摘要:

JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, SEPTEMBER 1987, VOL. 2 553 Trace Enrichment and Determination of Sulphate by Flow Injection Inductively Coup led Plasma Atomic Emission Spectrometry" Alan G. Cox and Cameron W. McLeod Department of Chemistry, Sheffield City Polytechnic, Sheffield SI I WB, UK Douglas L. Miles and Jennifer M. Cook Hydrogeolog y Research Group, British Geological Survey, Walling ford, Oxfordshire OX 10 SSB, UK A rapid and sensitive method for the determination of sulphate based on flow injection inductively coupled plasma atomic emission spectrometry (FI - ICP-AES) has been developed. A microcolumn of activated alumina (acidic form) was used in the FI manifold to pre-concentrate sulphate before ICP-AES detection at 180.73 nm. Linear calibration was established over the concentration range 0-1000 pg 1-1 and the limit of detection based on a 2-ml sampling volume was 2.8 yg I-'.The relative standard deviations at 10 and 1000 pg 1-1 were 7.0 and 0.8%, respectively. Determinations of sulphate in natural waters and boiler feed water were demonstrated. Keywords: Sulphate determination; on-line pre-concentration; flow injection; inductively coupled plasma; at om ic emission spectrometry There is considerable interest in the development of meth- odology for the determination of sulphate because of the importance of this ion in industrial and environmental systems. In water analysis, automated spectrophotometric and turbidimetric procedures are widely used for determinations at concentrations greater than about 5 mg 1 - 1 .1 More recent developments include procedures based on ion chromato- graphy,* flow injection (FI)-? and inductively coupled plasma atomic emission spectrometry (ICP-AES) .4 All these tech- niques lack the sensitivity necessary for quantification at the sub-mg 1-1 level and hence, for ultratrace determinations, sample pre-concentration is required. For examplt, in the analysis of boiler feed waters of nuclear power plant, on-line concentrator columns in conjunction with ion chromato- graphy are currently in use.5 As evidenced by recent literature, FI techniques are increasingly being used in atomic spectrometry for on-line sample pre-concentration .6-7 It has been shown that a FI manifold with a microcolumn of activated alumina provides a novel route for pre-concentration of a wide range of oxyanion species prior to measurement by ICP-AES.8 Sulphate was not investigated in these initial studies because of the lack of a vacuum spectrometer, but the high affinity of alumina for sulphate is well known."lO In this work the possibility of utilising a microcolumn of acidic alumina for on-line trace enrichment of sulphate prior to ICP-AES measurement at the S 180.73-nm line is considered.Performance characteristics for the FI - ICP-AES procedure are documented and the potential for quantification at the pg 1- 1 level is demonstrated. Experimental Reagents and Materials Standard solutions of sulphate were prepared daily by appropriate dilutions of a stock solution of potassium sulphate (1000 mg 1-1. S042-). Nitric acid (0.01 M) and ammonium solution (2.0 M) were prepared from concentrated reagents (BDH Chemicals, Aristar grade).Calcism (1000 mg I - I ) , phosphate (1000 mg 1-1) and chloride (1000 mg I-]) standard solutions were prepared from high-purity salts. High-purity water (Millipore) was used throughout in solution prepara- * Presented at the 1987 Winter Conference on Plasma and Laser Spectrochemistry, Lyon, France, 12th-16th January, 1987. tion. Water samples (natural waters and boiler feed water), acidified to pH 2 with nitric acid, were stored in pre-cleaned polyethylene (Nalgene) containers. Standard sea water (sul- phate concentration, 2781 mg 1-1) was obtained from the Institute of Oceanographic Sciences, Wormley, Hertford- shire, UK. Activated alumina (BDH Chemicals, Brockman grade 1, acidic form, particle size 75-120 pm) was used for column packing.Instrumentation and Procedure The ICP spectrometer (ARL 34000C Quantovac) utilised the S 180.73-nm line and instrument operating conditions were as previously reported.4 The FI manifold, described elsewhere, 1 1 consisted of a peristaltic pump, a rotary injection valve and a microcolumn of activated alumina (2.5 cm, i.d. 1.5 mm). Transient signals were displayed on a x - t chart recorder (Servoscribe RE 541.2) and standard instrument software was used in signal integration (integration time, 15 s). In operation, sample solutions were passed through the microcolumn of acidic alumina for specified time periods at a flow-rate of 1 ml min-1 and sulphate was retained on the column.An ammonia solution (200 yl, 2 M) was then injected to elute sulphate into the ICP. Residual sulphate was removed from the column by a repeat injection (200 PI) of the basic solution before further samples were processed. A carrier stream of nitric acid (0.01 M) was used to maintain column acidity. Sulphate recovery was,checked by calculating the ratio of peak areas for injection (200 pl) of a standard sulphate solution with and without the microcolumn in the manifold. Results and Discussion Adsorption/Desorption of Sulphate Sulphate underwent reproducible and quantitative adsorp- tionldesorption on the alumina microcolumn in a manner analogous to that previously reported for a range of oxyan- ions.8 A typical emission - time response corresponding to the deposition/elution cycle is given in Fig.1. The acidity of the carrier stream, sample pH and nature and concentration of the eluent were critical experimental parameters but rigorous optimisation studies on these aspects were not attempted as high analyte recoveries were realised in initial experiments for the conditions specified under Experimental. Both potassium hydroxide (concentration 3 0.5 M) and ammonia (concentra-554 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, SEPTEMBER 1987, VOL. 2 200 In C 3 4- 1 s .- : 100 In .- E W I I 1 I 1 I 0 20 40 60 80 100 120 Time/s Fig. 1. Emission (180.73 nm) versm time response for injection of sulphate solution (200 p1, 1 mg 1-1) and injection of ammonia solution (200 PI, 2 M) 500 r---+ 5 400 : t V .. : 300 .- In E 200 W .- 100 I D Time -+ 1 2 3 4 5 Sampling time/min Fig. 2.(a) Emission (180.73 nm) versus time response for elution of sulphate for various sampling times: A, 12 s; B, 30 s; C, 1 min; D, 2 min; E, 5 min. ( b ) Relationship between peak height reponse and sampling time tion 3 2 M) solutions were effective eluents giving recoveries of > 80% for a single injection (200 pl) of basic solution and this elution behaviour was similar to that previously reported8 for chromate and molybdate. Simultaneous monitoring of the A1 channel during elution did indicate chemical attack on the alumina microcolumn when potassium hydroxide was used, and hence ammonia solution was selected as the eluent in further work. Analytical Performance The calibration graph prepared from triplicate injections (2 ml) of standard solutions of sulphate (10,20,50,100 and 1000 pg I-l) gave good linearity (y = 0.015 x + 0.047, correlation coefficient 0.9993).The relative standard deviations (n = 7) at 20 and 1000 pg I-' were 7.0 and 0.8%, respectively. The limit of detection, calculated as three times the standard deviation of the background noise level, was 2.8 pg 1-1 and this represents an approximate 20-fold improvement over the conventional ICP-AES measurement.4 Extending the sam- pling time improved detection performance and as shown in Table 1. Analytical values (mg I-') for sulphate in waters Sample FI - ICP-AES ICP-AES Loch Fleet* (86/311) . . . . . . 6.6 7.8 Loch Fleet* (86/313) . . . . . . 4.0 4.1 BoxhallsLane* (861360) . . . . 15.5 15.6 Britty Hill* (86/364) .. . . . . 16.0 15.8 1 0 s sea water* (after200-folddilution) . . . . 14.2 13.0 Boiler feed watert . . . . . , 0.004 Boiler feed watert . . . . , . 0.010 - - * Sample volume, 200 p1. + Sample volume, 16 ml. 40 In C 3 c s 2 20 E .- In .- w C 1 mg 1-1 SO& 1000 mg I-' Ca2+ 1 2 Timeimin 3 Fig. 3. Emission 180.73 nm) versus time response for injection of standard solution 12 ml, 1 mg 1-1 S042- - 1000 mg 1-1 Ca2+) and injection of ammonia solution (200 p1, 2 M) Fig. 2 there was a linear relationship between peak height (and peak area) response and sampling time. The possibility of interferences associated with matrix constituents was examined because, as already noted,4 the S 180.73-nm line is subject to spectral overlap from a weak Ca transition. As shown in Fig.3 the alumina microcolumn did not retain calcium and provided an effective route for rapid removal of potentially interfering matrix cations. It was further confirmed in separate experiments, that the presence of relatively high concentrations of anions including phos- phate (1000 mg 1-1) and chloride (1000 mg 1-1) in standard solutions of sulphate did not affect analyte recovery. Determination of Sulphate in Waters Sulphate was determined in a range of water samples previously analysed by ICP-AES4 and the results are presen- ted in Table 1. The relatively good agreement between the two sets of data testifies to the over-all reliability of the FI procedure and indicates that the retention capabilities of the alumina microcolumn are not influenced by the presence of water matrix constituents.It should be borne in mind that the data for direct ICP analysis are representative of total sulphur concentrations but, given the comparability between the two sets of data, it can be concluded that sulphate is the predominant sulphur species in the samples investigated. (As pointed out in the peer review procedure, the fate of other sulphur species in the FI system has not been clarified and further investigations would be necessary before specificity can be assigned to the method.) In this short study no attempt was made to provide independent data for the boiler feed waters but it is clear that determinations at the pg 1-1 level are achieved through use of extended sampling times. A relatively poor sample throughput (ca.four samples per hour for 16 ml) is achieved for operation at sampling flow-rates of 1 ml min-1 , but recent studies indicated that sampling rates of 5 ml min-1 may be used without impairment of deposition efficiency. We are grateful to Dr. K. Tittle of the Central Electricity Generating Board for providing samples of boiler feed water.JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, SEPTEMBER 1987, VOL. 2 555 This paper is published with permission of the Director, British Geological Survey (NERC). 6. 7. Cresser, M. S . , Ebdon, L. C., McLeod, C. W., and Burridge, J. C., J. Anal. At. Spectrom., 1986, I , 1R. Ebdon, L., Cresser, M. S . , and McLeod, C. W., J. Anal. At. 1. 2. 3. 4. 5. References “Sulphate in Waters, Effluents and Solids, Methods for the Examination of Waters and Associated Materials,” HM Stationary Office, London, 1979. Hordijk, C. A., and Cappenburg, T. E . , J. Microbiol. Meth., 1985, 3,205. Van Staden, J. F., Fresenius Z. Anal. Chem., 1982, 312, 438. Miles, D. L., and Cook, J. M., Anal. Chim. Acta, 1982, 141, 207. Balconi, L., Pascali, R., and Signon, F . , Anal. Chim. Actu, 1986, 179, 419. Spectrom., 1987, 2, 1R. Cook, I. G., McLeod, C . W., and Worsfold, P. J . , Anal. Proc., 1986, 23, 5. Nydahl, F., Anal. Chem., 1954, 26, 580. Davies, J. E . , J. Oil Colour Chem. Assoc., 1971, 54, 425. Cox, A. G., Cook, I. G., and McLeod, C. W., Analyst, 1985, 110, 331. 8. 9. 10. 11. Paper J7l35 Received March 23rd, 1987 Accepted May 19th, 1987

ISSN:0267-9477    

DOI:10.1039/JA9870200553

出版商:RSC    

年代:1987

数据来源: RSC

摘要:

JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, SEPTEMBER 1987, VOL. 2 557 Use of a Glass Frit Nebuliser with a Helium Microwave-induced Plasma* Robert G. Stahl and Katherine J. Timminst Directorate of Quality Assuranceflechnical Support, Materials Centre, Ro yai Arsenal East, Woolwich, London SE 18 6TD, UK A glass frit nebuliser has successfully been linked to a low-power microwave-induced helium plasma at atmospheric pressures for the introduction of both aqueous and organic samples. Keywords: Microwave-induced plasma; sample introduction; atomic emission spectrometry; frit; nebulisation The microwave-induced helium plasma (MIP) is a very sensitive emission source for both metals and non-metals,' but, as it has a low thermal capacity, the plasma is easily extinguished by the introduction of foreign material and so has not, in general, found wide use in routine analysis.The exception and the major technique with which it is used2-4 is gas chromatography (GC), where the effluent may be readily split and interfaced into the gas stream. However, materials to be analysed are not always amenable to GC and although aqueous samples may be determined using electrothermal vaporisation (ETV), in this instance only transient signals5 are generated, and it is necessary at times to be able to integrate, or to scan a spectrum in order to look for the most sensitive line or for matrix interferences. Conventional nebulisation into an MIP has proved possible with low-pressure argon plasmas,hq7 high-8 or moderate-power air,Y argon") or helium11 plasmas, but not yet with low-power helium MIPS at atmospheric pressures.The nebuliser based on a glass frit, introduced by Layman and Lichte,l* has gas flow require- ments similar to those of an MIP. Its low sample flow-rates would be useful where sequential multi-element analysis is required, or where little sample is available and it has successfully been applied to moderate-power argon plas- mas13J4 and even to a very low-power (20 W) argon ~ 1 a s m a . l ~ However, this paper will demonstrate that it is also suitable for a helium plasma which is supported by a conventional, low-power cavity. 16.17 Experimental and Results The instrumentation (see Fig. 1) and general parameters used were as follows. Monochromator. Hilger Monospek 1000, grating 1200 lines mm-l, slits lOpm, PMT Hamamtsu R955, 1000 V, Brandenberg 472R, power supply, 1 : 1 image of MIP on slit.Readout. Tekman TE200 recorder, f.s.d. 0.3 s. MZP cavity. Beenakker'b TMoln and modified-aluminium Generator. EMS Microtron 200,75-100 W forward power. Nebufiser. Glass frit with 1-mm diameter glass capillary delivery tube, Watson-Marlow 2020U peristaltic pump with 0.019 mm tubing, Weir Maxireg 761 power supply to heating coil, 7.5 V, cooling, liquid nitrogen. The same basic design of the frit nebuliser12 as described previously was used except for three features. It was found to be more convenient to make the sample tube from the glass capillary of a micropipette, rather than from a stiff plastic TM010,17 silica plasma tube, 2.5 mm i.d. * Presented at the 1987 Winter Conference on Plasma and Laser T To whom correspondence should be addressed.Copyright Controller, HMSO, London, 1987. Spectrochemistry, Lyon, France, 12th-16th January, 1987. tube. The exit arm was broadened (see Fig. 2) to avoid the formation of droplets which inhibit free flow of the aerosol. The only sinters available in the UK were grades 3,4,4A and 5, BS1952: 1963. Of these, only the grade 4A (23 mm diameter, pore size 4-10 pm) proved to be suitable's; grades 3 and 4 did not give enough back pressure to produce a suitable aerosol and grade 5 allowed only a very small gas flow. Initially, the aerosol was taken straight into the MIP. The helium plasma could support this volume of material, but the silica tube was quickly eroded (ca. 1 h) (see Fig. 3) and the signals were very unstable. Reduction of the helium flow through the nebuliser and addition of an auxiliary flow immediately before the plasma prolonged the tube life to a few hours. The inclusion of a desolvation unit (a heated tube, followed by a cooled chamber) between the nebuliser and the auxiliary gas flow prolonged tube life to over a week and hence improved the stability due to the removal of part of the aqueous medium (up to 75% depending on the flow-rates used).The stability of the signal was also dependent on the combination of gas and sample flow-rates to the frit. If the carrier gas flow was too high, the sample was blown off the under surface before it could be replenished by the next pulse from the peristaltic pump, resulting in a marked oscillation of Wash Auxiliary helium - To monochromator - recorder Cold trap Frit nebuliser Fig.1. Block diagram of instrumentation Wash solution I aerosol Fig. 2. Glass frit nebuliser (size reduction 5 to 2)558 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, SEPTEMBER 1987, VOL. 2 Fig. 3. Eroded silica tube a) In In - ( b ) H HI NO 1 I I 1 410 420 430 440 450 Wavelengthh m Fig. 4. Spectrum for: (a) 10 pg ml-l indium in 2% HNO,: and ( b ) acid background. Conditions: He (frit), 700 ml min-1; He (auxiliary), 100 ml min-1; sample flow, 25 p1 min-1; recorder, 1 V; chart speed, 20 mm min-1; scan speed, 5 nrn min-1; range, 407-457 nm; and power, 85 W forward and 1 W reflected 40 20 > C c ‘5 0 50 - 500 1000 1500 2000 b ) 1 I I 1 1 1 400 600 800 1000 1200 1400 Auxiliary gas flow/ml min-’ Fig.5. Variation of emission from C1 lines with auxiliary gas flow: (a) CI I, 725.67 nm; and ( b ) CI 11,481.0 nm. Frit gas flows in ml min-1 indicated on each curve; sample flow-rate, 50 pl min-1 of 10% HCI; forward power, 75 W; and reflected power, 9 W CI II CI I I 1 // 1 I 1 Wavelengt h/n m Fig. 6. recorder, 5 V; chart speed, 20 mm min-l; scan speed, 5 nm min-I; and power, 85 W forward and 5 W reflected Carbon tetrachloride spectrum. Conditions: He (frit), 250 rnl rnin-1; He (auxiliary), 650 ml min - 1 ; sample flow, 5 pl min-1;JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, SEPTEMBER 1987, VOL. 2 559 the signal. A 50% variation in the intensity of the signal was observed in the C1 I (725.7 nm) and adjacent 0 I (725.4 nm) lines, when a 50 pl min-1 solution of HCI (1 + 49) was nebulised into an MIP of gas flows 500 (frit) plus 600 (auxiliary) ml min-1, 85-W forward power.The frit was also used while scanning spectra to search for sensitive lines. Fig. 4 shows the spectrum obtained from a 25 pl min-1 sample flow of a solution of 10 pg ml-1 indium in dilute HN03 (1 + 49) and the appropriate background spectrum. The 410.1- and 451.1-nm lines of indium may be readily observed. The linearity of the 451.1-nm line is better than two orders of magnitude and has a detection limit of 100 ng ml-1. This is to be contrasted with the detection limit obtained with ETV” (0.7 ng ml-1) and it is expected to be typical of the difference in sensitivity generally found between aqueous-nebulised and ETV-type techniques.12 The conditions used were found to be important, not only to avoid oscillations in the signal, but also to ensure that the species formed in the plasma were constant. As an example, the behaviour of C1 I (725.67 nm) and C1 I1 (481.06 nm) were studied using a sample flow-rate of 50 pl min-1 of HCl(1 + 9). Fig. 5 ( a ) indicates that the emission from C1 I decreases with an increase in the auxiliary gas flow, but increases with gas flow through the frit. (At 1.1 1 min-l He, the gas flow is too high for the frit and large droplets are formed.) Fig. 5(6) shows that the emission from C1 I1 was favoured by the increase in auxiliary gas flow, as well as by the frit gas. Organic solutions are readily vaporised through the frit ,I3 although no visible aerosol is formed.This is a much more easily controlled method of nebulising solvents into the MIP than simply sweeping the headspace vapour above the liquid into the gas stream. Fig. 6 shows a spectrum of CC14 at a sample flow-rate of 5 pl min-1. Conclusion The glass frit nebuliser is readily linked to a low-power helium MIP at atmospheric pressures, although it requires a desolva- tion system to obtain analytical results if aqueous solutions are used. While not as sensitive as other techniques, such as ETV, it does provide a continuous signal which may be integrated or scanned. The authors thank Mr. F. Plummer for making the glass frit nebulisers and Mr. F. Ward and his workshop for constructing the microwave cavities. 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17.18. 19. References Matousek, J. P., Orr, B. J., and Selby, M., Prog. Anal. At. Spectrosc., 1984, 7, 275. Olsen, K. B., Sklarew, D. S . , and Evans, J. C., Spectrochim. Acta, Part B, 1985, 40, 357. Goode, S. R . , Chambers, B., and Buddin, N. P., Spectrochim. Acta, Part B, 1985, 112, 329. Slatkavitz, K. J., Hooey, L. D., Uden, P. C., and Barnes, R. M., Anal. Chem., 1985, 57, 1846. Brooks, E. I., and Timmins, K. J . , Analyst, 1985, 110, 557. Hingle, D. N., Kirkbright, G . F., and Bailey, R. M., Talanta, 1969, 16, 1223. Fallgatter, K., Svoboda, V., and Winefordner, J . D., Appl. Spectrosc., 1971, 25, 347. Leis, F., and Broekaert, J. A . C., Spectrochim. Acta, Part B, 1984, 39, 1459. Urh, J. J., and Carnahan, J. W., Anal. Chem., 1985,57, 1253. Kollotzek, D., Tschopel, P., and Tolg, G., Spectrochim. Acta, Part B, 1984, 39, 625. Haas, D. L., and Caruso, J. A., ICP Inf. Newsl., 1982,8,379. Layman, L. R., and Lichte, F. E., Anal. Chem., 1982,54,638. Haas, D. L., Diss. Abstr. Int. B, 1984, 45, 1766. Haas, D. L., and Caruso, J. A., Anal. Chem., 1984, 56, 2014. Boss, C. B., and Matus, L. G., ICPZnf. Newsl., 1984,9, 543. Beenakker, C. I. M., Spectrochim. Acta, Part B, 1976,31,483. Matus, L. G., Boss, C. B., and Riddle, A. N., Rev. Sci. Instrum., 1983, 54, 1667. Westall, W. A., DGDQA Materials Technical Paper, No. 927, 1986. Timmins, K. J., J . Anal. At. Spectrom., 1987, 2, 251. Paper J7/43 Received April lst, 1987 Accepted May 7th, 1987

ISSN:0267-9477    

DOI:10.1039/JA9870200557

出版商:RSC    

年代:1987

数据来源: RSC

摘要:

JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, SEPTEMBER 1987, VOL. 2 561 Relationship Between Detection Limits and Mechanisms in Inductively Coupled Plasma Atomic Emission Spectrometry* Marthe Marichy, Monique Mermet and Jean-Michel Mermet Laboratory for Analytical Science, University Claude Bernard-L yon I, 69622 Villeurbanne, France Detection limits and optical properties of ionic lines in inductively coupled plasma atomic emission spectrometry are related, and from this relationship, two temperatures can be calculated, one for an energy sum below 16 eV (7900 K) and the other above (10000 K). The experiments carried out confirm the change in behaviour of lines around 16 eV and therefore the possible role of a charge-transfer process from the argon ions. Keywords: Atomic emission spectrometry; inductively coupled plasma; detection limits; charge transfer; tempera tu re Although a rare gas plasma seems to be a relatively simple medium, acquiring knowledge of the mechanisms in this type of source has been the aim of many studies and is not yet fully understood.Two review papers1.2 have emphasised the complexity of an argon inductively coupled plasma (ICP) because of the number of species present in the discharge: electrons, neutral argon Ar and Ar*, ionised argon Ar+ and AT+* and molecular argon AT2+ and AT2+*. Therefore, there is no predominant process but a combination of various mechanisms is occurring. Amongst these, a charge-transfer process from argon ions has been proposeds5 as one of the major mechanisms for the ionisation and excitation of an injected species X: Ar+ + X + Ar + X+* + AE The energy that is available is the ionisation energy of argon (15.76 eV) to which can be added the kinetic energy of the ion (ca.0.5 eV). It should be noted that there is another state (I = 1/2) close to the ion ground state with an energy of 15.94 eV. In addition to any other processes, charge-transfer processes will play a role for energies below 16 eV. This means that if this process is particularly important, a change in behaviour will be observed for lines having an energy sum (ionisation and excitation energies) below and above 16 eV. We have studied5 the behaviour of Cu and Mn near this limit as a function of the observation height. In a separate investigation, Myers and Tracy6 studied the line intensity change for a given variation in the carrier gas flow-rate.The percentage variation in the intensity for a 1% change in carrier gas flow-rate as a function of the energy sum is shown in Fig. 1. It can be seen that there is a regular variation up to the Zn I1 206.20-nm line and then a drastic change in the behaviour of the Cu I1 224.70- and Mg I1 279.08-lines (ie., for lines above 16 eV). In the present work, the behaviour of the Cu I1 224.70-nm line is different from that which we reported in reference 5, where the behaviour was rather similar to that of the lines below 16 eV (comparison of Cu I1 224.7, where E,,, = 15.95 eV, and Cu I1 213.5 where E,,, = 16.24 eV). Note that Blades7 found that Cu I1 224.7 behaved similarly to lines below 16 eV. It seems that the limit of the change in behaviour is not well defined.Similarly, preliminary results with plasma modulation8 indicated a change in line behaviour around the same 16 eV limit. In previous papers9.10 we have shown a relationship between the detection limits, deduced from ionic line measurements using the optical properties of these lines. For a concentration c, there is a corresponding signal S, a back- ground B and the noise of the background NB. According to * Presented at the 1987 Winter Conference on Plasma and Laser Spectrochemistry, Lyon, France, 12th-16th January, 1987. IUPAC criteria and for a Gaussian distribution of the variable, the detection limit cL is the concentration c where the signal to noise (of the background) ratio SNBR is equal to 3.11-13 S z 3 X N B .. . . . . . . (1) The noise is estimated by the root mean square (RMS) value that is equal to the standard deviation. The relative standard deviation of the background RSDB can also be used14.15 and S becomes equal to S=3XBXRSD, . . . . . . (2) The two approaches give equivalent results for the determina- tion of cL: CL = 3dSNBR = 3cRSD,/SBR . . . . (3) where SBR is the signal to background ratio. expression : Traditionally, the signal S is given by the line intensity S = K 1 ( g A n+/h 2) exp (-E,,,/k 7') . . (4) where K1 is a constant, g is the statistical weight, A the transition probability, n+ the ion density, h the wavelength, 2 the partition function and E,,, the excitation energy of the upper state of the optical transition, when a Boltzmann equilibrium is assumed. When expression (4) is applied to one element in a particular ionisation state, there is no problem I oBa I 1 1 I I t I I I I 8 9 10 11 12 13 14 15 16 I I 17 Energy sum/eV Fig.1. Percenta e change in ionic line intensities for a 1% change in carrier as as a function of energy sum (ionisation and excitation energie$ as reported by Myers and Tracy6562 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, SEPTEMBER 1987, VOL. 2 and an excitation temperature can be deduced from the equilibrium. As confirmed below, this is not so when applying expression (4) to several elements together and some dis- crepancies between results may be observed. Recently, a new relationship for the line intensity of atoms has been pro- posed’? S = K2 x A x exp (-Eio,lkT)/[exp (AE,,,IkT) - 11 . .(5) where K2 is a constant which depends on the operating conditions, Eion the ionisation energy of the atom and AEexc the energy difference between the two levels of the optical transition. In the UV - visible part of the spectrum exp(AE,,,lkT) >> 1 and S = K2 X A X exp - (Eion +AE,,,)IkT . . (6) In this work, it can be seen that, empirically, we found that the best results were obtained using the following relationship: S = K1 (g A n+lhZ) exp-(Eion + E,,,)/kT . . (7) In other words, the intensity is directly related to the energy sum. At the detection limit level, cL is related to the total density of particles in the form of atoms ( n ) and ions (n+): cLcc(n+n+)M . . . . - * (8) where M is the relative atomic mass.Considering the degree of ionisation a: a=n+/(n+n+) . . . . - (9) n+=K3aCL/M . . . . . . (10) and hence where K3 is a constant which takes into account the atomisa- tion efficiency and the spatial distribution of the species. By combining equations (2), (7) and (lo), we obtain: c1. = K4 (3 x B X RSDB x h X M X Z/CX X gA) exp(E,,,/kT) (11) where K4 is a combination of K 1 and K3. When selecting a limited wavelength range and under the same operating conditions, B and RSDB are considered as constant and it may be given as: l o g ( ~ g A C ~ l h M Z ) = f ( - E , , , l k T ) . . (12) where the slope is equal to 5040/T if E,,, is given in eV. We will see later that the temperature deduced from the slope is of the same order as the ionisation temperature obtained when using the Saha equation.Moreover, if charge transfer is the predominant process, this plot should exhibit a discontinuity around 16 eV. Therefore, the lines of several elements were selected covering a wide energy range in order to carry out an experiment to verify this. Experimental To obtain a large number of reliable data over a long period of time, a fully automated Perkin-Elmer Plasma I1 ICP system was used. The system includes two monochromators (equipped with interferometric gratings with 3600 and 1800 lines mm-1). For this particular experiment, only one monochroiqator (3600 lines mm-1) was utilised so as to obtain the same spectral band pass. The operating parameters are summarised in Table 1. The viewing height (15 mm) was the position where there was maximum emission from the ionic lines.We studied the influence of the integration time ton the limit of detection (Fig. 2). Between 0.1 and 1 s, the improvement corresponds to the 4 law. Above 2 s , improve- ment is no longer observed because of the signal drift. The best values are usually obtained near to t = 2 s. As we chose to use 15 replicates in order to approach the value of the standard deviation of the blank, an integration time of 0.4 s was selected. This is a good compromise between the total time of the experiment and detection limit values as a decrease in these values of only a factor two is observed when compared with the best values (with an integration time of 2 s). It should be noted that detection limit values are the results from experiments carried out over several months, and are within the uncertainty inherent to the determination of the detection limit (33% when SINS = 3).Similarly, the variation of the blank signal at a given wavelength was less than 5% for the duration of the experiment (the same torch and the same nebuliser were used). Results Line Selection In order to obtain a small variation in the background and therefore to allow a constant RSD of the blank, we worked in a reduced wavelength range. The 210-250 nm range was selected as it is a region rich in sensitive lines.” Only one line (Zn I1 206.191 nm) is slightly out of this range. Line selection is summarised in Table 2. Calculation of partition functions was carried out using polynomial expressions18 and are similarly given in Table 2 assuming an excitation temperature of 5000 K.The degree of ionisation was taken from reference 19 where an ionisation temperature of 7500 K and local thermodynamic equilibrium (LTE) were assumed. Although the ICP is not in LTE, it has been shown20 that the departure from LTE is not too great. Consequently, the variation in the values for the degree of ionisation is small, especially when 1 0.1 0.4 1.0 2.0 4.0 Integration time/s Fig. 2. time: A, Pb I1 220.353; B, In 1230.606; and C, Zn I1 206.191 nm Variation of detection limit (3a) as a function of integration Table 1. Operating parameters ICP system . . . . . . . . Generator . . . . . . . . Power . . . . . . . . . . Flow-rates: Outer gas . , . . Intermediate . . . . Carrier . . . . . . Viewing height .. . . . . . . Nebuliser . . . . . . . . Pump rate . . . . . . . . Monochromator . . . . . . Integrationtime . . . . . . PMT voltage . . . . . . . . Replicates . . . . . . . . Perkin-Elmer Plasma I1 27 MHz crystal controlled 1kW 15lmin-1 1 1 min-1 11 min-1 15 mm ALC Perkin-Elmer cross-flow 1 ml min-1 3600 lines mrn-1(160-400 nm, 0.4 s 800 V 15 20/20 ym slit widths)JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, SEPTEMBER 1987, VOL. 2 563 considering the uncertainty in the parameters, such as transition probabilities. Transition probabilities from differ- ent works may not be consistent and show variations in T; they must be determined using the same radiation source and the same method. This is why we chose to use the data of Corliss and Bozman,21 even though the accuracy of the data has been criticised because of incorrect estimations of the temperature, the electron number density and the spatial distribution of the signal. There are also some other errors.For instance, when considering W ionic lines, the W I1 224.875-nm line was rejected as its gA value is not consistent with the intensities observed with other W lines; although the gA value is ten times smaller than that for W I1 248.923 nm, the lines exhibit almost the same line intensity, with similar energy. Neverthe- less, we kept the Corliss and Bozman data as no other comprehensive set of data was available. Table 2 presents the Table 2. Relative atomic mass M , wavelength h, energy sum E,,,, statistical weight g and transition probability A (108 s-*), degree of ionisation (Y, partition function 2, value of log(agA/hMZ) used for this experiment and detection limits cL obtained in this work (A) and from reference 24 (B) Ag Ba Cd co c u Fe Hf In Ni Pb Pd Rh Sr Ta w Zn Element .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . , . . . , . . . , . . . . M . . , 107.9 . . . 137.3 . . . 112.4 . . . 58.9 . . . 63.5 . . . 55.8 . . . . 178.5 . . . . 114.8 . . . . 58.7 . . . . 207.2 . . . , 106.4 . . . . 102.9 . . . . 87.6 . . . . 180.9 . . . . 183.8 . . . . 65.4 hlnm 233.137 241.318 243.779 230.424 233.527 234.758 214.438 226.502 231.284 228.616 231.160 231.405 231.498 236.379 237.862 238.346 238.636 238.892 213.598 219.226 224.700 233.280 234.349 234.810 234.830 238.204 238.863 239.562 240.488 232.247 235.122 239.336 239.383 246.419 230.606 23 1.604 239.452 220.353 248.653 248.892 233.477 241.584 242.711 246.104 215.284 2 16.596 223.948 238.706 240.063 216.632 220.448 232.609 239.709 248,923 206.191 E,,,/eV 17.93 17.75 17.50 11.21 11.19 11.19 14.77 14.46 20.13 13.70 13.78 13.83 13.86 13.60 13.48 13.56 13.62 13.46 16.23 16.20 15.95 13.23 13.16 13.38 13.23 13.07 13.11 13.09 13.11 12.14 12.07 12.59 11.98 13.94 11.15 14.01 14.48 14.78 16.67 16.44 14.85 15.03 14.86 14.77 13.25 13.25 13.74 13.62 13.81 14.28 14.36 14.07 13.54 13.54 15.40 gA' loss-' 3 16 13 3.1 6.5 0.7 106 99 60 169 125 136 116 166 158 178 116 278 14 4.6 9.1 15 20 12 92 26 96 100 8.6 2.1 2.3 9.9 3.7 0.03 31 59 167 5.6 37 94 44 49 45 78 1.3 1.5 39 175 516 26 40 24 48 58 92 (Y 93 91 85 93 90 96 98 99 91 97 93 94 96 95 94 75 z 1 .o 4.2 2.0 29.6 1 .o 43.4 12.5 1.0 10.8 2.1 7.5 15.5 2.1 22.5 13.8 2.0 Log (oLgA/hMz) 6.04 6.76 6.66 5.33 5.64 4.66 7.27 7.22 6.99 6.58 6.46 6.50 6.43 6.57 6.55 6.60 6.41 6.79 6.97 6.47 6.76 5.40 5.53 5.30 5.16 6.18 5.63 6.20 6.21 4.60 4.63 5.26 4.83 5.74 4.05 6.56 7.00 5.75 6.24 6.64 6.05 6.08 6.04 6.27 5.49 5.55 5.61 6.23 6.70 5.65 5.83 5.58 5.87 5.94 7.40 CL (A)/ ng ml-1 1000 450 260 6 5 95 13 7 2200 13 22 32 48 22 20 27 44 11 60 75 18 40 19 64 48 6 32 9 15 29 35 58 30 52 96 15 50 126 255 88 82 210 185 135 75 47 120 48 34 285 150 115 45 65 48 CL (BY ng ml-' 20 15 25 0.5 0.4 4 0.7 2 2 2 2 2 2 4 1.5 0.7 0.9 0.9 9 2 3 1.5 1.5 5 10564 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, SEPTEMBER 1987, VOL.2 gA values and the values of log(cxgAlhM2) in arbitrary units for 16 elements and 55 lines.Note that the Zn I1 206.19-nm line has been selected so as to have a detection limit corresponding to an energy sum as close as possible to 16 eV. Experiments Experiments over a large period of time have shown that the value of the RSD of the blank was slightly better than 0.01 in the 210-250 nm range. Compared with the 230-nm central wavelength value, the relative variation in the background was 1.2 at 250 nm and 0.7 at 210 nm. As the variation in the RSD is proportional to the square root of the background signal, this variation is less than 10% between 230 and 250 nm and less than 15% between 230 and 210 nm. Therefore, we decided to select only one RSD value, equal to 0.01, for the determina- tion of the detection limits that are given in Table 2.These values have been inserted in equation (12), which has been plotted in Fig. 3. Although the scatter seems to be large, the most dispersed points correspond to a change in detection limits of only a factor of four. This is satisfactory when considering the uncertainty in the gA values. Up to 16 eV, the linear regression provides a slope which gives a temperature of 7900 K. This temperature is in good agreement with the one (7500 K) deduced from the electron number density measure- ments and the Saha equation using a 27-MHz generator and assuming LTE.19322 The highest value used for the linear regression is the detection limit from Zn I1 260.191 nm ( E = 15.40 eV). Above this value, a different slope is obtained and, although the uncertainty is large, a temperature of about 10000 K may be derived.This temperature is of the same order as the temperature of the free electrons, a similar value having been reported elsewhere23 taking into account the radiative recombination origin of the continuum. Below 16 eV, the charge-transfer process seems to be important in constrast to the higher energies where electron collisions remain the only process. As in the Myers and Tracy I I I I I I I I I 12 13 14 15 16 17 18 19 20 Energy sumlev Log(argAc,lhMZ) as a function of energy sum (ionisation and excitation energies) for lines reported in Table 2: 0, A ; e, Ba; (3, Cd; 0, Co; H, Cu; 0, Fe; 0 , Hf; 0, In; +, Ni; A , Pb; $: Pd; A, Rh; V, Sr; +, Ta; x, Zn; and m, W experiment,6 Cu I1 224.700 nm acts as a line that is not obtained through a charge-transfer process, in contrast to the results in reference 5.In order to illustrate the better fit of the results obtained using the energy sum, a similar plot taking into account only the excitation energy is given in Fig. 4. The scatter is large and no linear plot can be determined over the full energy range. Moreover, values associated with the lowest energies could provide a vertical slope, i.e., a temperature equal to zero. 0 0 I I I I I I 6 7 8 9 10 11 12 Excitation energyIeV Fig. 4. function of excitation energy of ionic lines Same experiment as that in Fig. 3, with log(oLgAc,lhMZ) as a I I I I 1 1 I 12 13 14 15 16 17 18 Energy sumlev Fig. 5. limits published by Boumans and Bosveld24 Log(agAc,lhMZ) as a function of energy sum using detectionJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY.SEPTEMBER 1987, VOL. 2 565 In order to confirm these results, we used other data from the literature. Only the work of Boumans and Bosveld24 provides enough data relevant to the lines reported in Table 2. Detection limits are given in Table 2 and the graph which is deduced from these data is shown in Fig. 5 . It can be seen that the scatter is not very important. The linear regression gives a temperature of 7800 K. Although only the Ag I1 lines have energies above 16 eV, it can be seen that the results do not fit with the linear regression obtained below 16 eV. This provides independent confirmation of the change in behaviour near this 16 eV limit.Influence of Parameters on Detection Limits In contrast to values obtained with ICP mass spectrometry, in atomic emission spectrometry detection limits cover several orders of magnitude. This can be explained by considering equation (11). Of the various parameters, g A and E (through the exponential) can vary over a large range (103 for the exponential and 104 for g A ) . Most of the time gA will be the predominant parameter. For instance, the In I1 230.606-nm line, although having a low energy, exhibits a poor detection limit. This is due to its very low g A value. A comparison with the Co I1 238.616-nm line shows that the 2 and exponential value ratio favour In, but this is not enough to compensate for the gA ratio which is almost 104. Therefore, a theoretical cL ratio of eight is obtained between Co and In, which is similar to the experimentally derived ratio of seven.Similarly, detection limits obtained with Ba I1 230.424 and Cd 226.502 nm are equal. The difference in energy (exponential ratio equal to 100) is compensated for by the g A and 2 ratios. Once again, the main parameter seems to be the transition probability. Conclusions For the ionic lines at least, it is possible to find a relationship between the detection limits using the optical properties of emitted lines, that confirms that a charge-transfer process is a mechanism to be considered below an energy sum of 16 eV. Temperatures deduced from the slope have values that are of the same order as the ionisation temperatures calculated from the Saha equation.This measurement allows the comparison of detection limits obtained with different operating param- eters (e.g., a different sample introduction system) with the same ICP system. A longer residence time would result in a higher temperature which means that any improvement will be more important at high energy than at low energy. On the other hand, when working at frequencies higher than 27 MHz, improvements will be observed mainly at low energies.25.26 Above 16 eV, a higher temperature is obtained which is linked to the free-electron temperature. This temperature is consistent with those determined using a Boltzmann plot above 16 eV.3-5 An extrapolation above 16 eV of the linear regression obtained below this limit provides detection limits worse than the ones observed here (one order of magnitude for Ag and Pd).More accurate results can be found if a large set of reliable gA values were available. No similar relationship was obtained for atomic lines. One of the possible reasons for this is connected with the large uncertainty in the neutral atom densities ( a ) , as high degrees of ionisation (a = 1) are usually observed in the ICP. For ionic lines, the possible departure from LTE does not significantly modify the value for the degree of ionisation20 as most elements are almost fully ionised. However, poor accuracy is obtained because of the number of atoms which are not ionised. Although we used the energy sum in expressions (7), (11) and (12), the physical reasons for doing this are not yet fully understood. References 1.2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. de Galan, L., Spectrochim. Acta, Part B, 1984, 39, 537, Davies, J., and Snook, R. D., J . Anal. A t . Spectrom., 1986, 1 , 325. Batal, A., and Mermet, J. M., Can. J . Spectrosc., 1982,27,37. Raaijmakers, I. K. M. M., Boumans, P. W. J. M., Van der Sijde, B., and Schram, D. C., Spectrochim. Acta, Part B, 1983, 38, 697. Goldwasser, A . , and Mermet, J. M., Spectrochim. Acta, Part B, 1986, 41, 725. Myers, S. A., and Tracy, D. H., Spectrochim. Acta, Part B , 1983, 38, 1227. Blades, M. W., University of British Columbia, Vancouver, Canada, personal communication. Olesik, J. W., Reitz, K., and Williamsen, E . , paper presented at the 13th FACSS meeting, St Louis, MO, USA, September 1986.Mermet, J. M., C.R. Acad. Sci., Ser. B, 1977, 284, 319. Mermet, J . M., and Trassy, C., Spectrochim. Acta, Part B, 1981, 36, 269. Winefordner, J . D., Fitzgerald, J. J., and Omenetto, N., Appl. Spectrosc., 1975, 29, 369. Alkemade, C . Th., Snelleman, W., Boutilier, G. D., Pollard, B. D., Winefordner, J. D., Chester, T. L., and Omenetto, N., Spectrochim. Acta, Part B, 1978, 33, 383. Long, G. L., and Winefordner, J . D., Anal. Chem., 1983, 55, 713A. Kaiser, H., Spectrochim. Acta, 1947, 3, 40. Boumans, P. W. J. M., McKenna, R. J., and Bosveld, M., Spectrochim. Actu, Part B, 1981, 36, 1031. Thelin, B., and Yngstrom, S . , Spectrochim. Acta, Part B, 1986, 41, 403. Winge, R. K., Peterson, V. J., and Fassel, V. A., Appl. Spectrosc., 1979, 33, 206. de Galan, L., Smith, R., and Winefordner, J . D., Spectrochim. Acta, Part B, 1968, 23, 521. Houk, R. S . , Anal. Chem., 1986, 58,97A. Blades, M. W., Caughlin, B. L., Walker, Z . H., and Burton, L. L., Prog. Anal. Spectrosc., 1987, 10, 57. Corliss, C. H., and Bozman, W. R . , “Experimental Transition Probabilities for Spectral Lines of Seventy Elements,” NBS Monograph 53, Washington DC, USA, 1962. Kalnicky, D. J., Fassel, V. A . , and Kniseley, R . N., Appl. Spectrosc., 1977, 31, 137. Batal, A , , Jarosz, J., and Mermet, J. M., Spectrochim. Acta, Part B, 1981, 36, 983. Boumans, P. W. J. M., and Bosveld, M., Spectrochim. Acta, Part B, 1979, 34, 59. Capelle, B., Mermet, J. M., and Robin, J., Appl. Spectrosc., 1982, 35, 102. Michaud-Poussel, E., and Mermet, J. M., Spectrochim. Acta, Part B, in the press. Paper 57/24 Received February 19th, 1987 Accepted May 15th, 1987

ISSN:0267-9477    

DOI:10.1039/JA9870200561

出版商:RSC    

年代:1987

数据来源: RSC

摘要:

JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, SEPTEMBER 1987, VOL. 2 567 Measurement of True Gas Kinetic Temperatures in an Inductively Coupled Plasma by Laser-light Scattering* Plenary Lecture Kim A. Marshallt and Gary M. HieftjeS Department of Chemistry, Indiana University, Bloomington, IN 47405, USA Laser-light Rayleigh scattering has been used to measure the localised gas density and from it the gas temperature in an inductively coupled plasma. The magnitude of the locally averaged gas temperature was found to be somewhat higher than that measured by Doppler-broadening methods; short-term relative standard deviations for the scattering measurements were usually less than 10%. Gas-temperature measurements were performed as a function of incident r.f. power and spatial position in the plasma and in the presence or absence of introduced aerosol.Keywords: Laser-light scattering; gas kinetic temperatures; inductive1 y coupled plasma The inductively coupled plasma (ICP), although used widely, is still not well understood. If a spectroscopic source such as the ICP is close to thermodynamic equilibrium (TE), all energy distributions used to determine its temperature should return similar values.’ This is found not to be so in the ICP. Instead, in a given plasma zone, a trend of decreasing temperature is found for different thermometric specie^^-^ with: Te > Tion ’ Tex > Tgas where Te is the electron temperature, T,,,, is an ionisation temperature, T,, is an excitation temperature and Tgas is the plasma gas kinetic temperature.The departure of the ICP from being a “thermal source” can be illustrated by the gap between its gas temperature (5000 K)s and its electron temperature (11 000 K).4,6 Research efforts have often focused on the determination of various temperatures in the ICP in order to understand better this state of dis-equilibrium .3-4,7-10 Excitation processes in the ICP are associated with the high-energy end of this temperature spectrum; in fact, partial equilibrium seems to exist between the ionisation process and excitation events close in energy to the ionisation limit.2~9,*0 Whereas the electron temperature is probably more useful in characterising excitation processes which occur once the analyte is in its molecular or atomic form, the gas temperature is the determining parameter during the earlier processes of desolvation and vaporisation.Gas temperatures in the ICP are therefore of great interest and have been reported by others.sJ1.12 Methods previously used for the approximate determina- tion of gas kinetic temperatures in an ICP are based on the Doppler broadening of emission lines5.12 and on the relative intensities of OH rotational lines.” The measured gas temperatures in these studies range from 3000 to 8000 K. It is noteworthy that this large range is not due to the variance between the two methods, to differing conditions in the plasma nor to different workers, but results mainly from the use of alternative emission lines in the Doppler method. In these studies, the gas kinetic temperature determined from the Doppler width of argon lines was approximately 2000 K higher than those determined from lines of other species.-5.l2 ~~~ * Presented at the 1987 Winter Conference on Plasma and Laser ?.Present address: Instruments SA, Inc., 173 Essex Avenue, $ To whom correspondence should be addressed. Spectrochemistry, Lyori, France, 12th-16th January, 1987. Metuchen, NJ 08840, USA. This disparity is due, in large part, to the fact that Stark broadening of the Ar lines was not considered.5 Both of these previously used methods for determining gas temperatures are prone to significant errors. With the Doppler-broadening approach, the computational problems are severe and measurement noise is especially troublesome. If spatial resolution is desired, the Doppler profiles must be experimentally determined at a number of lateral positions in the plasma.The lateral intensity data at each wavelength segment of the Doppler-broadened profiles must then be Abel inverted to generate the corresponding radially resolved information. This process carries with it all of the problems and assumptions of Abel inversion. Such difficulties include the necessary assumption that the ICP is cylindrically symmet- rical, the problem that Abel inversion is noise sensitive and the fact that it is least accurate near the centre of symmetry. The radial data obtained from Abel inversion must then be used to reconstruct the spatially resolved spectral profiles needed to make the gas-temperature calculations. Workers who used the Doppler methods.12 have so far avoided these computational problems by reporting only laterally integrated gas temperatures.The Doppler method suffers also from the need to deconvolute instrumental, Stark- and pressure-broadening contributions from the measured line (Voigt) spectral profiles before accurate gas temperatures can be determined.5 Instrumental broadening depends on the “quality” of the measuring system and is caused by the finite slit width, diffraction effects, optical aberrations and resolving power of the detection apparatus. The determination of gas temperatures by measurements of the intensity of rotational emission lines is also prone to error. First, the method measures rotational rather than true translational (gas kinetic) temperatures. Therefore, it is necessary to assume that these two values are equivalent.Moreover, the accuracy with which one knows the transition probabilities of the measured rotational lines is critical. Furthermore, the emission intensities of the rotational transi- tions must be Abel-inverted if spatial resolution is desired. Finally, the method is limited to those regions of the plasma where the molecular fragments which produce the monitored emission remain intact. In the hotter regions of the plasma, the observed rotational bands can become unacceptably weak because of the lowered concentration of the emitting frag- ments. In the present investigation, a new approach involving the measurement of Rayleigh-scattering intensities is used to determine gas temperatures in the ICP. Rayleigh scattering of568 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, SEPTEMBER 1987, VOL.2 incident light occurs only from very small particles, and is proportional in intensity to the concentration of the scatterers in the observation volume. In the ICP, where ground-state argon atoms are by far the most prevalent scatterers, the Rayleigh-scattering signal is a measure of the localised argon number density. Through application of the ideal-gas law, this number density can be related to the gas kinetic temperature. The Rayleigh-scattering approach to the measurement of gas temperatures offers several significant advantages. Because the observation volume is defined by the overlap between an incident laser beam and the cone of light accepted by the detection optics, the scattering signals need not be AbeI inverted to provide spatial resolution. Also, the accuracy of the determined temperatures does not depend on tabulated transition probabilities and there is no need for the deconvolu- tion of instrument or Lorentzian broadening.The accuracy of the Rayleigh-scattering method depends mostly on how well the scattering instrument can be calib- rated. This calibration can be accomplished most conveniently by measuring the scattering signals produced by gases with known and widely different Rayleigh-scattering cross-sec- tions. In this study, room-temperature argon and helium were used to calibrate the instrument over a temperature-equi- valent range from 300 to above 18000 K. The details of this calibration procedure are provided and preliminary tempera- tures reported for several zones in an ICP.Experimental Probe Laser The laser-based scattering system used in this investigation was originally designed for measuring Thomson-scattering signals. Thomson scattering is useful for determining electron concentrations and energies in the ICP. Details on this system can be found elsewhere,13 but a summary here is warranted. A Q-switched ruby laser (Model K1 laser head and Model KQS2 Q switch, Korad Division of Hadron, Santa Monica, CA, USA) with an output power of approximately 20 MW (0.5 joules per pulse) provides the incident 25-11s light pulse at 694.3 nm. A vertical laser polarisation was selected to maximise the scattering signal. The laser beam is focused by a two-lens optical system. The first lens focuses the laser to a 0.4-mm spot at a pinhole stop, while the second lens images this spot into the plasma.With this arrangement, the laser beam is focused in the plasma to a 1.3-mm point which determines the spatial resolution of the system. A series of beam stops is arranged along the laser path to reduce stray light. After passing through the plasma and a series of beam stops, the laser beam is sequentially attenuated by two beam dumps. Inside the torch housing, a viewing dump is placed across the plasma from the detection optics in an effort to reduce further the stray light in the system. Detection Optics A rotating-mirror optical chopper is used to protect the photomultiplier tube (PMT) detector from the intense back- ground emission produced by the ICP. During operation of this chopper, a 10 X 10-cm flat mirror (Melles Griot, Irvine, CA, USA), rotating at 3600 rev min-1, sweeps the image of the plasma past the entrance slit of the monochromator.The laser pulse is synchronised with the resulting 25-ys optical gate. The 1-m Czerny - Turner monochromator (Model HR-1000 Instruments SA, J-Y Optical Systems Division, Metuchen, NJ, USA) is equipped with a 2400 grooves mm-1 holographically ruled grating. A 25-channel fibre-optic array is mounted in the focal plane of the monochromator. The centre channel of this array is positioned at the ruby-laser wavelength (694.3 nrn) and monitors Rayleigh scattering; the remaining channels are used for Thomson-scattering measure- ments and are not important in the work reported here. Detectors and Detection Electronics The centre-output pigtail (Rayleigh channel) of the fibre-optic array is connected to a locally fabricated PMT housing containing a Hammatsu R928 PMT operated at 1200 V.A slide selector between the PMT and the output pigtail contains two different neutral-density filters (with neutral densities of 1.0 and 2.0, respectively). The particular filter that is used depends on the scattering medium being observed ( L e . , plasma, room-temperature argon or room-temperature helium). The output current from the PMT is collected by a gated integrator (Model No. 4130A Evans Electronics, Berkeley, CA, USA) whose gate width (normally 400 ns) is controlled by a gate-delay module (Model No. 4145-2 Evans Electronics) synchronised to the Q-switch trigger of the laser pulse.A 400-ns gate width was chosen here because of the jitter between this Q-switch synchronising pulse and the appearance of the 25-ns laser pulse. The voltage output from the gated-integrator card is digitised by a 12-bit analogue-to- digital converter (ADC) (Model No. DT5712 Data Trans- lation, Marlboro, MA, USA). The digitised output is read by a 6502 microprocessor (Model 7510 Cubit Division, Proteus Industries, Mountain View, CA, USA), and sent to a host laboratory computer (IBM-9000, IBM Instruments, Danbury, CT, USA) for storage and data analysis. Calibration Considerations The use of measured Rayleigh-scattering intensities to deter- mine the plasma-gas temperature requires that the response of the detection system be calibrated. Ideally, standards of two or more volumes of argon gas, at widely different and well known temperatures, would be used in this calibration.Although one such standard is available (room-temperature, atmospheric-pressure argon), it is difficult to produce a well thermostatted gas volume at a significantly higher tempera- ture (any means of containment might change the stray-light component of the scattering signal). As will be emphasised later, it is important that the stray-light levels in the scattering system be similar during both the calibration and gas-tem- perature measurements. Fortunately, calibration substitutes for equivalent volumes of argon gas at several temperatures are available. If one measures scattering intensities from gases with widely varying Rayleigh-scattering cross-sections, it is equivalent to measur- ing argon at different temperatures.For example, helium (a gas with a small scattering cross-section) at room temperature and atmospheric pressure produces a scattering signal equal to that from argon (which has a much larger scattering cross- section) at a lower number density (higher temperature). Conversely, a gas with a large scattering cross-section (e.g., C02), can be related to a lower-temperature volume of a gas with a smaller cross-section (argon). This calibration procedure can be treated simply in a quantitative manner. The polarised Rayleigh-scattering cross- section of a gas (oR) can be calculated from the following relationship14 when n is the refractive index of the gas, nR is the number density at which the refractive index is measured and LO is the wavelength of the incident light.From equation (1) and the respective refractive indices (nAr = 1.000281 and nHc = 1.000036) ,15 the Rayleigh cross-section for ground-state argon is 1.816 X 10-27 cm2 and that for helium is 2.980 x 10-29 cm* at a scattering wavelength of 694.3 nm. Of course, these cross-sections pertain to ground-state argon and helium; because we intend to monitor Rayleigh scattering in a plasma, one might question whether excited- state Rayleigh scattering is important. Indeed, it has been shown that excited-state argon can have significantly greater569 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, SEPTEMBER 1987, VOL. 2 Rayleigh-scattering cross-sections than ground-state argon.16 For example, the 1s5 and 2p10 states of argon have scattering cross-sections on the order of 105-times greater than that of the ground-state. However, even in a plasma where these states are populated by a 9000 K excitation temperature, their total number density is only about 2.5 x 10-7 that of the ground-state population. The Rayleigh-scattering intensity from these two excited states is therefore expected to contribute less than 1% to the total observed scattering signal. These calculations are based on the Boltzmann equation and excited-state energies of 9.31 X 104 and 1.04 x 105 cm-1, respectively. In this study, scattering measurements on helium were always carried out at room temperature so that excited- state helium populations are obviously insignificant.The scattering system is calibrated by measuring the scattering signals from room-temperature (300 K), atmos- pheric-pressure argon and helium. The small cross-section of helium produces a scattering signal equivalent to the same volume of room-temperature argon at a pressure equal to one atmosphere times the ratio of the respective cross-sections: 1.641 X 10-2 atmospheres (12.47 Torr). An argon number density equivalent to this pressure would exist also in a volume of argon gas at atmospheric pressure and at a temperature of 18280 K. Therefore, the two calibration points, obtained by measuring the scattering signals from room-temperature argon and helium, span the expected range of plasma-gas temperatures. Conveniently, Rayleigh scattering is linearly related to both the concentration of scatterers and to the scattering cross- section.This can be seen in the following relationship p, = PooRnRL . . . . a . (2) where P, is the radiant power scattered into 4n steaadians, L is the interaction length of the laser with the observation volume, Po is the incident laser power, nR is the number density of Rayleigh scatterers and oR is the Rayleigh-scatter- ing cross-section. Because of this linear relationship, one can expect the line established by the two-point calibration described above to provide a reliable instrument calibration. If the scattering system used to make these measurements were completely free from stay light, the scattering signal from a room-temperature , atmospheric-pressure volume of argon would be 60.94-times that of a helium scattering signal measured at the same temperature and pressure.Of course, if this assumption were valid, there would be no need to calibrate the system with two gases; a single-point calibration would suffice. We must, however, use a two-point calibration because our Ar and He scattering signals contain a finite stray-light component. Because of the multiple reflection or scattering events, this stray light might be somewhat depolarised with respect to the incident (laser) light and the Rayleigh scatter- ing. The magnitude of this component can be calculated from the negative x-intercept of a plot of scattering intensity versus equivalent argon pressure (Torr). The units of such a calculated stray-light component are in Torr argon.That is, the plot reveals that the stray-light component is equivalent to a Rayleigh-scattering signal from a volume of argon at the pressure given by the negative x-intercept. In the system described above, this stray-light component has been deter- mined to be 10 Torr. Because a 6000 K atmospheric-pressure argon plasma will produce a Rayleigh-scattering signal equi- valent to about 38 Torr argon at room temperature, this stray-light component would produce a significant measure- ment error if it were ignored in the calibration scheme; a two-point calibration is therefore required. Another valid concern one might have is whether the incident laser beam might be absorbed to an extent sufficient to change the measured gas kinetic temperature. Fortunately, Rayleigh scattering actually probes the gas density and not directly its temperature. Calculations indicate that gas density cannot change during the 25-11s laser pulse.Even if the kinetic energy (temperature) of the atoms in the scattering volume were perturbed by the laser, the measured gas density would not be affected on such a time scale and the inferred temperature would be valid. System Calibration During calibration of the scattering-measurement system it is necessary to maintain atmospheric-pressure helium in the scattering volume, and in a way that does not add to the stray-light component of the scattering signal. This goal was accomplished with the gas restrictor described in detail in reference 13. The restrictor is made of black-anodised aluminium, and in operation is mounted on top of the ICP torch.The restrictor has a 10-mm gap along the laser path which provides ready clearance for the 1.3 mm diameter laser beam, but is small enough to retain a flowing stream of helium gas and rninimise infusion of surrounding atmospheric gases. Preliminary scattering measurements were performed over a range of helium flow-rates to determine the flow necessary to produce an atmospheric-pressure pocket of helium inside the restrictor. As the helium flow was increased, the scattering signal decreased because the Rayleigh cross-section of helium is more than 50-times smaller than that of air. At flow-rates of 10-14 1 min-1 of helium, the scattering signal reaches a minimum; this higher value was used in subsequent calibra- tions.The two specific flows chosen for calibrating the system were 10 1 min-1 of argon introduced into the coolant-flow port of the ICP torch and a total flow of 14 1 min-1 of helium, 12 1 min-1 of which were introduced into the coolant-flow and 2 1 min-1 into the plasma-flow ports of the ICP torch. No nebuliser flow was used in either instance. Conveniently, the chosen argon flow-rate is not critical because the Rayleigh- scattering cross-section of air is almost identical with that of argon. In fact, air could be used to calibrate the system although argon reduces the chances of scattering from dust suspended in the air. Other Instrumentation The plasma power supply used in this study is a 2.5-kW, 27.12-MHz radiofrequency generator (Plasma Therm Model HFP 2500). All gas flows were metered by a mass-flow controller (Tylan, Carson, CA, USA).The ruby laser head was cooled to 15 "C by de-ioinised water from a recirculating cooler (Neslab Model No. RTE-SB, Neslab Instruments, Newington, NJ, USA) The ICP torch was of a demountable low-flow, low-power design similar in geometry and internal size to that described by Rezaaiyaan and Hieftje.17 A glass concentric nebuliser and a modified Scott-type spray chamber were used in the sample-introduction system. The spray chamber was mounted outside of the torch housing, and a long connecting tube was used between the spray chamber and the torch aerosol - gas introduction port to help reduce the number of large droplets that enter the plasma. Results and Discussion Rayleigh-scattering determinations of gas kinetic temperature were made in plasmas at applied radiofrequency powers of 750,875 and 1000 W (Fig.1 j . The solid line was obtained while an aerosol of de-ionised water was being introduced into the nebuliser-gas flow. The broken line in Fig. 1 pertains to a plasma without introduced aerosol. Although the gas temperature at an applied power of 750 W appears to be higher when an aerosol is present than when it is not, the uncertainty of the measurements is greater than the apparent temperature difference. (The plasma was somewhat less stable at the lower power.) In contrast, the relative570 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, SEPTEMBER 1987, VOL. 2 10000 F f 9000 c h 8000 c) .- u 7000 6000 c3 5000 .- 700 t ,4 Without water ,’ 800 900 1000 incident powerNV Fig.1. Effect of applied power and aqueous aerosol introduction on measured gas kinetic temperatures in the ICP. The solid line is for a plasma with an aqueous aerosol introduced into the nebuliser-gas stream and the broken line corresponds to the plasma without a water aerosol. All data were obtained in the central channel of the ICP and 10 mm above the load coil. Error bars indicate relative standard deviation of the mean of measurements at each setting $ 7500 2 a 3 c c : ’i 6500 C Y Ln m .- (3 5500 T [/ 1 With water ‘\ 1 \ \T 0 I I I 1 0 5 10 15 20 Height above load coil/mrn Fig. 2. Dependence of measured gas kinetic temperatures on height above the load coil. Data obtained on the plasma axis for a plasma sustained at 875 W of applied power.The solid line is for a plasma with an aqueous aerosol introduced into it and the broken line is for the same plasma in the absence of aerosol standard deviation of the means of the values at higher incident-power levels are all between 2 and 3%, so the differences between these higher power values are significant. A test shows that at 875 W the difference is significant at the 90% level, and at 1000 W the difference is significant at the 99.9% level. The general trends shown in Fig. 1 are as expected; higher applied plasma power produces a higher gas temperature and an entrained aerosol cools the plasma. This cooling effect is not as great as might be expected, and there is some evidence in data shown later that entrained water might keep the plasma from cooling as quickly after it leaves the load-coil region.Fig. 2 shows the variation of gas temperature measured on the plasma axis as a function of height above the load coil for a plasma at a power of 875 W both with and without aerosol introduction. The data presented in this figure are the average of three sets obtained on different days. Because it was found in this preliminary investigation that day-to-day variations were greater than those within a given experimental run, these averaged values reveal any trends that depend on experimen- tal variables. Unfortunately, these day-to-day variations were exaggerated by the use of different ICP power supplies and nebulisers on successive days. To provide a better indication of the attainable precision of the technique, the error bars shown in Fig.2 represent the la deviations that were common within each of the three data sets. The relative standard deviations are 5-6% for any given data point on a particular day. It is felt that much of this variation is due to source s h 7500 3 c 4- 0 C Y v) rn .- 5 6500 .- U I I 0 I T l \ T L 5500 I 1 I I I 1 0 5 10 15 20 Height above load coilhm Fig. 3. Gas temperatures determined at several vertical heights above the load coil and 3 mm off-axis, for an 875-W incident r.f.-power plasma. The solid line is for an aerosol-containing plasma and the broken line is for a plasma without an aerosol flicker, as the signal measured in this approach represents the gas density during a relatively short period of time (25 ns). Again, the trends found in Fig.2 are not surprising. One would expect the aerosol-containing plasma to be cooler than the same plasma without water aerosol. Interestingly, however, the gas temperature of the dry plasma declines more rapidly with height than when the plasma contains water vapour. This trend suggests again that the added water vapour keeps the plasma from cooling as quickly as it would without water. Similar data are presented in Fig. 3, but for vertical zones 3 mm radially from the central axis of the plasma. Again, the solid line corresponds to the plasma with an aqueous aerosol being introduced into it and the broken line to the same ICP without aerosol. The trends indicated in Fig. 3 are interesting, especially when compared with Fig. 2. When an aerosol is introduced into the plasma, the off-axis plasma gas seems to cool faster than when the ICP contains no aerosol.Perhaps the greater heat capacity of water vapour (and its fragments) in the aerosol channel retains heat there (see Fig. 2), and permits the surrounding gases to be cooled more rapidly as the plasma propagates upwards. A comparison of the gas temperatures determined here with those reported by previous w0rkers5.11~12 shows that our values are in best agreement with those obtained through the measurement of Doppler-broadened argon lines. For instance, Human and Scott5 reported a Doppler temperature of 7530 K for a 1000-W aerosol-containing plasma, 10 mm above the load coil (LC). In our study, on-axis measurements under the same conditions and height produce a gas temperat- ure of 8200 K (Fig.1). Human and Scott also reported values of from 4380 to 4960 K under the same conditions but using Ca and Sr lines for the Doppler measurements. Obviously, there is great disparity in temperatures measured from the Doppler width of different emission lines. The measurements reported here were obtained on a low-flow, low-power torch. It is therefore not surprising that the temperatures are higher than in a larger conventional torch at similar powers. Moreover, the values determined here are more likely to be biased high than low, because of likely errors in the calibration procedure. If the gas restrictor used to calibrate the system produces a significant amount of stray light, the “high temperature” calibration point obtained with helium would produce a larger signal than it should.The relatively lower stray-light contribution to the scattering intensity in the plasma itself would then be interpreted as a reduction of Rayleigh scattering, and thus an increase in the gas temperature. Conversely, water vapour that is present in the plasma gases, because its Rayleigh-scattering cross-section is largerJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, SEPTEMBER 1987, VOL. 2 571 than argon, would produce a negative error in the measured gas temperatures. However, because water vapour should comprise only a small fraction (1%) of the gases in the central channel of the ICP, this factor would probably be insignifi- cant. The gas-temperature values reported here provide a ther- mal picture of the ICP which is not too surprising.The aerosol channel, while extremely hot, is somewhat cooler than the surrounding plasma gas. As the gases rise, a thermal equilib- rium is approached between this central channel and its surroundings. In a plasma that contains no aerosol, this equilibration might be complete as low as 5 mm ALC. However, if the central channel contains aqueous aerosol, the equilibrium is not achieved so low in the plasma. In fact, the central channel then cools more slowly than the surrounding plasma gases and is therefore hotter at 15 mm ALC than are the gases 3 mm off-axis. This last effect is not very different from reported observa- tions that are based on excitation temperatures as a ther- mometric probe. For example, Furuta and Horlick18 dis- played a vertical map of iron excitation temperatures both on- and off-axis (2.5 mm) in an aerosol-containing ICP.Their map, while quantiatively different from ours, is qualitatively quite similar. Their excitation temperature measured on the plasma axis increased continuously until about 20 mm ALC where it approached equilibrium with the off-axis excitation temperature. Although our gas temperatures decrease over this vertical range, the on- and off-axis temperatures merge. Interestingly, the on-axis excitation temperature of Furuta and Horlick was higher at 20 mm ALC than the corresponding off-axis excitation temperature. This same phenomenon was observed in the gas-temperature measurements reported here. In conclusion, Rayleigh scattering appears to be a viable approach to obtaining gas temperatures in the ICP.The present work is, however, preliminary and further studies are necessary. In particular, an investigation of the effects of the various gas flow-rates on the spatial gas temperature profiles is of great interest. In addition, the effect of analytes on the thermal behaviour of the ICP would be informative. Not only are these further studies to be performed in our laboratory, but improvements being made in the scattering instrument should reduce the uncertainty in the measured gas tempera- tures to less than 3%. These improvements include the use of a Nd : YAG pulsed probe laser instead of the present ruby laser. This YAG laser is pulsed at 20 Hz and will allow many more scattering measurements to be obtained in the same (or less) time.Improvements from the ability to signal average are therefore anticipated. Supported in part by the National Science Foundation through grant CHE 83-20053, by the Office of Naval Research and by American Cyanamid. The authors are indebted to George E. Ewing for his generous loan of the ruby laser used in this study. References 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. Resnick, R., and Halliday, D., “Physics,” John Wiley, New York, 1960, Chapter 21, p. 526. Furuta, N., Spectrochim. Acta, Part B , 1986, 41, 1115. Boumans, P. W. J. M., and de Boer, F. J., Spectrochim. Acta, Part B , 1977, 32, 365. Kornblum, G. R . , and de Galan, L., Spectrochim. Acta, Part B , 1977, 32, 17. Human, H. G. C., and Scott, R. H . , Spectrochim. Acta, Part B , 1976, 31, 459 Huang, M., Marshall, K. A , , and Hieftje, G. M., Anal. Chem., 1986, 58, 207. Alder, J. F., Bombelka, R. M., and Kirkbright, G. F., Spectrochim. Acta, Part B , 1980, 35, 163. Furuta, N., Spectrochim. Acta, Part B , 1985, 40, 1013. Walker, Z., and Blades, M. W., Spectrochim. Acta, Part B , 1986, 41, 761. Blades, M. W., Caughlin, B. L., Walker, Z . H . , and Burton, L. L., Prog. Anal. Spectrosc., 1987, 10, 57. Kawaguchi, H., Ito, T . , and Mizuike, A . , Spectrochim. Acta, Part B , 1981, 36, 615. Hasegawa, T . , and Haraguchi, H., Spectrochim. Acta, Part B , 1985, 40, 123. Marshall, K. A., and Hieftje, G. M., Spectrochim. Acta, Part B , submitted for publication, 1987. Kunze, H. J., in Lochte-Holtgreven, W., Editor, “Plasma Diagnostics,” North-Holland, Amsterdam, 1968, Chapter 9, p. 587. Weast, R. C., Editor, “CRC Handbook of Chemistry and Physics,” 67th Edition, CRC Press, Cleveland, 1986, p. E-360. Vriens, L., and Adriaansz, M., J . Appl. Phys., 1974,453,4422. Rezaaiyaan, R., and Hieftje, G. M., Anal. Chim. Acta, 1985, 173, 63. Furuta, N., and Horlick, G . , Spectrochim. Acta, Part B , 1982, 37, 53. Paper J7f6.5 Received June lst, 1987 Accepted July 30th, 1987

ISSN:0267-9477    

DOI:10.1039/JA9870200567

出版商:RSC    

年代:1987

数据来源: RSC

摘要:

JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, SEPTEMBER 1987, VOL. 2 573 Laser-en hanced lonisation Spectroscopy in Flames and Plasmas* Plenary Lecture Gregory C. Turk Center for Analytical Chemistry, US National Bureau of Standards, Gaithersburg, MD 20899, USA Recent developments of laser-enhanced ionisation (LEI) spectroscopy are reviewed. This method utilises tunable dye lasers to enhance the rate of collisional ionisation of specific metal atoms in aflame or other atom reservoirs. It can be utilised for trace metal analysis, or for fundamental studies of flames and plasmas. Among the topics highlighted here is the use of double-resonance laser excitation methods to generate three-dimensional atomic LEI spectra. Also discussed are recent results on the use of laser-induced ionic fluorescence as an optical detection method for LEI.An updated list of LEI limits of detection compiled from the literature is included. Keywords: Laser-enhanced ionisation; laser-induced fluorescence; oil analysis; ion-electron recombination; inductively coupled plasma ionisation interference This paper reviews the method of laser-enhanced ionisation spectroscopy (LEI), with particular emphasis on develop- ments which have taken place since the publication of other reviews of this subject.1-3 Notable areas of recent research activity have been in the further use of double-resonance excitation,4.5 progress on the use of LEI with other atom reservoirs besides the flame610 and in exploiting the unique properties of LEI for the diagnostic study of fundamental properties of the and the inductively coupled plasma (ICP).’ In addition, the application of LEI to real sample analytical problems has continued.15’7 Theory The term laser-enhanced ionisation refers to the enhanced rate of collisional ionisation of atoms (or molecules) which occurs in a flame when those atoms undergo intense photo- excitation from a laser source.The laser excitation results in significant population of the upper level of the electronic transition to which the laser is tuned. As these excited atoms require less collisional energy from the flame to be ionised, the rate of such ionisation is very much greater than ionisation from the ground state. In an air - acetylene flame, those atoms which receive 4 eV of energy by the absorption of photons from an ultraviolet laser will undergo collisional ionisation at a rate eight orders of magnitude greater than the ground-state atoms.This enhanced ionisation can be detected in a flame by placing biased electrodes in the flame and measuring the increased current that flows between them. Thus LEI is an opto-galvanic method.18.19 It is also related to the technique known as resonance ionisation spectr6scopy (RIS) .20 Reson- ance ionisation spectroscopy also utilises laser excitation of a specific atomic electronic transition, but then relies on photo-ionisation from a second photon rather than collisional ionisation, and can thus be performed in collision-free environments. The terminology which has evolved regarding the different ionisation based measurements is quite confus- ing. From a strictly semantic point of view one could consider LEI to be a form of RIS, as the ionisation in LEI occurs when the laser is in resonance with an electronic transition, and RIS could also be considered an opto-galvanic technique as detection occurs electrically in response to optical excitation.This situation is further confused by the fact that both mechanisms of ionisation (collisional and photo-ionisation) can occur together in some situations.5.21 * Presented at the 1987 Winter Conference on Plasma and Laser Spectrochernistry, Lyon, France, 12th-16th January, 1987. Equipment and Applications A variety of different pump sources have been utilised for the dye lasers used in LEI spectroscopy, including continuous- wave argon and krypton ion lasers, and copper vapour, nitrogen, Nd : YAG and excimer lasers and pulsed flashlamps.The excimer and Nd : YAG pumped lasers are presently the most commonly used, largely for their reliability and wavelength flexibility. Under optimum conditions, it is possible to ionise all of the analyte atoms within the laser beam, but when short pulse duration (5-10 ns) lasers such as the Nd : YAG or excimer laser are used, very fast ionisation rates are required. Such high ionisation rates can only be achieved by populating excited states which are within approximately 1 eV of the ionisation limit. For elements with low ionisation potentials this is rather easy to achieve, but for high ionisation potential elements double-resonance stepwise excitation is req~ired.5~22723 This technique utilises two independently tunable, synchronised and co-linear dye lasers to “step ladder” up to a high lying electronic level using two transitions such that the upper level of the first step transition is the same as the lower level of the second step transition.The requirement for a double resonance greatly enhances the selectivity of the method. The major advantage of LEI is its high sensitivity. When proper excitation conditions are met, it is possible to ionise all of the analyte atoms within the laser irradiated portion of the flame and to detect the ionisation with high efficiency. Table 1 gives a current listing of the best limits of detection for LEI published by six research groups for 39 elements. For 23 of these elements, limits of detection less than or equal to 200 pg ml-1 have been measured, with ten of these below the level of 50 pg ml-1. As mentioned earlier, the double-resonance stepwise exci- tation technique offers improved spectroscopic selectivity.This is illustrated in Figs. 1-3 which are spectra recorded for the determination of Ni in a heavy oil flash distillate sample,14 an important analysis for the control of catalyst poisoning in petroleum refining. The spectrum shown in Fig. 1 was recorded using single-photon excitation in the vicinity of the Ni analysis line at 300.249 nm. Also seen in the figure is another Ni line at 300.363 nm, two Fe lines at 300.095 and 300.303 nm and an unidentified line at approximately 300.28 nm. The sample contains 12.6 pg g-1 of Ni and has been diluted by a factor of 100 into a solvent of 50+50 V/V xylene-butanol giving a Ni concentration of 126 ng ml-1 which is aspirated into the air-acetylene flame.Both the sensitivity and selectivity are marginal in this instance for accurate analysis. A much improved situation is seen in Fig. 2574 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY. SEPTEMBER 1987. VOL . 2 Table 1 . LEI limits of detection Element Ag . . . . . . . . A1 . . . . . . . . As . . . . . . . . Au . . . . . . . . Ba . . . . . . . . Bi . . . . . . . . Ca . . . . . . . . Cd . . . . . . . . c o . . . . . . . . Cr . . . . . . . . c s . . . . . . . . c u . . . . . . . . Fe . . . . . . . . Ga . . . . . . . . In . . . . . . . . K . . . . . . . . Li . . . . . . . . Lu . . . . . . . . Mg . . .. . . . . Mn . . . . . . . . Mo . . . . . . . . Na . . . . . . . . Ni . . . . . . . . Pb . . . . . . . . Rb . . . . . . . . Sb . . . . . . . . s c . . . . . . . . si . . . . . . . . Sn . . . . . . . . Sr . . . . . . . . Ti . . . . . . . . TI . . . . . . . . Tm . . . . . . . . v . . . . . . . . w . . . . . . . . Y . . . . . . . . Yb . . . . . . . . Zn . . . . . . . . 1st or single 2nd step step wavelength/ wavelength/ nm 328.1 309.3 278.0 242.8 307.2 306.8 422.6 228.8 252.1 252 455.5 324.8 302.1 294.4 451.1 766.5 670.8 308.2 285.2 279.5 319.4 589.0 300.2 283.3 287.8 301.9 288.2 284.0 293.2 320.0 291.8 297.3 318.4 283.1 298.4 267.2 213.9 nm 421.1 479.3 585.7 466.2 591.7 453.1 571 . 0 460.3 521.5 568.3 576.5 600.2 597.0 377.57 396.6 Laser* E F E Y F F Y Y Y E N Y E E N K E F E Y F E Y Y N E F F Y E F E F F E F E Y LOD$/ Flamet ng ml-1 A N A A A A A A A A P A A A A H A N A A N A A A P A N N H A N A N N A N A A 0.07 0.2 3000 1 0.2 2 0.03 0.1 0.08 0.2 0.004 0.7 0.08 0.04 0.0009 0.1 0.0003 0.2 0.003 0.02 0.003 0.08 0.09 0.1 0.2 0.3 0.01 1 0.008 200 0.9 300 10 2 1 10 50 40 Reference 5 24 25 22 26 26 15 22 22 27 28 22 29 29 30 31 5 32 29 33 24 5 22 22 28 25 32 24 22 25 24 34 32 24 25 32 25 35 * F = flashlamp pumped dye laser.Y = Nd:YAG pumped dye laser. N = nitrogen pumped dye laser. K = krypton ion pumped t A = air . acetylene. N = nitrous oxide . acetylene. H = air . hydrogen and P = propane -butane . air . $ Limit of detection based on three times the standard deviation of a single measurement for a signal equivalent to the background .continuous-wave dye laser and E = excimer pumped dye laser . 0.25 1 1 300 300.1 300.2 300.3 300.4 300.5 Wavelengthlnm Fig . 1 . LEI spectrum for the determination of Ni in heavy oil flash distillate using single-photon excitation near the Ni analysis line at 300.249 nm where the same laser is scanned over the same spectral region while aspirating the same solution. but with a second laser tuned and fixed at the 561.469-nm wavelength which corre- sponds to a Ni transition from the 4p level at 33501 cm-1 to the 5 s level at 51306 cm-1. 10273 cm-1 below the ionisation limit . The lower level of this second laser transition is the same as the 7 1 6 - .. i!? 5 - F 4 - $ 3 - C 3 F (D I 1 300.4 300.5 300 300.1 300.2 300.3 Wavelengthlnm Fig . 2 . Stepwise excitation LEI of Ni in heavy oil flash distillate.scanning the first step wavelength with the second step wavelength fixed at 561.48 nm upper level of the Ni line at 300.249 nm . These two transitions are thus directly connected. and. as seen in the figure. a 35-fold increase in sensitivity is achieved at this double- resonance wavelength. improving both sensitivity and selec- tivity . Even greater selectivity can be achieved when the second step laser is scanned with the first step laser fixed at575 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, SEPTEMBER 1987, VOL. 2 7 - 6 - ul .: 5 t 4 - E + 3 - 3 4- .- - 2 - 1 - - O t I I I I I I I 56 1 561.2 561.4 561.6 561.8 562 Wavelengthhm Fig. 3. Stepwise excitation LEI of Ni in heavy oil flash distillate, scanning the second step wavelength with the first step wavelength fixed at 300.25 nm A Fig.4. Three-dimensional double-resonance LEI spectrum of 50 ng ml-I Ni. Reproduced with permission from Appl. Spectrosc., 1986,40, 1146\ 300.249 nm, as seen in Fig. 3, as there are no interfering lines observed in the wavelength region of the second step laser. Note that the base line observed in this situation is the single-photon Ni LEI signal induced by the first step laser and thus is not detrimental to either the selectivity or the limit of detection. Figs. 2 and 3 can be thought of as two planes of a three-dimensional spectrum of the first step wavelength versus the second step wavelength versus the LEI signal. Such a three-dimensional spectrum is shown in Fig. 4, which was recorded with the aid of a computer-controlled LEI spec- trometer.4 The wavelength regions studied are the same as for Figs.2 and 3, but a standard 50 ng ml-1 Ni solution was the aspirated sample rather than the diluted oil sample. The tall peak occurs when both wavelengths are on connected resonances. Two ridges orthogonal to the first step wavelength axis are observed. These are single photon Ni lines, the same as in Fig. 1. No corresponding ridge is seen for the second step wavelength, as there is negligible thermal population of the lower level of the second step Ni transition. A smaller peak is observed at the intersection of the two disconnected Ni transitions at 300.363 nm (880-34163 cm-1) and 561.479 nm (33501-51306 cm-I), which is an indication of some mixing between the levels at 34163 and 33501 cm-1.Intersecting with the large double resonance peak, another ridge can be seen running diagonally to the wavelength axes. A clearer example of such a diagonal ridge is shown in Fig. 5 for the element Ba. In this instance the spectrum has been plotted as a contour rather than a three-dimensional projec- tion, and shows the wavelength regions around the Ba double 553.40 553.45 553.50 553.55 553.60 553.65 553.70 Wavelength l/nm Fig. 5. Contour plot of the stepwise excitation LEI spectrum of Ba. (Reproduced with permission from Appl. Spectrosc., 1986, 40, 1146) Fig. 6. Three-dimensional LEI spectrum recorded for high-alloy steel SRM 1289a for the determination of Co. The double-resonance peak for Co is seen in the foreground corner.Dissolved solution concentration of Co is 350 ng ml-I. The diagonal ridge is caused by two- hoton excitation of Fe. (Reproduced with permission from AppfT Spectrosc., 1986, 40, 1146) resonance at 553.55 nm (CL18060 cm-1) and 494.73 nm (18060-38267 cm-1). Three ridges clearly intersect the double-resonance peak. As before, a ridge orthogonal to the first step wavelength is observed due to single-photon excitation, and in this instance there is sufficient thermal population of the lower level of the second step transition at 18060 cm-1 to observe a ridge which is orthogonal to the second step wavelength axis, and a very obvious diagonal ridge. Along this diagonal ridge, the sum of the energies of the two laser wavelengths remains constant and equal to the top level of the Ba double resonance at 38267 cm-1, which is being populated by means of two-photon excitation. This two- photon event uses a virtual level at an energy which varies along the ridge and which is being resonantly enhanced by the real Ba level at 18060 cm-1. At the intersection of this diagonal with the other ridges we have the special situation where the virtual intermediate level becomes a real level, and a much stronger signal is observed.A compound spectral feature such as the two-photon diagonal ridge is very difficult to be cognisant of without the ability to measure a three- dimensional spectrum. A practical example showing the occurrence of another diagonal ridge is seen in Fig. 6, which is a three-dimensional LEI spectrum recorded for the determi- nation of Co in a high-alloy steel Standard Reference Material (SRM 1289a).A 1 g 1-1 solution of the steel sample was aspirated, giving a solution Co concentration of 350 ng ml-l.576 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, SEPTEMBER 1987, VOL. 2 In the foreground of the figure is the Co double-resonance peak at 252.14 and 591.68 nm (0-39649-56546 cm-1). Near the Co peak is a two-photon diagonal ridge which originates from a large Fe double resonance at 285.285 and 590.71 nm, which is well off the scale of the figure. No interference of the Co measurement occurred in this example, but the possibility of interference from very weak diagonal two-photon signals of major elements in a sample does exist, and background correction would need to be performed using a three-dimen- sional spectrum.Flame and Plasma Diagnostics In addition to its use for chemical analysis, LEI has recently been exploited as an informative diagnostic method for the study of fundamental processes which occur in flames and plasmas. Axner and Berglind have recently utilised LEI to measure the Stark effect on the Rydberg levels of atoms in the flame.36 Laser-enhanced ionisation is a uniquely sensitive approach for the study of Rydberg levels because they lie just below the ionisation limit and atoms excited to such levels are rapidly ionised. Variation of the electric field applied between the LEI probe electrodes was shown to cause Stark shifting of transitions to Rydberg levels. Measurement of such Stark shifts can be used to map the distribution of the electric field in the flame.This will be very useful in the design of optimum electrode configurations for LEI detection. A different sort of flame diagnostic method was reported by Omenetto et a1.,12 who have determined the lifetime of a metastable level of thallium in the air-acetylene flame by means of an LEI measurement which utilises two disconnec- ted T1 lines in a double-resonance excitation. The experiment uses the T1 line at 276.79 nm to connect the 2P1,2 ground state to the level at 36118 cm-1. From this level an indirect population of the 2P312 metastable level at 7793 cm-1 occurs either by collision or by emission of a photon at 352.91 nm. A second laser tuned to 291.83 nm then connects the metastable level with a highly excited level at 42049 cm-1, from which rapid collisional ionisation occurs. By addition and variation of a time delay between the first and second lasers, the lifetime of the intermediate metastable level was determined to be 81 ns.When LEI is utilised together with laser-induced fluor- escence (LIF) , a particularly valuable diagnostic capability is achieved. One such example is the optical detection of LEI by means of laser-induced ionic fluorescence of the resulting ions, as recently reported by Turk and Omenetto.11 In this work the ionisation of Sr in the air-acetylene flame was laser-induced using the Sr atom resonance at 460.7 nm and photoionisation of the resulting excited state Sr using laser light at 308 nm from the XeCl excimer laser which pumps the dye laser. The resulting Sr ions were then detected by laser excitation of Sr+ at 421.55 nm and observation of the Sr+ fluorescence at 407.77 nm.By delaying the ionic fluorescence probe laser relative to the ionising lasers, the temporal fate of the laser produced Sr+ ions in the flame could be studied. The results of such an experiment are given in Fig. 7, which shows the Sr+ ionic fluorescence signal as a function of time delay between the probe and ionising lasers. At zero delay, the increased concentration of Sr+ relative to the natural level of Sr+ in the flame is observed. As delay is added, a fast decay of Sr+ is initially observed, consuming about 85% of the laser-produced ions, This fast decay, which occurs with an exponential time constant of 58 ns, has been attributed to chemical reaction of the Sr+ and was found to increase in rate under oxygen-rich flame conditions.After the fast decay subsides, the remaining 15% of the laser-produced ions decay at a much slower rate as ion-electron recombination takes place. Fig. 8 demonstrates the effect of different amounts of Cs matrix on the recombination of Sr+, showing the expected increased rate of recombination resulting from the increased electron density in the flame caused by thermal ionisation of c s . The effect of easily ionisable elements (EIEs) on ion - electron recombination in flames has been well under- stood for many years, but this is not so for the inductively coupled plasma (ICP). The primary difficulty in the study of interferences caused by EIEs in the ICP is that there are a variety of effects besides that of changing the ionisation equilibrium. Optical detection of laser produced Sr+ ions allows direct measurement of the rate of ion - electron recombination, and as changes in the recombination rate cause shifts of the ionisation equilibrium, a direct study of such an effect can be made.’ In Fig.9, the decay of laser-produced Sr+ ions in the ICP is shown, using the same procedure as in the flame. The data were collected with 1 mm diameter laser beams aligned through the centre of the plasma, 15 mm above the load coil. Fluorescence from the plasma centre was collected using a lens at a right angle to the laser beams. The plasma power level was 0.9 kW for this measurement, but similar results were obtained at higher power levels. Unlike in the flame, no fast decay due to chemical reactions with Sr+ is !+ \ f +t+ \ +s ++ - 200 0 200 400 600 800 Time delayins Fig.7. Observed time decay of strontium ion concentration in the air - acetylene flame following laser-induced ionisation of strontium. The time delay refers to time elapsed between the firing of the fluorescence probe laser from the moment of firing of the ionising laser. (Reproduced with permission from Appl. Spectrosc., 1986,40, 1085) Fig. 8. Effect of varying amounts of caesium on the observed recombination rate of strontium ions in the flame. Caesium concentra- tions: A , 100; B, 300; and C , 1000 mg 1-1. The concentration of strontium is 10 mg 1-I. (Reproduced with permission from Appl. Spectrosc., 1986, 40, 1085)JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, SEPTEMBER 1987, VOL.2 577 0 10 20 30 40 Time delayip Fig. 9. Sr; and B, 100 mg 1 I Sr + 1000 mg I - ’ Li Recombination of strontium ions in the ICP: A, 100 mg I - ’ observed in the inert argon atmosphere of the ICP. The figure shows recombination of Srf ions both with and without the presence of a matrix of 1000 mg 1-1 of Li. The effects of increased recombination rate and decreased level of non-laser produced Sr+ ions caused by the Li matrix are clear. It is expected that this type of measurement will be very useful in the study of EIE effects in the ICP and other plasmas. Future Directions Further research and development of LEI is expected to continue both in fundamental and applied areas. In the applied area, the use of LEI as an element-specific detector for liquid chromatography is presently being explored. The analysis of high-purity materials is another application area in which LEI may prove useful.The coupling of LEI to laser-induced fluorescence measurements has been shown to be beneficial for fundamental studies and this will continue to be explored. Fluorescence dip spectroscopy,37 which measures the loss of fluorescence signal resulting from ionisation, is such an area. The fluorescence measurement of ion - electron recombination in the ICP has already yielded some interesting observations, but further work needs to be carried out under a wide variety of ICP conditions. In addition, this technique should be applicable to other plasmas. Diagnostic techniques such as these will be useful in achieving the desirable goal of developing LEI measurements in other atom reservoirs. References 1.Travis, J. C.. Turk, G. C., DeVoe, J . R . , Schenck, P. K., and Van Dijk, C. A . , Prog. Anal. At. Spectrosc., 1984, 7, 199. 2. Green, R. B., in Piepmeier, E. H . , Editor, “Analytical Applications of Lasers,” Wiley-Interscience, New York, 1986, Chapter 3. Travis, J . C., Turk, G . C., and Green, R. B., Anal. Chem., 1982,54, 1006A. Turk, G. C., Ruegg, F. C., Travis, J. C., and DeVoe, J. R., Appl. Spectrosc., 1986, 40, 1146. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22, 23. 24. 25. 26. 27. 28. 29. 30. 31. 32. 33. 34. 35. 36. 37. Omenetto, N., Smith, B. W., and Hart, L. P., Fresenius 2. Anal. Chem., 1986, 324, 313.Turk, G. C., and Watters, R. L., Jr., Anal. Chem., 1985, 57, 1979. Turk, G . C., Axner, O., and Omenetto, N., Spectrochim. Acta, Part B, in the press. Magnusson, I., Axner, O., Lindgren, I., and Rubinsztein- Dunlop, H., Appl. Spectrosc., 1986, 40, 968. Magnusson, I., Sjostrom, S . , Lejon, M., and Rubinsztein- Dunlop, H . , Spectrochim. Acta, Part B, 1987,42, 713. Bykov, I. V., Skvortsov, A. B., Tatsii, Yu. G., and Chekalin, N. V., J . Phys. Colloq., 1983, C7, 345. Turk, G. C., and Omenetto, N., Appl. Spectrosc., 1986, 40, 3 085. Omenetto, N., Berthoud, T., Cavalli, P., and Rossi, G . , Appl. Spectrosc., 1985, 39, 500. Axner, O., Lejon, M., Magnusson, I . , Rubinsztein-Dunlop, H . , and Sjostrom, S . . Appl. Opt., in the press. Turk, G. C., Havrilla, G . J., Webb, J.D., and Forster, A. R., in Lyon, W. S . , Editor, “Analytical Spectroscopy,” Elsevier, Amsterdam, 1984, pp. 63-68. Turk, G. C . , and De-Ming, Mo, in Koch, W. F., Editor, “Methods and Procedures Used at the National Bureau of Standards to Prepare, Analyze, and Certify SRM 2694, and Recommendations for Use,” NBS Special Publication 260-106, US Government Printing Office, Washington, DC, 1986, pp. 30-33. Havrilla, G. J., and Carter, C. C., Appl. Opt., in the press. Salcedo Torres, L. E., Zorov, N. B., and Kuzyakov, Y. Y., J . Anal. Chem. USSR (Eng. Transl.), 1982, 36, 1016. Green, R. B., Keller, R. B., Luther, G . G., Schenck, P. K., and Travis, J. C., Appl. Phys. Lett., 1976, 29, 727. Camus, P., Editor, “Optogalvanic Spectroscopy and its Appli- cations,” J .Phys. Colloq., 1983, C7. Young, J. P., Hurst, G. S . , Kramer, S. D., and Payne, M. G . , Anal. Chem., 1979, 51, 1050A. Curran, F. M., Lin, K. C., Leroi, G. E., Hunt, P. M., and Crouch, S. R . , Anal. Chem., 1983, 55, 2382. Turk, G . C., DeVoe, J. R . , and Travis, J . C., Anal. Chem., 1982, 54, 643. Gonchakov, A. S . , Zorov, N. B., Kuzyakov, Y. Y., and Matveev, 0. I., Anal. Lett., 1979, 12, 1037. Messman, J. D . , Schmidt, N. E . , Parli, J . D., and Green, R. L., Anal. Chem., 1985, 39, 504. Axner, O., Magnusson, I . , Peterson, J . , and Sjostrom, S . , Appl. Spectrosc., 1987, 41, 19. Turk, G. C . , Travis, J. C., DeVoe, J . R., and O’Haver, T. C., Anal. Chem., 1978, 50, 817. Berthoud, T., Camus, P., Drin, N . , and Stehle, J . L . , J . Phys. Colloq., 1983, C7, 389. Chaplygin, V. I., Kuzyakov, Y. Y., and Zorov, N . B., Spectrochim. Acta, Part B, Supplement, 1983, 38, 386. Axner, O., Lindgren, I., Magnusson, I . , and Rubinsztein- Dunlop, H., Anal. Chem., 1985, 57, 773. Bykov, I. V . , Chekalin, N. V., andTikhomirova, E. I.,J. Anal. Chem. USSR (Eng. Transl.), 1986, 40, 1579. Havrilla, G. J . , Weeks, S. J., and Travis, J. C., Anal. Chem., 1982,54, 2566. Peters, R. A . , and Green, R. B., presented at the 185th ACS Meeting, paper No. 209, Seattle, WA, March 24, 1983. Turk, G . C., Travis, J . C., and DeVoe, J. R., J . Phys. Colloq., 1983, C7, 301. Omenetto, N . , Berthoud, T., Cavalli, P., and Rossi, G., Anal. Chem., 1985, 57, 1256. Havrilla, G . J . , and Choi, K. J., Anal. Chem., 1986, 58, 3095. Axner, O., and Berglind, T., Appl. Spectrosc., 1986,40, 1224. Omenetto, N., Turk, G . C., Rutledge, M., and Winefordner, J. D . , Spectrochim. Acta, Part B, 1987, 42, 807. Paper J7l56 Received April 27th, 1987

ISSN:0267-9477    

DOI:10.1039/JA9870200573

出版商:RSC    

年代:1987

数据来源: RSC

摘要:

JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, SEPTEMBER 1987, VOL. 2 579 Thermal Lensing Spectrophotometry of Uranium(V1) with Pulsed Laser Excitation* Nicolo Omenetto, Paolo Cavalli, Guglielmo Rossi, Giovanni Bidogliot and Gregory C. TurkS Joint Research Centre, Chemistry Division, 21020-lspra ( Varese), Italy A double-beam thermal lensing experiment based upon a pulsed dye laser as the excitation beam and a He - Ne laser as a probe beam is described. The technique has been used to detect the small absorptions given by uranyl ions in aqueous solution. The system is capable of detecting absorption coefficients as low as 5 x 10-7 cm-1. Moreover, by changing several dyes, an absorption spectrum can be obtained at concentration levels (ca. 4 x l O - 6 ~ ) far below (ca. IO3x) those obtained by conventional absorption spectrophotometry, thereby allowing the direct study of the chemical equilibria involved.The sensitivity of the present apparatus is limited by short-term fluctuations of the probe laser. Keywords : Pulsed tunable dye lasers; thermal lensing spectrophotometry; uranium; chemical speciation A knowledge of the migration mechanism of actinide ions in ground waters is essential to elucidate radionuclide reactions in geochemical processes such as rock formation or migration from nuclear waste repositories. The sensitivity of classical analytical techniques is generally too low for the direct detection of dissolved species at realistic concentration levels, as no chemical treatment or enrichment processes can be performed. Moreover, the determination of the oxidation states of these species, or their chemical speciation, is of major interest for the understanding of their physico-chemical behaviour at such low concentration levels.Thermal lensing spectrophotometry1.2 and photoacoustic spectroscopy3-4 have been proposed as techniques for the study of the behaviour of the actinides in water. Both methods possess very high detection sensitivity and can, in principle, provide directly the absorption spectrum of the ionic species present in solution, characterised by a given oxidation state. The aim of this paper is to describe a dual-beam system, based on pulsed laser excitation, which has been used for the detection and speciation of uranyl ions in a carbonate - perchlorate system. This is of interest in studies of the complexation behaviour of uranium.5 Theoretical Considerations The theory for the thermal lens effect is well described in the literature for both continuous wave (CW) and pulsed laser excitation616 and therefore only its essential principles will be summarised below.The thermal blooming or thermal lens effect is a phenom- enon associated with the temperature rise in a sample solution, caused by absorption of radiation. When a laser, tuned to an appropriate absorption band and characterised by a Gaussian intensity distribution, is focused on the sample, the sample is heated most strongly at the beam centre where the intensity is greatest. The corresponding variation of the refractive index makes the solution behave as a diverging lens, as usually the increase in temperature lowers the refractive index and therefore shortens the optical path at the beam centre.Under laser illumination, the sample temperature reaches an equilibrium value when the rate of heat input from the laser is just balanced by the rate of heat conduction out of the illuminated sample. Therefore, there is a fundamental “response time” or characteristic “time constant” for the * Presented at the 1987 Winter Conference on Plasma and Laser t Radiochemistry Division, Ispra. $ On leave from: Center for Analytical Chemistry, National Spectrochemistry, Lyon, France, 12th-16th January, 1987. Bureau of Standards, Gaithersburg, MD, USA. photo-thermal effect which depends on the spot size of the laser beam and on the density, specific heat and thermal conductivity of the sample solution.This time constant can vary from the order of several microseconds to seconds. The thermal lens effect is conveniently monitored in the far field by measuring the change in the spot size, i.e., by sampling the intensity of the beam at its centre. If w is the beam size at the measuring target, the relative intensity change at the beam centre (f&) reflects the relative change in w2 at time zero and at steady-state conditions ( t = w), and is given by the relationship9 Ib,(t = 0) - 1bc(t = w) - - - - AZbc - Ibc(f = O0) Ibc ~ 2 ( t = 00) - ~ 2 ( t = 0) A w ~ . . . . (1) - - - w y t = 0) W2 or, in the parabolic approximation, by AZbc -2.3 P (dnldT) Ibc h k . A . . . . (2) where, P (W) is the CW laser power, (dnld7‘) is the variation of the refractive index of the solution with temperature, h (cm) is the wavelength of the radiation, k (W cm-1 K-1) is the thermal conductivity and A (E ECI) is the absorbance of the solution where E (1 mol-1 cm-1) is the molar absorptivity, C (mol-1) the concentration and 1 (cm) the geometrical absorp- tion path length.The factor multiplying the absorbance term in the right- hand side of equation (2) is usually referred to as the “enhancement in sensitivity” relative to conventional spectro- photometry. Pulsed laser excitation can similarly be used with the advantage that experiments can be performed when high peak powers are important, as for example in non-linear absorption experiments.’ The beam quality of a pulsed dye laser is poorer than that of a CW laser and its spatial intensity distribution departs from a purely Gaussian one.On the other hand, the spectral coverage of the pulsed laser is larger than that of CW lasers. The total effective heating energy, H, of the pulsed laser is given by the relationship -- - H = P(t)dt . . . . * ( 3 ) i where T is the duration of the laser pulse (s) and P(t) its temporal power distribution (W). If we assume a rectangular pulse, the energy per pulse will be given by the product of the peak power and the pulse duration. Thus it can be assumed that the effective heating power of the pulsed laser is the energy per pulse equivalent to that delivered by a CW laser in a time approximately half of the time constant characteristic of580 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, SEPTEMBER 1987, VOL.2 the sample.10 For the uranyl solution considered here, this time is ca. 60 ms with the absorption cell placed at the focus of the excitation beam (see later); as a consequence, a 3 mJ per pulse laser would be equivalent to a 100 mW CW laser. With a pulsed excitation, a double-beam arrangement is the necessary choice. In this arrangement, also called the “pump- and-probe” configuration, one laser (the pump) forms the thermal lens while another laser (the probe) interrogates the effect. A theoretical approach to the dependence of the signal I Fig. 1. Theoretical behaviour of the measured signal intensity at the beam centre as a function of the relative positions of the pump and the probe beams. w , Probe beam size in the far field; wo minimum waist of the probe beam; and woe,,: minimum waist of the pump (excitation) beam.(a) Both waists coincident; ( b ) probe beam waist after pump beam waist; and (c) probe beam waist before pump beam waist. The sample cuvette is depicted as a diverging lens P” Beam strength on the relative positions of the waists of the two laser beams has been described in the literature17 and the results can be summarised as follows. The absorbing cuvette is placed at the waist of the excitation beam where the diverging lens is formed. In this instance, the maximum irradiance is incident on the sample. The waist of the probe beam can be independently adjusted to be at the position of maximum sensitivity, i.e., at f l Z C from the cell, where 2, is the confocal distance of the probe beam ( Z , = mv2,,/hp,). When the waist of the probe beam is placed before the cell, i.e., before the diverging lens, the already diverging probe beam expands further as it propagates towards the detector and therefore the intensity measured at the beam centre decreases.When, on the other hand, the waist of the probe beam is located after the cell, the diverging lens moves the waist of the probe beam further along the beam path with the result that the beam has less distance in which to diverge and the effect is noticed as an increase in the intensity at the beam centre. Therefore, the behaviour of the lens is antisymmetric with respect to the waist of the excitation beam and a differential response can be obtained when two identical absorbing cells are placed on each side of this waist.If one cell contains the analyte plus the solvent and the other cell contains the pure solvent, the net response is only due to the analyte.’6,17 This behaviour is depicted schematically in Fig. 1. Experimental The dual-beam arrangement used in this work is shown schematically in Fig. 2. An excimer laser (Model EMG-102, Lambda Physik, Gottingen, FRG) pumped a tunable dye laser (Jobin-Yvon, Longjumeau, France) which uses the eight-cell configuration in the oscillator and in the amplifier as described by Bos.18 The excimer laser was operated with XeCl at 308 nm Filter laser 0) 3 C m v, - a I I / al C 2 a Y- \ Mirror Aperture 0 = lmm I He - Ne laser dr Boxcar (NB) Scope - Signal Current a rn plifier T - t- Fig.2. Diagram of the experimental set-up: F, filter; PD, photodiode; and IF, interfaceJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY. SEPTEMBER 1987. VOL. 2 581 and provided ca. 130 mJ per pulse, part of which was used to pump the dye laser, which was characterised by a peak power of several hundred kilowatts in a pulse of ca. 10 ns and by a spectral band width of ca. 0.02 nm. The dye-laser output (the excitation beam) after traversing a beam splitter was focused down to a spot size of about 50 pm diameter in the middle of the absorbing solution by a quartz lens cf = 25 cm). A first cut-off filter placed near the cell blocked the beam while a second filter placed in the far field prevented any further light, from the excitation beam, from reaching the detector. A small fraction of the beam was reflected into a pyroelectric detector (Photochemical Research Associates, Ontario, Canada) which then monitored continuously the laser energy for calibration purposes.‘The probe beam was produced in a 5 mW He-Ne laser (Model 05-LHR-151, Melles Griot, Irvine, CA, USA) which was independently focused by another lens (‘f = 25 cm) before the absorbing cell so that a negative effect was measured in the far field. The maximum signal strength was found experimen- tally by translating the absorbing cell and/or the lens while recording the signal. All the optical components could be accurately adjusted in the X , Y and 2 planes so as to make the excitation and probe beam co-linear. Care was also taken not to reflect the beams directly back into the laser cavities, to avoid unwanted oscillations in the intensity.The intensity at the probe beam centre was measured by a photodiode located behind a 1.5 mm diameter pinhole at ca. 3 rn from the absorbing solution. The photodiode current, converted into a voltage and suitably amplified, was fed into a digital storage oscilloscope (Model 468, Tektronix, Beaver- ton, NJ, USA) and into a box-car integrator (Models SR 250 and SR 245, Stanford Research, Palo Alto, CA, USA) which was interfaced with a computer (HP Model 98165, Hewlett- Packard, San Diego, CA, USA). The dye laser was triggered externally at a repetition rate of 10 Hz. This rate was sufficiently low to allow the complete thermal recovery of the solution between the laser pulses. Fig. 3. Oscilloscope traces of the thermal lensing signals (inverted) obtained at 448 nm.Dye laser, coumarin 450,2.8 mJ per pulse; scope sensitivity, 50 mV per division; time base, 0.5 ms per division; current to voltage converter, lo9 V A-1; number of averages, 128. The shape of the signal is due to the detection circuit and should not be taken as the time constant of the thermal lens. (a) Blank (0.1 M NaHCO, + 0 . 4 ~ NaC10,); and (6) Uvr, 5 x M t - m C 01 v) .- Results and Discussion Evaluation of the Instrumental Sensitivity In order to check the sensitivity of the double-beam configura- tion, several solutions containing different concentrations of UV1 in a perchlorate and perchlorate - carbonate medium were tested. The molar absorptivities of these solutions vary from 7.5 to 25 1 mol-1 cm-1 at the respective peak absorption wavelengths.19 As stressed before 2 the analytical capability of the technique has to be judged on the basis of the minimum observable absorption coefficient and not on the minimum concentration detectable. Fig. 3 shows two typical oscilloscope pulses obtained with the blank and a 5 x 10-7 M U V I solution. These signals are also processed by the box car integrator and the corresponding recorder tracings are shown in Fig. 4. It is interesting that the noise in the signals is the same as the base-line noise which is obtained by blocking the dye-laser beam in front of the absorbing cuvette. The spike observed in the blank signal is probably due to an instrumental artifact. From these data the uranium solution results in a signal to noise ratio of 45 which gives a detection limit (signal to noise ratio = 2) of 2 X 10-8 M.Taking into account the molar absorptivity (E = 25 1 mol-1 cm-1) at this wavelength, the minimum detectable absorption coefficient is 5 x 10-7cm-1. If a more stable probe beam is available ( i e . , with less than 1% short-term noise) this limit can be lowered even further. The antisymmetric behaviour of our dual-beam arrange- ment is clearly borne out in Fig. 5 where the pulses shown have been obtained sequentially with a 10-4 M uranium solution. Here, the cuvette is placed as usual at the waist of the dye-laser beam and the waist of the probe beam, by adjusting its Time - Fig. 4. Recorder tracings of the signals shown in Fig. 3 and processed by the boxcar integrator.Number of pulses averaged, 300; laser repetition rate, 10 Hz. Both signals recorded for about 10 min: (a) blank; and (b) UVI, 5 x 1 0 - 7 ~ Fig. 5. Antisymmetric behaviour of the thermal lens signal in the dual-beam arrangement. ( a ) Probe beam focused after the absorbing cell; and ( b ) probe beam focused before the absorbing cell. Uvl concentration, 10-4 M ; sensitivity, 50 rnV per division; time base, 0.5 ms per division focusing lens, is formed before and after the cuvette. The location of the maximum signal is found experimentally. As one can see from the figure, almost total compensation of the effects is achievable. A differential set-up with CW lasers €or the study of lanthanides in solution has recently been described.2o582 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, SEPTEMBER 1987, VOL.2 Measurement of the Absorption Spectrum The high sensitivity of the thermal lensing technique makes it possible to obtain an absorption spectrum, characteristic of the degree of oxidation of the ion and of its chemical form, at concentrations which are totally inaccessible to ordinary spectrophotometry. In this respect, pulsed laser excitation is preferable to CW excitation, as the spectral coverage obtained with pulsed dye lasers is larger than that of CW dye lasers. In addition, with our eight-cuvette configuration, it is simple and rapid to change from one dye to another. However, it is fair to observe that the theoretical claim that the thermal lensing technique provides an absorption spec- trum over a large wavelength interval has been experimentally demonstrated only with incoherent light.21 Indeed, even with the similar technique of pulsed optoacoustic spectroscopy3.4 only a major absorption feature, i.e., a 10-15 nm absorption band, was published as it was representative of the absorption spectrum of the actinide ion investigated.The following considerations apply to the laser system used here but should also be of more general validity. (i) The useful spectral range of a single dye is restricted to that in which a ratioing procedure gives a sufficient signal to noise ratio; as a result, many dyes are needed to cover an interval of, say, 100 nm. (ii) The energy delivered to the absorbing solution depends on the size of the laser at the waist and, for diffraction limited beams, this parameter changes with wavelength and with the laser divergence, which is different for each dye used. (iii) The position of the reference detector may influence the shape of the absorption spectrum; this results from the fact that the polarisation of the laser output might change from one dye to another and possibly within the same dye,22 therefore changing the fraction of radiation reflected towards the detector when the arigle of reflection is greater than 15".As a result, taking an absorption spectrum over a large wavelength interval is neither simple nor practical. Fortu- nately, many interesting spectral features are narrow enough to allow the use of a single dye. As a test of the feasibility of the technique in obtaining a spectrum, the uranium - carbonate - perchlorate system was chosen as its spectrum is known to be characterised by several peaks in the region 400-500 nm.19 Fig.6 shows a conventional absorption spectrum taken at 4 x 10-3 M U V I in a solution containing NaC104 (0.4 M) and NaHC03 (0.1 M). The corresponding thermal lens spectrum shown in Fig. 7 was obtained at a concentration of 10-4 M and with four different dyes, as indicated in the figure. The intensities of the I 1 U 400 450 500 Wavelengthlnm Fig. 6. Absorption spectrum of a 4 X 1 0 - 3 ~ Uvl solution in a carbonate - perchlorate medium (0.1 M NaHCO, + 0.4 M NaClO,) obtained by conventional spectrophotometry Wavelengthh m Fig. 7. Thermal lensing spectrum of a ~ O - , M UvI solution in a carbonate - perchlorate medium (0.1 M NaHC03 + 0.4 M NaC10,).The four laser dyes indicated are: S1, stilbene 1; S420, stilbene 420; C440, coumarin 440; and C450, coumarin 450 4.0 r------ 1.5' 1 I I I 430 440 450 460 470 480 Wavelengthhm Fig. 8. Partial thermal lensing spectrum of the complex uranium - carbonate at a 4 X 10-6 M UVI concentration; dye, coumarin 450 individual spectra were scaled relative to each other according to the different laser energies. The similarity between the two spectra (Figs. 6 and 7) is evident. However, as stressed before, the process of changing the dyes which give different output characteristics may well result in slightly different thermal lensing parameters so a quantitative comparison between the two spectra (e.g., peaks ratio) has not been attempted at the present stage of this research.Work is now in progress in order to characterise the spatial profile of the pump beam fully and to evaluate the waist and the confocal parameter of both beams. Then it will be possible to compare accurately the theory with the experi- ment. At present however, we are studying the chemical equilibria of the uranium - carbonate system by restricting the spectrum to the last two peaks shown in Fig. 7, i.e., those obtainable with a single coumarin dye (C450). The sensitivity of the apparatus is sufficient to obtain this partial spectrum at a concentration below the limit imposed by solubility considera- tions.23 Fig. 8 shows a typical result obtained at a concentra- tion of 4 x 1 0 - 6 ~ . Conclusions Several conclusions may be drawn from this work, as follows.(i) The sensitivity of the thermal lensing technique clearly surpasses that of ordinary spectrophotometry by many ordersJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, SEPTEMBER 1987, VOL. 2 583 of magnitude. A detection limit even lower than 10-7 cm-1 is within the reach of the dual-beam technique used in our work. (ii) Pulsed laser excitation does not degrade the sensitivity of the method. In addition, the wavelength coverage of the pulsed dye laser is larger than that of CW dye lasers. (iii) Because the beam quality of the pulsed laser is poorer than that of a CW laser, care has to be taken to derive intrinsic properties of the absorbing solution from theoretical relation- ships which hold strictly for TEM, Gaussian beams.(iv) A complete absorption spectrum of the solution investigated can indeed be obtained by changing several dyes. However the over-all procedure is neither simple nor practical. (v) When a single absorption feature is sufficient for speciation purposes, then the thermal lensing and the complementary photoacous- tic techniques are the only choice for studying the chemical behaviour of species in solutions characterised by very low absorption coefficients of ca. 10-4 cm-1 or lower. G. C. Turk thanks the Joint Research Centre authorities particularly the Education Training Service of Ispra for a visiting scientist fellowship grant. References 1. Beitz, J. V., and Hessler, J. P., Nucl. Technol., 1980,51, 169. 2. Berthoud, T., Mauchien, P., Omenetto, N., and Rossi, G., Anal.Chim. Acta, 1983, 153, 265. 3. Schrepp, W., Stumpe, R., Kim, J. I., and Walther, H., Appl. Phys. B, 1983, 32, 207. 4. Stumpe, R., Kim, J. I., Schrepp, W., and Walther, H., Appl. Phys. B., 1984, 34, 203. 5. Grenthe, I., Ferri, D., Salvatore, F., and Riccio, G., J. Chem. SOC., Dalton Trans., 1984, 2439. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. Hu, C., and Whinnery, J. R., Appl. Opt., 1973, 12, 72. Twarowski, A. J., and Kliger, D. S . , Chem. Phys., 1977, 20, 253. Bailey, R. T., Cruickshank, F. R., Pugh, D., and Johnstone, W., J . Chem. SOC., Faraday Trans. 2, 1980, 76, 633. Harris, J. M., and Dovichi, N. J . , Anal. Chem., 1980,52,695A2. Fang, H. L., and Swofford, R. L., in Kliger, D. S., Editor, “Ultrasensitive Laser Spectroscopy,” Academic Press, New York, 1983, p. 175. Sheldon, S. J., Knight, L. V., and Thorne, J. M., Appl. Opt., 1982, 21, 1663. Mori, K., Imasaka, T., and Ishibashi, N., Anal. Chem., 1982, 54, 2034. Bialkowski, S . E., Appl. Opt., 1985, 23, 2792. Carter, C. A., and Harris, J. M., Appl. Opt., 1984, 23, 476. Perry, J. W., Ryabov, E. A., and Zewail, A. H., Laser Chem., 1985, 1, 9. Dovichi, N. J., and Harris, J. M., Anal. Chem., 1980,52,2338. Berthoud, T., Delorme, N., and Mauchien, P., Anal. Chem., 1985, 57, 1216. Bos, F., Appl. Opt., 1981, 20, 3553. Rabinowitch, E., and Belford, R. L., “Spectroscopy and Photochemistry of Uranyl Compounds,” Macmillan, New York, 1964. Berthoud, T., and Delorme, N., Appl. Spectrosc., 1987,41,15. Stone, J., Appl. Opt., 1973, 12, 1828. Bos, F., personal communication, 1986. Bidoglio, G., Cavalli, P., Omenetto, N., and Tanet, G., “2nd International Conference on the Chemistry of Lanthanides and Actinides,” Lisbon, Portugal, 1987, Znorg. Chim. Acta, in the press. Paper J7l18 Received February 9th, 1987 Accepted March 16th, 1987

ISSN:0267-9477    

DOI:10.1039/JA9870200579

出版商:RSC    

年代:1987

数据来源: RSC

摘要:

JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY. SEPTEMBER 1987, VOL. 2 585 Inductively Coupled Plasma Fourier Transform Spectrometry: A New Analytical Technique? Potentials and Problems* Plenary Lecture Lynda M. Faires Los Alamos Nationai Laboratory, Los Aiamos, NM 87545, USA The search for new methods in analytical chemistry has led to investigations at several laboratories into the feasibility of combining the inductively coupled plasma (ICP) source with a Fourier transform spectrometer (FTS) to make spectrochemical measurements in the ultraviolet and visible regions by ICP-FTS. Fourier transform spectrometers have the potential of fulfilling many of the criteria of an ideal detection system for analytical atomic spectroscopy, but also have characteristic properties and limitations which must be carefully considered in the development of ICP-FTS as a practical new analytical technique.This paper presents an overview of the major potentials and problems in ICP-FTS. Keywords : Fourier transform spectrometry; inductively coupled plasma; analytical atomic spectroscopy Analytical atomic spectroscopy might be described as a continual search for the “ideal” source and the “ideal” spectrometer. Although many researchers have pointed out that the inductively coupled plasma (ICP) is not an ideal source, it has been accepted over the past decade as an extremely useful and effective source and is the current source of choice for multi-element analysis in most contemporary laboratories. The ideal spectrometer might be described as possessing the following characteristics: wide total spectral range; simultaneous and comprehensive wavelength coverage over any selected band pass; high accuracy of wavelength measurements; high accuracy and linearity of intensity measurements; large dynamic range; variable resolution up to full resolution of the physical line widths in the source; good sensitivity; fast data acquisition and processing; ease of operation; and reasonable cost.The search for the ideal spectrometer has historically included the development and application of spectrographs, monochromators, polychromators and photodiode array spec- trometers. Each of these types of spectrometer has met some of the ideal criteria and had unique advantages to offer the analyst, but each has also had its own characteristic set of limitations.Now several laboratories in the United States, Canada and Europe are investigating the possibility of applying Fourier transform (FT) spectrometers to spectro- chemical measurements in the ultraviolet and visible spectral and commercial instrumentation suitable for ICP-FTS is being developed.2(&22 The question to be discussed in this paper is: will inductively coupled plasma Fourier transform spectrometry (ICP-FTS) prove to be a practical new analytical technique, and what are its potentials and prob- lems? Theory Analytical atomic spectroscopists usually work with wavelength-dispersive instrumentation. In order to apply Fourier transform spectrometry to the problems of analytical chemistry, a new way of thinking about spectral measure- ments must be adopted; one must learn to think multiplex.1 Multiplexing is a method of information processing. Indepen- dent pieces of information are encoded so that they can be simultaneously transmitted or received over common infor- mation pathways. The multiplexed information must later be decoded so that each independent piece of information is uniquely and correctly retrieved. These steps are illustrated in * Presented at the 1987 Winter Conference on Plasma and Laser Spectrochemistr y , Lyon , France, 12t h- 16t h January, 1987. Fig. 1. In ICP-FTS, the independent pieces of information are the intensities of light at various frequencies emitted by the ICP source. The spectrometer encodes this information in the form of an interferogram. The computer decodes the informa- tion by performing the Fourier transform.A Fourier transform spectrometer is based on the design and operational principles of a Michelson interferometer, as shown in Fig. 2. Light from the source (ICP) enters the aperture of the spectrometer and is collimated. A beam splitter divides the collimated beam into two parts, which travel along the two arms of the interferometer and are reflected by mirrors back to the beam splitter and recombined. The resulting beam is then focused on to a single detector. If one or both of the mirrors move, the optical paths in the two arms are not equal. Constructive and destructive interferences then occur in the recombined beam, according to the optical path difference and the frequencies of light in the source.If the mirrors move as a function of time, then each frequency of light present in the source is modulated, and the time- - Information in source Encode by interferometer Detect by single detector Decode by computer Information in useful form Fig. 1. Multiplex processing of spectral information. Reproduced with permission of The American Chemical Society from Anal. Chem., 1986,58, 1023A586 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, SEPTEMBER 1987, VOL. 2 dependent signal at the detector is a record (an interferogram) of the modulation of all the frequencies present in the source. In terms of data domains, the interferogram is the time- domain half of a Fourier transform pair, and the mathematical process of the Fourier transform produces the frequency- domain half of the pair, known more familiarly as the spectrum.The logical units for reporting spectral information from FT spectrometers are frequency and wavenumber because the Fourier transform inverts the physical units of measurement. If the interferogram is measured in units of time (s), then the transformed spectrum is given in units of frequency (s-1). If the interferogram is measured in units of distance (cm), then the transformed spectrum is given in units of wavenumbers (cm-1). It is useful to recall that 1 v * * (1) a=-=- h c * * * * . . where a is wavenumber in cm-1; h is wavelength in cm; v is frequency in s-1; and c is the speed of light in cm s-1. The principles of Fourier transform spectrometers and the mathematics of the Fourier transform are described in detail in references 1 and 23-30. Reference 1 describes the principles of the Fourier transform with particular emphasis on its applica- tion to analytical atomic spectroscopy, and reference 17 M - Mirror - I Mirror R ef e re n ce detector I- Aperture- Fig.2. Michelson interferometer. Reproduced with permission of The American Chemical Society from Anal. Chem., 1986,58, 1023A 20 15 >. v) C a, .- 4- .- 10 a, (D a, .- c - a 5 0 provides a quantitative comparison of Fourier transform spectrometers and scanning monochromators for atomic emission spectrometry. Experimental Over the past five years, studies at Los Alamos National Laboratory have investigated the ICP as a source for high-resolution FTS. These studies have included an assess- ment of the analytical applicability of ICP-FTS,2.6 a vertical profile of Fe I excitation temperatures in the ICP,s line width and line shape analysis of Fe I in the ICP,4 ICP argon emission in the near-IR,3 and population distributions and oscillator strength values for Mo I in the ICP.lSJ1 These studies were performed using the 1-m Fourier transform spectrometer in the McMath solar telescope at the National Solar Observa- tory, Kitt Peak, AZ, USA. The experimental details and results of these studies have been previously reported in the corresponding references.Los Alamos National Laboratory is currently engaged in building a state of the art Fourier transform spectrometer1.32 for applications in the ultraviolet, visible and near- to far-infrared spectral regions. As part of a national facility for high-resolution spectroscopic research, this instrument will serve many functions in basic and applied chemistry, physics and materials science, including continuing studies focusing on the development of new and advanced techniques in analytical chemistry such as ICP-FTS.Discussion Potentials The results of investigations at this laboratory indicate that the most significant potentials of ICP-FTS as an analytical technique are as follows: (i) the simultaneous and comprehen- sive wavelength coverage of the selected band pass; (ii) the ability to achieve high resolution relatively easily in a compact instrumental system; and (iii) the high degree of accuracy of both wavelength and intensity measurements. As a multiplex instrument, the Fourier transform spec- trometer simultaneously records all the spectral information within the selected band pass.Therefore, the transformed spectrum includes all the emission lines from atomic and molecular species in the source, as well as the background and base-line characteristics of the spectrum. Fig. 3 shows a 10-nm segment of the ICP-FTS spectrum of a multi-element solution. 28600 28800 29000 29200 Wavenumbericm- 29400 I 1 I I I I 349.550 347.1 23 344.729 342.368 340.038 Waveleng t h/nm Fig. 3. ICP-FTS spectrum of multi-element solutionJOURNAL O F ANALYTICAL ATOMIC SPECTROMETRY, SEPTEMBER 1987. VOL. 2 587 A Many lines of several different elements are present, along with argon lines from the plasma gas. The spectral band pass is determined by the combination of selected optical and electronic filters and components and the detector response.Choice of the proper combination of these instrumental components can provide a band pass as small as a few or as large as several hundred nanometres. The total spectral information within that selected band pass is comprehensively recorded by the spectrometer and permanently available stored in the interferogram or the transformed spectrum. Unlike wavelength-dispersive systems, such as the poly- chromator or monochromator which require that only a few analytical wavelengths be preselected, Fourier transform systems allow the selection of the wavelengths for analysis to be made at any time before, during, or after the recording or the spectral information. Thus, the selection can be based on an examination of the content of the actual spectrum, and the maximum flexibility of choice is allowed.Qualitative analysis is therefore easily achieved, as the comprehensive emission spectrum of all species in the source will be recorded within the selected band pass. Spectral interferences are usually easily identified, as the presence of key lines of elements can alert the analyst (or the computer) to possible interferences at other wavelengths. Computerised cross-reference wavelength tables can be used for an initial qualitative analysis as well as for flagging potential spectral interferences. Alternate line selection in the event of spectral interference is usually available, as all the emission lines of all the analytes were recorded in the band pass.The computer system can cross-reference potential spectral interferences based on the initial qualitative survey and suggest the best interference-free lines for the analysis, based on stored reference spectra and the actual sample composition. This might be called true “dynamic analytical line selection.” I (a) 8000 16000 ). C a, C ”, 12000 +- 8000 4- m 3 - 5 4000 u 2 0 L I I I I I 0 20 40 60 80 1 20 1 (cl 0 20 40 60 80 Point number Fig. 4. Multiple line analysis: (a) conventional analysis using only one analytical line of Fe I in the ICP spectrum; ( b ) multiple line analysis of the same sample spectrum using the weighted intensities of 12 spectral windows centred at 12 major emission lines of Fe I to acquire an accumulated intensity; (c) accumulated intensities of the same 12 spectral windows in a sample spectrum where Fe is not present Because all the spectral information within the band pass is available, it is also possible to develop new data processing methods using more than one of the emission lines of an analyte for quantitative analysis.A method of “multiple line analysis” can be used to improve the sensitivity of ICP-FTS. First a reference spectrum is recorded for the analyte of interest, and a reference set of prominent emission lines is selected and stored in a computer file with wavelengths and relative intensities. This same reference file may then be used in all subsequent multiple line analyses for that element. Next the sample spectrum is recorded. Spectral windows of a given number of data points, centred at each reference line wavelength, are extracted from the sample spectrum, and all points in each window are multiplied by 8 weighting factor based on the relative intensity of that reference line in the reference spectrum.The windows are added together, point for point, so that a single window results for the analyte of interest, with an accumulated intensity for the multiple lines. Noise in the base line surrounding each reference line will tend to average, while the weighted intensities of the reference lines will directly add to an accumulated line intensity at an improved peak-signal to base-line noise ratio, resulting in an improvement in sensitivity over single line analysis. This method is illustrated in Fig. 4 for Fe I, where (a) shows the single line intensity and ( b ) shows an accumulated line intensity for 12 prominent lines. If a given analyte is absent in the sample, this method will result in zero accumulated intensity, as shown in Fig.4(c), and may be used as a check for the absence of analytes in complex or noisy spectra. The second major advantage of ICP-FTS is the high resolution which is relatively easy to achieve in a very compact instrumental system, compared with wavelength-dispersive ~pectrometers.15~17 The resolution of the transformed spec- trum is determined by the maximum optical path difference (OPD) in the two arms of the interferometer, which is determined by the extent of the movement of the mirrors. The resolution R (in wavenumbers) is equal to the inverse of twice the maximum optical path difference L (cm): 1 R=&J=- .* (2) 2L“ . . . . The resolution is constant across the spectrum in wavenumber units, but it varies across the spectrum in wavelength units. The resolution in wavelength units 6h for any given wavelength h can be easily calculated using the relationship so 61 o h The variation of resolution in wavelength units for a fixed OPD of 10 cm is shown in Table 1. Long path differences (large interferometers) are required to achieve high resolution of the longer wavelengths in the infrared. Much more compact systems are sufficient for high resolution of the shorter wavelengths in the visible and ultraviolet. Some values of resolution (in both wavenumber units and wavelength units at 300 nm) as a function of OPD are given in Table 2.Very high resolution for the ultraviolet and visible is attained with rather small optical paths compared with those required for wavelength-dispersive instruments. A portion of the multi- element spectrum from Fig. 3 is expanded in Fig. 5 to show the high resolution actually achieved in the FT spectrum. . . . . . . . * (3) _ - _ - Table 1. Resolution as a function of wavelength for a fixed optical path difference of 10 cm A Rlnm 200 nm 0.0002 500 nrn 0.001 25 1 pm 0.005 5 pm 0.125 10 pm 0.5 20 pm 2.0588 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, SEPTEMBER 1987, VOL. 2 Table 2. Resolution as a function of optical path difference (OPD) OPD/cm 0.25 0.50 1 .00 2.50 5.00 10.00 25.00 50.00 100.00 Rlcrn-’ 2.0 1 .0 0.5 0.2 0.1 0.05 0.02 0.01 0.005 R at 300 nm/nm 0.018 0.009 0.0045 0.0018 0.0009 0.00045 0.000 18 0.00009 0.000045 20 15 z cn C a, c..- c .- c10 a, m a, .- c - m 5 28760 28770 28780 28790 Wavenumbericm - ~~ 347.606 347.485 347.364 347.234 Wavelengthinm Fig. 5. High resolution of ICP-ITS spectrum. A 0.5-nm segment of the multi-element spectrum in Fig. 3 is plotted on an expanded wavenumber axis to illustrate the quality of resolution obtained with a 6.4-cm optical path difference. The incomplete resolution of the Mn I1 and Co I lines (peak to peak separation of 0.009 nm) illustrates spectral overlap due to the physical line widths in the source and is not due to insufficient instrumental resolution How much resolution is needed for ICP-FTS? Ideally one would like to resolve fully the physical line widths and shapes in the source without distortion or broadening due to the instrumental function.A conservative guide line to follow”) is that this is achieved when the FTS resolution R is at least three times smaller than the physical line width. Emission line widths depend on several factors including the temperature of the source, the relative atomic mass of the element and the wavelength of the transition. Using the results of previous high-resolution studies of the ICP,4-33 the optical path difference required for full resolution of various ICP lines can be calculated. For most medium relative atomic mass ele- ments in the ultraviolet to the visible, an OPD of 5-7 cm will provide full resolution. For very heavy elements such as the actinides, an OPD up to 11-14 cm is required.High resolution was previously considered very difficult and costly to achieve using wavelength-dispersive instruments. The development of Fourier transform spectrometers for the ultraviolet and visible now provides the analytical chemist with the opportunity to work with high-resolution and high-quality spectra in a practical system. High resolution reduces the number of spectral inter- ferences in a given analysis to those which are due to the actual physical line overlaps. This greatly simplifies many analytical problems, especially those involving very complex spectra such as rare earth and actinide elements. It is possible that some time-consuming sample pre-treatment and separation steps may be eliminated. High resolution is also an extremely valuable tool for basic research and spectrophysical studies of an excitation source such as the ICP.The ability to measure line widths and shapes provides information for the study of excitation and energy transfer mechanisms. High resolution can be used for the study and measurement of hyperfine structure and isotope shifts as well as for isotope abundance measurements, especially in geological and nuclear chemistry applications. The third major advantage of ICP-FTS is the high degree of accuracy which can be attained for both the wavenumber and the relative intensity measurements. Only a single internal laser is required to set the entire wavenumber scale of the spectrum. With reasonable care and good FTS instrumenta- tion, the wavenumber accuracy for high and moderate strength lines can be as good as 0.001 cm-1 (of the order of 10-5 nm at 300 nm) or better.Accuracy of this quality has very positive benefits for both qualitative and quantitative analysis and has especially significant implications for the development of entirely new approaches to data treatment such as multiple line analysis or correlation techniques, in either the spectral or the Fourier domain,34 which depend on accurate and very reproducible measurements. Intensity measurements are also accurate and linear. Relative intensities across the entire recorded spectrum can be easily corrected for instrumental and optical response, and the corrected relative intensities can be very valuable for spectrophysical studies such as tempera- ture determinations5 and population distributions.18 An obvious application of ICP-FTS which combines all three of these major advantages is the production of a new set of standard wavelength tables for ICP emission of the elements.35 The ability to measure comprehensively the entire spectrum at full resolution of the physical line widths and with high accuracy of wavenumbers and corrected relative intensi- ties is unique to ICP-FTS. Such a set of reference wavelength tables could provide a valuable resource to the users of all ICP instruments. If these tables were computer based, then spectral identification and cross-referencing for spectral inter- ferences could be programmed for any ICP system. Further- more, if the tables were measured at full resolution, then users of low-resolution ICP systems could program a calculation to generate tables based on their own instrumental function and predict the spectral interferences which would occur as a result of instrumental broadening on their systems. Some prelimi- nary work on the production of such tables has been completed. 3 .4 , ~ Problems Any new technique or method offers both potential (advan- tages) and problems (limitations). Studies at this laboratory have indicated that the three major problems in the develop- ment of ICP-FTS as a practical analytical technique are: (i) a multiplex disadvantage which may occur and degrade detec- tion limits in some analytical situations; (ii) the mathematical phenomenon of aliasing which may result from undersampling of the interferogram; and (iii) the more stringent optical and mechanical tolerances which are required for interferometry at the short wavelengths in the visible and ultraviolet regions.Whereas the realisation of the multiplex advantage was a major factor in the utility of Fourier transform spectrometry in the infrared (FTIR), the application of FTS in the ultraviolet and visible may actually involve a multiplex disadvantage, which must be carefully considered in the development of ICP-FTS. The multiplex advantage is a gain in signal to noise ratio of the FT spectrum compared with that achieved for comparable measurement time by a sequential scanning wavelength-dispersive spectrometer. This advantage is only realised in detector-noise limited situations (such as com- monly occur in the infrared): that is, the noise at the detector is independent of the signal intensity. Ultraviolet and visible measurements are usually photon- or source-noise limited: that is, the noise at the detector is dependent on the signal intensity. As FTS is a multiplex technique, all the source emission intensity is recorded simultaneously at a singleJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY. SEPTEMBER 1987, VOL.2 589 detector. In the Fourier transform process, the noise at any given frequency in the source is distributed throughout the resulting spectrum. One important implication is that source noise from strong emission lines may increase the over-all base-line noise in the transformed spectrum and thereby degrade the detection limits for other analytical species in the sample.Detection limits in ICP-FTS are definitely, and sometimes dramatically, dependent on the total sample composition. In order to evaluate the possible extent of this multiplex disadvantage in analytical applications, studies were per- formed at this laboratory to approximate “worse case” situations. The results of these studies2 demonstrated that matrix elements of the alkaline earths such as calcium, which exhibit intense emission in a few spectral lines, can produce a serious degradation of detection limits for other analytes in the sample, even when present as a matrix element at relatively low concentration levels. By comparison matrix elements of the transition-type such as iron and nickel, which exhibit their emission in several less intense lines, produce only moderate degradation of detection limits for other analytes in the sample even when present as a matrix element at moderately high concentration levels. The conclusion and warning is that the multiplex disadvantage can be a serious problem in the use of ICP-FTS for analytical applications, and the further charac- terisation and understanding of the causes and effects of noise in the FT spectrum must be given careful consideration.2.” The “worse case” problems may luckily be the easiest to solve, as simple optical filtering can eliminate the one or two strong lines of the alkali or alkaline earth elements causing the degraded detection limits for the other elements in the sample, while preserving most of the comprehensive wavelength coverage advantage of FTS.In less simple instances, some degree of limiting the comprehensive wavelength coverage to a narrower band pass may be a necessary compromise solution in order to achieve the desired detection limits for some analyses. Instead of using one comprehensive FT spectrum for the total analysis, it may be necessary to use one comprehen- sive and one or more band-limited FT spectra. Some results based on this approach have been reported.12.14 Another possible means for overcoming the multiplex disadvantage is the use of multiple line analysis, described above. Unlike ICP-MS, ICP-FTS does not appear at this time to promise significant breakthroughs in sensitivity, Even under optimised conditions, detection limits for ICP-FTS will probably be comparable to, but not significantly better than, those attained with wavelength-dispersive systems.The second major problem in the development of Fourier transform spectrometry for the ultraviolet and visible spectral regions is the phenomenon of aliasing,’J3-30 which must be well understood and carefully considered in the applications of ICP-FTS. In order to perform the Fourier transform using a digital computer, the continuous interferogram must be sampled and digitised. A simple method uses a reference laser to generate a monochromatic beam of radiation travelling through the interferometer to a reference detector (Fig. 1). Each full cycle of laser light modulation at the reference detector is used to trigger the sampling of the interferogram. If a He - Ne laser is used as the reference, the sampling interval S is equal to 0.6328 pm.It is a mathematical property of the Fourier transform’.*?-’() that the transformed spectrum has both a positive and a negative image, symmetric about zero wavenumbers on the spectral axis. It is also a mathematical property of the Fourier transform of a digitised interferogram that both the positive and negative images are replicated along the wavenumber axis at intervals of l/S wavenumbers (IS800 cm-* in the simple situation of the He - Ne reference laser as described above). It is possible for overlapping of these positive and negative images and their replicas to occur; this situation is “aliasing.” High-frequency information will appear to be “folded-over” into the low-frequency regions of the spectrum.Fig. 6 illustrates an aliased spectrum resulting from the overlap of images and replicas along the wavenumber axis. The problem for the analytical chemist is that it is not possible to determine whether a line originated from a positive or negative image or an overlapped replica without prior knowledge of the source spectrum (which is usually not available in the analysis of unknown samples). The assignment of wavenumbers to the resulting lines in an aliased spectrum is neither sequential or unique. Furthermore, the overlapping effect can also increase spectral interferences and base-line noise in the analytical spectrum. The range of wavenumbers that can be uniquely recovered from the interferogram without aliasing is called the free spectral range (FSR).This range extends theoretically from zero wavenumber (infinite wavelength) to some upper wave- number (short wavelength) limit equal to 1/2S, where S is the sampling interval. When the sampling interval S is made smaller, the free spectral range is made greater to include higher wavenumber (shorter wavelength) spectral informa- tion. If all the spectral information in the source is contained within the free spectral range, there will be no aliasing (overlapping) in the transformed spectrum. This situation can be ensured if the source is band-limited to the wavenumbers contained in the free spectral range. However this is not a practical solution in ICP-FTS using a He - Ne reference laser as all the visible and ultraviolet emission lines would be excluded.The alternative approach is to increase the free spectral range (to shorter wavelengths) to contain all the spectral information in the source by making the sampling interval sufficiently small. There are several methods22332 of decreasing the sampling interval, including sub-dividing the He - Ne laser frequency. Table 3 shows the free spectral range as a function of sampling intervals based on the He - Ne laser. The interferogram must be sampled at intervals of the order of one-eighth the He-Ne frequency in order to recover un- aliased spectra in the ultraviolet region. The phenomenon of aliasing and the practical desirability of avoiding it in FTS applications for analytical spectroscopy can be considered major pragmatic obstacles in the development of ICP-FTS as a new analytical technique.The third major problem in the development of ICP-FTS is the general subject of instrument design and engineering. d I i b a f A1 3 - L 2 - 1 + FSR + - - 1 0 - -2 - -3 2s 2s 2s 2s 2s 2s Wavenumbericrn ~ Fig. 6. Aliasing of an FT spectrum. A , True spectral information in the source; B, positive and negative images of the true spectral information, symmetric about zero on the wavenumber axis; C, first replicas of the positive and negative images of the true spectral information centred at + l/S and - 1/S wavenumbers; D, result of the overlapping of positive and negative images and their replicas and shows the appearance of the transformed spectrum. All the true spectral information is present within each interval of the free spectral range 1/2S, but in overlapped fashion which does not allow sequential and unique wavenumber assignment, Reproduced with permission of The American Chemical Society from Anal.Chern., 1986, 58, 1023A590 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, SEPTEMBER 1987, VOL. 2 Table 3. Free spectral range as a function of sampling interval Short Multiple Sampling wavelength of He - Ne interval/pm FSWcm-1 limithm 1 0.6328 7900 1265.6 f 0.3164 15800 632.8 4 0.1582 3 1600 316.4 R 0.0791 63200 158.2 1 1 Because FTS depends on interference effects of the recom- bined light beams, the instrumental design criteria become much more severe at the short ultraviolet wavelengths than for the infrared, and state of the art technology is required in optics, mechanics, electronics and computer science.The general requirements have been discussed elsewhere1~*5-17.22.32 and are highly demanding but definitely within the scope of contemporary technology. Conclusions It can be seen that ICP-FTS offers both powerful potentials and unique problems to the analytical chemist. No new technique ever totally replaces those already in use, but rather provides a complementary method with some advantages and some inherent limitations. The multiplex approach is rela- tively unfamiliar in analytical atomic spectroscopy. Before its implications are understood and realised, its unique charac- teristics must be carefully studied and considered. The field of ICP-FTS is currently being recognised as a possible new and practical analytical technique, The possibili- ties for research in this field are rich.Any developments which reduce the noise in the source will reduce the detrimental effects of the multiplex disadvantage and improve analytical sensitivity. This may include new designs for torches and nebulisers in general or specifically for FTS applications. 19 New methods of sample introduction will be needed, either to reduce source noise resulting from the nebulisation process, or to find ways to introduce solid samples for ICP analyses with the steady-state signal required for FTS. Versatile and flexible methods of band-limiting the spectral information from the source will be needed to address the issues of multiplex disadvantage and aliasing. Ideally, a band-pass selector would be desirable which would allow the analyst to select a band pass of any extent over any range.Applications of pre- dispersion, post-dispersion and multiple detectors will find uses in ICP-FTS. The realm of data treatment and data analysis is awaiting clever and innovative approaches using, for example, multivariate and correlation techniques in either the Fourier or the spectral domains. Now that the feasibility of Fourier transform spectrometry in the visible and ultraviolet regions has been demonstrated, the frontier is open for the development of entirely new excitation sources36 whose properties may be even better suited to the multiplex approach. This work was performed at Los Alamos National Labora- tory, Los Alamos, NM, USA, and at the National Solar Observatory, Kitt Peak, AZ, USA.I would like to acknow- ledge and express sincere appreciation to the following persons, without whose contributions this work could not have been accomplished: Dr. T. M. Niemczyk, University of New Mexico; Dr. B. A. Palmer, Dr. R. Engleman J r . , Dr. P. Cunningham, Los Alamos National Laboratory; and Dr. J. W. Brault, National Solar Observatory. 1. 2. 3. 4. 5 . 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. 27. 28. 29. 30. 31. 32. 33. 34. 35. 36. References Faires, L. M., Anal. Chem., 1986, 58, 1023A. Faires, L. M., Spectrochim. Acta, Part B, 1985, 40, 1473. Faires, L. M., Palmer, B. A., Engleman, R., Jr., and Niemczyk, T. M., Spectrochim. Acta, Part B, 1985, 40, 545. Faires, L. M., Palmer, B. A., and Brault, J .W., Spectrochim. Acta, Part B, 1985,40, 135. Faires, L. M., Palmer, B. A,, Engleman, R., Jr., and Niemczyk, T. M., Spectrochim. Acta, Part B, 1984, 39. 819. Faires, L. M., Palmer, B. A., Engleman, R., Jr., and Niemiczyk, T. M., SPIE J. , 1983, 380, 396. Horlick, G., Hall, R. H., and Yuen, W. K., in Ferraro, J. R., Basile, L. J . , and Mantz, A., Editors, “Fourier Transform Infrared Spectroscopy,” Volume 3, Academic Press, New York, 1982, pp. 37-81. Irfan, A. Y., Thorne, A. P., Bolander, R. A., and Gebbie, H. A., J. Phys., E: Sci. Instrum., 1979, 12, 472. Marra, S., and Horlick, G., Appl. Spectrosc., 1986, 40, 804. Ng, R. C. L., and Horlick, G., A p p f . Spectrosc. 1985,39,834. Ng, R. C. L., and Horlick, G.. Appf. Spectrosc., 1985,39,841. Stubley, E. A . , and Horlick, G., Appl.Spectrosc., 1985, 39, 800. Stubley, E. A., and Horlick, G., Appf. Spectrosc. 1985, 39, 805. Stubley, E. A., and Horlick, G., A p p f . Spectrosc., 1985, 39, 811. Thorne, A. P., Harris, C. J., Wynne-Jones, I., Learner, R. C. M., and Cox, G., J. Phys. E., 1987, 20, 54. Thorne, A. P., Anal. Proc., 1985, 22, 63. Thorne, A. P., J . Anal. A t . Spectrom., 1987, 2, 227. Brault, J. W., and Faires, L. M., presented at the 1986 Winter Conference on Plasma Spectrochemistry, Kona, Hawaii, paper No. 147. Davies, J., and Snook, R. D., presented at the 1986 Winter Conference on Plasma Spectrochemistry, Kona, Hawaii, paper No. 145. Goulter, J . E., Algeo, J. D., Tikkanen, M. W., and Routh, M. W., presented at the 1986 Winter Conference on Plasma Spectrochemistry, Kona, Hawaii, paper No. 176. Thorne, A. P., Harris, C. J., and Wheaton, J. E. G., presented at the 1986 Winter Conference on Plasma Spectrochemistry , Kona, Hawaii, paper No. 90. Snook, R. D., and Grillo, A., A m . Lab., 1986, Nov. Mertz, L., “Transformations in Optics,” Wiley, New York, 1965. Bracewell, R. N., “The Fourier Transform and Its Applica- tions,” McGraw-Hill, New York, 1965. Horlick, G., Appl. Spectrosc., 1968, 22, 617. Bell, R. J., “Introductory Fourier Transform Spectrometry,” Academic Press, New York, 1972. Brigham, E. O., “The Fast Fourier Transform,” Prentice Hall, Englewood Cliffs, NJ, 1974. Griffiths, P. R., “Transform Techniques in Chemistry,” Plenum Press, New York, 1978. Lam, R. B., Wieboldt, R. C., and Isenhour, T. L., Anal. Chem., 1981, 53, 889A. Brault, J. W. , “High Resolution Fourier Transform Spectro- scopy; Proceedings of the 15th Annual Advanced Course of the Swiss Society of Astrophysics and Astronomy,” Geneva Observatory, Sauverny, Switzerland, 1985. Whaling, W., and Brault, J . W., Technical Report 1987A, NSF Grant AST-83-00948, to be published. Palmer, B. A., and Engleman, R., Jr., “Proceedings of the 1985 International Conference on Fourier and Computerised IR Spectroscopy,” SPIE J., 1985, 553, 413. Faires, L. M., PhD Thesis, Los Alamos National Laboratory Report, 1983. Parsons, M. L., Faires, L. M., Palmer, B. A., and Lyon, R. S., presented at FACSS, 1985, Philadelphia, PA, paper No. 69. Faires, L. M., ICP I f . Newsl., 1984, 10 449, Bieniewski, T. M., and Hull, D. E.. presented at the 1986 Winter Conference on Plasma Spectrochemistry, Kona, Hawaii, paper No. 65. Paper J7l32 Received March 20th, 1987

ISSN:0267-9477    

DOI:10.1039/JA9870200585

出版商:RSC    

年代:1987

数据来源: RSC

摘要:

JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, SEPTEMBER 1987, VOL. 2 I 591 7 r Preliminary Results with a High-resolution Inductively Coupled Plasma Fourier Transform Spectrometer* Dominic E. M. Spillane,t Richard D. Snook,S Anne P. Thorne and J. E. G. Wheaton§ Department of Physics, Imperial College of Science and Technology, London S W7 ZBZ, UK This paper presents some preliminary results obtained by monitoring the optical emission from an argon radiofrequency inductively coupled plasma using a Fourier transform spectrometer. The system is shown to be capable of obtaining high (picometre) resolution data over a large wavelength range (>I00 nm) in a relatively short period of time (ca. 6 min). The possibility of monitoring several wavelengths simultaneously in order to provide an averaged emission value for a particular element is analysed and results are presented to show that this can offer some improvement in signal to noise ratio.Keywords: Fourier transform spectrometry; high-resolution instrumentation; inductively coupled plasma; atomic emission spectrometry The use of Fourier transform spectrometry (FTS) can, in principle, provide significant advantages for the analyst over conventional grating spectrometers. For analytical atomic spectrometry in the ultraviolet region this may be interpreted as the ability to provide rapidly a high-resolution scan over a wide spectral range (e.g., using the instrumentation described below, a 6-min scan can yield spectral data over the region 32 OOS64 000 cm-1 with a resolution of 0.2 cm-1). This type of performance should allow rapid multi-element analyses to be performed with the inductively coupled plasma (ICP) using a single-detector system. Previous work in the field has been comprehensively reviewed by Faires.1 This work, however, has either employed high-resolution systems which are large and essentially immobile, being designed and built for astrophysical research'-4 or low-resolution systems often of limited wavelength range.5-9 Recent work at Imperial Col- legel0 has led to the development of a compact, mobile FTS capable of high- (sub-picometre) resolution spectroscopy. The advantages of applying such an instrument to optical emission have been discussed by Thornell and this paper presents some preliminary results obtained using an ICP.Instrumentation The instrumental configuration is shown schematically in Fig. 1. The ICP used was a commercially available system (Model BTP 1500, RF Applications Ltd., Eastbourne, East Sussex, UK), the torch box being mounted so that the plasma axis was co-axial with the FTS optics to facilitate axial viewing. The central channel of the plasma was imaged via a lens and a plane mirror on to the entrance aperture of the FTS. The signal from the photomultiplier tube (PMT) was taken via a filter and analogue to digital converter (ADC) to a microcom- puter (PDP 11/34, DEC Ltd., Reading, Berkshire, UK). Unless otherwise stated the operating conditions were as given in Table 1. Sample introduction was achieved by headspace sampling of a solid (for vapour sample introduc- tion) or by pneumatic nebulisation using a Meinhard nebuliser for solution samples._____ ~~ * Presented at the 1987 Winter Conference on Plasma and Laser Spectrochemistry, Lyon, France, 12th-16th January, 1987. t To whom correspondence should be addressed. Present address: Department of Applied Chemistry and Life Sciences, Polytechnic of North London, Holloway Road, London N7 8DB, UK. 4 Present address: Department of Instrumentation and Analytical Science, University of Manchester Institute of Science and Technology (UMIST), Manchester M60 lQD, UK. 0 Present address: Chelsea Instruments Ltd., Unit 9, Avonmore Business Centre, Avonmore Road, London W14 8TS, UK. Fig. 1. optical signal; and solid lines, electrical signals Schematic of the instrumental arrangement: broken line, Table 1.Operating conditions for the ICP-FT spectrometer Sample introduction: Gas flow-rate . . . . . . . . . . 1 1 min-1 Gas pressure (liquid samples only) . . . . 38 p.s.i. ICP forward power . . . . . . . . . . 800 W Coolant gas flow . . . . . . . . . . 10 1 min-1 Auxiliary gas flow . . . . . . . . . . 0 I min- I FTS entrance aperture (diameter) . . . . 2.4 mm PMT power supply . . . . . . . . . . 500 V Results and Discussion The aim of the investigation was to establish the potential of the system for measurement of atomic emission signals from an inductively coupled plasma. For the preliminary measure- ments reported in this paper, the system was not optimised with respect to the operating conditions of either the FTS or the ICP. The objective was to assess the linearity of response and relative noise and detection limits for Fe.Initial measure- ments were made using vapour samples in order to distinguish noise intrinsic to the ICP from that arising from nebulisation of the sample. This was achieved by headspace sampling from a vessel containing a small amount of molybdenum hexacarbonyl crystals. Fig. 2 shows a typical portion of the molybdenum spectrum, the resolution was 0.2 cm-1 and the spectral range was in principle 180-310 nm, but the use of a cut-on filter limited the detected range to 240-310 nm. (This is the spectral range of all measurements in the following discussion.) The data were collected in 6 min. The spectrum is at sufficiently high resolution to allow lines from different elements to be distinguished by their Doppler widths.This can be seen from Fig. 3 which shows a carbon emission line (C I) at592 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, SEPTEMBER 1987, VOL. 2 0.70 0.60 m c. .- 5 0.50 2 F 5 0.40 s 5 0.30 m 4- .- c. C m c 0.20 0 v) .- - .- 0.10 Wavenurnbericrn- Fig. 2. Section of the spectrum acquired by vapour [Mo(CO),] sample introduction Wavenumber/crn --I Fig. 3. C I line from the spectrum recorded for Mo(CO), 247.86 nm. Comparison with the Mo I1 emission lines in Fig. 1 shows clearly the dependence of line width on mass. An examination of a logarithmic plot of the spectrum indicated that the background level is at 5 X 10-4 V compared with the peak signal of 1 V. The noise level on the signal is at approximately the same level (peak to peak) as the background level.Thus the maximum signal to noise ratio (SNR) for the spectrum is 2 x 103. This performance is slightly worse than that obtained by Thorne et al. 10 using an iron hollow-cathode lamp (HCL) as a source for the FTS. Their measurements were taken on a photon-noise limited system. The emitted intensity from the ICP was considerably higher than that for the HCL whereas the SNR was fractionally lower. This would appear to indicate that the molybdenum - ICP system is source-noise limited and may thus be taken as a guide to performance for the results presented below. It should be noted, however, that an absolute performance measure is difficult to obtain as the SNR depends critically on the spectral density of the source under examination.In this instance, Mo and Fe have broadly similar spectral densities over the region viewed and a comparison is possible. Having established the above as a performance criterion for the system as currently configured without a nebuliser, the response of the system for liquid (solution) samples was assessed. Solutions were made up containing 1000,100,10 and 1 p.p.m. of iron together with a blank and nebulised into the plasma. Initial measurements were made at a lower resolution (1 cm-1) than above, which enabled each of the five sample spectra to be acquired in 1 min. Fig. 4 shows a typical section of the spectrum for the iron solutions showing a number of lines, the strongest of which is the Fe I1 line at 259.940 nm (the section covers the region 259.50&263 .OOO nm).Examination of these spectra indicated that the base-line noise was independent of analyte concentration up to 100 p.p.m. and limited the maximum SNR to ca. 102. The signals obtained were generally poor and non-linear with iron concentration. The iron spectrum could not be detected at the 1 p.p.m. level. The measurements were then repeated at higher resolution for iron concentrations of 100 p.p.m. and below. These show a considerable enhancement of the iron signal relative to the background and result in an improvement of the SNR, the maximum being ca. 4 x 102. This had the effect of considerably linearising the iron signals but quantification to the 1 p.p.m. level still proved difficult with measurement at a single wavelength, An attempt was therefore made to improve the calibration by averaging the signal from several iron lines.This could be achieved by noting the wavenumbers of the linesJOURNAL O F ANALYTICAL ATOMIC SPECTROMETRY, SEPTEMBER 1987, VOL. 2 1.00 593 I- Wavenum ber/cm -’ Fig. 4. Section of the spectrum recorded while nebulising 1000 p.p.m. of Fe in aqueous solution giving a suitably large signal in the more concentrated solutions and using the excellent wavenumber reproducibility (0.001 cm-1) to obtain measurements at the same set of wavenumbers for the lower concentrations, even though the iron emission lines were essentially undetectable. These signals could then be averaged together to produce an improved SNR for the measurement as a whole. This method is somewhat less straightforward than it might immediately appear, as the following analysis shows. It is assumed that the noise is distributed evenly throughout the spectrum and can be measured by a standard deviation a on each point.This is to be expected in FT spectroscopy1JJ and will certainly be true for low signal intensities. If an iron line of intensity 1, exists at a particular wavelength h, then SNR = Ida If there are n lines at hl to hn each of intensity lo then by averaging SNR = nJl,/a In general, intensities vary and the intensity of the ith line may be written as ai x Zo (0 < ai < l), taking the line at hl to be the most intense. In this instance Therefore to maximise SNR, the function n i = 1 2 a i / d must be maximised and if n i = 1 Z ai < ni the SNR will actually degrade.The averaging process also improves the reliability of the base line. If the region around a peak can be assumed to be level, then all the neighbouring points may be averaged to reduce the noise on the base-line level. For the purposes of data manipulation, the spectra are stored in blocks of 512 points and so an expected blank ( E ) was computed by taking a 512-point average of the actual blank around each measurement point for iron. The uncertainty in the E values made a negligibly small contribution to the over-all SNR, which was held to be dictated by the averaged iron value. Table 2. Wavelengths of points, a, and Zailnl values Point No. (E i) 1 2 3 4 5 6 7 8 9 10 Wavelength/ nm 259.940t 259.940t 261.187t 276.750 259.837 26 1.187t 260.709 258.588 275.574 26 1.762 a1 1 0.573 0.364 0.341 0.328 0.299 0.291 0.285 0.283 0.180 1 n 2 a,/d* = I 1 1.11 1.12 1.14 1.17 1.19 1.21 1.23 1.25 1.25 * Value given is cumulative and is computed for the particular point and those above it in the table. t The emission lines are more than one measurement point wide.For the strongest lines (at 259.940 and 261.187 nm) two points were taken within the same line for each measurement. Table 3. Calibration data for iron solutions, single point and averaged measurements Averaged Emitted intensity emitted intensity Iron concentration, at 259.940 nm, for ten points, p.p.m. arbitrary units arbitrary units 100 7500 7569 10 868 808 1 69 72 Blank 0 0 The following example illustrates this analysis practically. Ten points were selected from the iron spectrum obtained for 100 p.p.m.of iron at the higher resolution. Table 2 indicates the wavelengths of the points chosen, the detected intensities and the computed values for ai and for Zai/nJ. It can be seen that the expected enhancement of the SNR levels off as i increases, and the SNR would probably be degraded by inclusion of any weaker lines. Table 3 shows the calibration figures obtained using the strongest single line and the averaged values, respectively, taking the expected blank E as base line in both instances. The linear correlation coefficients for the single-point and averaged values are 0.99987 and 0.99998, respectively. This would seem to confirm the gain from the averaging procedure. Whether this is fortuitous in view of the 5 to 4 over-all improvement in SNR requires further investigation.594 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, SEPTEMBER 1987, VOL.2 Conclusion From this preliminary study it is possible to conclude that the instrument is capable of performing rapid high-resolution measurements on emission signals from the ICP. True line width measurements can be made over a wide spectral range for many elements in a single 6-min scan. Quantitative measurements are possible but are at present limited in range by the non-optimised instrumental configuration. Quantita- tive measurements are best carried out at high resolution (ideally a resolution at which the band of interest is just fully resolved) and can be improved by judicious use of simul- taneous measurement of the emission signal at several wavelengths. It should be stressed that this work is not an attempt to set definitive detection limits for the system.A number of instrumental parameters require optimisation and work is currently in progress to carry these changes out. A more complete characterisation of the system will be reported at a later date. The authors acknowledge the assistance of the Department of Quality Assurance of the Ministry of Defence for the provision of equipment and support for one of us (D. E. M. S.). 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. References Faires, L. M., Anal. Chem., 1986, 58, 1023A. Faires, L. M., Palmer, B. A,, Engleman, R., Jr., and Niemczyk, T. M., SPIE J., 1983, 380, 396, Faires, L. M., Palmer, B. A., Engleman, R., Jr., and Niemczyk, T. M., Spectrochim. Acta, Part B, 1984, 39, 819. Faires, L. M . , Palmer, B. A., Engleman, R., Jr., and Niemczyk, T. M., Spectrochim. Acta, Part B, 1985, 40, 545. Stubley, E. A., and Horlick, G., Appl. Spectrosc., 1985, 39, 800. Stubley, E. A., and Horlick, G., Appl. Spectrosc., 1985, 39, 805. Stubley, E. A., and Horlick, G., Appl. Spectrosc., 1985, 39, 811. Ng, R. C. L., and Horlick, G., Appl. Spectrosc., 1985,39,834. Ng, R. C. L., and Horlick, G., Appl. Spectrosc., 1985,39,841. Thorne, A. P., Harris, C. J., Wynne-Jones, I., Learner, R. C. M., and Cox, G., J. Phys. E . , 1987, 20, 54. Thorne, A. P., J . Anal. A t . Spectrom., 1987, 2, 227. Paper J7l.59 Received May 8th, 1987 Accepted June loth, 1987

ISSN:0267-9477    

DOI:10.1039/JA9870200591

出版商:RSC    

年代:1987

数据来源: RSC

摘要:

JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, SEPTEMBER 1987. VOL. 2 595 Role of Aerosol Water Vapour Loading in Inductively Coupled Plasma Mass Spectrometry* Robert C. Hutton and Andrew N. Eaton VG Elemental, (Division of VG Isotopes Ltd.), Ion Path, Road Three, Winsford, Cheshire CW7 3BX, UK The influence of aerosol water loading has been shown to be of great significance in inductively coupled plasma mass spectrometry (ICP-MS). The resultant spectra from both singly charged M+, oxide MO+ and doubly charged M2+ species have a strong dependence on the water loading in the plasma. Similarly the levels of polyatomic species, whether derived directly or indirectly from water components (O+, H+) are also significantly improved when loadings are lowered. At low water loadings, the mean ion energies are reduced to low but reasonably constant levels (ca. 4 eV) for a wide range of elements facilitating more predictable plasmas and minimising mass dependent ion focusing.Keywords: Water loading; inductively coupled plasma source mass spectrometry; cooled spray chamber To a first approximation the argon inductively coupled plasma (ICP) is an ideal source for mass spectrometry (MS). Mainly singly charged monoatomic ions (M+) are produced for most elements in the Periodic Table. Sensitivity is very high, typically > l o 6 counts per second can be expected for solutions of 1 pgml-1 for most elements. This combined with back- ground levels of about 10 counts per second yields detection limits at the pg ml-1 level. Although the isotopic spectra are very simple, some spectral interferences are observed.Iso- baric overlaps between isotopes of the same nominal mass can be found, but, in general, alternative isotopes are available for analytical use. Interferences from other species may also be observed at low levels.'-2 These are due to oxide and hydroxide species (MO+ and MOH+) and also, less signifi- cantly, doubly charged species (M2+) are observed for elements whose second ionisation potentials are below 17 eV. Comprehensive lists of possible polyatomic species have been documented by Gray3 and also by Tan and H ~ r l i c k . ~ One of the advantages of the ICP is its ability to accept samples from a variety of sources provided that the sample can be converted into an aerosol for transport into the plasma.Solution nebulisation is the most common technique employed, but alternative techniques where sample vapour is generated by laser ablation,5 electrothermal vaporisation,6 chemical vapour generation,' arc discharges8 or even as solid suspensions9 have been employed in ICP-MS. The ICP, whether used for atomic or mass spectrometry, has a limited tolerance of water vapour.1" This has been documented with reference to atomic spectrometry, where it has been postulated that undissociated water vapour at (3000-4000 K) represents a considerable thermal buffer in the plasma injector.11.12 In ICP-MS the situation is more complex. Ions are principally extracted from an atmospheric plasma (where ionisation temperature Ti may be ca. 7500 K) into a low pressure (ca. 2 mbar) interfacial region situated before a quadrupole mass filter maintained at 10-6 mbar.It has been postulated that many of the less desirable spectral features are derived from events occurring within this low pressure interface region.1.13.14 Hence, the nature of the plasma and its relationship with this interface would be expected to have a substantial influence on the resulting mass spectrum.15 This paper describes how many spectral features in ICP-MS can also be influenced by the solvent water loading in the plasma and that the mechanisms may be traced to reactions within the interface. * Presented at the 1987 Winter Conference on Plasma and Laser Spectrochemistry , Lyon, France, 12th-16th January, 1987. Experimental The work described herein was performed on a standard commercially available ICP-MS instrument (VG Plasma- Quad, VG Isotopes, Winsford, UK).The ICP consisted of a two-turn asymetrically grounded load coil and Fassel-type torch with power derived from a 2.0 kW, 27 MHz RF generator (Henry Radio, USA). A glass, water-cooled, Scott-type spray chamber was used and the wall temperature of this was controlled to k0.1 "C from a Nesslab RTE 9DD recirculator. This allowed the spray-chamber temperature to be varied from -20 to 100 "C if required. The nebuliser used was a Babington type16 which typically operated at 0.750 1 min-1 with a back pressure of 30 p.s.i. The interface consisted of a nickel sampler (Nicone) with a 1 .O-mm orifice, and a nickel skimmer (Nicone, mini-type) with a 0.75-mm orifice. The sampler - skimmer spacing was 4.5 mm. Typical multi-element conditions used were: 1.3-kW power, 12 1 min-1 coolant and 0.5 1 min-1 auxiliary flows.The plasma was sampled 10 mm from the load coil. The ion lenses were tuned on 100 ng ml-1 115In, 59C0 and 208Pb to give a uniform response across the mass range. Under these conditions instrument response was typically 2 x 106 counts s-1 per pg ml-l. Solutions were introduced via an eight roller peristaltic pump (Ismatek CH) at a rate of 0.9 ml min-1. Results and Discussion The effect of aerosol water loading on the spectral characteris- tics observed in ICP-MS were measured by incrementally adjusting the spray chamber temperature from 0 to 30 "C, the latter temperature being typical of the instrument cabinet temperature in the vicinity of the ICP.This temperature variation raises the vapour pressure of the water in the spray chamber by almost six times (Fig. 1). Using the continuous measurement technique described by Maessen et al. 12 the water loading in the plasma changed from 0.075 mg s-1 at 0 "C to 0.392 mg s-1 at 30 "C under the conditions used, an increase of over five times. The spectral features observed to be dependent on aerosol loading were: (i) M+ signal intensity; (ii) MO+ and M*+ levels; (iii) polyatomic ion levels; (iv) residual sampler material contribution; and (v) ion kinetic energy. M+ Signal Intensity On cooling the spray chamber from 30 to 0 "C, the ion sensitivity increases dramatically. From Fig. 2, it can be seen that for high mass elements the increase is of the order of 2-3 whereas for low mass elements the increase can be as much as 6-7 times.596 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, SEPTEMBER 1987, VOL.2 35 I 1 / B I / /- I 0 30 25 20 15 10 5 0 Te m pera t u re/"C Fig. 1. Variation in water vapour pressure with temperatures at atmospheric pressure17 / A / J 7 I I I I I I I I I I 1 I' / 35 25 20 15 10 5 0 Spray chamber temperatureiT Fig. 2. Change in M+ (relative to 30 "C) response with spray chamber temperature: A, YBe, 24Mg and *'Al; B, W o , llsIn, '"Ba and 2"gPb; C, 232Th and 2 3 W . 100 ng ml- 1 of each solution; results are averages of five runs Table 1. Dependence of experimentally measured oxide level on M-0 bond energy" under typical operating conditions Oxide level, YO Element Bond energy/kJ mol - I 1.0-2.0 Th, U, Ce, La 750-850 0.5-1.0 Ta, Hf, W, 670-7.50 0.1-0.5 Ti, V, Y, Mo, 460-670 Nb, Zr, Pr, Si Ru, Ba, Al, rare earths 0.01-0.1 Ca, Sr, platinoids <460 <0.01 Pb, Rb, Cs, Co, <460 Fe, Cr, Ni, Cu, Zn 1 1 I I 1.1 1.2 1.3 1.4 1.5 PowerikW Fig.3. (B) spray chamber temperature Power optimisation responses for 14We at 0 "C (A) and 30 "C Once partially desolvated, the plasma is probably con- siderably hotter, extra energy being available which pre- viously was required to dissociate residual water vapour. At below 8 "C there seems little increase indicating that little residual water vapour is then removed. Temperature optimi- sation curves for "We+ shown in Fig. 3 indicate that at lower water loadings, 100 W less power is required to obtain the maximum signal.However, at 30 "C the increased power still does not return the 140Ce+ signal to the same level as that obtained in the same system with the chamber operated at 0 "C. A sharp maximum is obtained for 0 "C at 1250 W whereas a less well defined maximum at 1350-1400 W was observed for 30 "C. MO+ Levels Under typical multi-element operating conditions, the levels of MO+ are low.1 The oxide levels predictably depend on the M-0 bond strength as can be seen in Table 1 with elements such as Ce, Th and U exhibiting the highest levels. These oxide levels can be reduced by reducing the carrier gas fl0~18J9; however, this results in a reduction in sensitivity and for the experiments detailed in this work, typical multi- element operating conditions were used, optimised to give an even ion response across the mass range.Oxide species have been postulated to be formed via recombination reactions in the extraction process. It is unlikely1 that such oxides would be present in the plasma or in the boundary layer to any significant extent. This, however, may not be so when the oxygen population is increased at high water loadings. From Fig. 4 it can be seen that the oxide levels fall by a factor of approximately two for the most refractory species when chamber temperature is reduced from 30 to 0 "C. Typically MO+/M+ levels for Th and U fall from about 4.5 to 2.0% whilst that for Ce falls by a similar level from 2.3 to 1.1%. In a dry plasma, typical of that observed in laser ablation work, the oxide levels are reduced by a further order of magnitude, as observed for WO+ and UO+ in Table 2.20 In such a system the only oxygen present will be that due to either impurities in the argon gas, entrainment of atmospheric oxygen or ingress of oxygen from the boundary layer.M2+ Doubly Charged Levels The doubly charged ion response is potentially much less of an analytical problem than are oxides. Elements such as barium with a low second ionisation potential ( < l o eV) will give M*+/M+ ratios of ca. 2%. This is lower than can be expected from the Saha equation alone. Whilst thermal ionisation within the plasma is expected to be the dominant process, additional collisional ionisation occurring in the plasmaJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, SEPTEMBER 1987, VOL. 2 597 4.5 t 'n 2.5 t 1 .o 25 20 15 10 5 0 Spray chamber temperature/"C Fig.4. and Th Effect of spray chamber temperature on oxide levels of U, Ce 2.0 I I I I 30 25 20 15 10 5 0 Spray chamber temperature/"C Fig. 5 . Effect of spray chamber temperature on the Ba* +/Ba+ ratio expanding beyond the orifice involving the ions in the tail of ion energy distribution*.l4 has been postulated. Fig. 5 shows how the water loading can reduce this to below 1% indicating perhaps a contribution from water to the ionisation process. Polyatomic Ion Species Although the spectra in ICP-MS are intrinsically very simple consisting primarily of singly charged species with some low levels of oxide and doubly charged species, some generic polyatomic ion species are observed at low mass ( 4 0 u) which are associated with argon and solvent species.Species such as ArO+ and ArN+ may be formed by ion - molecule condensation reactions on the surfaces within the expansion chamber and careful attention to sampling design13 can reduce these cluster ions. However, reduction in water loading is also critical in obtaining a cleaner background spectrum. Fig. 6 indicates the dependence of ArO+ on solvent loading where it can be reduced to ca. 20 ng ml-1 equivalent when measured using W o as an internal standard. Dimeric species such as ATZ+ are also observed in the background spectrum and Gray3 has shown how these species may be formed within the shock wave in the expansion region. It is not directly obvious how a reduction in water loading could be directly related to dimer formation; it is, however, possible to postulate three-body collision processes in which oxygen may have a dominant role.21 Certainly from Fig.7 it can be seen that there is a strong dependence of AT2+ species on water loading and improve- ments up to an order of magnitude can be evidenced at low water loading levels. Table 2. Comparison of oxide levels in desolvated (spray chamber temperature 0 "C) plasma and dry plasma MO+/M+ Desolvated plasma Dry plasma UO+/U+ 2.05 0.32 wo+/w+ 0.8 0.06 100 20 c *, \ \ \ 'x I I I 1 I " 30 25 20 15 10 5 0 Spray chamber temperature/"C Fig. 6. Effect of spray chamber temperature on polyatomic ion ArO+ level at mass 56. ArO+ level measured in ng ml-l relative to "Co at 10 ng mlki 3000 - - ' 2000 E u) C a, a, t , - - N G 1000 I I I I 1 30 25 20 15 10 5 0 Spray chamber tem peratu re/"C Fig.7. AT2+ level expressed in ng ml-1 relative to "Co at 10 ng ml-I Effect of spray chamber temperature on ATZ+ at mass 80. Residual Sampler Contribution In early prototype ICP-MS instruments, a residual discharge (termed the pinch effect)22 resulted in some material from the sample cone being observed in the spectrum. In current commercial instruments, load coil grounding configurations are used which minimise this effect.15-23 Under certain circumstances, some small amounts of orifice material can still be observed. Erosion may also however be attributable to chemical attack and lowering water loadings certainly reduces this to <0.5 ng ml-1 on the instrument used in this work as shown in Fig.8. Hence the lifetimes of the sampling orifice have been found to be increased to the order of 4-5 months with rigorous use. Ion Kinetic Energies The reduction in M2+ with water loadings is indicative that on the instrumentation used here, the ion kinetic energies may also be similarly reduced. Mean ion kinetic energies,598 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, SEPTEMBER 1987, VOL. 2 ~~ Table 3. Influence of spray chamber temperature (water) on mean ion kinetic energy Spray Mean ion kinetic energylev temperature 27Al 5To 11% 2OXPb 30 “C 12 12 10 12 0 “C 3 4 4 5 Dryplasma - 4 chamber - - I I I I I 1 30 25 20 15 10 5 0 Spray chamber temperaturei’c Fig. 8. Influence of spray chamber temperature on level of residual nickel from sampler system measured by applying a retarding potential to the d.c.quadrupole rod bias, are seen to fall from 10-12 eV across the mass range, to 4-5 eV at the lower water loadings, as shown in Table 3. These latter values are similar to those obtained in a dry plasma20 and are consistent with observations made by other workers.14 Although the absolute magnitude of the energies is subject to some error (+ 1-2 eV) the qualitative trend is quite clear, i.e., low, but non-mass dependent values of ion energies are produced at reduced water loadings. This trend is in contrast to work reported by Fulford and Douglas24 who, with their system, reported values which varied from 2 to 9 eV across the mass range. The dependence of ion energies with water loading may be seen as contradictory to observations made by Olivares and Houk,25 who, using a prototype instrument, observed ion energies of 4-5 eV both with and without aerosol loading.However, in their system an ultrasonic nebuliser with desolva- tor was employed, this being experimentally analogous to the cooled spray chamber arrangement used here. Bearing this in mind, the actual value for ion energies obtained appear to be encouragingly similar in both works despite the different operating conditions. Conclusions The advantages of operating the ICP ion source with low and constant aerosol water loadings or in fact with dry aerosol are many fold. The background spectrum from polyatomic ions is reduced, resulting in potentially improved detection for problematic elements such as Fe and Se. Oxide levels are similarly reduced and ion sensitivity is much improved, consistent with the plasma being more efficient at ion production when the water, which acts as a thermal buffer, is reduced.Ion-energy measurements indicate that low ion energies (ca. 5 eV) can be obtained with low water loadings using this particular interface and that these are constant across the mass range. This is a particularly attractive situation as it minimises mass discriminatory ion focusing effects and allows for simpler ion focusing. The idea of totally desolvating the aerosol, however, does not at this stage appear a practical proposition for real analysis. However, operating the spray chamber at 10 “C with low solvent flows does provide a very practical alternative. It may be that future work can be directed towards improving aerosol transport to the ICP particularly with regard to the gas to liquid ratios in the aerosol. 1.2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21 22. 23. 24. 25. References Gray, A. L., and Williams, J . G., J. Anal. A t . Spectrom., 1987, 2, 81. Vaughan, M. A., and Horlick, G., Appl. Spectrosc., 1986,40, 434. Gray, A. L., Spectrochim. Acta, Part B, 1986, 21, 151. Tan, S. H., and Horlick, G., Appl. Spectrosc., 1986, 40, 445. Gray, A. L., Analyst, 1985, 110, 551. Gray, A. L., and Date, A. R., Analyst, 1983, J08, 1033. Russ, G. P., Bazan, J. M., and Date, A. R., Anal. Chem., 1987, 59, 984. Jiang, S. J., and Houk, R. S., Anal. Chem., 1986, 58, 1739. Williams, J . G., Gray, A. L., Norman, P., and Ebdon, L., J. Anal. A t . Spectrom., 1987, 2, 469. Trassy, C., and Mermet, J. M., “Les Applications Analytiques des Plasmas HF,” Technique et Documentation (Lavoisier), Paris, 1984, Chapter 9, p. 160. Tang, Y. Q., and Trassy, C., Spectrochim. Acta, Part B, 1986, 41, 143. Maessen, F. J. M. J., Kreuning, G., and Bake, J., Spectro- chim. Acta, Part B, 1986, 41, 3. Hutton, R. C . , British patent application, No. 86102463. Gray, A. L., Houk, R. S . , and Williams, J . G., J. Anal. At. Spectrom., 1987, 2, 13. French, J. B., Douglas, D. J., Spectrochim. Acta, Part B, 1986, 41, 197. Ripson, P. A. M., and de Galan, L., Spectrochim Acta, Part B, 1981, 36, 71. Weast, R. C., Editor, “Handbook of Chemistry and Physics,” 56th Edition, CRC Press, Cleveland, OH, 1975-1976. Horlick, G., Tan, S. H., Vaughan, M. A., and Rose, C. A., Spectrochim. Acta, Part B, 1985, 40, 1555. Zhu. G., and Browner, R. F., Appl. Spectrosc., 1987,41,349. Tye, C. T., and Gordon, J. S . , VG Internal Technical Report, 1987. Gray, A. L., personal communication. Houk, R. S . , Fassel, V. A., and Svec, H. J., Dyn. Mass Spectrom., 1981, 6, 234. Gray, A. L., J. Anal. A t . Spectrom., 1986, 1, 247. Fulford, J. E., and Douglas, D. J., Appl. Spectrosc., 1986, 40, 971. Olivares, J. A., and Houk, R. S . , Appl. Spectrosc,, 1985, 39, 1070. Paper J7l22 Received February 17th, 1987 Accepted May 2lst, 1987

ISSN:0267-9477    

DOI:10.1039/JA9870200595

出版商:RSC    

年代:1987

数据来源: RSC

摘要:

JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, SEPTEMBER 1987, VOL. 2 599 System Optimisation and the Effect on Polyatomic, Oxide and Doubly Charged Ion Response of a Commercial Inductively Coupled Plasma Mass Spectrometry Instrument* Alan L. Gray and John G. Williams Department of Chemistry, University of Surrey, Guildford, Surrey GU2 5XH, UK The practical problems of setting up and operating a commercial inductively coupled plasma mass spectrometry (ICP-MS) instrument for routine analysis are discussed and the effect of choice of operating parameters on performance described. With the instrument used, a VG PlasmaQuad, the major parameter variables, plasma power and gas flows and the aperture- load coil spacing are normally kept constant for routine use. Under these conditions a sufficiently uniform response to a range of elements of different chemistries and atomic mass is obtained for the operator to use the instrument without repeated optimisation throughout the working day even though a variety of different sample types and matrices may be run.At fixed plasma power and spacing the only other critical parameter is the carrier gas flow-rate which is adequately stabilised in the instrument used by a precision mechanical flow regulator. The effect upon performance of variation of this flow-rate is shown for elemental response, doubly charged ion ratio and polyatomic ions and it is demonstrated that setting up for maximum elemental response minimises the effect of these interfering species. In practice carrier gas flow-rate is also normally fixed and setting up the instrument consists only of minor trimming of the ion-lens potentials for optimum response which is usually carried out when starting up for a day's running.Operated in this way the instrument is used for routine analysis by a variety of users on a wide range of matrices. An indication is given of the performance obtained under these conditions. Keywords: Inductively coupled plasma mass spectrometry; optimisation; routine analysis; parameter variation; interferences Mass spectrometry (MS) with an inductively coupled plasma (ICP) ion source is a cross disciplinary technique between ICP emission and quadrupole mass spectrometry and few potential users have experience of both. The technique is often seen as an alternative route from atomic absorption spectrometry (AAS) to multi-element analysis instead of through induc- tively coupled plasma atomic emission spectrometry (ICP- AES) and thus not all new users have any plasma and still fewer mass spectrometry experience.It is now widely appre- ciated that the detection limits achieved in real matrices may be considerably higher than the idealised values reported from 10-s integrations on laboratory standards. The principal cause of this degredation is spectroscopic interference with the isotope responses of interest which cannot be resolved by the quadrupole mass analyser. These arise from three main sources: (i) polyatomic ions resulting from ion - molecule reactions between major species in the plasma which occur during the extraction process or in the expansion stage; (ii) analyte oxide ions resulting from incomplete dissociation in the plasma, recombination in the boundary layer or ion - molecule reactions during ion extraction; and (iii) doubly charged ions of analyte or matrix species. The extent of these problems, normally most serious below 80 u, is found to depend on many factors which include extraction geometry, operating parameters for plasma and nebuliser systems and the nature of the acid and sample matrix. Some recent publications,1-3 describing, for one of the commercially available instruments, extensive studies of the effect of parameter variations on performance and of the system optimisation needed for different elements, may have created an impression that the technique is inherently difficult to operate for multi-element analysis.Such studies are important in investigating and developing the technique and have an important role in establishing operating mechanisms, * Presented in part at the 1987 Winter Conference on Plasma and Laser Spectrochemistry, Lyon, France, 12th-16th January 1987. but the report of an exhaustive study can generate some confusion in the mind of the inexperienced. Some aspects of the behaviour reported are so different from the authors' own experience on a another system that it appears that many of the effects may be characteristic of system design rather than of the technique as a whole. It was therefore considered worthwhile to record the behaviour of the system they use and their own operational practice.The effect of operating parameter variation on performance is discussed and the process of optimisation described for the preferred operating conditions selected for routine analysis. The instrument concerned has now been in use in the laboratory at Surrey for over 18 months and is used by six or seven operators for a wide range of different analyses. Experimental The instrument in question (VG PlasmaQuad ICP Mass Spectrometer, VG Isotopes, Winsford, Cheshire, UK) is in all respects but two a standard version at the time it was manufactured and uses an RFA plasma generator and torch box with a three-turn load coil. It was supplied with an IBM PC-XT computer. The two changes made to the instrument since it was delivered are the fitting of a Negretti and Zambra APR pressure regulator on the gas feed to the nebuliser as a precision flow controller and the use of a water-cooled spray chamber, cooled by tap water.The gas supply to the instrument as a whole is also maintained at constant pressure by a precision regulator, the Negretti and Zambra Type R182, and with this combination a mass flow controller has not been found necessary. Both Jarrel Ash cross flow and a Meinhard concentric nebuliser have been used in these studies with essentially identical results, for brevity only those with the concentric nebuliser are reported here. A high solids de Galan - Babington-type nebuliser from Van der Plas Products is also used where sample conditions require it. For the multi-parameter studies the instrument was oper- ated in the normal way as used for routine analysis, except that plasma parameters were varied from the default values shown600 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, SEPTEMBER 1987, VOL.2 I 400 Table 1. Normal system parameters for multi-element operation - 260 I 1 7 2 4 0 1 1, 220 ACO Plasma power . . . . . . . . . . Reflected power . . . . . . . . Coolant argon flow . . . . . . . . Auxiliary flow . . . . . . . . . . Carrier flow . . . . . . . . . . Spraychambertemperature . . . . Pumpedsample uptake rate . . . . Skimmer diameter . . . . . . . . Load coil - aperture spacing . . . . Aperture diameter . . . . . . . . 1300 W c2ow 14 1 min-1 0 0.73 1 min-l 13 "C 1.1 ml min-1 1 .O mm 0.7 mm 10 mm in Table 1 by changing the central channel gas flow in steps at three power levels in turn.Other plasma gas flows, aperture - load coil separation and sample uptake rate (pumped) were not changed. Reference to gas flow below therefore always refers to central channel flow. Multi-element samples were used at 1 pg ml-1 for the variation of M+ response with flow-rate. For polyatomic, oxide and doubly charged ions, solutions containing the elements shown in Table 2 were run at concentrations of 1 and 10 pg ml-1. Chlorine values were obtained from dilute HCl at 1 pg ml-1 and 10 mg ml-1. Solutions were introduced in turn at each setting. The lens system was optimised for each new set of flow and power levels. Integral response was recorded in the scanning mode across a mass range from 25 to 260 u using 2048 MCA channels and a 1-min integration time for each run.The choice of elements in the samples was made to provide a range of second ionisation energies and monoxide bond strengths. The yields of polyatomic ions were monitored by observing three species, ArO+, C10+ and ArAr+, which are among the most prominent and troublesome occurring and are believed to be representative of the behaviour of most of those observed on this system. Interface The interface used in this instrument is of the type developed at Surrey4-6 in which the plasma load coil is grounded at the end closest to the aperture. Apertures of 1 mm diameter are normally used with miniature skimmers of 0.7 mm diameter. It is routine practice to clean the aperture at the start of each days' run. The skimmer is cleaned when its' surface shows any significant deposit.This may be as frequently as daily when running samples with a high solids content. Both are cleaned with a proprietary stainless-steel cleaning preparation, Pol- aris, applied with a damp tissue. After cleaning, residues are removed in acetone in an ultrasonic bath. Discussion Plasma Optimisation The performance of the ICP mass spectrometer interface is dependent on parameters such as plasma power, carrier gas flow-rate, aperture - load coil separation and water content of the carrier gas flow and aerosol, and these are mutually interdependent. Arbitrary variation of any of these can produce confusing results but there is no necessity in normal operation for such variation, and the best results are obtained by keeping all these parameters fixed.The water content of the carrier gas is an important parameter because H+ and O+ together contribute a substan- tial fraction of the total plasma electron population7 and it is therefore important that this is kept stable. While the use of a peristaltic pump keeps the H20 introduced as aerosol constant, evaporation from water surfaces in the spray chamber at temperatures above about 25°C can add con- siderably more vapour to the carrier gas than the aerosol. This variable has therefore been eliminated by stabilising the temperature of the spray chamber with a water jacket. This Y 3501 I \ c 300 n~ 250 2? 200 ? 150 V I '0 100 a $ 50 €r 0 t I P b \ 0.3 0.5 0.7 0.9 1.1 1.3 Nebuliser flowil rnin-1 Fig. 2. System response as Fig. 1 for six elements from Rb to U may, with advantage, be kept just above freezing-point, but at Surrey mains water is used for cooling.At a typical operating temperature for the uncooled spray chamber in this instru- ment of 35 "C and under the operating conditions usually used, a nebuliser efficiency of 1% produces a calculated water loading of the carrier gas of 45 mg 1-1 of which 21 mg 1-1 is due to evaporation. When the spray chamber is cooled to 10°C a higher gas flow is normally used and these levels are reduced to a 16 mg 1-1 total, of which evaporation contributes 6 mgl-I. The aperture - load coil spacing is normally set at 10 mm on this system and not varied. This spacing is related to the plasma operating parameters as it is important that dissocia- tion and ionisation are as complete as possible before the carrier gas flow reaches the aperture.As the spacing is reduced or the carrier gas flow increased the ion response rises as more ground-state ions are extracted but this carries an increased risk of incomplete dissociation of major matrix species. In an extreme instance an operator seeking the maximum signal on setting up with an aqueous standard without a significant matrix, may make the system vulnerable to the matrix of a real sample when the initial radiation zone (IRZ) is pushed forward so far that it reaches the aperture, allowing undissociated material to be extracted.8 Operation close to the brink of this situation is not necessary with this instrument. At the spacing used, high sensitivity is obtained at flow-rates which keep the IRZ more than 5mm from the aperture and high matrix concentrations do not cause increased oxide levels.Elemental response With the spray chamber temperature stabilised and a fixed value of aperture - load coil spacing a choice of central channel gas flow and power has to be made. The effect of varying gas flow on M+ response at a power level of 1300 W, the level normally used, is shown in Figs. 1 and 2. Ion-lens potentials (as discussed below) were re-optimised for each new flow-JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, SEPTEMBER 1987, VOL. 2 601 1500 W 1 1300W I X I I I I 1 0.4 0.6 0.8 1 .o 1.2 Nebuliser flow/l min-l Variation of ArO + response with carrier gas flow-rate at the Fig. 3. three POH . levels shown ik I I 0.4 0.6 0.8 1 .o 1.2 Nebuliser flowil m i n - l Ratio ArO+/Co+ from Fig.3 Fig. 4. 17 2 % c 50 n ?? 0 -. - ln + u n - 0.4 0.6 0.8 1 .o 1.2 Variation of CIO+ response with carrier gas flow-rate at the Nebuliser flowil min Fig. 5. three power levels shown rate. A similar pattern of response versus flow-rate is obtained at other power levels from 1000 to 1500 W with the optimum flow increasing with power. The elements used for M+ response variation span the Periodic Table and show a wide range of chemical behaviour but it will be seen that they show optimum response at very similar flow-rates. Therefore, there is no need with this system to change the flow-rate for optimum response for different elements. Polyatomic ions In order to relate the levels experienced to analytical problems the practice has been adopted of quoting the size of the interfering polyatomic peaks in terms of equivalent concentra- tion, referring them to the adjacent monoisotopic peak of Co, usually included in the test matrix solution.Very little interference is observed with Co so that it forms a useful reference. This then provides a reasonable guide to the significance of the polyatomic interferences to be expected from the matrix concerned. 0.4 0.6 0.8 1 .o 1.2 Nebuliser flowil min-' Fig. 6. Ratio CIO+/Co+ (1% HCI) from Fig. 5 - . . Nebuliser flowil min-' Fig. 7. rate at the three power levels shown Variation of argon dimer AT2+ response with carrier gas flow- 5 4 4 0 0 3 2 .s 2 N m a 1 0 /T"" 0.4 0.6 0.8 1 .o 1.2 Nebuliser flowil rnin-l Fig. 8. Ratio Ar2+/Co+ from Fig. 7 The yields of polyatomic ions were monitored by observing three species, ArO+, ClO+ and ArAr+, which are among the most prominent and troublesome observed and which are believed to behave in ways representative of most of those observed on this system.The responses of the three polyatomic ions examined are given for three power levels both as the ion intensity and as the ratio to the intensity of the Co ion from a 1 pg ml-1 solution. Figs. 3 and 4 show the results for ArO+. Apart from a high peak at 1500 W the intensity of the ArO+ ion changes comparatively little above 0.6 1 min-1 as the flow-rate increases. As a result the relative level of this ion compared with the Co signal is least when the response is optimised at the normal flow-rate, a sharp trough being seen at ca.0.8 1 min-1. The C10+ response shown in Figs. 5 and 6 shows a similar sharp fall in both C10+ and ClO+/Co+ values but as the C10+ response continues to fall above 0.8 1 min-1 there is little increase in the corresponding ratio. The argon dimer response is shown in both forms in Figs. 7 and 8. It shows a similar hump602 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, SEPTEMBER 1987, VOL. 2 13 ? - 2 12 C 11 010 0. a 9 - v, 8 - c 0 0. .- c .- c 7 - 2 6 - Table 2. Oxide formation (MO+/M+) at 0.73 1 min--' nebuliser flow for elements with a range of oxide bond strengths Bond Ratio MO+/M+ strength/ Element kJmol-1 l1OOW 1300W 1500W - - - - Rb . . , . . . . . 255 C1 . . , . . , . . 272 Bi . . . . . . . . 343 Cr . . . . . . . . 427 Tm . . . . . . . . 557 Ho .. . . , . . . 623 Ce . . . . . . . . 795 Th . . . . . . . . 854 * ND, not detectable. ND* 0.0087 ND 0.0008 0.0019 0.0033 0.0122 0.0213 ND 0.0096 ND 0.001 1 0.0016 0.0030 0.0129 0.0187 ND 0.0269 ND 0.0023 0.0023 0.0047 0.0179 0.0278 at flow-rates below the optimum for the signal but falls sharply at high flows and stays low at all powers. These three ions, and their associated isotopes produce some of the most serious interferences but these plots show that operating conditions may be chosen to minimise their impact without losing sensitivity significantly. It is clearly important that operating conditions are stable but as the parameters concerned are gas flow and plasma power this is readily achieved. Oxides The behaviour of the oxide response from elements of increasingly refractory nature is shown for the worst case, the highest flow-rate, as a function of bond strength in Fig.9. Tabulated values for a flow-rate of 0.73 1 min-1, which is close to the sensitivity optimum, are shown in Table 2. Apart from C1, the values for which suggest a different origin for this ion, the values for the non-refractory elements of low bond strength such as Bi and Rb are below the background9 for solutions of 10 pg ml-1 and lower. At the highest flow-rate the increased values for Ce and Th show the effect of reduced dwell time in the central channel and the resulting poorer dissociation of the most refractory oxides, although even here the oxide levels for the rare earth elements Tm and Ho, and others with lower bond strengths, are still below 4%.The variation with flow-rate for the three most refractory elements is shown in Fig. 10 for the usual power level of 1300 W. The sharp rise at the highest flow-rate is evident but at the flow-rate for peak response shown in Figs. 1 and 2 of 0.73 1 min-1 acceptably low oxide levels are obtained without any special precautions. The effect of power level is seen in Fig. 11 for the most refractory element Th. Higher power results in slightly higher 200 400 600 800 Fig. 9. Variation of oxide response ratio MO+/M+ with bond strength at highest carrier gas flow-rate used, 1.13 1 min-I and at the three power levels shown Nebuliser flowil min-l 0.4 0.6 0.8 1 .o 1.2 Nebuliser flowil min-' Fig. 10. Variation of MO+/M+ for refractory elements with carrier gas flow-rate.Normal operating power 1300 W; 1 pg ml-1 solutions 0.3 1 1500 W/: I 0.4 0.6 0.8 1 .o 1.2 Nebuliser flowil min-' Fig. 11. power levels Variation of ThO+/Th+ with carrier gas flow-rate at three 4 ' I I I 0.4 0.6 0.8 1 .o 1.2 Nebuliser flow/l min-l Fig. 12. Variation of mean ion stopping potential with carrier gas flow-rate for In+ ions at the three power levels shown 14 I 5' I I I 0.4 0.6 0.8 1 .o 1.2 Nebuliser flowil rnin-' As Fig. 12 but for U+ ions Fig. 13.JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY. SEPTEMBER 1987, VOL. 2 603 oxide levels at all flows although the effect at the normal working position is small. This may be attributable to reduced ion energy producing less collisional dissociation in the interface, but this is not well supported by plots of mean ion stopping potential shown in Figs.12 and 13. This was measured by observing the analysing rod decelerating poten- tial, or pole bias, required to reduce the signal rate for the ion concerned from the normal level observed to background. The mean stopping potential used is the mean of this value and the potential at full signal rate. A fuller discussion of this can be found in references 10 and 11. For both In and U the potential, which gives a measure of extracted ion energy,IO reaches a plateau at about 0.8 1 min-1 and does not show a strong dependence on power or mass over the region where the oxide level is changing most rapidly. Doubly charged ions The doubly charged ion response for the two elements used having the lowest second ionisation energies, Ce and Ba, is shown as a function of flow-rate in Figs.14 and 15. The M2+ integral is shown as a ratio to the Co+ integral at the same concentration for convenience so the ratio shown does not correspond exactly to the ratio M2+/M+ often quoted. The 0.4 0.6 0.8 1 .o 1.2 Fig. 14. Variation of doubly charged ion ratio Ce2+/Co+ with carrier gas flow-rate Nebuliser flow4 min- 0.4 0.6 0.8 1 .o 1.2 Nebuliser flow/l min-1 Fig. 15. As Fig. 14 but for Ba2+/Co+ true ratio is difficult to determine as it is not clear whether the ion transmission, and hence the mass discrimination, of the over-all system is a function of ion mass or mlz ratio. There is always a problem therefore in defining doubly charged ion yield in ICP-MS systems and this should be borne in mind when comparing reported results.Interference considerations are best aided by comparison with the ions affected and expression in relation to the Co+ response is a convenient form. The response shows a familiar pattern, as also reported elsewhere,l0,12,13 with a sharp rise for Ba at the highest flow-rate. Again at the optimum flow-rate for elemental response acceptable levels are obtained. The values obtained at 1300 W for elements with a range of second ionisation energies are shown in Table 3. With the water loading and aperture-load coil spacing fixed, the remaining interdependent variables are then readily optimised and are normally left unchanged for analytical work. The operating power level is usually set at 1300 W. A fixed setting for this makes the optimisation of the flow-rate simpler but similar patterns of response versus flow-rate are obtained over the useful power range of 1000-1500 W, with the optimum flow increasing with power.Little trouble is thus experienced in choosing the optimum carrier gas flow-rate for the selected operating power. Flow-rate regulation to better than 0.1% obtained with the controller used is more than adequate to ensure stable operation at the desired figure. Once set the flow-rate and power are kept constant for all routine operations with the chosen nebuliser. Similar behaviour is observed for concentric and high solids nebulisers although the optimum flow-rate may be different. The plasma operating parameters normally used are shown in Table 1. Ion Energy and Ion Lens Optimisation The potential assumed by the ICP central channel above ground depends on the extent of the RF capacitative coupling from the load coil to the plasma.A variety of coil grounding arrangements may be used to control the plasma poten- tialll,l”14 which may lie between 0 and about 20 V in practical operating systems. As this sets the potential at which the ions are formed and hence the energy at which they enter the interface, it also affects the operation of the ion-lens system needed to produce an ion beam along the axis of the quadrupole mass analyser. The final energy of ions reaching the quadrupole is of little importance provided that the spread of energy is only a few eV. The pole bias of the analysing rods may be used to off-set any excess of ion energy.The most significant consequences of the extracted ion energy lie in ion-lens performance and in collisional ionisation effects within the interface. The mass dispersion of an electrostatic lens of the type used in ICP-MS instuments is zero for all ions of constant energy. Table 3. Ratio M*+/Co+ against nebuliser flow and ion stopping potential. Forward power 1300 W Element Ne buliser flow/ I min-1 EtIV 0.46 4.53 0.59 6.09 0.73 8.58 0.86 11.40 1.13 13.50 * gill, Second ionisation energy. t E , mean stopping potential Co+. $ ND, not detectable. Ba Ce Zr Pb Pt c s Rb Ei“*/eV 10.00 10.85 13.13 15.03 18.56 25.08 27.50 0.0015 0.0006 ND$ ND ND ND ND 0.0143 0.0053 0.0003 ND ND ND ND 0.0573 0.0188 0.0022 0.0003 ND ND ND 0.0469 0.0223 0.0027 0.0001 ND ND ND 0.1625 0.0532 0.0158 0.0018 ND ND ND604 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, SEPTEMBER 1987, VOL.2 For an elemental analysis system, where the ions to be focused simultaneously have a mass range from 6 to 238 u, this is a desirable property. However, the dispersion is dependent on the ion energy, a property which is used in the well known electrostatic energy analyser, or “electric sector ,” of double focusing mass spectrometers, and in the “Bessel box” type of energy analyser used with quadrupole instruments. For the uniform transmission desirable it is therefore necessary that the energy spread across the mass range required is as small as possible. This is commonly achieved in magnetic instruments because after leaving the source the ions are accelerated through a high potential difference so that the relative energy spread from the source is comparatively small.The ICP ion source potential however is close to ground and in the centre-tapped load coil design of Douglas15 the effective source potential is probably below 1 V.16 As the plasma gas containing the ions, which form about 0.1% of the total, is extracted by the aperture, both ions and neutrals acquire a common supersonic velocity. As their velocity is the same, the heavier ions gain more energy than the light ones from this expansion. Fulford and Douglas have17 shown that with a very low source potential, where the principal source of ion kinetic energy is the supersonic expansion during the extraction, the ion energies observed range from 2 to 9 eV across the elemental mass range and are affected very little by the carrier gas flow-rate.With an asymmetrically grounded load coil, choice of coil geometry and plasma operating parameters enables plasma potential to be set within a range of values from 2 to >20V. Although the value obtained is dependent on both carrier gas flow-rate and plasma power these are very easily stabilised and 20 1 0.4 0.6 0.8 1 .o 1.2 Nebuliser flowil min-’ Fig. 16. flow-rate for four ions of different masses; power 1300 W Dependence of mean ion stopping potential on carrier gas Ion Quadrupole Lens system 4 detector analyser Aperture Out Defl. skimmer and ‘w Pole * L4 L3 L2L1 Coll. bias DA Ext. Fig. 17. Schematic diagram of the ion extraction, analysis and detection system of the VG PlasmaQuad.Coll. and Ext. are collector and extractor electrodes, DA the differential aperture and L1-4 the lens cylinders. Pole bias is applied to the analysing rods do not need to be changed at all by the operator in the course of analysis. At low potentials a similar dependence on mass is found for ion energy to that reported by Fulford and Douglas but as the potential increases it has more influence on the ion energy and only a small spread of energy across the mass range remains at the usual operating flow, as is shown in Fig. 16. Here the variation of mean ion stopping potential on the analysing rods (closely related to mean ion energy) with carrier gas flow-rate is shown for the Surrey PlasmaQuad for four ions ranging in mass from YBef to 2 3 W + . Above the usual operating flow-rate the ion energies diverge again but with the lighter ions now showing the higher energies.Operation at flow-rates where the spread of ion energies is small thus makes it simpler to optimise a lens system for a level mass response as the lens potentials needed are then almost independent of mass. Optimisation is achieved with a multi-element standard containing elements with a range of masses which shows the response across the mass range. The settings are usually checked each morning when the system is started up, a process which takes only a few minutes, and remain valid as long as the carrier gas flow-rate is not changed. As the same flow-rate is close to the optimum for all elements there is no reason to change it and the system may be used for analysis in any part, or over the whole, of the spectrum without resetting either the gas flow or ion lens.Although the ion lens control panel displays a set of 12 control potentiometers, optimisation in practice is much simpler than it at first appears. A diagram of the ion extraction, analysis and detection system of the instrument is shown in Fig. 17 from which the electrodes on which the potential is adjusted may be identified. Two controls are unused. Of the remainder, the differential aperture potential remains fixed and the quadrupole entrance aperture and the entrance rod bias levels are usually set at zero. The analysing rods are usually given a small fixed positive bias. The remaining six controls are concerned with beam formation and focusing and operate in pairs, being adjusted in order along the ion path.The lens is normally tuned while monitoring the Co+ ion, as the settings are usually less sharp than at higher mass, and then retrimmed at Bi+ and Co+ in turn for level response. Provided that this is carried out systematically from the front to the back lens components, little difficulty is experienced in obtaining these settings. A typical set of response values obtained after setting up in this way at the parameters given in Table 1 is shown in Table 4. If a different nebuliser, such as a high solids type, requiring a different flow-rate, is substituted, the lens settings are quickly re-optimised for the new flow-rate by following this procedure. No difficulty is experienced with the controller used in re-setting any desired flow, although this must be monitored by the pressure drop across the nebuliser using a Table 4.Elemental response of Surrey PlasmaQuad. Normalised for 100% abundance and referred to Co = 1 .O; all elements have E,’ < 8 eV and are assumed to be 100°/o ionised; aerosol gas flow 0.73 I min I ; power 1300 W: Meinhard nebuliser Element Response Li . . . . . . . . 0.43 A1 . . . . . . . . 0.39 c o . . . . . . . . 1 . 0 Rb . . . . . . . . 0.43 In . . . . . . . . 0.87 Ba . . . . . . . . 0.63 Ce . . . . . . . . 0.81 w . . . . . . . . 2.26 Pb . . . . . . . . 1.39 Bi . . . . . . . . 1.09 u . . . . . . . . 1.34 Mg . . . . . . . . 0.32JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, SEPTEMBER 1087, VOL. 2 605 calibration graph. Although not considered necessary for stability, a mass flow controller may have the advantage of providing a direct read out of flow-rate.Collisional ionisation within the extracted gas may be a disadvantage of a high plasma potential if the resulting ion energy becomes high enough to generate significant numbers of doubly charged ions. This may be expected at energies much above 10 eV, the lowest value of Ei” (Ba), when M2+ yields may greatly exceed those expected from the plasma alone. lonisation of aperture corrosion products may also become significant if ion energy is too high. However, at potentials of about 1OV or less which give a reasonably uniform ion energy these effects are very small. Slight corrosion of the nickel aperture occurs in the presence of the water content of the aerosol and some ionisation of this can occur at the higher plasma potentials.Under normal operating conditions the consequent level of nickel ions seen in a blank spectrum is well below the 1 ng ml-1 level and does not significantly affect the detection limit for nickel or cause any reduction of aperture lifetime which with nitric acid solutions is usually measured in months. Resolution Controls The mass resolution of a quadrupole, MIAM, is not a constant, as with magnetic instruments, but the ratio of peak width (usually at 50% peak height, although it may also be expressed in terms of 10% peak height) to the separation between adjacent peak centres is usually constant and the resolution is often expressed as U A M . At any particular mass the usual MIAM resolution may thus be obtained by multiply- ing the UAM value by M .The resolution of a quadrupole is thus often quoted as a multiple of M. 24 26 2a 30 m z !/% 34 Fig. 18. Part of the full range spectrum on setting up standard at 1 pg ml-1 showing the response obtained with normal resolution settings for quantitative analysis m z Fig. 19. As Fig. 18 but for region of spectrum around “We+ Both high and low mass resolution controls are provided in the PlasmaQuad, each having its main effect over part of the spectrum. As the setting of resolution affects the ion trans- mission of the quadrupole they do also to some extent provide a means of balancing the response across the mass range but this is not their proper function and they should be set to give uniform resolution at the required level. This is usually achieved on a multi-element sample by setting the low mass control first then adjusting the high mass resolution to match.The VG 12-12s quadrupole used has a specified resolution of 2.5 M (at 50% peak width) but is usually operated at rather less than this, at about 1.6 M . This lower setting is fully adequate for most quantitative analyses, a part of the spectrum of the setting up standard including 1 pg ml-1 of A1 is shown in Fig. 18. The interference from 27 u to 26 u is below 10-4. A similar plot at Ce is shown in Fig. 19 where the vertical scale has been greatly expanded to show the small distinct peak obtained from 0.22 ng ml-l of La in the Specpure Ce at 1 pg ml-1. Although higher resolution may be obtained if required, at the expense of lower sensitivity, it is rarely needed and the resolution controls are normally never touched.Relaxation of the resolution is not necessary as sensitivity is adequate at these settings. These spectra were obtained in the scanning mode but the instrument may always be used, at the cost of some degree of “blindness,” in a peak-jumping mode where the spectrometer is set successively to the top of the wanted isotope peaks. The use of the peak-jumping mode does not change the effective peak shape and the response from the fringes of the peak is still present, although it is no longer visible. It does, however, tend to reduce the interfer- ence between adjacent masses as the central response peak is never set to the intermediate region between the peaks where overlap may occur.The abundance sensitivity is usually measured in this way, and is defined as the ratio of the response at M + 1 to the response from an ion occurring at M . An effective illustration of abundance sensitivity may be obtained from the spectrum shown in Fig. 20 where the response to a solution containing 10 ng ml-1 of Mn in a matrix of 0.1% Fe is shown. The 55Mn peak is completely resolved between the very large peaks of 54Fe and 5hFe. Quadrupole peaks are not usually symmetrical, extending further on the low mass side, but even so the response falls to zero between ”Fe and 55Mn so that the Mn peak is completely free from overlap from Fe. As the concentration of Mn is 10-5 of that of Fe and the 56Fe isotope is 91.66% abundant, the top of the Mn peak represents the abundance sensitivity level of 1.09 X 10-5.Most of the region between the 54 and 56 peaks is actually well below the 10-6 level and the area response for Mn is thus free from Fe interference. 10 50f 56Fe I I 54 55 56 mi z I 57 Fig. 20. of 1OOOpgmlk’Feand 1Ongml-1 Mn.SensitivityofMn+ ion5 X counts s-1 per pg ml-1 Spectrum obtained with increased resolution on a solution606 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, SEPTEMBER 1987, VOL. 2 The response shown in Fig. 20 required some increase of resolution over that normally employed and that which was used for Figs. 18 and 19, but even so the response to 10 ng ml-1 of Mn was still high at more than 5000 counts s-1 which is the equivalent of 5 x 105 counts s-1 per pg ml-1, The random background level of between 5 and 10 counts s-1 is thus the equivalent of less than 20 pg ml-1 under these conditions.Conclusions The data shown demonstrate the range of response variation of polyatomic, oxide and doubly charged ions obtained by parameter variation on the VG PlasmaQuad system used for routine analysis at the University of Surrey. The ions selected for display are commonly quoted as sources of interference and it may be seen that by maloperation of the system high levels of these ions may be produced. It is also evident however that if the system is set up in the usual way for maximum elemental response the operator may be confident that the levels of these ions observed will be close to the minima obtainable and that very careful optimisation for specific analytical problems is not needed.The most sensitive parameters affecting the response are carrier gas flow-rate and plasma power and both these are normally stabilised to better than 1%. Neither of these need or should be varied in the course of analysis of normal matrices, whatever element range is scanned or which peaks are chosen for peak jumping. Thus, with this system the operator is not unknowingly at the risk of unexpectedly high levels of interference occuring. An attempt has also been made to show how the Plasma- Quad instrument is set up and operated for day to day routine analysis by a number of different operators. These vary from graduate students within the University to research workers in oceanography and geology from other universities and research establishments.With such a wide range of users, many of whom are not analytical chemists, it is essential that the instrument be operated in a well defined mode that does not require continual expert intervention either to change operating conditions for different analytical tasks or to maintain performance during the running time of each user. Operation in this way is the basis of the continued funding of the facility and it has been found possible to provide such a service without compromising performance. In addition to this role the instrument is used for contract analysis and for the development of both methods and instrumentation. This imposes a heavy demand on instrument availability but this now exceeds 90%. In particular the data handling demand has required the installation of a second computer system using the machine software to free the instrument for analysis. The operation of the Surrey PlasmaQuad as an analytical facility is supported in part by The Natural Environment Research Council and one of us (A. L. G.) also acknowledges their support and permission to publish this paper. J. G. W. acknowledges support from the Ministry of Defence. 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. References Horlick, G., Tan, S. H., Vaughan, M. A., and Rose, C. A,, Spectrochim. Acta, Part B, 1985, 40, 1555. Vaughan, M. A., and Horlick, G., Appl. Spectrosc., 1986,40, 434. Tan, S. H., and Horlick, G., Appl. Spectrosc., 1986, 40, 445. Date, A. R., and Gray, A. L., Spectrochim. Acta, Part B, 1983, 38, 29. Gray, A. L., and Date, A. R., Analyst, 1983, 108, 1033. Gray, A. L., Spectrochim. Acta, Part B, 1985,40, 1525. Alder, J . F., Bombelka, R. M., and Kirkbright, G. F., Spectrochim. Acta, Part B, 1980, 35, 163. McLaren, J. W., Beauchemin, D., and Berman, S. S . , J. Anal. A t . Spectrom., 1987, 2, 277. Gray, A. L., and Williams, J. G . , J. Anal. At. Spectrom., 1987, 2, 81. Olivares, J. A . , and Houk, R. S., Anal. Chem., 1985,57,2674. Gray, A. L., Houk, R. S., and Williams, J. G., J . Anal. At. Spectrom., 1987, 2, 13. Gray, A. L., Spectrochim. Acta, Part B, 1986, 41, 151. Gray, A. L., J . Anal. A t . Spectrom., 1986, 1, 247. Gray, A. L., Fresenius 2. Anal. Chern., 1986, 324, 561. Douglas, D. J., US Pat., 4 501 965, 1985. Houk, R. S., Shoer, J. K., and Crain, J. S., J . Anal. At. Spectrom., 1987, 2, 283. Fulford, J . E., and Douglas, D. J., Appl. Spectrosc., 1986,40, 971. Paper J7l47 Received April 9th, 1987 Accepted May 21st, 1987

ISSN:0267-9477    

DOI:10.1039/JA9870200599

出版商:RSC    

年代:1987

数据来源: RSC