JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY FEBRUARY 1991 VOL. 6 9 Determination of Fluorine in Urine and Tap Water by Laser-excited Molecular Fluorescence Spectrometry in a Graphite Tube Furnace With Front-surface Illumination David J. Butcher Richard L. Irwin Junichi Takahashit and Robert G. Michel* Department of Chemistry University of Connecticut Storrs CT 06269 USA Fluorine was determined in urine and tap water using laser-excited molecular fluorescence spectrometry (LEMOFS) with an unmodified atomic absorption graphite tube furnace. Molecular fluorescence from magnesium fluoride (excited at 268.94 nm detected at 358.82 nm) was collected using front-surface illumination and detection. A frequency doubled excimer pumped dye laser operating at 500 Hz was used for excitation. This is the first report of the use of front-surface illumination for molecular fluorescence.After optimization of the chemical furnace and laser conditions the detection limit of 0.3 pg of fluorine (as fluoride) was from two to six orders of magnitude better than other methods commonly used for the determination of fluorine and two orders of magnitude more sensitive than the best previously reported LEMOFS detection limit. The linear dynamic range of the technique was five orders of magnitude. Significant interferences from other ions (Na+ H+ CI- Br-) were observed. The sensitivity for the determination of fluorine in a freeze-dried urine standard reference material and in tap water by LEMOFS was sufficiently high to allow the samples of be diluted by a factor of 100 to remove the interferences.Good agreement with certified values was obtained with an analytical precision of 7-1 1 %. Keywords Laser-excited molecular fluorescence spectrometry; electrothermal atomizer; fluorine determination Fluorine is a ubiquitous element that is present in minerals water and many animal tissues and has been shown to be es- sential to human life.' It is routinely added to drinking water as fluoride at a level of 1 mg 1-I. However it is necessary to monitor the fluoride concentration carefully because higher levels of fluoride have been shown to cause bone disorders. A variety of methods have been developed for the determi- nation of fluorine. These include the fluoride ion-selective electrode (ISE) photometric techniques ion chromatography atomic spectroscopy and molecular spectroscopy.The fluoride ISE which was first developed by Frant and ROSS,* is the most widely accepted This method which measures only free fluoride in solution requires the addition of a reagent so- lution called the total ionic strength adjustment buffer (TISAB). This reagent serves to maintain a constant high ionic strength to regulate the pH and to release fluorine from its complexes. For practical analysis the detection limit for fluoride is approximately 0.019 mg 1-I (190 ng) with a sample volume of 10 mi and the linear dynamic range for the ISE is four orders of magnitude. Hallsworth et al.s reported a fluoride ISE method that required a 1 pl solution volume. The solution was confined as a thin layer between the fluoride electrode and a calomel electrode mounted directly below it.An absolute de- tection limit of 10 pg was reported and the linear dynamic range was four orders of magnitude. A variety of photometric techniques have been used for the determination of fluorine.' As with the ISE methods the pho- tometric techniques monitor only free fluoride present in solu- tion. The most sensitive photometric method employs sulphonated alizarin fluorine blue ( potassium 3-[N&-di(car- box ymethyl)aminomethyl]- 1,2-dihydroxyanthraquinone-5-sul- phonate} (AFBS)?' with a detection limit of 10 yg 1-' (10 ng). The major advantage of the AFBS photometric method com- pared with the fluoride ISE is improved reproducibility. Disad- vantages include longer analysis time and less tolerance to interferences. Ion chromatography has been used to separate and measure * Present address Department of Chemistry and Physics Western t On leave from Advanced Technology Institute Ryowa Building 4F $ To whom correspondence should be addressed. Carolina University Cullowhee NC 287234 USA.6- I Kamda Surugadai 3 Chiyoda-Ku Tokyo I0 I Japan. the concentrations of a variety of anions including fluoride with a conductivity detection system.R Fritz et aL9 employed a special anion-exchange column with a potassium hydroxide eluent that did not require a suppressor column. The detection limit for fluoride was 1.5 bg 1-I (150 pg) and the linear dynamic range was four orders of magnitude. Ion chromato- graphy is particularly well suited to water analysis because a variety of anions can be determined.Atomic spectrometry is widely used for the determination of metals and metalloids because of the high sensitivity achieva- ble and ability to detect the total amount of all chemical forms of an element. Compared with metals and metalloid elements little research has been performed on investigations into the use of atomic spectrometry for the determination of fluorine. Dittrich'O has summarized the reasons why fluorine has not been commonly determined by atomic spectrometry. Firstly the most sensitive resonance wavelength is at 95 nm which is far into the vacuum ultraviolet. Secondly fluorine is the most electronegative element and therefore forms stable molecules which are difficult to dissociate. Bond and O'Donnelll' determined fluorine by an indirect atomic absorption spectrometric (AAS) method in an air-coal gas flame.The fluorine concentration was monitored by meas- uring the depression of the magnesium AA signal at 285.2 nm. This technique has a variety of disadvantages including poor sensitivity with a detection limit of 0.2 mg 1-I a short linear dynamic range (one order of magnitude) and suffers severe in- terferences from sulphate and phosphate. Gelhausen and Carnahanl* have recently determined fluorine by atomic emission spectrometry in a helium microwave- induced plasma (He MIP). Using direct ultrasonic nebulization of fluoride solution the detection limit for fluorine was 4 ppm at 685.6 nm. The linear dynamic range was 2.5 orders of magnitude. An alternative method to atomic spectrometry is the deter- mination of fluorine by monitoring the spectroscopic proper- ties of diatomic molecules consisting of fluorine and a metal atom introduced as a reagent in a conventional atom cell."' The most widely studied method has been molecular absorp- tion spectroscopy (MAS) which was first investigated by Tsunoda et ~ 1 .l ~ for the determination of fluorine. They used a deuterium arc to excite aluminium fluoride (AlF) molecules in an air-acetylene flame or above a heated graphite rod. A10 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY FEBRUARY 199 I VOL. 6 detection limit of 24 pg I-' as fluoride was obtained with the flame and 21 pg as fluoride with the graphite rod. The method was successfully applied to the determination of fluorine in an agricultural standard reference material and in organofluorine compounds.More recently Tsunoda et al.I4 investigated the relative sen- sitivities of four alkaline earth monofluorides MgF CaF SrF and BaF. A tungsten lamp was used to excite the molecules vaporized from a graphite rod. Strontium fluoride was shown to be the most sensitive molecule. The relative sensitivities of seven monofluorides have also been studied." Aluminium monofluoride was shown to be the most sensitive with a 1% absorbance signal at 20 pg followed by indium monofluoride 110 pg gallium monofluoride 160 pg strontium fluoride 380 pg calcium monofluoride 500 pg magnesium monofluoride 1.5 ng and lithium monofluoride 182 ng. Tsunoda et a1.16 also demonstrated that a platinum hollow cathode lamp (HCL) could be used to excite AIF at 227.45 nm above a graphite rod.The sensitivity with the platinum HCL and with the continuum source were approximately the same.I6 They used this procedure successfully to determine fluorine in ~ r i n e ~ ~ . ' ~ blood serumt7 and milk.'* Other workers have investigated the determination of fluorine by AlF absorbance measurements with Pt HCL exci- tation in a graphite tube furnace. Venkateswarlu et al." used this method for the determination of fluorine in fluorinated organic molecules. They reported that the organic fluorine had to be converted into inorganic fluorine with a sodium biphenyl reagent in order to obtain accurate results. Itai et al.20 investi- gated the effects of chemical modifers and furnace material upon the determination of fluorine by AIF absorbance.They concluded that magnesium nitrate or barium nitrate were ade- quate chemical modifiers and that glassy carbon and synthetic carbon were the most suitable furnace materials. Gomez et al.2' used AlF absorption to determine fluorine in drinking water and sea-water. The MAS results were in good agreement with results obtained with a fluoride ISE. Dittrich and co-workers22-26 investigated the determination of fluorine by molecular absorption spectrometry employing a wide variety of metallic reagents and deuterium arc excitation in a graphite furnace. The detection limit for fluorine using gallium fluoride (GaF) was 1.6 ng.22 Indium monofluoride and aluminium monofluoride were reported to provide fluorine de- tection limits of 0.8 and 2.2 ng respectively.23 A more recent publication demonstrated that a detection limit of 5 ng could be obtained using MgF for the determination of and a review article2s summarized the MAS results obtained. Dit- trich et a1.26 also developed a method for the determination of fluoride by graphite furnace MAS that included an extraction method to separate and pre-concentrate the fluoride.In order to obtain higher sensitivity Dittrich et ai.27 devel- oped a new technique for the determination of fluorine that in- volves laser-excited molecular fluorescence spectrometry (LEMOFS) in a graphite tube furnace. They employed the molecular fluorescence of magnesium fluoride (MgF) and transverse illumination which requires additional ports in the graphite furnace to allow passage of the laser beam at a right angle to the fluorescence axis [Fig.l(a)]. Detection limits of 10 and 40 pg were obtained using resonance (268.9-268.9 nm) and non-resonance fluorescence transitions (268.9-359.3 nm) respectively. A more modem approach to collect fluorescence from a graphite tube furnace is front-surface illumination,2x which is the collection of fluorescence at 180" to the direction of the laser beam [Fig. I(h)]. A mirror through which a hole is drilled to allow passage of the laser radiation is used to collect ff uorescence along the bore of the tube. There are two primary advantages of front-surface illumination compared with trans- verse illumination. Firstly the probe volume which is the volume illuminated by the laser and observed by the detector is approximately ten-times larger for front-surface illumina- To monochromator t LensT7 To monochromator I P Lens I Fluorescence Probe volume Mi!ror T Fig.1 atomizers ( a ) transverse illumination and ( h ) front-surface illumination Two illumination arrangements of LEMOFS in graphite tube tion. Therefore signals are approximately one order of magni- tude bigger than those using transverse illumination.29 Second- ly front-surface illumination does not require modification of the graphite furnace which means that modem furnace techno- logy that has been developed for the determination of trace elements by atomic absorption spectrometry can be used di- rectly for molecular fluorescence. The determination of fluorine by LEMOFS of magnesium monofluoride (MgF) in a graphite tube furnace with front- surface illumination is described in this paper.This first report of LEMOFS with front-surface illumination demon- strates improved sensitivity compared with previous work. The amounts of reagents the furnace conditions and the laser power were optimized to provide the maximum magnesium monofluoride signal. The effects of a variety of potential in- terferent ions were studied. The sensitivity of this method was sufficiently high to allow the samples to be diluted by a factor of 100 which reduced interferences and maintained detection limits that were still 2-3 orders of magnitude better than conventional methods for the determination of fluorine. The potential of this method for real sample analysis was investi- gated by the analysis of a urine standard reference material and tap water.Experimental Instrumentation A schematic diagram of the equipment used is shown in Fig. 2 and the components are listed in Table 1 . The majority of the work was performed with an excimer pumped dye laser and aJOURNAL O F ANALYTICAL ATOMIC SPECTROMETRY FEBRUARY 199 I VOL. 6 11 I Mirror I Furnace power I Fig. 2 Schematic diagram of the instrumentation for LEMOFS Table 1 Instrumentation Componenthlodel number Manufacturer Primary laser system- Excimer laser/800XR Dye laser/DL- 19P Frequency doubler/5- 12 Secondary laser- system- Tachisto Needham MA USA Molectron Santa Clara CA USA h a d Northvale NJ USA Excimer laser/EMG 104 Dye laser/FL 3002 Frequency doublem37- MSC Lambda Physik Acton MA USA Lambda Physik 1 Lambda Physik Other- components- Boxcar averagedl 65 162 Pre-amplifier/VV IOOBTB Monochromator/H- 10 Graphite furnace/HGA-500 Pyrolytic graphite coated graphite tube/BO 109322 PMTD893QB - 3 50 (Lot NO.11-1 2-68) L’vov platform/l09324 Autosampler/AS-40 AT compatible micro- Computer interfaceJlT280 1 -A Triggering circuitry Data processing software Bandpass filter/B390 (Lot NO. 9-05273-164) computer/System 200 PAR Princeton NJ USA LeCroy Spring Valley NY USA Thorn-EMI Fairfield NJ USA ISA Metuchen NJ. USA Perkin-Elmer Norwalk CT. USA Perkin-Elmer Perkin-Elmer Perkin-Elmer Dell Austin TX USA Data Tanslation Marlboro MA USA Laboratory constructed Asyst Software Rochester NY USA Hoya Tokyo Japan frequency doubler operating at 80 Hz.The maximum power at the MgF excitation wavelength (268.94 nm) was 10 pJ per pulse. Fluorescence was detected at 358.82 nm. A second excimer pumped frequency doubled dye laser system was used to optimize the laser power and to obtain the best detection limit. This laser system was operated at 500 Hz with a maximum power of 200 pJ per pulse. For both laser systems a small portion of the excimer laser beam was used to trigger a boxcar averager that was used to process the fluorescence signal from a PMT. Coumarin 540A ( 1 H,4H-2,3,5,6-tetrahydro-8- (trifluoromethyl)quinolizino[9,9a 1 -gh]coumarin) was em- ployed as the laser dye at a concentration of 16 mmol dm-3 dis- solved in methanol. The frequency doubled output was passed through a telescope to adjust the beam size to 3-5 mm in dia- meter before being passed through the atomizer.The atomizer was a Perkin-Elmer HGA-500 graphite tube furnace equipped with an AS-40 autosampler and a L’vov platform. Fluorescence was collected from the graphite tube furnace using front-surface illumination” [Fig. 1 (h)]. A 2 in diameter plane mirror with a in hole in the centre through which the laser beam passed was positioned in front of the furnace at an angle of 45” with respect to the excitation axis. A 10 cm focal length lens and a 5 cm focal length lens were used to collect the fluorescence. The diameter of both lenses was 5 cm. The longer focal length lens was used after the mirror rather than one of shorter focal length because of the physical problem associated with positioning the mirror between the furnace and the lens. The resultant fluorescence was collected by a gated detec- tion system consisting of a photomultiplier tube (PMT) a pre- amplifier and a boxcar integrator. Data from the boxcar were collected by a computer which plotted the fluorescence signal as a function of atomization time and calculated the integrated fluorescence. A colour bandpass filter Hoya B390 (Hoya Tokyo Japan) was used to discriminate against stray laser ra- diation The maximum transmittance (75%) of this filter was at 390 nm; the full width at half maximum was 140 nm.Detection Limits and Calibration Graphs The detection limits were determined after subtraction of the blank signal by extrapolation of the calibration graphs to a signal level equal to three times the standard deviation of 16 measurements of the blank noise.w The measurement of the blank noise was performed either with the laser tuned to the analytical wavelength (the on-line measurement) or with the laser tuned 0.1 nm away from the analytical wavelength (the off-line measurement).Calibration graphs were established by using 20 p1 aliquots of aqueous standard solutions. Attenuation of the LEMOFS signal was necessary to ensure a linear response of the PMT during the construction of calibration graphs. The attenuation was carried out by inserting calibrated neutral density filters between the two lenses. The attenuation of the laser power was done by inserting calibrated neural density filters before the mirror. Standard Solutions and Samples The water supply at the University of Connecticut contains a significant amount of fluoride.Consequently even after de- ionization of the water with ion-exchange columns the concentration of fluoride in the water was approximately 100 pg 1-I. Therefore the de-ionized water was further purified by passage through a sub-boiling quartz distillation unit (Quartz and Slice Paris France) which reduced the fluoride concentration to 10 pg I-’. The standard fluorine (from NaF) magnesium [from Mg(NO,),] and barium [from Ba(NO,),] solutions were pre- pared daily on a class 100 clean air bench by serial dilution of a 1 g I-’ stock solution. Standard solutions with concentrations less than 1 mg I-’ were stored in polyethylene bottles. The stock solutions were made from high-purity salts (Spex Indus- tries Metuchen NJ USA).Two freeze-dried urine samples were obtained from the Na- tional Institute of Standards and Technology (SRM 267 1 A). These were a low fluoride sample and an elevated fluoride sample. These samples were reconstituted with de-ionized water as directed. Results and Discussion Chemical and Spectroscopic Properties of Magnesium Monofluoride (MgF) Searcy3’ considered the formation of diatomic molecules in a chapter on high temperature inorganic chemistry. High tem- perature was defined as a temperature at which entropy differ- ences have a significant influence in determining the reaction equilibria of interest. Under these high-temperature conditions fluorine was shown to form stable diatomic molecules with all elements in the Periodic Table except for the rare gases.Magnesium monofluoride (MgF) which was first identified12 JOURNAL O F ANALYTICAL ATOMIC SPECTROMETRY FEBRUARY 1991 VOL. 6 by Datta,32 has been characterized by the presence of three band systems:33 A211 t X,C+; B2C+ t X2C+; and C2C+ c X2C+; where X2Z+ is the ground state. The A2n t X2Z+ system which is composed of three series of doublets is the most intense system with the highest sensitivity at the 358.82 nm (0,O) vibrational transition.33 The B2C+ t X,Z+ band system has the highest sensitivity at the 268.94 nm (0,O) vibrational transition.33 These two systems have been observed in a direct current (d.c.) arc at 1150 "C3' The C,C+ t X,C+ system is much less intense than the other two with maximum sensitivity at the 234.78 nrnL(O,O) transition.Fuwa and c o - ~ o r k e r s ~ ~ ~ ' ~ employed the 358.82 nm wave- length for the determination of fluorine by MgF molecular ab- sorbance spectrometry (MAS) in a graphite furnace. They reported 1% absorbance at 1.5 ng. Dittrich and co- w o r k e r ~ ~ ~ - ~ ~ . ~ ~ . ~ ~ compared the sensitivity of the 268.94 and 358.82 nm wavelengths for graphite furnace MAS. The former band was approximately twice as sensitive as the latter.25 Both groups reported that MgF was approximately two orders of magnitude less sensitive than AlF by MAS which is the most sensitive molecule for fluorine determination. Dittrich et al.27 determined fluorine by LEMOFS of MgF in a graphite tube furnace with transverse illumination [Fig. l(a)]. They employed MgF rather than AIF because the AIF wavelength (227.45 nm) was inaccessible with their laser system.They obtained a detection limit of 45 pg using a non- resonance transition (excited at 268.94 nm detected at 358.82nm). A detection limit of 1Opg was achieved using resonance fluorescence (excitation and detection at 268.94 nm). For this work MgF was employed for the determination of fluorine by LEMOFS. Aluminium fluoride was not used because the primary laser system could not provide the AIF wavelength. Since performing the bulk of this work it became 8000 4000 .- 2' f feasible to determine aluminium fluoride with the secondary laser system although we did not do this. Excitation was per- formed at the MgF B2C+ t X,C+ (0,O) transition (268.94 nm) and non-resonance fluorescence was detected at the A2n t X2C+ (0,O) transition (358.82 nm).Front-surface illu- mination was employed to improve the sensitivity of the in- strument and to allow the use of unmodified atomic absorption graphite tube furnaces. The non-resonance transi- tion was employed to reduce stray light noise. Chemical Optimization Dittrich and c o - w o r k e r ~ ~ ~ - ~ ~ - ~ ~ summarized procedures to opti- mize the chemical conditions for the determination of non- metals (X) by measuring molecular absorbance of a diatomic molecule (MX) where M is a metallic reagent. The chemical parameters to be optimized included the amount of the metal- lic reagent magnesium; the amount of acid or base required to control the pH; and the amount of chemical modifier barium or strontium.Dittrich and c o - w o r k e r ~ ~ ~ ~ ~ ~ ~ ~ ~ demonstrated that excess of metallic reagent (Mg) should be introduced into the furnace to react with all the analyte non-metal (F). They also proposed that the formation of diatomic molecules in a graphite furnace involves gas-phase combination of the metallic reagent and the analyte. A reaction that competes with the formation of MgF is the formation of magnesium difluoride (MgF,). If the gas phase contains excess of magnesium relative to the fluorine then the equilibrium that controls the system is given by31 (1) Complete conversion of fluorine into MgF is likely by use of excess of magnesium. The results obtained for the optimiza- tion of the MgF signal as a function of the amount of magne- sium introduced (as magnesium nitrate) are summarized in Mg(g) + MgF,(g) + 2MgF(g) 200 loo v 1x104 lxlo-z 1x10' 1x102 1x10' Amount of OH-(as NaOH)/pg Amount of Srlpg Amount of B4pg Fig. 3 Optimization of the reagents for LEMOFS.(a) Effect of magnesium added as Mg(NO,) on the MgF fluroescence signal. Experimental condi- tions NaF (10 ng as F); Ba(N03)? (1.65 pg as Ba); atomization temperature 1800 "C; char temperature 800 'C; and laser power 10 pJ per pulse. ( h ) Effect of hydroxide added as NaOH on the MgF fluroescence signal. Experimental conditions NaF (10 ng as F); Mg(NO,) (20 pg as Mg); atomization temperature 1800 "C; char temperature 800 "C; and laser power 10 yJ per pulse. (c) Effect of barium added as Ba(NO& on the MgF fluorescence signal. Experimental conditions NaF (10 ng as F); Mg(NO,) (20 pg as Mg); atomization temperature 1800 "C; char temperature 800 "C; and laser power 10 pJ per pulse.( d ) Effect of strontium added as Sr(NO& on the MgF fluorescence signal. Experimental conditions NaF (10 ng as F); Mg(N03) (20 pg as Mg); atomization temperature 1800 "C; char temperature 800 'C; and laser power 10 pJ per pulseJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY FEBRUARY 1991 VOL. 6 13 Fig. 3(a). A maximum fluorescence signal was obtained with 20pg of magnesium which is within a factor of two of Dittrich’s optimized amount of lOpg.,’ The reason for the decrease of the MgF signal when more than 20 pg of magne- sium were added to the furnace is not yet known. Work is being done in this laboratory to explore more fully the mecha- nism of MgF formation in a graphite tube furnace.Dittrich’O investigated the effect of pH upon the formation of diatomic molecules and demonstrated that basic conditions enhance the formation of many molecules. This enhancement was attributed to suppression of the high-temperature hydroly- sis of hydrated salts that may be formed during drying of the sample. lo Dittrich and co-workers10~2s~27 also reported that the molecu- lar fluorescence signal can be enhanced by the introduction of an additional metal compound a chemical modifier that forms an insoluble salt with the analyte anion. It was stated that the chemical modifiers provide a nearly simultaneous evaporation of the fluoride and the magnesium. These workers employed Ba(OH) or Sr(OH) which maintained basic conditions to eliminate hydrolysis and allow coincidental evaporation.However Ba(OH) or Sr(OH) were not employed in the present work because it would have been necessary to keep these solutions under an atmosphere of nitrogen in order to prevent precipitation of BaCO or S I - C O ~ . ~ ~ Consequently the effect of pH upon the MgF signal was investigated by the addi- tion of hydroxide (added as sodium hydroxide) and the amount of barium (added as barium nitrate) or strontium (added as strontium nitrate) was optimized separately. The investigation of the effect of sodium hydroxide upon the MgF fluorescence signal is shown in Fig. 3(h). When amounts of hydroxide of less than approximately 1 pg were in- troduced into the furnace no change in the MgF fluroescence signal was observed because MgF does not form a hydrated The addition of large amounts of hydroxide (greater than 10 pg) reduced the MgF signal probably because the hydrox- ide competed with the fluorine for magnesium decreasing the amount of MgF formed in the graphite furnace.Dittrich et ~ 1 . ~ ~ did not investigate the effect of sodium hydroxide upon the MgF signal for LEMOFS probably because a basic compound (barium dihydroxide) was added as the chemical modifier. Either barium nitrate [Ba(NO,),] or strontium nitrate [Sr(NO,),] were used as the chemical modifier rather than the hydroxides because these solutions were stable in air. The ad- dition of barium nitrate caused a significant enhancement in the MgF signal [Fig. 3(c)] with a maximum enhancement of 100 times for the introduction of 1-2 pg of barium.For routine analytical work 1.65 pg of barium were added to the furnace. The addition of larger amounts of barium to the furnace caused suppression of the MgF signal which was probably due to increased formation of BaF relative to MgF. Dittrich et al.27 reported that the optimized amount of barium (added as barium hydroxide) was 13 pg which is a factor of ten higher than the optimized amount of barium nitrate obtained in the present experiments. These data indicate that barium nitrate is a more effective chemical modifier than barium hydroxide. Strontium nitrate was also shown to enhance the MgF signal [Fig. 3(6)] to give a maximum 10-fold increase in signal size with the addition of 10 pg of strontium.The MgF signal was suppressed with the addition of very large amounts of stron- tium probably due to an increase in the formation of SrF rela- tive to MgF. Barium was used as a chemical modifier rather than strontium because a larger increase in signal was obtained with a lower amount of reagent. Dittrich et ~ 1 . ~ ~ did not use strontium as a chemical modifier for LEMOFS. The mechanism for the enhancement of the MgF signal by barium has not been investigated experimentally. Pertinent thermodynamic and physical properties of fluorine barium and magnesium compounds that are likely to be formed in the graphite furnace are given in Table 2 in order to help propose a mechanism for the enhancement induced by barium. Dit- trichIo suggested that the formation of diatomic molecules in- volves gas-phase reaction between the metallic reagent (Mg) and the analyte (F).In the absence of a chemical modifier fluorine is believed to evaporate at a lower atomization tem- perature or temporally sooner in the graphite furnace than magnesium. The observed enhancement in the presence of barium was attributed to more nearly simultaneous evapora- tion of magnesium and fluorine. This enhancement may be caused either (i) the evaporation of fluorine at a higher atomi- zation temperature or temporally sooner in the graphite furnace or (ii) the evaporation of magnesium at a lower atom- ization temperature or temporally later in the graphite furnace. Dittrich et al.,’ suggested that this enhancement is caused by the preferential precipitation of BaF relative to MgF after ad- dition of the reagents to the furnace followed by nearly simul- taneous vaporization of BaF and Mg.Although barium does form an insoluble fluoride the solubility of barium difluoride is greater than the solubility of magnesium difluoride (Table 2). Thus there should be little formation of BaF,(s) in the graphite furnace following addition of the reagents which to some extent invalidates Dittrich’s precipitation hypothesis. Another approach to investigating the signal enhancement by barium is to consider the atomization mechanisms of barium and magnesium in graphite tube furnaces. Although magnesium is a relatively volatile metal with a boiling-point of 1090 “C the recommended atomization temperature for magnesium is 2400 “C.377.38 The mechanism of atomization for magnesium has been investigated because magnesium nitrate is commonly added to graphite furnaces as a chemical Table 2 Physical and thermodynamic properties of barium and magnesium compounds.All data were taken from reference 35 unless indicated otherwise Compound Boiling-point/”C Solubility/g 1-’ HF*/kcal mol-’ HEtkcal mol-I T,,SI”C Ba BaO BaF2 NaF 5aC2 1090 3600 2239 1640 2000 2137 - - - 0.0062 0.076 - 1.2 - 0 -143.8 -263.5 0 - 133.4 -286.9 - 42.0 69.8s - - * HF heat of formation. -t HE heat of evaporation. 4 T,,,. formation temperature. 5 From reference 36. 4 From reference 33.14 JOURNAL O F ANALYTICAL ATOMIC SPECTROMETRY FEBRUARY 1991 VOL. 6 modifier. Slavin and c o - w ~ r k e r s ~ ~ . ~ ~ proposed that Mg(N0,)2 is converted into magnesium oxide in the solid phase prior to vaporization.Magnesium oxide is non-volatile (Table 2) with a boiling-point of 3600 "C. A possible reason for the MgF signal enhancement is the conversion of relatively non-volatile magnesium oxide by barium (or barium compounds) into more volatile magnesium metal (boiling-point 1090 "C) causing more coincidental evaporation of magnesium and fluorine. The determination of barium in a graphite tube furnace has been shown to be inhibited by formation of carbide.3x4 Frech et al.39 reported that the tailing observed in barium peaks in graphite furnace atomic absorption may be caused by carbide formation. StyriP proposed that the mechanism involved the reaction BaC,(s) +BaC,(g) +Ba(g). Although we have no experimental evidence to support this perhaps barium carbide formation is involved in the enhanced sensitivity of MgF (2) This mechanism would cause vaporization of magnesium at a lower temperature and temporally sooner in the graphite furnace atomization step.This would cause more coincidental evaporation of magnesium and fluorine and a larger MgF signal. Although these processes may influence the mechanism a rigorous study of all the species present their concentrations and their temporal appearance is necessary to define the chemi- cal reactions involved in the enhancement of MgF by barium. These studies are currently being performed in this laboratory. BaC + 2Mg0 + 2Mg + Ba + 2CO Furnace Optimization The atomization and char temperatures were optimized to obtain the maximum MgF fluorescence signal with the front- surface illumination system.Dittrich ef al.27 reported that an atomization temperature of between 2700 and 3000 "C was re- quired to obtain optimum sensitivity for MgF fluorescence for the determination of fluorine. Fig. 4(a) shows the atomization obtained using optimized amounts of magnesium nitrate (20 pg as magnesium) and barium nitrate (1.65 pg as barium). The fluorescence signal increases with temperature from 1300 to 1800 "C slightly drops off between 1800 and 2100 "C and then is relatively constant up to 2700 "C. The optimum temperature of 1800 "C for the instrument used was approximately lo00 "C lower than Dittrich et al. re- ported for their transverse optical axes. This relatively low atomization temperature tends to support the hypothesis that fairly volatile magnesium and fluorine were vaporized rather than other less volatile species such as magnesium oxide.The atomization temperatures required for LEMOFS would be ex- pected to be the same (within k200 "C) for transverse and front-surface illumination because of previous experience with transverse illumination for laser-excited atomic fluorescence spectrometry (LEAFS).4' Therefore the higher atomization temperature in Dittrich's magnesium fluoride work was probably due to other aspects of the instrumentation and protocol that could not be identified rather than in the use of transverse illumination. Dittrich ef al.27 did not report the use of a charring step in their LEMOFS work. The results obtained for optimization of the charring step using the optimized amounts of Mg (20 pg) and Ba (1.65 pg) and an atomization temperature of 1800 "C are shown in Fig.4(h). The maximum fluorescence signal was obtained with charring temperatures of between 800 and 1 100 "C. The use of temperatures greater than 1 100°C caused a decrease in signal with temperature due to the removal of analyte during the charring step. A charring temperature of 800 "C was used for the remainder of this work. The optimized furnace programme is shown in Table 3. Laser Optimization The laser power was optimized by obtaining a saturation curve 1200 lo00 800 f I 8 01 J I I I I I 6 1200 -1400 1600 1800 2000 2200 2400 2600 2800 i I 0 200 400 600 800 loo0 1200 1400 1600 Char temperature/"C Fig.4 Optimization of furnace parameters for LEMOFS. (a) Effect of atomization temperature on the magnesium fluoride fluorescence signal. Experimental conditions NaF (10 ng as F); Mg(NO,) (20 pg as Mg); Ba(NO,) (1.65 pg as Ba); char temperature 800 "C; and laser power 10 pJ per pulse. (h) Effect of char temperature on the magnesium fluoride fluorescence signal. Experimental conditions NaF (10 ng as F); Mg(NO,) (20 pg as Mg); Ba(NO,) ( 1.65 pg as Ba); atomization temperature 1800 'C; and laser power 10 pJ per pulse Table 3 LEMOFS of MgF Furnace programme for the determination of fluorine by Internal flow*/ Step T /"C Ramp/s Hold time/s ml min-' Dry 200 20 40 300 Char 800 1 30 300 Cool 20 1 10 300 Atomize 1800 O t 3 0 Clean 2700 1 5 300 COO1 20 1 20 300 * Argon.f Maximum power heating (approximately IS00 'C SKI) Table 4 Optimized chemical. furnace and laser conditions for MgF for the determination of fluorine by LEMOFS Chemical conditions Furnace conditions Laser conditions 20 pg of Mg [added as Mg(N0,)2]; 1.65 pg Ba [added as Ba(NO,),] Atomization temperature 1800 "C; char temperature 800 "C Laser power 100 pJ per pulse at 268.94 nm (Fig. 5 ) with the more powerful laser system and the optimum chemical (Mg 20 pg; and Ba 1.65 pg) and furnace conditions (Table 3). Fig. 9 shows that the fluorescence signal was direct- ly proportional to the laser power up to 100 pJ per pulse. The best LEMOFS detection limit was obtained by use of a laser power of I 0 0 p.I per pulse that saturated the MgF transition. A summary of the optimum chemical furnace and laser con-JOURNAL O F ANALYTICAL ATOMIC SPECTROMETRY FEBRUARY 199 I VOL.6 15 1000 I 0.01 ' I I 0.1 1 10 100 lo00 Laser power/@ per pulse Fig. 5 Saturation curve for magnesium fluoride which demonstrates the effect of laser power on the MgF signal. Experimental conditions NaF (10 ng as F); Mg(N0,)2 (20 pg as Mg); Ba(NO,)? (1 -65 pg as Ba); atomization temperature 1800 'C; and char temperature 800 'C 10 1 I 0 1 2 3 t l S Fig. 6 Temporal fluorescence signal obtained for 8 ng of F. Experimen- tal conditions NaF (10 ng as F); Mg(NO,) (20pg as Mg); Ba(NO,) (1.65 pg as Ba); atomization temperature 1800 "C; char temperature 800 "C; and laser power 10 pJ per pulse .- a 2500 (I) E .r 2000 1,l I i i 2G.W 268.25 268.50 268.75 269.00 269.25 Unm Fig. 7 Spectral scan for magnesium fluoride which is a plot of MgF signal w x u s the laser wavelength.Experimental conditions NaF (10 ng as F); Mg(N03)2 (20 pg as Mg); Ba(N03)? (1.65 pg as Ba); atomization temperature 1800 "C; char temperature 800 "C; laser power 10 pJ per pulse at all wavelengths; and approximate laser line width 0.003 nm ditions for the determination of fluorine by LEMOFS using MgF is given in Table 4. A LEMOFS signal of 8 ng of fluoride under these optimized conditions is illustrated in Fig. 6. MgF Excitation Profile Fig. 7 shows an excitation profile for MgF which is a plot of excitation wavelength obtained by tuning the laser versus relative fluorescence signal. The observed bands matched the I I .c 1x10~ C c .- Q 1x10' 8 2 Q) 1x10-' 3 = .- CI - Q K iX10-' L 1 1 I 1 1 lXlO-' 1x10-2 1x10' 1x102 1x10' Amount of fluorinelng Fig. 8 Calibration graph for fluorine by magnesium fluoride fluorescence which is a plot of MgF fluorescence signal versus amount of fluorine (as NaF) added. Experimental conditions NaF (10 ng as F); Mg(NO,) (20 pg as Mg); Ba(NO,) (1.65 pg as Ba); atomization tempera- ture 1800 "C; char temperature 800 "C and laser power 10 pJ per pulse reference values compiled by Rosen,33 at 268.94 268.65 268.38 and 268.13 nm.These four bands corresponded to the (O,O) (1 ,l) (2,2) and (3,3) vibrational transitions respectively associated with the B2Z+ t X,C+ electronic transition. All ana- lytical work was performed with the laser tuned to 268.94 nm which is the peak of the most intense (0,O) vibrational transition. Detection Limits and Linear Dynamic Range A calibration graph for the determination of fluorine by LEMOFS is shown in Fig.8. Using the optimized chemical and furnace conditions and the maximum laser power from the primary laser system (10 pJ per pulse) at a repetition rate of 80 Hz a detection limit of 2 pg of fluorine was obtained. Using the secondary laser system at a repetition rate of 500 Hz the laser power was optimized (100 pJ per pulse) to obtain a detection limit of 0.3 pg. The linear dynamic range (LDR) was five orders of magnitude which is the first LDR reported for LEMOFS. The observed improvement in sensitivity between the two lasers can be predicted from signal to noise ratio considera- t i o n ~ . ~ " ~ ~ The laser power of the secondary laser was ten times larger than the primary laser system which between 10 and 100 pJ would lead to a 10-fold increase in the signal.Owing to increased laser stray light the signal to noise ratio is propor- tional to the square root of the laser power so the detection limit should improve by a factor of 6 0 = 3.2.2R.42 The repeti- tion rate of the secondary laser was five times higher than that of the primary laser system. The signal to noise ratio is propor- tional to the square root of-the repetition rate so an improve- ment in detection limit of 45 = 2.2 was expe~ted.?~.~~ The over- all predicted improvement in the detection limit was therefore 7 which is in good agreement with the experimentally ob- served improvement of 6.6. A summary of all LEMOFS detection limits for the determi- nation of fluorine by MgF fluorescence is given in Table 5.Dit- trich et al." obtained a detection limit of 40 pg using transverse illumination wall atomization and a non-resonance (268.9- 359.4 nm) transition. Using the same illumination geometry and sample introduction the fluorine detection limit was im- proved by a factor of four (10 pg) by use of a resonance (268.94-268.94 nm) transition. In this work with front-surface illumination and non-resonance fluorescence a detection limit of 2 pg was obtained with a laser power of 10 pJ per pulse at a laser repetition rate of 80 Hz. Using the same illumination geo- metry and the same transition the detection limit was improved by a factor of 6.6 (0.3 pg) by the use of a more modem laser system which provided 100 CLJ per pulse at a repetition rate of 500 Hz.The results reported here are 5-20 times more sensitive16 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY FEBRUARY 199 I VOL. 6 ~~ Table 5 Summary of LEMOFS detection limits obtained with MgF fluorescence for the determination of fluorine Detection limit Researchers Illumination Transit ion/nm pg ng ml-I Dittrich ef a/.* Transverse 268.9-268.9 10 0.5 Dittrich ef al.* Transverse 268.9-359 40 2 This work? Front surface3 268.9-359 2 0. I This work? Front surfaces 268.9-359 0.3 0.015 * Taken from reference 27. Experimental conditions Mg(NO& (10 Fg as Mg); Ba(OHI2 (26 pg as Ba); atomization temperature 2700-3000 'C; and charring temperature none. t This work. Experimental conditions Mg(N03)2 (20 pg as Mg); Ba(NO& ( I .65 pg as Ba) atomization temperature 1800 'C and charring temperature 800 'C.$ A laser power of 10 pJ per pulse was used at a repetition rate of 80 Hz. J A laser power of 200 pJ per pulse was used at a repetition rate of 500 Hz. Table 6 Comparison of detection limits obtained by common techniques used for the determination of fluorine Detection limit ~~ Technique Method Pg Pg I-' LEMOFS* MgF 0.3 0.015 MAST AIF 20 1 Spectrophotometryt Sulphonated alizarin 1x104 10 ISES TISAB water 1 . 9 ~ lo5 19 * This work by LEMOFS in a graphite tube furnace with platform atomization t Taken from reference 15 by molecular absorption spectrometry with a platinum $ Taken from reference 6 by spectrophotometry with sulphonated alizarin fluorine 5 Taken from reference 4 with an ISE with an aqueous total ionic strength adjust- and front-surface illumination.hollow cathode lamp and wall atomization. blue reagent. ment buffer. than previously reported LEMOFS results for fluorine. A comparison of the best LEMOFS detection limit obtained (0.3 pg) with other methods for the determination of fluorine is given in Table 6. The best molecular absorption spectrometry detection limit for fluorine is 20 pg obtained by Tsunoda et aE.I5 which is two orders of magnitude less sensitive than the LEMOFS detection limit. A commonly used spectrophoto- metric method involves the formation of a complex between free fluoride ion and sulphonated alizarin fluorine blue. Leonard and Murray6 reported a detection limit of 10 ng which is almost five orders of magnitude worse than the LEMOFS detection limit.The most widely used technique for the determination of free fluoride is the ISE. For routine analy- sis the absolute detection limit for fluoride is 190 ng,4 which is six orders of magnitude worse than the LEMOFS result. These data show that LEMOFS is considerably more sensitive than the methods commonly used for the determination of fluorine. Effects of Other Ions on the MgF Signal Dittrich et al.27 investigated the effects of concomitant salts upon the MAS and LEMOFS signal sizes for MgF. Severe signal depressions were observed in the presence of excess of sodium chloride and bromide at levels 10-1OOO times higher than the level of fluorine. Here the effects of a variety of salts on the MgF fluorescence signal were investigated for the front- surface approach where interferences might be expected to be lower than the transverse approach.Open atomizers such as a graphite tube furnace modified for transverse illumination have been shown to be susceptible to vapour phase interfer- ences in AAS3' and in LEAFS.2n The effect of Na+ added as NaNO on the MgF fluorescence signal is demonstrated in Fig. 9(a). The optimized chemical conditions were used with 10 ng of fluorine at an atomization temperature of 1800 "C. The fluorescence signal was sup- pressed when more than 100 ng of sodium were added to the sample which represented a 10-fold excess on a per mass basis. The suppression was probably due to a reduction in the formation of MgF and an increase in the formation of NaF in the presence of large amounts of sodium. Similar behaviour was observed when Na' was added in the form of NaOH.Fig. 9(h) demonstrates that there was no suppression of the fluorescence signal until a 200-fold excess of sodium relative to fluorine was present. The effect of H' added as nitric acid upon MgF fluorescence was also investigated. Using the optimized chemical and furnace conditions the fluorescence signal de- creased when a 100-fold excess of nitric acid was present [Fig. 9(c)] which suggests that it may be difficult to determine fluorine in samples that have been dissolved by acid digestion unless steps are taken to remove the acid. The influence of anions on LEMOFS was also investigated. Fig. 9(4 is a plot of MgF fluorescence signal versus amount of C1- added as NH,Cl.A suppression of the fluorescence signal was observed when a 100-fold excess of chloride relative to fluoride was present. Similar behaviour was observed when bromide (added as ammonium bromide) was introduced as a 100-fold excess of bromide suppressed the MgF signal [Fig. 9 (e)]. These anions probably suppressed the MgF signal by competing with fluorine for magnesium. As the amount of MgCl or MgBr was increased there was insufficient magne- sium to obtain the maximum amount of MgF. In summary all ions that were investigated caused a sup- pression of the MgF signal when in excess at levels of between 10 and 200 times. The suppressions reported here with front-surface illumination occurred at approximately the same concentration as those reported by Dittrich et a/.*' with transverse illumination. These data indicate that an unmodified graphite furnace was not more effective at reducing vapour phase interferences for LEMOFS than a furnace modified for the use of transverse illumination.The suppression due to cations (Na' H') probably occurred because the suppressing cation competed with magnesium for fluorine which caused a decrease in the level of MgF fluorescence. At this point it is unclear why sodium added as the nitrate caused a greater signal suppression than sodium added as the hydroxide. The anions probably suppressed the MgF fluorescence signal by competing with fluorine for the magnesium. It was concluded from this study that a variety of ions depress the fluorescence signal which indicated that it may be difficult to use LEMOFS in the presence of a sample matrix.Accordingly the technique was tested by performing some real sample analyses. Real Sample Analysis The first sample studied was a National Institute of Standards and Technology (NIST) Standard Reference Material (SRM) 267 1 a Freeze-Dried Urine. Two samples are provided with this SRM a control level and an elevated level. The concentra-A c c .c 4Ooo - P) g 3000 - 8 2 2000 - 2 5 1000 - > .- * 1 1 tion of fluorine present in the tap water at the University of Connecticut was also determined by LEMOFS. This result was compared with a value obtained with a fluoride ISE. The results obtained for the determination of fluorine in the control level Freeze-Dried Urine sample along with the certified value are given in Table 7.Initially the analysis was carried out by diluting the urine sample ten times. The experi- mental value for fluorine concentration obtained by LEMOFS of 0.259 k 0.044 mg I-' was more than a factor of two lower than the certified value of 0.55 k 0.030 mg I-'. After diluting the sample 30-fold an experimental value of 0.373 mg I-' of fluorine was obtained which was significantly closer to the certified value than the previous result. Finally when the sample was diluted 100 times the fluorine concentration was determined by LEMOFS to be 0.541 f 0.036 mg I-' which was shown to be the same as the certified value by Student's t- test. This study demonstrated that LEMOFS could be used for real sample analysis if the matrix was diluted sufficiently to reduce interferences from ions.This would seem to contradict the interferences studies shown previously which indicated that the mass ratio of the interfering species to the analyte was the important parameter as that ratio is not changed by dilu- tion. However it is possible that at high concentrations of matrix constituents the ratio of interferent to analyte is the governing factor while at low concentrations the absolute amount of the interferent is the governing factor. The results for all of the real sample analyses are summar- ized in Table 8. As was discussed above good agreement with the certified value was obtained by LEMOFS for the control level urine sample by diluting the sample 100 times. Analysis (b) ( C) 800 - 400 200 - 2000 - - I I I Table 7 Determination of fluorine by LEMOFS in NIST SRM 2671a.Freeze-Dried Urine. Molecular fluroescence of MgF excitation at 268.94 nm detection at 359 nm. Experimental conditions Mg(NO,) (20pg as Mg); Ba(NO,) ( I .65 pg as Ba); atomization temperature. 1800 "C; char- ring temperature 800 "C; laser power 10 pJ per pulse. The data are k one standard deviation 5Ooo P) = 1ooo- Q 2.s 3000(1 :f 2000- S E > .- + 1000- K 0 - .- CI Sample Concentration of F/mg I-' RSD '24 2500 2000 1500 lo00 - 500 - 0 ( dl (4 1 I I Certified value 0.550 & 0.030 10-fold dilution 0.259 k 0.044 30-fold diiution 0.373 f 0.053 100-fold dilution 0.541 f 0.036 - 16.9 14.2 6.7 of the elevated level urine sample using a 100-fold dilution by LEMOFS also gave a value which was statistically the same as the certified value.The analytical precision of the LEMOFS measurements was between 7.1 and 10.7%. The concentration of fluorine in the University of Connecti- cut tap water was determined by LEMOFS to be 0.70 k 0.05 mg I-' with an RSD of 7.1%. This result agreed well with a value of 0.77 k 0.09 mg I-' that was obtained with a fluoride ISE. Conclusions Laser-excited molecular fluorescence spectrometry in a graph- ite tube furnace with front-surface illumination was shown to be an extremely sensitive method for the determination of fluorine. Fluorine was determined by the molecular fluorescence of magnesium fluoride in a commercially avail- able atomic absorption graphite furnace. For the first time molecular fluorescence was collected at 180" to the direction of the laser beam from an unmodified atomic absorption graphite furnace using front-surface illumination.Advantages of front-surface illumination include improved sensitivity and the use of modern furnace technology that has been developed for atomic absorption. Barium was shown to enhance the MgF fluorescence signal by two orders of magnitude although the mechanism of this enhancement was uncertain. The detection limit for fluorine was 0.3 pg which is from two to six orders of magnitude more sensitive than commonly used methods. The LDR for LEMOFS was five orders of magnitude which is the first to be reported. The analytical precision was between 7 and 11%. However LEMOFS was shown to be severely af- fected by interferences from other ions which suppressed the fluorescence signal.In spite of these interferences it was pos- sible to show that LEMOFS can be used to determine fluorine accurately in real samples by 100-fold dilution of the sample. The dilution capability can only be employed when using in- strumentation such as that used here. that gives sufficient sensitivity. The LEMOFS technique has considerable potential for the determination of ultratrace amounts of non-metals in samples. This work is continuing in our laboratory to deter- mine the mechanism of the enhancement of the MgF signal by barium and to use LEMOFS as a highly sensitive18 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY FEBRUARY 1991 VOL. 6 Table 8 Results of analysis by LEMOFS of NIST SRM 267 la Free-Dried Urine and University of Connecticut tap water Sample Fluorine concentration/mg I-' LEMOFS Certified value ISE* LEMOFS? RSD % 0.54 k 0.047.4 NIST SRM 267 la Freeze-Dried Urine control level 0.55 k 0.03 - NIST SRM 267 la Freeze-Dried Urine elevated level 5.7 +- 0.3 - 5.6 +_ 0.6 10.7 University of Connecticuttap water - 0.77 k 0.09 0.70 k 0.05 7.I * Fluoride ISE with an acetate total ionic strength adjustment buffer. The data are It one standard deviation. t Molecular fluorescence of MgF excitation at 268.94 nm detection at 359 nm. Experimental conditions Mg(NO& (20 pg as Mg); Ba(NO&( 1.65 pg as Ba); atomization temperature I 8 0 'C charring temperature 800 'C; laser power 10 pJ per pulse. The data are f one standard deviation. method for the determination of other non-metals e.g.chlo- rine bromine and phosphorus. It is possible that the detec- tion limit for fluorine could be improved by using the AIF molecule rather than MgF which would allow further flexibility in dilution of solutions in order to remove interferences. We would like to thank Klaus Dittrich for helpful discussions prior to the start of this work which was supported by the National Institutes of Health under grant number GM 32002. R. G. M. was supported by a Research Career Development Award from the National Institute of Environmental Health Sciences under grant number ESOO130. D. J. B. and R. L. 1. were each supported by State of Connecticut High Technology Graduate Fellowships. References I 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 Campbell A.D. Pure AppI. Chem. 1987,59,695. Frant M. S. and Ross J. W. Anal. Chem. 1968,40 I 169. Nicholson K. and Duff E. J. Analyst 1981. 106,985. Coetzee J. F. and Martin M. W. Anal. Chem. 1980,52,2412. Hallsworth A. S. Weatherall J. A. and Deutsch D. Anal. Chem. 1976,48 1660. Leonard M. A. and Murray G. T. Analyst 1974,99,645. Deane S . F. Leonard M. A. McKee V. and Svehla G. Analyst 1978,103 1 134. Small H,. Stevens T. S. and Bauman W. C. Anal. Chem. 1975 47 1801. Fritz J. S. Du Val D. L. and Barron R. E. Anal. Chem. 1984 56 1177. Dittrich K. CRC Crit. Rev. Anal. Chem. 1986 16,223. Bond A. M. and O'Donnell T. A. Anal. Chem. 1968,40,560. Gelhausen J. M. and Camahan J. W. Anal. Chem.. 1989,61,674. Tsunoda K. Fujiwara K. and Fuwa K. Anal. Chem. 1977 49 2035.Tsunoda K. Chiba K. Haraguchi H. Chakrabarti C. L. and Fuwa K. Can. J. Spectrosc.. 1982,27,94. Tsunoda K. Haraguchi H. and Fuwa K. Spectrot-him Acta. Part 6 l985,40 1651. Tsunoda K. Chiba K. Haraguchi H. and Fuwa K. Anal. Chem. 1979,51,2059. Chiba K. Tsunoda K. Haraguchi H. and Fuwa K.. Anal. Chem. 1980,52,1582. Takatsu A. Chiba K. Ozaki M. Fuwa K. and Haraguchi H. Spectrochim. Acta Part B 1984,39,365. 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 Venkateswarlu P. Winter L. D. Prokop R. A. and Hagen D. F. Anal. Chem. 1983,55,2232. Itai K. Tsunoda H. and Ikeda M. Anal. Chim. Acta 1985 171 293. Gomez M. G. Corvillo M. A. P. and Rica C. C. Analyst 1988 113 1 109. Dittrich K. Anal. Chim. Acta 1978,97,69. Dittrich K. Anal. Chim. Acta. 1979 111 123. Dittrich K. and Vorberg B. Anal. Chim. Acfa 1982 140,237. Dittrich K. Vorberg B. Funk J. and Beyer. V. Spectrochim A m Part 6 1984,39,349. Dittrich K. Shkinev V. M. and Spirakov B. K. Talanta 1985 32 1019. Dittrich K. Hanisch. B. and Stark H. J. Fresenius Z . Anal. Chem. 1986,324,497. Butcher D. J. Dougherty J. P. Preli F. R. Walton A. P. Wei G.-T. Irwin R. L. and Michel R. G. J. Anal. At. Spectrom. 1988,3 1059. Wei G.-T. Dougherty J. P. Preli F. R. Jr. and Michel R. G. J . Anal. At. Speca-om. 1990 5 249. Long G. L. and Winefordner J. D. Anal. Chem. 1983,55,7 13A. Searcy A. W. in Progress in Inorganic Chemistry ed. Cotton F. A. Interscience New York 1963 vol. 3 pp. 49-127. Datta S. Proc. R. Soc. London Ser. A 1921,99,436. DonnPes Spectroscopiques Relatives aux Molecules Diatomiques ed. Rosen B. Pergamon Oxford 1970. Cotton F. A. and Wilkinson G. Ahpanced Inorganic Chemistry A Comprehensive Text John Wiley New York 1980. Handbook of Chemistry and Physics ed. Weast R. C. CRC Press Cleveland OH 57th edn. 1976. Emeleus H. J. Fluorine Chemistry ed. Simons J. H. Academic Press New York 1950 vol. I pp. 1-76. Slavin W. Carnrick G. R. and Manning D. C. Anal. Chem. 1982 54,62 I . Slavin W. Graphite Furnance AAS A Source Book Perkin-Elmer Nonvalk CT 1984. Frech W. Lundberg E. and Cedegren A. frog. Anal. At. Spectrosc. 1985,8,257. Styris D. L. Anal. Chem. 1984,56 1070. Dougherty J. P. Preli F. R. Jr. and Michel R. G. J. Anal. At. Spectrom. 1987,2,429. Bolshov M. A. Zybin A. V. Kolshnikov V. G. and Vasnetsov M. V. Spectrochim Acta Part B 1981.36,545. Paper 0102084B Received May 11th. 1990 Accepted August 30th. I990