] A. Savitzky and M. J. E. Golay, Anal. Chem., 1627,36 ( 1964).201[92] B. V. L'vov, A. V. Novichikhin, and L. K. Polzik, Spectrochim. Acta,289, 47B ( 1992).[93]P. Hocquellet, Spectrochim. Acta, 719, 47B ( 1992).[94]W. Frech, E. Lundberg, and A Cedergren, Prog. Analyt. Atom.Spectrosc., 257, 8 ( 1985).[95] C. Duval, Inorganic Thermogravimetric Analysis, Elsevier, NewYork, 1963.[96]J. McNally and J. A. Holcombe, Anal. Chem., 1105, 59 ( 1987).[97] R. W. Fonseca, J. McNally, and J. A. Holcombe, Spectrochim. Acta,79, 48B ( 1993).[98]W. L. Winterbottom and J. P. Hirth, The Vaporization Kinetics of SolidSilver, Ed.; E. Rutner, P. Goldfinger, and J. P. Hirth, Condensation andEvaporation of Solids; Science Publishers, New York, 1964; p 348.[99]J. W. McLaren and R. C. Wheeler, Analyst, 542, 102 ( 1977).[100]D. C. Manning, W. Slavin, and S. Myers, Anal. Chem., 2375, 51 ( 1979).[101]S. G. Salmon, R. H. Davis, Jr., and J. A. Holcombe, Anal. Chem.,324, 53 ( 1981).[102]G. F. R. Gilchrist, C. L. Chakrabarti, J. P. Byrne, and M. Lamoureux, J.Anal. At. Spectrom., 175, 5 ( 1990).[103]S. Imai and Y Hayashi, Anal. Chem., 772, 63 ( 1991).[104]W. C. Campbell and J. M. Ottaway, Talanta, 837, 21 ( 1974).[105]R. E. Sturgeon, C. L. Chakrabarti, and C. H. Langford, Anal. Chem.,1792, 48 ( 1976).202[106] M. Suzuki, K. Ohta, T. Yamakita, and T. Katsuno, Spectrochim. Acta,679, 36B ( 1981).[107] G. F. R. Gilchrist, C. L. Chakrabarti, J. T. F. Ashley, and D. M. Hughes,Anal. Chem., 1144, 64 ( 1992).[108] R. E. Sturgeon, D. F. Mitchell, and S. S. Berman, Anal. Chem.,1059, 55 ( 1983).[109] D. A. Bass and J. A. Holcombe, Anal. Chem., 974, 59 ( 1987).[110] J. Drowart, R. Colin, and G. Exsteen, U.S. Govt. Res. Develop. Rept.,20, 40 ( 1965).[111] R. E. Sturgeon and S. S. Berman, Anal. Chem., 1268, 57, ( 1985).[112] A. Cedergren, W. Frech, and E. Lundberg, Anal. Chem.,1382, 56 ( 1984).[113] D. A. Bass and J. M. Holcombe, Anal. Chem., 578, 60 ( 1988).[114] G. F. R. Gilchrist, C. L. Chakrabarti, and J. P. Byrne, J. Anal. At.Spectrom., 533, 4 ( 1989).[115] R. E. Sturgeon and S. N. Willie, J. Anal. At. Spectrom., 339, 7 ( 1992).[116] B. Smets, Spectrochim. Acta, 33, 35B ( 1980).[117] J. Aggett and A. J. Sprott, Anal. Chim. Acta, 49, 72 ( 1974).[118] J. McNally and J. M. Holcombe, Anal. Chem., 1918, 63 ( 1991).[119] C. Th. J. Alkemade, W. Snelleman, G. D. Boutilier, B. D. Pollard,J. D. Winefordner, T. L. Chester, and N. Omenetto, Spectrochim. Acta,383, 33B ( 1978).203[120] M. S. Epstein and J. D. Winefordner, Prog. Analyt. Atom. Spectrosc.,67, 7 ( 1984 ).[121] R. E. Sturgeon, S. N. Willie, V. T. Luong, and S. S. Berman, J. Anal. At.Spectrom., 635, 5 ( 1990).[122] C. H. Corliss and W. R. Bozman, Experimental Transition Probabilitiesfor Spectral Lines of Seventy Elements, NBS Monogrpg # 53, US Dept.Commerce., 1962.[123] R. E. Sturgeon, S. N. Willie, V. T. Luong, and S. S. Berman, J. Anal. At.Spectrom., 19, 6 ( 1991).[124] H. Falk, E. Hoffmann, and Ch. Ludke, Prog. Analyt. Atom. Spectrosc.,417, 11 ( 1988).[125] J. M. Harnly, D. L. Styris, and N. E. Ballou, J. Anal. At. Spectrom.,139, 5 ( 1990).[126] W. M. G. T. van den Broek and L. de Galan, Anal. Chem.,2176, 49 ( 1977).[127] J. M. Ottaway and F. Shaw, Appl. Spectrosc., 12, 31 ( 1977).[128] W. Slavin, and D. C. Manning, Anal. Chem., 261, 51 (1979).[129] B. Welz, S. Akman, and G. Schlemmer, J. Anal. At. Spectrom.,793, 2 ( 1987).[130] M. W. Blades, Ph.D. Thesis, University of Alberta, Canada, 1980.[131] M. W. Blades and G. Horlick, Spectrochim. Acta, 861, 36B ( 1981).[132] R. R. Williams, Anal. Chem., 1638, 63 ( 1991).[133] G. Torsi and G. Tessari, Anal. Chem., 1812, 45 ( 1973).204[134] D. C. Gregoire, C. L. Chakrabarti, and P. C. Bertels, Anal. Chem.,1730, 50 ( 1978 ).[135] N. G. Zhou, W. Frech, and L. de Galan, Spectrochim. Acta, 225, 39B( 1984 ).[136] H. Falk, CRC. Crit. Rev. in Anal. Chem., 29, 19 ( 1988).[137] P. Schramel, Anal. Chim. Acta, 414, 72 ( 1974).[138] R. E. Sturgeon, C. L. Chakrabarti, I. S. Maines, and P. C. Bertels, Anal.Chem., 1240,47 ( 1975).[139] R. E. Sturgeon, C. L. Chakrabarti, and P. C. Bertels, Anal. Chem.,1250, 47 ( 1975).[140] J. M. Harnly, J. Anal. At. Spectrom., 43, 3 ( 1988).[141] E. H. Piepmeier, Anal. Chem., 1296,48 ( 1976).[142] J. M. Laeven and H. C. Smit, Anal. Chim. Acta, 77, 176 ( 1985).[143] E. Voigtman App/. Spectrosc., 237, 45 ( 1991).[144] U. Kale and E. Voigtman Appl. Spectrosc., 1636, 46 ( 1992).[145] G. R. Komblum and L. de Galan, Spectrochim. Acta, 71, 32B ( 1977).[146] A. Batal, J. Jarosz, and J. M. Mermet, Spectrochim. Acta,983, 36B ( 1981).[147] D. Littlejohn, J. N. Egila, R. M. Gosland, U. K. Kunwar, and C. Smith,Anal. Chim. Acta, 71,250 ( 1991).[148] J. P. Matousek, Prog. Analyt. Atom. Spectrosc., 247, 4 ( 1981).205[149]C. Veillon and M. Margoshes, Spectrochim. Acta, 503, 23B ( 1968).[150]P. W. J. M. Boumans and F. J. DeBoer, Spectrochim. Acta,355, 31B ( 1976 ).[151]T. D. Hettipathirana and M. W. Blades, J. Anal. At. Spectrom., in press.[152]J. D. Stafford and J. A. Holcombe, J. Anal. At. Spectrom., 35, 3 ( 1988).[153]C. W. Huie and C. J. Curran, Jr., Appl. Spectrosc., 1307,42 ( 1988)[154]R. Herrmann and C. T. J. Alkemade, Chemical Analysis by FlamePhotometry, Wiley-Interscience, New York, 1963.[155]G. F. Larson, V. A. Fassel, R. H. Scott, and R. N. Kniseley, Anal. Chem.,238, 47, ( 1975).[156]D. J. Kalnicky, V. A. Fassel, and R. N. Kniseley, Appl. Spectrosc.,137, 31 ( 1977).[157]M. W. Blades, and G. Horlick, Spectrochim. Acta, 881, 36B ( 1981).[158]R. Rezaaiyaan, J. W. Olesik, and G. M. Hieftje, Spectrochim. Acta,73, 40B ( 1985).[159]L. J. Prell, C. Monnig, R. E. Harris, and S. R. Koirtyohann,Spectrochim. Acta, 1401, 40B ( 1985).[160]M. S. Hendrick, M. D. Seltzer, and R. G. Michel, Spectrochim. Acta,335, 41B ( 1986).[161]J. P. Matousek, B. J. Orr, and M. Selby, Spectrochim. Acta,415, 41B ( 1986).[162]S. Dahai, Z. Zhanxia, Q. Haowen, and C. Mingxiang, Spectrochim. Acta,391, 43B ( 1988 ).206[163] C. LeBlanc and M. W. Blades, J. Anal. At. Spectrom., 99, 5 ( 1990).[164] E. J. Czobik and J. P. Matousek, An al. Chem., 2, 50 ( 1978).[165] W. Slavin, G. R. Carnrick, and D. C. Manning, Anal. Chem.,163, 56 (1984).[166] R. H. Eklund and J. A. Holcombe, Anal. Chim. Acta, 97, 109 ( 1979).[167]J. A. Holcombe, G. D. Rayson, and N. Akerlind, Jr., Spectrochim. Acta,319, 37B ( 1982).[168] W. Frech, and A. Cedergren, Anal. Chim. Acta, 57, 88 ( 1977).[169] K. Visser, F. M. Hamm, and P. B. Zeeman, Appl. Spectrosc.,620, 6 ( 1976).[170] J. G. T. Regan, and J. Warren, Analyst, 447, 103 ( 1978).[171] M. Tominaga, and Y. Umezaki, Anal. Chim. Acta, 279, 139 ( 1982).[172] J. P. Matousek and K. G. Brodie, Anal. Chem., 1606,45 ( 1973).[173] E. J. Czobik and J. P. Matousek, Talanta, 573,24 ( 1977).[174] D. L. Tsalev, V. I. Slaveykova, and P. B. Mandjukov, Spectrochim. ActaRev., 225, 43 ( 1990).

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

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"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
http://iiif.library.ubc.ca/presentation/dsp.831.1-0059353/manifest

Comment

Related Items