138 Analyst, February, 1983, Vol. 108, $p. 138-144 Glassy Carbon Tubes in Electrothermal Atomisation - Atomic-absorption Spectrometry Leo de Galan, Margaretha T. C. de Loos-Vollebregt and Rene A. M. Oosterling Laboratorium UOOY Analytische Chemie, Technische Hogeschool, Jaffalaan 9, 2628 BX Delft, The Netherlands Preliminary results are presented on the use of glassy carbon tubes of standard dimensions for electrothermal atomisation - atomic-absorption spectrometry. In comparison with pyrolytically coated graphite tubes, the glassy carbon tubes show similar limits of detection provided that the same atomisation temperature can be realised. Glassy carbon tubes promise the following advantages : more uniform heating ; resistance against chemical attack ; a life- time of several thousands firings ; and a remarkably constant sensitivity.Keywords : Glassy carbon ; electrothermal atomasation ; atowic-absorption spectrometry The impressive performance of electrothermal atomisation (ETA) - atomic-absorption spectro- metry (AAS) for trace analysis relies to a large extent on the material from which the atomiser tube is fabricated. Traditionally, graphite has been and still is widely used, because it is strong, easily machinable and chemically pure, it has good electrical and thermal conductivity and it can be heated to 3300 K. A disadvantage of graphite is its permeability and affinity for chemicals. This property is exploited in adsorption columns, but forms a serious source of interferences in ETA - AAS, as has been documented in a recent and extensive review by Slavin and 1Manning.l The porosity of graphite can be reduced by the application of protec- tive coatings.Pyrolytic coatings, originally applied iiz sit% but nowadays supplied by the manufacturer, have shown more promise. Rapid deterioration of the pyrolytic coating and the graphite substrate under chemical attack from perchloric acid and ammonium nitrate has been ~bserved,~ but appears to have been overcome by improvements in the coating t e c h n i q ~ e . ~ Better resistance to chemical attack may be expected when solid pyrolytic graphite is used. This is confirmed by recent work on interference reduction through the use of platforms made from pyrolytic g r a ~ h i t e . ~ Complete ETA tubes have also been made entirely from pyrolytic graphite, but as a result of the anisotropy of the material only in experiments with very fast heating.6 However, there is an alternative carbonaceous material that is isotropic and lends itself easily to the fabrication of complete tubes.This is called glassy carbon or vitreous carbon. The material derives its name from its glassy surface, its strength and hardness, its impermea- bility to gases and its resistance to chemical attack. The latter two properties combined with acceptable electrical and thermal conductivity warrant exploration of glassy carbon as ETA material. The preparation of glassy carbon for use as ETA tubes has been described in a patent.' Tubes of standard dimensions can be obtained from various manufacturers and chemical purity, which was insufficient some years ago,8 has been improved lately.The review by Van der Linden and Diekerg is valuable for the information it contains of the material itself. Only a few references have been found to the application of glassy carbon in ETA - AAS. The potential utility was demonstrated by Yanagisawa and Takeuchi,lo who used a strip of glassy carbon enclosed by a Pyrex dome to observe signals from copper and strontium. In a later study by the same group,ll a glassy carbon tube was used to measure degrees of atomisation in ETA - AAS, but the analytical potential was not discussed. In the meantime, Volland et aZ.12 investigated different carbonaceous tube materials and found that their glassy carbon tubes were thermally too unstable to last more than a few firings.Prech and co-workers, who like Yanagisawa and Takeuchi used Tokai material, have been more successful. In a largely fundamental study on the atomisation of phosphorus the use of a glassy carbon tube was briefly mentioned.13 In another, more detailed study on arsenic Metal-lined tubes2 are restricted in upper temperature. Glassy carbon is widely used in electrochemical analysis.DE GALAN, DE LOOS-VOLLEBREGT AND OOSTERLING 1 39 some interesting differences between glassy carbon and ordinary graphite were observed.14 In glassy carbon arsenic is released at a lower temperature, to a large extent as molecules, so that the analytical sensitivity is lower than in graphite tubes. On the other hand, the atomic-absorption peak is much more symmetrical in glassy carbon, because volatilised arsenic is not re-deposited, as is confirmed by radioactivity measurements.In the absence of contrary information it seems that in all of these s t ~ d i e s l ~ - ~ ~ the graphite and the glassy carbon tube have mutually similar dimensions and heating characteristics. In view of the different electrical and thermal properties of the two materials, this is surprising and raises questions as to the nature of the glassy carbon used. It appears that glassy carbon is a potentially highly suitable material for electrothermal atomisers, but has not yet been thoroughly investigated. Therefore, we have started a systematic study of glassy carbon for ETA - -4AS. Some preliminary results are reported in this paper. Experimental Glassy carbon tubes of standard dimensions (length 28 mm, 0.d.8 mm, i.d. 6 mm) were obtained from Le Carbone-Lorraine (Paris, France), Tokai (Tokyo, Japan) and Sigri Elektro- graphit GmbH (Meitingen, Germany). Holes were drilled for sample introduction and the ends were machined to fit the tubes into a standard HGA 500 housing. All experiments were performed with Perkin-Elmer equipment, i.e., a Model 5000 atomic-absorption spectro- meter, a Model 500 power supply for ETA and a Model AS 40 autosampler. Data were fed into a Model 3500 Data Station, modified by us to display both the net signal and the back- ground absorbance. For comparison, standard Perkin-Elmer pyrolytically coated tubes were used with a temperature programme taken from the manufacturer's manual. Unless otherwise stated maximum power heating was used with either tube.Standard chemicals of analytical- reagent grade were introduced as aqueous solutions in a volume of 20 pl. Over a period of 3 months most experiments were carried out with a single tube from Le Carbone-Lorraine. So far only a few data have been obtained with the Tokai and Sigri tubes. Hard-copy output was recorded with an HP 7225 A plotter. Results and Discussion Heating Characteristics of Glassy Carbon Some relevant properties of different brands of glassy carbon are compared with time of ordinary graphite in Table I. The mechanical properties are largely similar, i . c . , a density of 1.5 g ~ m - ~ , a strength of 500 N cm-2 and a thermal expansion coefficient of 2 x 10k6 K The glassy carbon described in tlie patent' still has a high porosity and significant gas permeability.However, after thermal treatment, as has been done for the other two entries in Table I, glassy carbon has a much lower porosit; and gas permeability than ordinary graphite. Finally, glassy carbon also differs from ordinary graphite in having a lower thermal conductivity and a higher electrical resistivity. For electrothermal atoinisers of equal dimensions this leads to different heating characteristics. The data in Table I refer to room temperature. IVhereas tlie electrical resistivity o f ordinary graphite is virtually independent of temperature, Fig. 1 shows that the resistivity of glasI;. carbon decreases with increasing temperature. In fact, at common atomisation temperatures the resistivity may come even closer to that of graphite.This behaviour has several consequences. TABLE I PROPERTIES OF GLASSY CARBON Gas Thermal Electrical Deiisitv/ Porosity, permeability/ conductivity/ resistivity/ 0 cn12 s ' \\' c111-l K 10 12 cni l<eference hhterial g c111-~ 0' Graphite . . .. 1.7 -20 0.1-1 0 1 0.6-1 15 . . 1-1 4 -50 10-3-10-7 0 . 1 1 0 7 . . 1.5 0 35 -10-1' 0 0s 5 16 . . 1.45 0.3 10-7-10 9 0.1 3.5 17 Glassy carbon140 DE GALAN et a2. : GLASSY CARBON TUBES Analyst, Vol. 108 With increasing temperature the local resistance decreases and heat dissipation becomes more pronounced at the ends. The low thermal conductivity reduces the heat losses to the cooled electrodes. As a result, the combination of a low thermal conductivity and a negative temperature coefficient of the electrical resistance may produce a more uniform heating of the glassy carbon tubes.When power is supplied to the tube, it will initially heat up in the centre. Le Carbone-Lorraine 4- m ll Graphite G I , , I 1000 2 000 3 000 0 Tern peratu re/K Fig. 1. Variation of electrical resistivity of glassy carbon and graphite with temperature, from information supplied by the manufacturers.15-1’ Although this assumption has not yet been verified by actual temperature measurements along the tube, it might explain the absence of lead deposits at the ends of the tube, a phenomenon that is frequently observed with graphite tubes. Of course, the smooth and impenetrable surface of the glassy carbon also minimises the formation of deposit^.'^ For voltage-controlled power supplies the comparatively high resistance of glassy carbon predicts a relatively low heating rate, at least initially.Indeed, when the ashing step is deleted from the temperature programme, a slow initial heating is readily observed. With increasing tube temperature the decreasing resistance will enhance the heating rate. Conse- quently, when an ashing step is included] as is customary in ETA - AAS analysis, the effect of slower heating during atomisation will be small. According to the manufacturer’s specification, glassy carbon, grade V 25, from Le Carbone- Lorraine can be used up to 2800 K, whereas the grade GC-30s Tokai material can be heated to 3300 K. No upper limit is specified for the Sigri tube, but after a few runs with maximum power (i.e., 2500 K, see below), this tube was destroyed.In this study very high tempera- tures were not realised. The HGA 500 power supply is limited to a maximum voltage of 7 V, designed to heat a graphite tube to 3300 K. The higher resistance of the glassy carbon tube lowers the power dissipation and hence the final tube temperature. An estimate of the final temperature of the glassy carbon tube was obtained with the facilities of the HGA 500. The voltage control corresponds to a temperature scale calibrated for graphite. With the maximum power mode the voltage is regulated by the signal from a photodiode focused at the ETA tube. The surface of well used glassy carbon tubes looks similar to graphite tubes. Therefore, for the present, we assume that glassy carbon and graphite have equal emissivity.I t is now possible to draw up a temperature scale for glassy carbon. For example, with a graphite tube installed let the photodiode be adjusted to have the power supply heat the graphite tube to 2300 K. Next, the graphite tube is replaced by a glassy carbon tube and the voltage is increased until the photodiode lights up again. At this voltage (corresponding to 3000 K on the graphite scale) we know that the glassy carbon tube has reached a tempera- ture of 2300 K. The curve was confirmed by direct temperature measurements.18 Obviously, for a certain setting of the HGA 500 power supply graphite is heated to a higher temperature than glassy carbon, in agreement with the differences in electrical resistivity. The maximum temperature reached for glassy carbon is In this way, the complete graph in Fig.2 was constructed.February, 1983 I N ELECTROTHERMAL ATOMISATION AAS 141 2500 K. As has been noted in the introduction, previous reports on the use of glassy carbon tubes in ETA - AAS1°-14 devote little attention to the heating characteristics of glassy carbon. However, recently Styris and Kayelg published a heating curve for a vitreous carbon tube with a wall thickness of 2 mm and a length of 39 mm using another commercial power supply. The heating curve rises fairly slowly and levels off at 2500 K. There are two ways to reach higher temperatures: one is to decrease the resistance of the glassy carbon tube by increasing the wall thickness and the other, more attractive, possibility is to increase the upper voltage of the power supply to, e.g., 15 V.3000 I CJ I I I Fig. 2. Temperature reached with glassy carbon and graphite tubes of equal dimensions for the same setting of the HGA 500 power supply. Analytical Results I t can be seen that during the first ten firings the peak height and hence the analytical sensi- tivity increase steadily and then become constant. This phenomenon was observed each day. Possibly the structure of glassy carbon is slightly modified during the first few firings and returns to its original state slowly overnight. Curiously, it is necessary to execute a sequence of ten (blank) firings. Simply heating at maximum power for an equivalent amount of time is not sufficient. However, with this daily pre-treatment the sensitivity remains remarkably constant.Over a period of 3 months and a succession of more than 2500 firings the sensitivity remained constant to within 5%. The glassy carbon tubes from Le Carbone-Lorraine, Tokai and Sigri show equal sensitivity for the determination of cadmium. Moreover, the same sensitivity has been obtained with two different tubes from Le Carbone-Lorraine. When inspected visually, no degradation of the inner surface of the tubes is observed. In contrast, the outside of the tube reveals some Fig. 3 presents the peak height of the cadmium signal observed in successive firings. The reasons for this behaviour are not clear. U 0 0.4 4 z 2 0.2 Y I 1 500 2 500 a O ' 2 4 6 8 1 8 1 0 0 500 ' Number of firings Fig. 3. Variation of analytical sensitivity with number The measure- of firings of a single glassy carbon tube.ments extended over 3 months.142 A~zal_vst, VoL. 108 deposit of carbon dust. So far, we have not been able to determine the lifetime of a glassy carbon tube. The specimen used for the data in Fig. 3 was destroyed when the plastic tube alignment tool was accidently left in the sample injection hole during atomisation. However, over this period the tube was subjected to heavily salted and strongly acidic solutions (1 M perchloric acid). Although chemical interferences were noted, the tube did not suffer and the sensitivity was restored when aqueous solutions were analysed again. These observations demonstrate the resistance of glassy carbon to chemical attack. Sensitivity data for a number of elements are presented in Table I1 for pyrolytically coated graphite tubes and Le Carbone-Lorraine glassy carbon tubes. A few elements were also run on glassy carbon tubes from Tokai and Sigri.No significant variations in sensitivity were observed between tubes from different manufacturers. The results obtained with the pyrolytically coated tubes using the temperature programme advised in the Perkin-Elmer manual were generally and sometimes significantly better than the sensitivity checks provided in that manual. DE GALAN et al. : GLASSY CARBON TUBES The data in Fig. 3 refer to pure aqueous solution. This demonstrates that the instrument is functioning properly. TABLE I1 CHARACTERISTIC CONCENTRATIONS (pg 1-1 FOR 0.0044 ABSORBANCE UNIT) I N GLASSY CARBON AND PYROLYTICALLY COATED TUBES Wavelength/ Elemcnt nm Ag .. . . 328.1 A1 . . . . 309.3 Bi . . . . 223.0 Cd . . . . 228.8 co . . . . 240.7 Cr . . . . 357.9 c u . . . . 324.7 Fe . . . . 248.3 Mn . . . . 279.5 Ni . . . . 232.0 Pb . . . . 283.3 Sb . . . . 217.6 Si . . . . 251.6 sn . . . . 224.6 Ti . . . . 365.3 v . . . . 318.4 Pyrolytically coated graphite T a r a c t c r i s t i c ’ Tat concentration 1500 0.03 2 700 0.5 1400 1 1500 0.1 2 500 0.25 3 000 0.15 2 500 0.4 2 300 0.4 2 200 0.1 2 600 0.3 2 300 1 2 300 1.1 2 300 4.5 3 000 4.5 2 850 2.2 3 000 57 Glassy carbon r-----Characteristicl T*at concentration 1850 0.07 2 500 0.8 1850 2 2 350 0.1 2 500 0.7 2 500 0.7 2 500 0.9 2 500 1.3 2 500 0.3 2 500 5 2 500 1 2 500 1.2 2 500 3 2 500 2.2 2 500 200 2 500 11 * Limited to 2500 K (see Fig.2). 7 Ordinary graphite. When the data for glassy carbon are considered, two general conclusions can be drawn. Those elements which atomise at a relatively low temperature (c.R., Cd, Mn, Sb) yield sensi- tivities that are comparable to the values observed with pyrolytically coated graphite tubes. Differences of up to a factor of two in favour of pyrographite may be attributed to the slower heating of glassy carbon, especially at low atomisation temperatures. On the other hand, and not surprisingly, elements that require atomisation temperatures over 2500 K (e.g., Ni, Ti) yield appreciably poorer sensitivity with the glassy carbon tubes as used in this investigation. For some of these elements (Cu, V, Ti) incomplete volatilisation of the analyte was also apparent from memory effects observed in consecutive blank firings. Significant improvement may be expected when the atomisation temperature for glassy carbon tubes can be increased over its present upper limit of 2500 K as discussed above.A pronounced influence of the tube material is also apparent from the differences in peak shape observed between pyrographite and glassy carbon. Some examples are shown in Fig. 4. These signals were obtained with the atomisation temperatures indicated (cf., Table 11) and following a brief ashing step at 800 K to enhance the heating rate of the glassy carbon tube. For the 28 mm long tubes used in this study the removal of the analyte from the atorniser is probably fast enough to identify the peak shapes as the analyte introductionFebruary, 1983 I N ELECTROTHERMAL ATOMISATION AAS 143 0.4 0.2 0 0.4 0.2 $ 0 10 p.p.b.Cd Glassy carbon 2350 K Pyrog raphite 1500 K 1 1 20 p.p.b. Cu 0.2 I Glassy carbon I 2500 K Pyrographite 2500 K 0.1 I I 1 2 0 1 2 3 0 1 2 3 4 5 L 0 2 0.6 a 0.3 0 0.6 0.3 0 *O0 P.Pmb* si Glassy carbon Pyrographite Pyrographite 3000 K 2300 K 0 1 2 0 1 2 3 4 5 Time/s Fig. 4. Examples of atomic-absorption signals measured in glassy carbon and pyro- lytically coated graphite tubes. Atomisation temperatures are indicated. function.20 The peak shape is then determined by a combination of analyte release from the tube wall and atomisation reactions in the gas phase. The rapid release of the very volatile cadmium reflects the non-adhesive character of glassy carbon, even more so than pyrolytically coated graphite.For the less volatile antimony this effect is not observed, probably as a result of the lower heating rate of glassy carbon. A t the present stage of investigation it is too early for a fruitful discussion of peak shapes, because for most elements the temperature programme is not optimal. This is certainly so for copper, where observed memory effects stress the need for higher atomisation temperatures. Some preliminary experiments on matrix effects demonstrated the presence of chemical interferences. However, a closer investigation has been postponed until the glassy carbon tubes can be heated to optimuin temperatures. A blank signal for silicon was observed with the Sigri tube. According to the manufacturer's specification ,17 the Tokai tubes contain aluminium, iron and silicon at the parts per million level and copper, magnesium and boron at sub-parts per million trace levels.No blank signal was observed for copper, but aluminium, iron and silicon gave strong blank signals that persisted even after several hundred firings: aluminium, 0.26; iron, 1.5; silicon, 0.1 absorbance unit. In contrast, the glassy carbon from Le Carbone-Lorraine is remarkably pure. For none of the elements listed in Table I1 was a blank signal observed with tubes from Le Carbone-Lorraine. The purity of the glassy carbon materials deserves attention.8 Conclusion The results obtained so far have shown that glassy carbon is a very interesting material for Tubes of customary dimensions electrothermal atomisers in atomic-absorption spectrometry.can be obtained from several sources and at a similar cost to graphite tubes.144 DE GALAN, DE LOOS-VOLLEBREGT AND OOSTERLING The full potential of glassy carbon can be appreciated only when the tube dimensions or, preferably, the ETA power supply has been adapted to the different electrical and thermal properties of glassy carbon. With such modifications it should be possible to reach tube temperatures of 3300 K relatively fast. Peak shapes and detection limits might then be comparable to or better than those observed in ordinary or pyrolytically coated tubes. It would then be possible to exploit the major advantages of glassy carbon observed so far, i.e.. the great resistance to chemical attack, the extremely long lifetime of the tube and, probably most important, the constancy of the analytical sensitivity during the tube’s lifetime.The financial support of one of the authors (M.C.T. de L.-V.) and the instrumentation put at our disposal by Perkin-Elmer, Norwalk, CT, USA, are gratefully acknowledged. Dr. W. Slavin of the same company verified the temperature dependence of the glassy carbon tube shown in Fig. 2. Prof. A. Cedergren, Dr. W. Frech and Dr. E. Lundberg of the Depart- ment of Analytical Chemistry, University of Umei, Sweden, are thanked for helpful dis- cussions and comments. Useful information on the electrothermal properties of glassy carbon was provided by ing. J. W. M. van Uffelen from the Department of Electrical Engineering, Technische Hogeschool, Delft . 1. 2.3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. References Slavin, W., and Manning, D. C., Prog. Anal. At. Spectrosc., 1982, 5, 243. L’vov, B. V., and Pelieva, L. A., Can. J . Spectrosc., 1978, 23, 1 . Montgomery, J. R . , and Peterson, G. N., Anal. Chim. Acta, 1980, 117, 397. Slavin, W., Manning, D. C., and Carnrick, G. I<., Anal. Chem., 1981, 53, 1504. Slavin, W., Carnrick, G. R., and Manning, D. C., Anal. Chem., 1982, 54, 621. Chakrabarti, C. L., Wan, C. C., Hamed, H. A . , and Bertels, P. C., Anal. Chern., 1981, 53, 444. Knippenberg, W. F., Lersmacher, B., Lydtin, H., and Schelhas, K., Offenlegungsschrift N E 2702189, Frech, W., personal communication. Van der Linden, W. E., and Dieker, J . W., Anal. Chim. Acta, 1980, 119, 1 . Yanagisawa, M., and Takeuchi, T., Anal. Chim. Acta, 1974, 23, 364. Kitagawa, K., Ide, Y . , and Takeuchi, T., Anal. Chim. Acta, 1980, 113, 21. Volland, G., Kolblin, G., Tschopel, P., and Tolg, G., 2. Anal. Chem., 1977, 284, 1 . Persson, J. A., and Frech, W., Anal. Chim. Acta, 1980, 119, 75. KoreEkova, J., Frech, W., Lundberg, E., Persson, J . A., and Cedergren, A., Anal. Chim. Acta, 1981, Manufacturer’s information, Ringsdorf, Bonn-Had Godesberg, Gcrniany, 1982. Manuiacturer’s information, Le Carbone-Lorraine, Paris, France, 1982. Manufacturer’s information, grade GC-30 material, Tokai, Tokyo, Japan, 1982. Slavin, W., and Manning, D. C . , personal communication. Styris, D. L., and Kaye, J. H., Anal. Chem., 1982, 54, 864. Van den Broek, W. M. G. T., and de Galan, L., Anal. Chem., 1977, 49, 2176. BRD, 1978; US Pat., 4 204 769, 1980. 130, 267. Received July 14th, 1982 Accepted September 20th, 1982