JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY NOVEMBER 1994 VOL. 9 1273 Packed Glassy Carbon Tube Atomizer for Direct Determinations by Atomic Absorption Spectrometry Free from Background Absorption* Kuniyuki Kitagawa,t Masahide Ohta Tetsuya Kaneko and Shin Tsuge Department of Applied Chemistry School of Engineering Nagoya University Furo-cho Chikusa-ku Nagoya 464-01 Japan An electrothermal atomizer consisting of a long glassy carbon tube packed with activated charcoal was evaluated for the direct determination of trace elements in biological materials by atomic absorption spectrometry in the absence of any background absorption. The maximum temperature of the atomizer of 2280°C was maintained continuously by an electric power supply of about 7 kW. Under such isothermal conditions few chemical interferences and high accuracy (relative error < 8%) and precision (RSO < 3.5%) were obtained in the direct determination of trace elements (Cu Mg and Mn) in National Institute of Standards and Technology biological Standard Reference Materials by solid sampling atomic absorption spectrometry. Keywords Electrothermal atomizer; solid sampling; biological material; atomic absorption spectrometry; background absorption Several electrothermal atomizers with the ability to separate analytes have been reported.'-'' Separation of atomic vapours had been reported in earlier studies on super high temperature gas chromatography by Sokolov and c o - ~ o r k e r s .~ ~ - ' ~ As in gas chromatography in general separative atomizers can be categorized into two types open tubular at~mizersl-~ and packed column atomizer^.^-'^ The former type has been applied to the separation of atomic vapours which would reduce inter-element effects.Conversely the latter type has been mainly applied to the removal or separation of non- atomic species from the atomic vapours which overcomes the problems associated with background absorption in the direct analysis of biological samples using electrothermal atomization. An open tubular molybdenum atomizer operating at a high temperature of up to 1800°C has been applied to the sequential separation of metal vapours in the analysis of National Institute of Standards and Technology (NIST) biological Standard Reference Materials ( SRMS).~ However a charring stage is indispensable when using this type of atomizer in order to remove the organic matrices.Therefore the charring temperature and time must be optimized. This is also the case for direct analyses using conventional electro- thermal atomizers. On the other hand separative column atomizers (the packed do not require the inclusion of a charring stage. However they have a limitation as to the temperature attainable which is below about 1400 "C since an alumina tube is used for the column. More recently a molybdenum tube atomizer packed with tungsten powder has been applied to the separation of atomic vapours at tempera- tures of up to 1820°C.'5 In the present study a glassy carbon tube (Tokai Carbon GC30) was employed as the heating material so as to obtain higher temperatures above 2000 "C.Glassy carbon is less permeable than graphite and does not form carbides. For this reason a glassy carbon tube has been used to determine the diffusion coefficients of metal vapours." From the direct analysis of NIST biological SRMs the accuracy of the proposed procedure was evaluated in conjunction with a study of the capability of the glassy carbon tube atomizer packed with activated charcoal to remove background absorption. * Presented in part at the 54th Symposium of the Japanese Society t Present address Research Center for Advanced Energy for Analytical Chemistry Mito Ibaraki Japan ( 1993). Conversion Nagoya University Nagoya Japan. Experiment a1 Apparatus A schematic diagram of the glassy carbon tube atomizer developed for the present study is shown in Fig.1. A glassy carbon tube (Tokai Carbon GC30) of 6.5 mm i.d. 8.6 mm 0.d. and 20 cm in length (1) is supported vertically by ring (4) and cone ( 5 ) graphite electrodes which are fixed at upper and lower stainless-steel discs respectively. The glassy carbon tube Fig. 1 Schematic diagram of the glassy carbon tube atomizer 1 Tokai Carbon GC30 glassy carbon tube (20 cm x 6.5 mm i.d. x 8.6 mm 0.d.); 2 0.67 g of activated charcoal (30-60 mesh); 3 graphite stopper 1.5 mm in thickness with 10-20 holes of 0.8 mm in diameter; 4 graphite ring cut into two halves; 5 graphite cone; 6 water-cooled stainless- steel chamber 64 mm id.; 7 graphite sample cup (30 p1 inner volume); 8 tungsten wire of 0.8 mm diameter; 9 sample inlet; 10 stainless-steel plunger of 2 mm diameter; 11 and 12 water inlets; 13 and 14 argon inlets; 15 AAS signal observation holes of 2 mm diameter; 16 spring; 17 quartz window; 18 to monochromator; 19 to variable transformer; and 20 mica plate1274 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY NOVEMBER 1994 VOL. 9 is mounted in a cylindrical atomization chamber of stainless steel the wall of which (6) is water cooled (12).The glassy carbon tube is packed with 0.67g of activated charcoal of 30-60 mesh (2) the packing length being 45 mm. The bottom end of the packing is plugged with a graphite disc of 1.5 mm thickness (3) through which 10-20 holes with diameters 0.8 mm are drilled in order for the vapours eluting from the packing to pass through. A quarter of the way up from the bottom end of the glassy carbon tube two holes of diameter 1.5 mm are drilled to allow passage of the light beam in order for atomic absorption to take place.Argon purge gas flows outside the glassy carbon tube at a flow rate of 1 1 rnin-l. Argon carrier gas is forced to pass through the packing layer at a flow rate of 0.1-0.4 1 min-'. The glassy carbon tube is heated continuously by passage of a large alternating current of 220 A at a voltage of 32.5 V. A 30 pl graphite sample cup is suspended by a tungsten wire of 0.8 mm diameter (8) which is fitted to a stainless-steel plunger of 2 mm diameter (10). In order to avoid overheating the top of the glassy carbon tube and the bottom stainless- steel disc are water A conventional atomic absorption spectrometer was used for the atomic absorption spectrometric (AAS) measurements.A deuterium lamp was also used for monitoring background absorption. The radiation from the hollow cathode and deu- terium lamps were merged into a beam using a grid mirror. Phase rectification was employed for discrimination between the AAS and non-specific signals. The resulting intensity measurements were converted into absorbance by a microcom- puter programmed in a C-language (MSX C Version 2.1). The temperature of the glassy carbon surface was measured around the holes used for observation of the AAS signals by using a digital optical pyrometer. The instruments used for this work are listed in Table 1. Procedure Prior to measurement the glassy carbon tube atomizer was thermally equilibrated by passing current continuously for more than 15 min.After this has been done electrical control and switching for the drying and charring stages as is necessary for conventional electrothermal atomizers are not necessary. After removing the sample inlet screw (9) lop1 of a test solution were quickly added by a micropipette into the Table 1 Instrumentation Spectrometer Photomultiplier tube Power supply for lamps Electronics for signal processing Hollow cathode lamps Deuterium lamp Microcomputer Program Micropipette Shaker Optical pyrometer graphite cup suspended at the sample inlet position of the tube the screw was set the cup allowed to stand there for about 30 s in order to evaporate the solvent and then inserted into the hot region which is a few mm above the top of the packing layer by lowering the plunger.The drying process could be dispensed with for volumes of less than 10 pl. After 2500. 2410 2000 0 9 5 1500 2 !!? E r-" 1000 600 kd I t I I I I J 0.5 1 2 3 5 10 Electric power/kW Fig. 2 Dependence of the temperature (T) of the glassy carbon tube on the electric power (P) The temperature was measured around the AAS observation hole by a digital optical pyrometer O" 0 A Techtron AA4 Czerny-Turner mount Hamamatsu Photonics R106. 400-600 V f = 50 cm slit-widths 50-200 pm Laboratory-made current regulators 0 4 8 12 16 Time/s Fig. 3 Change in absorbance profile at 324.8 nm for Cu (100 ng) with carrier gas flow rate at STP (a) 130ml min-'; (b) 174 ml min-'; (c) 420 ml min-'; temperature of glassy carbon tube = 1985 "C Laboratory-made phase-sensitive rectifiers 3.6 I I Hitachi HLA-3 (Cu Mn Mg Cd and Zn) Hamamatsu Photonics L233 (Pb) current 2-8 mA Hamamatsu Photonics current d 12 mA Matsushita FS4600F fitted with a hard disk drive (Logitec SHD-40J) and a laboratory- made analogue-to-digital converter card (12 bits) designed by the electronics workshop by the Middle Branch of the Japan Society for Analytical Chemistry Version 1.2 (Japan ASCII) Laboratory prepared written in MSX-C Gilson digital micropipette Yamato MT-11 Minolta TR-630 2.1 x 2.0 2 a 1.5 2 1.0 in Y 2 I] a 1 0 100 200 300 400 500 Carrier gas flow rate/ml min-' Fig.4 Dependence of the absorbance peak height and area on carrier gas flow rate 0 peak area; 0 peak height. The conditions are the same as in Fig. 3JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY NOVEMBER 1994 VOL.9 1275 atomization the next run could be carried out after cooling the sample cup for about 2 min at the sample inlet port. A slurry method was used for sampling powders. An aliquot of NIST SRM 1571 Orchard Leaves or NIST SRM 1577 Bovine Liver was dispersed using a mechanical vibrator in 1 5 or 10 ml of a 0.1 moll-' HN03. The resulting suspension or slurry was taken into the graphite cup using the micropip- ette. In order to allow larger sample particles to be taken into the tip with good reproducibility the top of the micropipette tip was cut off to form a larger entrance bore of = 1.5 mm in diameter. The atomization procedure for the slurry samples was the same as that for the solution samples. Stock standard solutions of 1000 mg 1-l of Cu Mg and Mn were prepared by dissolving the pure metals in HN03 and those of Cd Pb and Zn were commercially available standard solutions (Wako Pure Chemicals). Test solutions were prepared by adjusting the acidity to 0.1 moll-' HNO just before use.All reagents used were of analytical-reagent grade. Results and Discussion Electric Power The electro-thermal performance of the glassy carbon tube atomizer developed was estimated by the following measure- ments. The dependence of the temperature on electric power is shown in Fig. 2. The slope of the rectilinear curve obtained on a logarithmic scale was found to be This suggests that radiative heat loss from the glassy carbon plays a major role in determining the temperature the slope should be approxi- mately % based on the Stefan-Boltzmann law if radiation alone regulates the temperature.The maximum temperature of 2280°C was obtained with a power of 7 kW the limit of the power supply. From Fig. 2 a temperature above 2400 "C which would be sufficiently high for atomization of most elements could be expected to be attained at 10 kW. In the present work the inner wall of the atomization chamber was not polished to a mirror finish however a temperature above 2400°C could expected to be obtained at a lower power after such a modification. 0.4 0.2 a 0 5 0 f! 2 0.4 0.2 Optimization of Carrier Gas Flow Rate The change in the profile of the atomic absorption peak for Cu uersus the carrier gas flow rate is demonstrated in Fig. 3 and the dependence of the peak height and area on the carrier gas flow rate in Fig.4. The peak absorbance becomes smaller at lower carrier gas flow rates (Fig. 4). This could be due to the interaction of the analyte vapour with the packing layer leading to a decrease in the recovery of the analyte atoms. With an increase in the carrier gas flow rate the peak absorbance increases and shows a roll-over effect. At the higher flow rates the residence time of the atoms becomes shorter leading to a decrease in the absorbance. Accordingly a carrier gas flow rate of around 170 ml min-' was considered to be optimum. However the optimum flow rate will change with the atomizer temperature and the type of analyte atoms. As seen in Fig. 4 the absorbance in terms of peak area also decreases in a fashion similar to that for the peak absorbance.This suggests that the optical thickness of the absorption path of 6.5 mm (15 in Fig. 1) is insufficient. Capability of Removing Background Absorption Typical absorbance peak profiles which were obtained with the glassy carbon tube atomizers in the presence and absence of the packing material heated at a relatively low temperature are shown in Fig. 5. The test sample was a solution of Cd to which a rotary vacuum pump oil and a liquid synthetic detergent (Natera Lion Oils and Fats) had been added as a matrix to cause background absorption and as a surfactant (and which also causes background absorption) respectively. Strong background absorption is observed for the open tubular atomizer [Fig. 5(c)]. On the other hand background absorp- tion is drastically decreased for the packed atomizer [Fig.5 ( d ) ] . Non-atomic species such as carbonaceous particulate matter resulting from the organic matrices are trapped as charcoal on the surface of the packing layer and any other eluting species are non-absorbing at the analytical wavelength. In addition the atomic absorption peak obtained with the packed atomizer [Fig. 5(b)] becomes greater than that obtained with the open tubular atomizer [Fig. 5(a)] the net area for the latter which is estimated by subtracting the 0 2 4 6 8 0 2 4 6 8 Ti me/s Fig. 5 Typical absorbance peak profiles (A=228.8 nm) with [(a) and (c)] open and [(b) and (d)] packed glassy carbon tube atomizers measured with (a) and (b) a hollow cathode lamp (atomic + background absorption); (c) and ( d ) a deuterium lamp (background absorption) Sample is 3 pl of 100 pg ml-' Cd solution + 5 pl of vacuum pump oil + 2 pl of synthetic detergent (Natera Lion Oils and Fats); carrier gas flow rate 240 ml min-' STP; temperature of glassy carbon tube = 1233 "C1276 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY NOVEMBER 1994 VOL.9 0 2 4 6 8 0 2 4 6 8 0 2 4 6 8 Time/s Fig. 6 Typical absorbance peak profiles obtained through repeated runs (a)-@) hollow cathode lamp; ( d ) - ( f ) deuterium lamp. The sample and conditions are the same as those in Fig. 5. The profiles were run in alphabetical order. Integrated absorbance 0.47 s for (a); 0.51 s for (b); 0.46 s for (c); 0.23 s for (d); 0.16 s for (e) and 0.09 s for (f) 2 4 6 8 0 2 4 6 8 l-ime/s Fig. 7 Typical absorbance peak profiles (1=217.5 nm) for Pb with [(a) and (c)] open and [(b) and (d)] packed glassy carbon tube atomizers measured with (a) and (b) a hollow cathode lamp (atomic + background absorption); (c) and ( d ) a deuterium lamp (background absorption).Sample is 3 ~1 of aqueous 20 pg ml-' Pb solution+ 5 pl of vacuum pump oil +2 pl of synthetic detergent (Natera Lion Oils and Fats); carrier gas flow rate 487 ml min-'; temperature of glossy carbon tube= 1534°C background absorption becomes smaller. This effect arises mainly because the degree of atomization becomes higher after the non-atomic species containing the analyte element comes into contact with the heated graphite packing which has a large surface area. An interesting phenomenon was observed for the open tubular glassy carbon tube atomizer.Typical peak profiles obtained through repeated measurements are shown in Fig. 6. The background absorption gradually decreases in magnitude run by run while the atomic absorption remains almost the same (see the values for integrated absorbances in the figure caption). This suggests that the inner wall of the glassy carbon tube gradually became coated with carbon resulting from the pyrolysis of the organic matrices. The coating layer of carbon acts as a packing layer decomposing non-atomic species such as carbonaceous particulate matter into carbon. This is ana- logous to wall-coated open tubular columns used for gas chromatography. Typical peak profiles of atomic and background absorption signals for a sample of Pb mixed with oil are shown in Fig.7. These were obtained with open tubular and packed glassy carbon tube atomizers heated to a moderately high temperature. The results are similar to those for Cd the background absorp- tion is removed by the packed atomizer. Robinson and co-workers have also pointed out the decrease in the back- ground absorption in the direct atomization of lead sulfate samples using a packed graphite a t ~ m i z e r . l ~ * ~ ~ However loss of atoms through leakage would be a potential problem in their atomizer since graphite is highly permeable. Glassy carbon is preferable for the atomizer material as it is much less porous. Peak profiles of typical atomic and background absorptions for Cu are shown in Fig. 8. These were obtained by direct 0.3 ( a ) 0.1 O.* I I 0 20 30 50 Time/s Fig.8 Typical absorbance peak profiles (3 = 324.7) for Cu on direct atomization of 0.25 mg NIST SRM 1577 Bovine Liver containing 100 ng of Cu with a packed glassy carbon tube atomizer with (a) a hollow cathode lamp; (b) a deuterium lamp; carrier gas flow rate 670 ml min-'; temperature of glassy carbon tube= 1750 "C atomization of NIST SRM 1577 Bovine Liver with the packed glassy carbon tube atomizer heated to 178OoC a higher temperature than that used for Pb. No background absorption exists [Fig. 8 ( b ) ] . In Fig. 9 are shown typical peak profiles for Mg. No background absorption occurs in the direct atomization of the NIST SRM 1571 Orchard Leaves with the packed glassy carbon tube atomizer [Fig. 9(e)]. Similar peak profiles wereJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY NOVEMBER 1994 VOL.9 1277 Fig. 9 Typical absorbance peak profiles (A=285.2 nm) for Mg on direct atomization of (a) and (d) aqueous solution (50 ng of Mg); (b) and (e) 8 pg of NIST SRM 1571 Orchard Leaves containing 49 ng of Mg using the packed glassy carbon tube atomizer; (c) and (f) 8 ng of NIST SRM 1571 Orchard Leaves containing 49 ng of mg using the open glassy carbon tube atomizer; (a)-(c) hollow cathode lamp; (d)-(f) deuterium lamp; carrier-gas flow rate 725 ml min-'; temperature of glassy carbon tube=2160 "C Table 2 Results for the NIST Standard Reference Materials Sample Element SRM 1571 Orchard Leaves* c u Mg Mn SRM 1577 Bovine Livert c u c u With standard Standard solution/ addition/ Certificate value/ CLgg-' Pg g-' Pg g-' 11.1 12.6 12+ 1 87.0 98.5 91 +9 138$ 193 f 10 150 0.64 0.59 0.62 & 0.02 - - * T > 2000 "C and RSD for measurement points < 3.5% n 2 3.t T= 1820°C and RSD for measurement points > 15% n 2 3. $ Peak height measurements (all other values from peak area measurements). observed for the test solution and the powder sample [Fig. 9(a) and (b)] although a slight distortion can be seen for the latter. Fig. 9(f) shows that the open tubular glassy carbon tube atomizer also has some ability to remove the background absorption as described for Cd. Peak broadening occurs in the packed atomizer [compare Fig. 9(b) and (c)]. This is mainly due to interaction of the analyte atoms with the packing layer. Thus from the point of view of peak broadening the open tubular glassy carbon atomizer would be preferable.Because of the smaller surface area for interaction however a longer tube length would be required for the open tubular column to remove the back- ground absorption when large amounts organic sample matrices are loaded. Direct Determinations of Cu Mg and Mn Analytical curves were prepared for standard solutions and for the NIST reference materials using the standard additions technique with the packed glassy carbon tube atomizer. For Cu a slight chemical interference was indicated by the differ- ence in the slope between the two types of calibration curves. High precision in terms of relative standard deviation (RSD) was obtained less than 3.5% RSD at most of the measurement points. The results are listed in Table 2.At a lower atomizer temperature of 1820"C the relative error exceeds 22% and the RSD is greater than 15% n 2 3 for the direct determination of Cu in NIST SRM 1577 Bovine Liver. The recovery could be low owing to retention on the packing layer. At a temperature above 2000"C much better precision and accuracy were obtained for the direct determination of Cu Mg and Mn in the NIST Orchard Leaves the RSD being <3S% and the relative error <8%. The packed glassy carbon tube atomizer when operated at a high temperature of above 2000°C proved to have the capability of removing background absorption leading to high precision and accuracy in the direct determinations of trace elements in powder samples. The sensitivities for 1 % peak height absorption were 3.3 ng for Cu 1.0 ng for Mn and 0.53 ng for Mg.Further improvement is necessary to obtain sufficient sensitivity for determinations at the pg kg-' level. The lifetime of the glassy carbon tube was longer than 200 h when used at temperatures of above 2000°C. The authors thank S. Takahashi and K. Tachibana for manufacturing the complicated atomization chamber and T. Watanabe and T. Imura for skillful glass-blowing. 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 References Ohta K. Smith B. W. and Winefordner J. D. Spectrochim. Acta Part B 1982 37 343. Ohta K. Smith B. W. and Winefordner J. D. Anal. Chem. 1982 54 320. Ohta K. Smith B. W. and Winefordner J. D. Microchem. J. 1985 32 50. Ohta K. Yamanaka N. Inui S. Winefordner J. D. and Mizuno T. Talanta 1992 39 1643. Yanagisawa M. Kitagawa K. and Tsuge S. Spectrochim. Acta Part B 1982 37 493. Yanagisawa M. Suzuki H. Kitagawa K. and Tsuge S. Spectrochim. Acta Part B 1983 38 1143. Kitagawa K. Takeuchi T. and Yanagisawa M. Anal. Sci. 1989 5 445. Kitagawa K. Mizutani A. and Yanagisawa M. Anal. Sci. 1989 5 539. Yanagisawa M. Ida K. and Kitagawa K. Anal. Sci. 1989,5,765. Yanagisawa M. Katoh K. and Kitagawa K. Anal. Sci. 1990 6 471. Kitagawa K. and Takeuchi T. Anal. Chirn. Acta 1973 67 457. Sokolov D. N. Vakin N. A. and Kalmykov Yu. D. Zauod Lab. 1969 35 158. Sokolov D. N. and Vakin N. A. Zauod Lab. 1970 36 1314. Sokolov D. N. J. Chromatogr. 1970 47 320. Ohta K. Yamanaka N. and Inui S. Analyst 1993 118 1031. Robinson J. W. Slevin P. J. Hindman G. D. and Wolcott D. K. Anal. Chim. Acta 1972 61 431. Robinson J. W. Rhodes L. and Wolcott D. K. Anal. Chirn. Acta 1975 78 474. Paper 4102092H Received April 8 1994 Accepted June 17 1994