JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY OCTOBER 199 1 VOL. 6 52 1 Tungsten-tube Electrothermal Atomizer WETA-90 Part 1. Design and Performance of the Atomizer Vaclav Sychra JiFi Doleial Robert HlavaC Libor PgtroS Olga VyskoCilova and Dana Kolihova Institute of Chemical Technology Technicka 5 166 28 Prague 6 Czechoslovakia Petr Puschel institute for Brown Coal 434 37 Most Czechoslovakia A new version of a transversely heated tungsten-tube electrothermal atomizer (denoted WETA-90) has been developed and tested for a number of elements. The performance of two types of tungsten tubes is compared. The capability of this system to reach stabilized temperature atomization is discussed. Characteristic masses for 23 elements are listed and compared with those obtained with graphite furnaces (heated graphite atomizers) following the original or modified stabilized temperature platform furnace concept.The WETA-90 atomizer features unique analytical parameters particularly for analytes which react with graphite and form analyte compounds of low volatility. Keywords Tungsten-tube atomizer; electrothermal atomization from tungsten surface; atomic absorption spectrometry; determination of refractory elements In a previous paper' a simple tungsten-tube atomizer was described consisting of two profiled tungsten strips forming a cylindrical cavity which could be accommodated in the workhead of a commercial Varian carbon rod atomizer (CRA-63 or CRA-90) and operated as an alternative to the graphite tubes and cups. The process of atom formation was also studied in this atomizer and it was demonstrated that solid phase or vapour phase thermal dissociation of analyte species is the most probable mechanism of atomization.2 Two years later further improvements were announced covering a new design of workhead and voltage and optical feedback circuits in the power supply.This resulted in the construction of an independent unit denoted as the WETA- 82 that was compatible with most existing atomic absorp- tion spectrometer^.^ This system featured an absence of carbide formation and memory effects homogeneous tem- perature distribution along the tube and a rapid and controlled heating resulting in high sensitivity and analytes of medium and low volatility were atomized under virtually isothermal condition^.^-^ Chakrabarti and co-~orkers*~~ studied both theoretically and practically spatial and temporal temperature distribu- tion on the tungsten-tube surface and in the gas phase in the tube.They reported a significant temperature gradient over the circumference of the tube and suggested possible improvements in the design of the atomizer with a view to making the temperature distribution more uniform. They also showed that the experimentally determined gradient between the temperature of the gas and the tube surface was smaller than that predicted. The WETA-82 atomizer was manufactured commercially in Czechoslovakia (Laboratory Instruments Prague) in 1984-1985. Despite the fact that almost 100 units have been sold and used routinely few papers dealing with analytical applications of this atomizer have been pub- l i ~ h e d .~ J ~ - l ~ Most of these applications cover carbide- forming elements and other analytes of low volatility such as Ba," rare earth e l e m e n t ~ ~ J ~ J ~ P5 and V.5 The intention of the present paper is to introduce a new model of transversely heated tungsten-tube electrothermal atomizer (denoted WETA-90). Compared with the WETA- 82 all parts of the system e.g. power supply workhead and the tube itself have been altered significantly taking into account the latest electrothermal atomizer technology and nearly ten years of experience with the operation of the original system. Experimental Instruments All measurements were carried out with a Varian Techtron AA 775 ABQ double-beam atomic absorption spectro- meter.The wavelengths of the analyte lines and lamp currents were as recommended by the manufacturer. A spectral bandwidth of 0.2 nm was used throughout. The atomizer surface temperatures were measured with a dual-wavelength pyrometer (Quotienten Pyrometer QP3 1 Leybold-Heraeus Hanau Germany). Dynamic tempera- ture measurements were made with a calibrated Ge photodiode. Absorbance-time and temperature-time pro- files were measured with a six-channel storage oscilloscope (Tesla OPD 600 ValaSskC MezifiEi Czechoslovakia). Sampling was performed manually with an adjustable 15 pl syringe (SGE Melbourne Victoria Australia). Reagents Specpure metals or compounds (Johnson Matthey Roys- ton Hertfordshire UK) were used for the preparation of stock solutions of the metals at a concentration of 1000 pg ml-l.All working solutions were prepared immediately before use by stepwise dilution of the stock solutions with doubly distilled de-ionized water and were acidified with nitric acid (except for Sb Sn and Zr where hydrochloric acid was used) to a final concentration of 1% v/v. Tungsten-tube Atomizer WETA-90 The WETA-90 consists of a power supply and control unit a workhead with the tungsten tube and a computer (Fig. 1). The workhead is permanently connected to the power supply and control unit by a supply cord which carries all the gas water power cables fibre optics and other electrical supplies. The power supply and control unit includes a three- phase transformer all power circuits supplying the power to the tungsten tube a computer-controlled gas control unit power supply for the microcomputer and part of the electronics including a 12 bit analogue-to-digital (ND) converter.The microcomputer (conventional 8 bit labora- tory-built based on an 8080A microprocessor) works in an operational system CP/M and has an internal RAM memory of 128 kbytes and an external memory consisting522 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY OCTOBER 199 1 VOL. 6 Power supply - 14 V-2500 A L I P a Atomizer workhead Constant current power supply 1 1 i A py 3 I L P a v) 4nalogue electronics 12 bit A/D converter 1 P - - * " - l l - - 1 0 n 4 0 \ I LUIILIUII~I \OV401 I 4 I Microcomputer (8080A) Fig. 1 Schematic diagram of a WETA-90 electrothermal atomizer of two 5 i in floppy-disk drives with a capacity of 2 x 360 kbytes.It also contains a single-chip microcomputer (con- troller) based on an 8048 microprocessor which is used solely to control the temperature of the tungsten tube. Instead of a voltage feedback as applied in the WETA-82 atomizer a resistance feedback is used to control the temperature of the tube in the range 40-1260 "C; the temperature in the range 1260-3200 "C is controlled with an optical feedback circuit incorporating a sensitive silicon photodiode specially made for this purpose. The perform- ance of the resistance feedback is illustrated in Fig. 2. Calibration dependences R= f(T) and U= f(T) (where R is the resistance of the tungsten tube Uis the output voltage of an amplifier of the optical feedback circuit and T is the temperature of the tungsten tube) for the resistance feed- 500 5 I + t K i I I 0 1 2 3 3.33 rlms Fig.2 Dependence of tungsten-tube current versus time -4 communication between microprocessors; B start of A/D conver- sion measured value is the zero line for correction of an amplifier offset; C switching on of a constant current (1 00 A) power supply; D start of A/D conversion measured value is voltage correspond- ing to a workhead temperature; E start of AID conversion. measured value is atomizer voltage corresponding to the resistance of a tungsten tube; F. switching off of the constant current ( 100 A) power supply; G calculation of time needed for supplying power to the tungsten tube H switching on of a power supply ( 14 V-2500 A) for time calculated in step G; I. calculation of real temperature values from measured data communication between microproces- sors; J switching off of the power supply ( 1 4 V-2500 A); and K.end of the control cycle back and the optical feedback respectively are stored in the memory of the microcomputer. At the beginning of each temperature cycle the resistance of the tungsten tube and the temperature of the workhead (including the tempera- ture of the tube) are first measured; from the data obtained the dependence R =f(v is recalculated for this particular temperature cycle. During temperature regulation of the tungsten tube by means of the optical feedback the temperature of the tube is measured at intervals of 560 ps; if it is higher (or lower) than the required temperature the power supply (1 4 V-2500 A) is either switched on or off.An automatic recalibration of the optical feedback circuit can be readily realized. During this recalibration the micro- computer gradually sets two different temperature values within the range of optical feedback but utilizes the resistance feedback for the temperature control. The tem- perature values obtained are then used for the correction of the dependence U= f(v. Both feedbacks feature high stability and reproducibility of temperature settings 2 4 K at 11O"C k 2 0 K a t 1OOO"Cand +40Kat300OoC.The accuracy of the tube temperature when different tubes are used is within f 5 K (t- 2% above 250 "C) .of the pre-set temperature value provided that the mass ratio of the upper and lower strips of the tube is within k 1% tolerance. The computer software allows 15 steps of the tempera- ture programme to be written by the operator.In each step the operator can select final temperature time (or heating rate) necessary to reach the pre-set temperature hold time mode of introduction of the sheath gas the composition of the sheath atmosphere and three independent commands for external instruments and accessories. The heating rate of the tube can be set continuously in the range 0.01-30 K ms-l. The ratio of argon to hydrogen in the sheath gas may be altered from 0 to 0.75 in 0.05 increments. Each programme can be arbitrarily modified and stored on the floppy disk. The tungsten tube is located in a water-cooled and gas- tight workhead which is shown schematically in Fig. 3. The tube is heated transversely to its longitudinal axis.The workhead is opened by swinging away the right-hand sidewall by means of a cam disc. Unlike the WETA-82 workhead this workhead features so called 'free clamping' of the tungsten tube laid in the recess of the contactJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY OCTOBER 199 1 VOL. 6 523 electrodes which prevents rapid deformation of the tube owing to dilation of the material. In order to decrease the temperature gradient over the circumference of the tube to a minimum value the distance between the contact workhead electrodes was increased compared with the WETA-82 workhead. Thus the cooling action of the water- cooled contact electrodes on the wings of the tube was diminished. Two iris diaphragms mounted at the sidewalls of the workhead and fitted with circular quartz windows serve for rapid and precise alignment of the optical beam through the tungsten tube; the diaphragm close to the entrance slit of the monochromator can efficiently prevent the excess of radiation from the hot tube from falling on the entrance slit.b Fig. 3 WETA-90 tungsten-tube workhead The workhead enables two modes of introduction of the sheath atmosphere into the compartment where the tube is placed. The sheath gas can flow from the bottom part of the workhead external flow and/or from cones directed to both ends of the tube internal flow. The role of the internal gas flow is to help to remove products of drying and pyrolysis from the inner volume of the tube. Contact sensors measuring the resistance of the tube for the resistance feedback are carefully insulated from the other parts of the workhead in order to overcome the distortion of measured resistance values due to contact resistance between the contact area on the workhead electrode and the tube itself. The important parts of the workhead are also a sensor measuring the instantaneous temperature of the workhead and a fibre optic monitoring the radiation from the end of the lower part of the tungsten tube.The tungsten tube itself consists of two profiled tungsten strips 0.127 mm thick and 20 mm wide manufactured by highly sophisticated technology in Metallwerk Plansee (Reutte Tirol Austria).14 The strips form a tube 20 x 6 mm i d . A sampling hole is drilled in the centre of the upper strip. A sampling microboat pressed into the bottom part of the lower strip can accommodate up to 25 pl of the sample solution. A standard tube is shown in Fig.4(a). The physical dimensions of the tungsten tube are very similar to those of most commercial graphite furnaces. The WETA-90 workhead can also accommodate modi- fied tungsten tubes which consist of the standard upper part (a) Fig. 4 Different designs of tungsten tubes (a) standard tube; ( b and c) modified lower strips of the tube Fig. 5 Tungsten tube modified lower part with sampling micro- boat and overflow rims and a modified lower strip [see Fig. 4(b) and (c) and Fig. 5].15 Because of the change of the effective cross-section for current in the modified wings of the lower strip these tubes are expected to exhibit the so called ‘autoplatform effect’ i.e. a delay in heating of the central part (microboat) of the lower strip.A similar effect was achieved by Lawson et all6 by ‘end heating’ (heating the forked supports) of an 18 mm CRA atomizer. The performance of these tubes is discussed under Results and Discussion. Results and Discussion Measurement of Surface Temperature Distribution in Modi- fied Tubes In order to verify the predicted performance of the modified tungsten tubes time-temperature dependences at5 24 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY OCTOBER 199 1 VOL. 6 five different test points (see Fig. 6) on the surface of the modified tungsten strip located for the purpose of these measurements in the upper part of the tube were investi- gated. Because of the response curve of the photodiode used for the dynamic temperature measurements these depen- dences could be measured with reasonable precision only at temperatures above 1400 K (Fig.7). From this figure it can be seen that throughout most of the heating cycle the temperature in the middle part of the strip where the sample is deposited test points 2 and 3 lags behind the temperature of the end of the strip test point 1. This temperature difference during the initial part of heating is typically 50-250 K the corresponding delay in time being typically 50-1 50 ms. The temperature at the test points 1,2 and 3 is relatively quickly balanced when the final pre-set tube temperature is reached and the power is cut off. The 4. fin 1 Fig. 6 Temperature test points on the modified lower tungsten strip I 1 2600 1 I 2 400 1600 0 I 0.2 0.4 0.6 0.8 1.0 1.2 t / S Fig.7 Time-temperature variation at different test points on the modified lower tungsten strip. Test points refer to those given in Fig. 6 3000 2500 0 0.2 0.4 0.6 0.8 1.0 tls Fig. 8 Temperature difference between test points 1 and 3 as a function of time and heating rate. Solid lines without points represent temperature variation at various heating rates and the solid lines with points represent temperature difference between test points 1 and 3 at various heating rates. A 1.5; B 5; and C 15 K ms-' 1 1 .o 0 C 2 u) 2 0 J3000 2000 1000 I I 1 10 0 1 2 3 tls Fig. 9 Copper (75 pg) atomization from A standard. tungsten tube; and B tungsten tube with modified lower strip (c). Line C represents the tube temperature variation fact that the ends of the tube are heated slightly faster than the middle part of the lower strip should prevent condensa- tion of the analyte at the ends of the tube and diminish the loss of analyte due to expulsion.Slight overheating at test point 4 as expected is due to an increase in the local thermal and/or electrical resistance at that point. The temperature differences between test points 4 and 5 and test points 1-3 indicate that there is no significant temperature gradient over the circumference of the tube. Fig. 8 shows the temperature difference between test points 1 and 3 as a function of time and heating rate. The higher the heating rate the higher the temperature differ- ence between test points 1 and 3 in the initial part of the heating cycle.The temperature distribution measured in the modified tungsten tubes should result in a shift in the time of appearance of the analyte i.e. in a shift of the analyte peak towards the region of stabilized temperature atomization as compared with the standard tubes. This is shown in Fig. 9 where the peaks for Cu obtained using the standard tube and from the modified tube are compared. The modified tube is recommended for the determination of analytes of high and medium volatility. Since the 0.300 0) C ; 0.200 2 a 0 0.100 n " 2400 2600 2800 3000 3200 7°C Fig. 10 Change in absorbance signal at the W 255.14 nm line with the atomizer temperature and hydrogen flow rate A 100; B 400; C 800; and D 1450 ml min-l. Flow rate of argon 2 1 min-'; heating rate 5 K ms-l; 20 ,ul H,O added prior to the atomizationJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY OCTOBER 199 1 VOL.6 525 Table 1 Typical experimental characteristic masses of elements Atomization Characteristic mass/pg temperature/ Element Wavelength/nm "C WETA-90* HGA-STPF method? Ag A1 As Ba Bi Cd Cr c u DY Er Ga Mn Ni Pb Sb sc Sr Sn Tm V Y Yb Zr 328.1 308.2 193.7 553.6 306.8 228.8 357.9 324.8 42 1.2 400.8 287.4 279.5 232.0 283.3 217.6 39 1.2 460.7 224.6 37 1.8 3 18.5 410.2 398.8 360.1 1800 2700 2400 2600 2200 1800 2600 2400 3000 3000 2400 2300 2600 1800 2500 3000 2600 2400 2800 3000 3100 2800 3200 0.8 9.7 1.5 7.4 0.2 0.7 0.8 5.5 6.2 0.7 2.8 3.1 4.4 0.14 2.5 1 1 1 1 19 11 23 40 600 0.6 1.2 11.8 15 22 4.3$ 0.4 2.4 2.5 6.5$ 17$ 1.7 7.5 7.7 8.7$ 0.8$ 3.1$ 12 22 20 24$ 38$ 1.7 - *Values based on pe?k height measurement under optimized heating rate and hydrogen flow rate conditions.?Peak area values. $Values obtained from ref. 17; all other values from refs. 1 8 and 19. modified tubes exhibit significantly shorter lifetimes com- pared with the standard tubes even when the use of very high heating rates and very high atomization temperatures are prevented for some complex analytical applications a rapid heating of the standard tube combined with chemical modification of the analyte is preferable to provide a shift in the analyte pulse. Determination of Tungsten in the Gas Phase of the Tube Recently a paper has been published13 by one user of the original system WETA-82 indicating that there is too much atomic tungsten in the vapour phase of the tube at a relatively low temperature which would cause spectral interferences.Such observations were not made in the present study as can be seen from Fig. 10. A relatively small absorption signal for tungsten was observed at the most sensitive tungsten line at 255.14 nm above 2800 "C. No non-specific absorption as a result of the presence of tungsten species was observed. Since the main precursor of the tungsten atoms and other tungsten species is believed to be tungsten trioxide the discrepancy in the paper mentionedI3 could be explained by the superior performance of the WETA-90 workhead i.e. by lower oxygen partial pressure in the sheath gas as compared with the WETA-82. Figures of Merit Since the application of high heating rates together with the non-porous nature of the tungsten surface results in the generation of absorbance peaks with half-widths which are very often less than 100 ms the ratio of peak height to peak area is much higher for the WETA-90 than for graphite furnaces and peak height is usually used for signal evalua- tion.Characteristic masses based on the peak height evaluation and atomization temperatures for 23 elements are listed in Table 1 and compared with those obtained with graphite furnaces (heated graphite atomizers HGA) utiliz- ing the original or modified stabilized temperature platform furnace (STPF) concept.17J8 The results for the WETA-90 are in most instances better (results are comparable for Dy and Y) than those for the HGA furnaces. When comparing the results for some carbide-forming elements it should be taken into account that the modified STPF concept assumes that the analyte is atomized from a tantalum platform inserted into a tanta- lum lined graphite furnace which is a very complicated and impractical procedure.Lifetime of the Tungsten Tube The lifetime of the tungsten tubes depends of the oxygen and nitrogen content in the protective atmosphere acidity of the solutions analysed sample matrix and the tempera- ture programme used. At 2500 "C the tube can be re-used 200-350 times provided that hydrogen ( 10% v/v) is added to the argon purge gas. Despite the improved clamping of the tube in the workhead slight distortion of the tube occurs after 200-300 hundred firings. The lifetime of the tungsten tubes is significantly reduced when samples with complex organic matrices e g .petro- leum samples that leave an appreciable amount of reactive carbon in the tube are analysed. Tungsten carbide formed in the furnace at high temperature rapidly changes the chemical and mechanical properties of the tube. These problems could be partly overcome by analysing such samples with a tungsten platform inserted into the tungsten tube. Problems encountered during the analysis of organic samples in the WETA-90 will be the subject of a separate paper. Conclusions The new design of the tungsten-tube furnace workhead and power supply and the addition of a microcomputer incor-526 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY OCTOBER 199 1 VOL. 6 porated in the WETA-90 electrothermal atomizer signifi- cantly improved the performance and ease of manipulation of the system.It seems to be one of the few tube atomizers to date which features simultaneously homogeneous surface temperature distribution along the tube and over the circumference of the tube spatial and temporal isothermal- ity almost total absence of memory effects for carbide- forming elements high sensitivity and considerable free- dom from matrix interferences. The real value of this system will probably be recognized when electrothermal atomic absorption spectrometric assemblies in which tung- sten and graphite tubes are easily interchangeable become available and instruments for use with atomic absorption with a fast response have been developed. Two prototypes of the WETA-90 are currently being tested one in this laboratory and the other in the applica- tion laboratory of Bodenseewerk Perkin-Elmer Uberlingen Germany.The capability of the WETA-90 to solve practical analytical problems will be discussed in future papers. References Sychra V. Kolihova D. VyskoEilova O. HlavaE R.. and Puschei P. Anal Chim. Acta 1979 105 263. VyskoEilova O. Sychra V. Kolihova D. and Puschel P. .4nal. Chim. Acta 1979 105 271. Puschel P. Formanek Z. HlavaE R. Kolihova D. and Sychra V. Anal. Chim. Acta 1981 127 109. Sychra V. Kolihova D. HlavaE R. Doleial J. Piischel P. and Formanek Z. in Wissenschaftliche Beitrage 'Analytiktrqf- .fen 1982' K.M.U. Leipzig 1983 p. 154. HlavaE R. Ph.D. Thesis Prague Institute of Chemical Technology 1986. 6 7 8 9 10 11 12 13 14 15 16 17 18 19 Ortner H.M. Birzer W. Welz B. Schlemmer G. Curtius A. J. Wegscheider W. and Sychra V. Fresenius 2. Anal. Chem. 1986,323 68 1. Sychra V. Kolihova D. HlavaE R. Doleial J. VyskoEilova O. Puschel P. Formanek Z. and Ortner H. M. paper presented to the X Conference on Analytical Atomic Spectros- copy Torufi Poland 1988. Chakrabarti C. L. Delgado A. H. Chang S. B. Falk H. Huton T. J. Runde G. Sychra V. and Doleial J. Spectro- chim. Acta Part B 1986 41 1075. Chakrabarti C. L. Delgado A. H. Chang S. B. Falk H. Sychra V. and Doleial J. Spectrochim. Acta Part B 1989 44 209. Komarek J. and Ganoszy M. Collect. Czech. Chem. Com- mun. 1991 56 764. Koiuinikova J. Chem. Listy 1984 78 1209. Zemberyova M. Ph.D. Thesis Comenius University Bratis- lava Czechoslovakia 1985. Krakovska E. J. Anal. At. Spectrom. 1990 5 205. Puschel P. Patent No. B1 174 728 Czechoslovakia 1978. Ortner H. M. Wilhartitz P. Doleial J. HlavaE R. Sychra V. and Puschel P. Patent Application No. PV-75-89 Czecho- slovakia 1989. Lawson S. R. Dewalt F. G. and Woodriff R. Prog. Anal. At. Spectrosc. 1983 6 1. L'vov B. V. J. Anal. At. Spectrom. 1988 3 9. L'vov B. V. Nikolaev V. G. Norman E. A. Polzik L. K. and Mojica M. Spectrochim. Acta Part B 1986 41 1043. Grobenski Z. paper presented at the X Conference on Analytical Atomic Spectroscopy Toruii Poland 1988. Paper 0/05615D Received December I4th I990 Accepted May 16th 1991