1138 Analyst, December, 1979, Vol. 104, pp. 1138-1150 Investigation of Atomiser Tube Design for Carbon Furnace Atomic-emission Spectrometry 0. Littlejohn" and J. M. Ottaway Department of Pure and Applied Chemistry, University #of Strathclyde, Cathedral Street, Glasgow, G1 1 X L Modifications to a standard graphite furnace tube, designed for atomic- absorption measurements, are shown to give iimproved performance for atomic-emission measurements of a number of elements. Four different modified tubes are compared with a standard Perkin-Elmer HGA-72 graphite tube with respect to their thermal characteristics and the detection limits of ten selected elements of varying atom-appearance temperatures. Different elements were found to give lowest detection liimits in tubes of different design.The results indicate the need for a new approach to tube design, specifically for atomic-emission measurements, in order that best detection limits can be achieved for all elements in a single graphite tube. Keywords ; Modified atomiser tubes ; atomic-emission spectrometry ; carbon furnace atomisation It has been suggested in previous publications1--6 that the application of commercial carbon furnace atomisers in atomic-emission spectrometry may, in certain analyses, be an attractive alternative to carbon furnace atomic-absorption spectrometry. The operation of the carbon furnace as an emission source facilitates the use of electrothermal atomisation in simul- taneous multi-element analysis.' However, a number of features of the design and operation of commercial atomisers are unsuitable for the measurement of optimum atomic-emission signals of many elements.With standard instrumentation, the maximum atomic-emission intensity occurs after the maximum atom concentration has been attained and when the tube temperature is still As atomisation and excitation occur in the same central section of the tube, the thermal energy available to populate energy levels depends greatly on the atom-appearance temperature. Maximum atom concentrations of volatile and medium-volatile elements therefore exist at temperatures much lower than the maximum available and it is often difficult to detect the atoinic emission of these elements in the presence of the rapidly increasing background signal. Carbon furnace atomic-emission spectrometry (CFAES) detection limits are dependent in part on the atomic emission to tube-background emission ratio.Both signals represent the summation of the photons emanating from each infinitely narrow cross-section of the tube vapour (analyte emission) and directly or indirectly, the tube surface (background emission). If a vapour phase temperature gradient exists along the tube, it is clear that the maximum atomic emission to background emission ratio and lowest detection limit will be achieved when the maximum possible concentration of atoms is formed in the hottest temperature zone, when that section of the tube has attained the maximum temperature available. This condition is generally not achieved with standard commercial furnaces but is fulfilled for L'vovlO and Woodriffll furnaces, and to an extent in T-shaped atomisers12 where the sample is introduced into a hot tube.To our knowledge, however, there are no reported analytical applications of these atomkers as emission sources, which is surprising considering the inherent advantages of the design of these furnaces. Although optimisation of the tube temperature enhances the detection of a number of elementsg the application of this procedure is limited to particular carbon furnace atomisers and is unsatisfactory for many volatile elements. To achieve universal improvements in the CFAES detection limits obtained with commercial atomisers, it appears necessary therefore to separate the processes of atomisation and excitation and to allow each to occur in separate sections of the tube assembly.I t would also be advantageous if the temperature and hence the background signal of the excitation zone were almost constant at the time of measure- * Present address : Imperial Chemical Industries Ltcl., Petrochemicals Division, P.O. Box 90, Wilton, Middlesbrough, Cleveland, TS6 8 JE.LITTLEJOHN AND OTTAWAY 1139 ment, as this would reduce errors introduced when subtracting the appropriate background intensity from the combined analyte plus background peak-height signal when the facility of automatic background correction is not available. Modifications to the design of a standard Perkin-Elmer HGA-72 atomiser tube, to allow more sensitive atomic-emission measurements, have been d e s ~ r i b e d . ~ ~ ~ ~ ~ ~ ~ Lower detection limits were obtained for a number of volatile elements by reducing the thickness of the carbon wall towards the ends of the tube, making the temperature of these end sections greater than the tube centre at any atomisation setting.13 Improvements were greatest for gallium, thallium and indium, but detection limits substantially lower than 1 pg ml-l for other elements such as lead, bismuth and cadmium were only achieved when a small sample cup was fitted to the centre of the modified tube a~semb1y.l~ An alternative modification that gave higher temperatures at the centre of the tube was observed to give sensitive emission signals only for elements of higher atom-appearance temperatures.8 The thermal and analytical parameters of these tube modifications were not considered in detail in the initial communications.A more rigorous investigation is therefore presented in this paper to establish the important aspects of tube design that influence the sensitivity of atomic-emission signals generated from a carbon furnace during sample volatilisation and atomisation. Five HGA-72 tube designs are considered and the analytical utility of each tube is assessed, where possible, by the measurement of tube wall and vapour temperatures and a consideration of the tube-temperature gradient. Detection limits for ten elements of varying volatility, namely lead, gallium, silver, tin, manganese, chromium, iron, titanium, molybdenum and gadolinium, are presented to indicate the useful range of application of each tube design. Experimental Instrument ation The instrument used for all measurements was a Perkin-Elmer HGA-72 carbon furnace atomiser mounted in a Perkin-Elmer 306 atomic-absorption/emission spectrophotometer and coupled to a Servoscribe RE541.20 potentiometric strip-chart recorder.The spectral band width of the spectrometer was set at 0.07, 0.14 or 0.2 nm as indicated in the text. The monochromator was adjusted to the required wavelength using the appropriate hollow- cathode lamp, which was then disconnected. If suitable lamps were not available, the monochromator was adjusted to the approximate wavelength and then peaked on the line using the atomic-emission signal obtained during the atomisation of a suitable standard solution. Sample solutions were transferred into the centre of the standard and modified HGA-72 tubes with 20- and 50-pl Oxford micropipettes.The solutions were dried a t 373 K for 40- 45 s and then at 400-900 K for a further 20 s, depending on the volatility of the element. The instrument settings that corresponded to these tube-wall temperatures varied with the design of the HGA-72 atomiser tube. All samples, however, were atomised a t maximum power (999 units) for 10 s unless otherwise stated, in an atmosphere of research-grade argon (99.996%). The interrupted gas flow facility (gas stop) was used to enhance atom-residence times and to allow a higher vapour temperature to be achieved during the lifetime of the atom population. The tube-wall temperatures achieved by the different tube designs during this atomisation sequence will be referred to in the text.Time-resolved atomic-emission signals were recorded on the strip-chart recorder set at a speed of 2 cm s-l, using a trigger mechanism that initiated the recorder motor a t the start of the HGA-72 atomisation stage. The Perkin-Elmer 306 spectrophotometer does not exhibit the facility of automatic wave- length modulation background correction. I t was necessary therefore to record first the combined analyte and background emission and then, on the same section of chart paper, the tube background signal only, for the same instrumental conditions. The background intensity at the time of the combined signal peak was then subtracted to give the net maxi- mum atomic-emission intensity. HGA-72 Atomiser Tube Design than most commercial atomisers.The Perkin-Elmer HGA-72 carbon furnace atomiser is more amenable to tube modification Different temperature gradients along the 53 mm long,1140 AnaZyst, VoZ. 104 8.6 mm i.d. tube can be created by reducing the 1.85 mm wall thickness a t a number of positions on the outside of the tube. In this work experiments were performed with the standard and four modified tubes. The “high-temperature’’ tube was prepared as described by Ottaway and Shaw,8 by removing 1 mm of carbon with a lathe from the outside of the standard tube up to 0.5 cm either side of the injection hole [Fig. 1 (A)]. The “tapered” tube was designed to reduce to a minimum the temperature gradient along the I-IGA-72 atomiser. Starting from a position 6mm either side of the central injection hole, the wall thickness was tapered towards the ends of the tube by removing up to 1 mm of ca.rbon over a 12-mm section on the external surface, as shown in Fig.1 (B). The initial “volatile-elements” tube, designed by Ottaway and Hutton,13 was prepared by removing 1.2 mrn (0.05 in) from a 12-mm (0.5-in) portion of the outside of the tube at each end, starting 6 nim (0.25 in) from the central injection hole. In this work however, only 0.75-1.0mm of carbon was removed from the end sections [Fig. 1 (C)]. The “cup” tube is an adaptation of the “volatile-elements” tube and has been described in detail in a recent Cornmuni~ation.~~ The bottom 5.0-mm section of a 9 mm long and 4.5 mm i.d. Varian CRA 90 atomiser cup was positioned 2.5 mm inside the volatile- elements tube through a 5.0 mm diameter hole cut in the centre of the tube beneath the injection hole.The cup was then fixed permanently in position in the modified tube by pyrolysis of the tube assembly at 2400 K for 10 min in a 5% methane - 95% argon gas atmosphere in the HGA-72 atomiser workhead [Fig. 1 (D)]. This pyrolysis procedure was also used to form a coating of pyrolytic graphite on the standard and modified tubes for the measurement of molybdenum, titanium and gadolinium emission signals. LITTLE JOHN AND OTTAWAY : INVESTIGATION OF ATOMISER TUBE Fig. 1. Modified HGA-72 carbon tubes used in this study: (A) the “high-temperature” tube; (B) the tapered tube; (C) the “volatile ele- ments” tube; and (D) the cup -tube. Temperature Measurements Tube-wall temperatures were measured with ;in Ircon, Series 1100, optical pyrometer as described previ0us1y.l~ The pyrometer was focused on the inside of the tube through the injection hole and the ends of the tube.The temperature during atomisation was recorded on a Servoscribe RE54130 chart recorder at a speed of 2 cm s-1. All vapour-phase tempera- tures were obtained from iron electronic excitation temperature measurements using the iron-atom lines at 370.56, 373.49, 373.71 and 392.29 nm using the procedure described in detail e1~ewhere.l~ Atomic-emission signals generated at these wavelengths during the atomisation of 20-4 aliquots of a 2.0 pg ml-l iron standard solution were recorded on the Servoscribe recorder at 2.0 cm s-l. Corrections for the variation in the spectral response of the spectrometer were applied to the emission intensities at various times during the atomisa- tion sequence and graphs of ln(lXij/giAij) against E , were constructed. From the slopes of -l/kT, the absolute temperature T can be obtained at each point in time.In this expression, I is the corrected emission intensity measured under conditions of negligible self-absorption at wavelengths Aij, g, is the statistical weight of the upper energy level i of energy E,, A i j is the transition probability and K is Boltzmann’s constant. Chemicals and Detection Limits Stock solutions of the required elements were prepared from reagents of the highest available purity by dissolving the appropriate atmount of a suitable salt in distilled water and nitric or sulphuric acid to give a final acid concentration of M.Working standards were prepared from these stock solutions when required.December, 1979 DESIGN FOR CARBON FURNACE ATOMIC-EMISSION SPECTROMETRY 1141 Detection limits were calculated from 7-10 injections of a standard solution giving an atomic emission to background emission ratio of between 1.0 and 2.0, and are defined as that concentration giving a signal above the background equal to twice the standard deviation of the maximum emission signals from the standard solution injections. Results and Discussion The modifications to the HGA-72 tube design described in this paper have been achieved by altering the thickness of the graphite wall at various positions along the tube. Before describing the effects that each of these alterations has on CFAES detection limits, it is of interest to note, for comparison, the thermal and analytical parameters of the standard tube design.Standard HGA-72 Tube The variation with time of the carbon wall temperature at the centre of a standard HGA-72 tube at a number of atomisation settings has been discussed in detail in a previous paper.g At each setting the vapour temperature is controlled by the temperature attained by the carbon wall and the gradient that exists along the tube surface. An indication of the temperature gradient between the centre of a standard tube and a point 10mm from the end is illustrated for atomisation settings of 600, 690 and 999 units in Fig. 2. The wall temperatures were obtained by focusing the Ircon, Series 1100, pyrometer on the inside of the tube wall through the injection hole and from the end of the furnace.There appears to be little variation in the temperature difference between the two positions at any time from 2 to 7 s during the heating cycle of each setting. The temperature of the carbon wall 10 mm from the end of the tube appears to be 200-280 K lower than the centre throughout each sequence and the effect that this gradient has on the vapour temperature at 620 units is shown in Fig. 3. As might be expected the apparent vapour temperature is averaged over the tube-wall temperature gradient but is only, at maximum, 120 K lower than the tempera- ture at the centre of the tube. A similar deviation for the HGA-72 has been reported previously,15 for atomisation at 999 units (maximum power).2800 - 2600 - 2400 - 2200 - C C 2000 L ' 1 1 1 I I I I I I I I 1 3 5 7 1 3 5 7 1 3 5 7 Time/s Fig. 2. Variation with time of the temperature at the centre (C) and 10 mm from the end (E) of a standard HGA-72 atomiser tube operated a t (a) 600, (b) 690 and (c) 999 units. Detection limits for the ten sample elements of varying volatility are given in Table I. All atomic-emission measurements were made a t an atomisation setting of 999 units, and this maximum power setting was chosen for each of the HGA-72 tube designs. The tempera- tures given in Table I are those of the carbon wall at the centre of the tube, measured with the Ircon optical pyrometer at the time of the maximum atomic-emission intensity. Discus- sion of the detection limit values will be considered in a later section, but it appears that the standard tube design, a t maximum power, is most suited to the determination of elements of medium to high atom-appearance temperatures. Although the maximum intensities of volatile elements occur, in general, before those of involatile elements, maximum atomic1142 LITTLEJOHN AND OTTAWAY: INVESTIGATION OF ATOMISER TUBE AnaZyst, VoL.104 I I 1 3 ‘5 7 9 Tirne/s Fig. 3. Comparison of the vapour temperature (0) with the wall temperature (0) at the centre (C) and 10 mm from the end (E) of a standard HGA-72 atomiser tube operated a t 620 units. emission for lead occurs after the maximum atomic emission for a number of less volatile elements, such as manganese, iron and tin. This may be related to the comparatively high concentration of lead standard solution required to obtain measurable signals.A shift in the time of the maximum emission intensity, further from the start of the atomisation sequence, has been observed for a number of elements, as the sample solution concentration is increased. This will be discussed in greater detail elsewhere.16 TABLE I CFAES DETECTION LIMITS FOR ELEMENTS OF VARYING VOLATILITY ATOMISED AT MAXIMUM POWER IN A STANDARD HGA-72 ATOMISER TUBE Central tube-wall Time of maximum temperature at time Detection limit using Wavelength/ atomic-emission of maximum atomic a 50-p1 aliquot/ Element nm* intensit y/s emission/K pg ml-l Lead . . . . . . Gallium . . . . Silver . . .. Tin . . . . .. Manganese . . . . Iron . . . . . .Chromium . . . . Molybdenum? . . Titanium? . . . . Gadolinium?. . . . 405.78 403.30 328.07 286.33 403.08 371.99 425.43 379.83 399.86 440.19 5.0 3.0 4.4 4.0 3.5 4.0 4.5 8.0 6.5 5.0 2 843 2 703 2 823 2 803 2 773 2 803 2 783 2 873 2 863 2 803 5.37 0.049 0.016 0.24 0.007 4 0.025 0.001 6 0.051 0.020 1.32 * Spectral band width 0.2 nm (except Gd, 0.14 nm; Mo and Ti, 0.07 nm). t Pyrolytically coated tube. High-temperature HGA-72 Tube The variation with time of the carbon wall temperature at the centre of the cut-down section of this tube is illustrated in Fig. 4 for a number of atomisation settings. The heating rate at the‘tube centre is much faster than for the standard tube design15 and exceeds 2500 K s-l at 550 units. Clearly the heating rate depends on the mass of carbon at the centre of the tube and will increase as the wall thickness is reduced.As might be expected, the atomisation settings give higher temperatures with the “high-temperature” tube than for the standard version. A maximum equilibrium temperature of 3000 K was obtained for this tube at 550 units. Increasing the control to 999 units did not give further increases in either the tube temperature or heating rate. At settings close to the maximum voltage, for example for 400-550 units, similar heating rates were observed, making the concept ofDecember, 1979 DESIGN FOR CARBON FURNACE ATOMIC-EMISSION SPECTROMETRY 1143 temperature optimisation applicable to the high-temperature tube, as discussed previo~sly.~ At the settings normally selected for atomic-emission analysis, the maximum temperature is reached within 2-3 s.In this period, however, there is a peak in the wall temperature at the tube centre, for a number of settings. This is most pronounced at 400-550 units (Fig. 4) and causes an equivalent peak in the background signals measured at these atomisation settings . 3100 I 2 900 450 2700 3 50 2 300 I ' I I I 1 3 5 7 Time/s Fig. 4. Variation with time of the wall temperature at the centre of a high-temperature HGA-72 atomiser tube operated a t settings between 350 and 550 units. As expected, the temperature gradient from the centre of the tube to a point 10 mm from the end is greater than for the standard tube, but is not altered significantly by changing the temperature setting as illustrated in Fig. 5 for 400, 450 and 550 units.In the initial seconds of the atomisation sequence, the gradient is as much as 600-800 K between the measurement points. However, as the equilibrium temperature is reached, the gradient is reduced to 300400K. The effect of this temperature gradient on the apparent vapour temperature is shown in Fig. 6 for atomisation at 500 units. When iron solutions were injected it was found that, during the initial drying sequence at 373 K for 40 s, some of the vaporised water condensed in the cooler end sections of the tube. It was therefore necessary 3100 2 700 Y 3 2300 2 --. L K l 6 1900 I- 1500 I I E I I 0 2 4 6 0 2 4 6 0 2 4 6 Time/s Fig. 5. Variation with time of the temperature a t the centre (C) and 10mm from the end (E) of a high-temperature HGA-72 atomiser tube operated at (a) 400, (b) 450 and (c) 550 units.to introduce a second drying stage at 873 K for 20 s to dry the tube completely. As with the standard tube design, the vapour temperature is averaged over the carbon wall tempera- ture gradient. Because the gradient is greater than in the standard tube, the vapour-phase temperature deviates more from the tube-centre temperature, in this instance by between 200 and 400 K.1144 LITTLEJOHN AND OTTAWAY: INVESTIGATION OF ATOMISER TUBE Analyst, Vd. 104 3000 2 800 Y --. L + 3 F 2600 a, a E l- 2400 2 200 0 2 4 6 8 lime/s Fig. 6. Comparison of the vapour temperature (0) with the wall temperature (m) at the centre (C) and 10 mm from the end (E) of a high- temperature HGA-72 atomiser tube operated at 500 units.Detection limits for the ten sample elements atomised in the high-temperature tube at 999 units (maximum power) are given in Table 11. As the tube attains the equilibrium temperature much faster than the standard tube the maximum atomic emission is reached within 4.0 s for all elements, irrespective of volatility. As expected the detection limits of elements of high atom-appearance temperature are enhanced when the high-temperature tube is used (shown here for gadolinium and in reference 8 for several other elements), while those of more volatile elements are degraded. The analytical utility of the high-temperature tube is somewhat restricted by the short lifetime of the tubes. Standard HGA-72 tubes can be used, in general, for more than 100 analyses, even when the furnace gas flow is interrupted during atomisation.The lifetime is reduced to about 30 injections when the carbon wall thickness is reduced at the tube centre. This can. be extended by forming a pyrolytic graphite coating on the tube after every 20-25 analyses, vvhich increases the lifetime by up to a factor of three times. TABLE I1 CFAES DETECTION LIMITS FOR ELE:MENTS OF VARYING VOLATILITY ATOMISER TUBE ATOMISED AT MAXIMUM POWER IN A HIGH-TEMPERATURE HGA-72 Time of maximum atomic-emission Element* intensity I s Lead . . . . . . Gallium . . . . . . Silver . . . . . . Tin . . .. .. Iron . . . . . . Chromium . . . . Manganese . . . . Molybdenumt . . . . Titanium? . . . . Gadolinium? . . . . 3.8 4.0 3.5 3.5 3.1 3.5 3.0 4.0 3.5 4.0 Central tube-wall -temperature a t time of maximum atomic emission/K 3 063 3 063 3 063 3 063 3 063 3 063 2913 2 993 2 953 2 973 * Wavelengths and spectral band widths as in Table I.7 Pyrolytically coated tube. Detection limit using a 50-pl aliquot/ pg ml-l 16.32 1.30 0.033 0.67 0.080 0.16 0.005 0.052 0.017 0.86 Tapered HGA-72 Tube This tube was designed in an attempt to reduce to a minimum the temperature gradient along the HGA-72 atomiser. Although it was difficult to reproduce accurately the same taper on different tubes, the temperature gradient from the central position to 8mm from~ ~ C 6 V Z b e Y , 1979 DESIGN FOR CARBON FURNACE ATOMIC-EMISSION SPECTROMETRY 1145 the end of the tube was usually reduced to a maximum of 100-150 K during the atomisation sequence. Because the temperature gradient is reduced, the vapour-phase temperature is very similar to that of the carbon wall at the centre and ends of the tube.Detection limits for the ten sample elements atomised at 999 units (maximum power) in the tapered tube are given in Table 111. Tapering the ends of the HGA-72 tube tends to give a lower equilibrium tube temperature at each atomisation setting than for the standard or high-temperature tube designs. This is illustrated in Fig. 7 for atomisation at 999 units. 2 700 2 600 Y 1 2500 3 + e 2400 P 2 300 Fig. 7. Comparison of the vapour temperature (A) with the wall temperature (-) at the centre (C) and 8 mm from the end (Ej of a tapered HGA-72 atomiser tube operated at 999 units. Consequently, the detection limits of the more involatile elements are degraded but as the temperature gradient is decreased, the detection limits of volatile and medium volatile elements are imprdved slightly.Tapering the tube does not appear to affect the lifetime of the tube significantly in comparison with the standard design. TABLE I11 ATOMISED AT MAXIMUM POWER I N A TAPERED HGA-72 ATOMISER TUBE CFAES DETECTION LIMITS FOR ELEMENTS OF VARYING VOLATILITY Tube-wall temperature atomic-emission maximum atomic a 5O-pl aliquot/ Time of maximum at time of Detection limit using Element* intensity 1s emission/K pg m - l Lead . . . . . . 4.0 2 693 2.3 Gallium . . .. . . 2.9 2 593 0.036 Silver , . . . . . 3.9 2 693 0.014 Tin . . .. . . 3.6 2 663 0.195 Manganese * . . . 3.0 2613 0.001 5 0.009 6 Iron . . . . . . 3.6 2 673 Chromiuni ... . 3.5 2 783 0.000 7 Molybdenum? . . . . 7.0 2 673 0.084 Tit ani urn t .. . . 5.5 2 653 0.047 Gadoliniunit . . . . 4.5 2 653 9.5 * Wavelengths and spectral band widths as in Table I. Pyrolytically coated tube. Volatile-elements HGA-72 Tube The initial volatile-elements tube design13 allowed the ends of the tube to be heated faster and to a higher temperature than the central section where the sample is atomised. At maximum power (999 units), the central section reached a maximum temperature of 2053 K and the end sections, of reduced wall thickness, 2443 K.13 The temperature gradient was thus reversed, compared with the standard HGA-72 tube design. Atoms formed in the central section were thought to diffuse into the hotter end-sections giving greater atomic1146 Analyst, VO,?.104 emission to tube-background emission ratios folr volatile elements than for the standard tube. The end sections are also nearer their maximum temperature when the atom popula- tion is present, and the temperature and background signal are therefore changing less rapidly, giving improved analyte signal reproducibilities. Significant improvements in detection limits were observed for a number of volatile elements but the application of this tube to the CFAES determination of elements with higher atom-appearance temperatures is limited by the comparatively low temperature of the central atomisation section of the tube wall. The concept of atomising samples into an already hot tube seems to be a positive step in the development of a furnace tube design specific to the requirements of carbon furnace atomic-emission analysis.Therefore, to extend the range of application of the volatile-elements tube design, only 0.75-1 .O mm of carbon was removed from the end sections of the tubes used in this study. By removing slightly less carbon than the original design, higher temperatures are achieved at both the end and centre of the tube. This is illlustrated by the optical pyrometer tempera- tures given in Fig. 8 obtained at an atomisation setting of 999 units. The apparent vapour temperature, estimated as before from iron electronic excitation temperatures, is considerably greater than the central tube-wall temperature in the initial seconds of the atomisation sequence. Consequently, the atomic emission of all elements with appearance temperatures less than 2273 K is enhanced compared with atomisation to the same central wall tempera- ture in a standard, high-temperature or tapered 11GA-72 tube.LITTLEJOHN AND OTTAWAY: INVESTIGATION OF ATOMISER TUBE 2600 Y \ 2500 3 + 2 f 2400 c 2 300 . - t + + c :&/ 1 I I I 0 2 4 6 E Time/s Fig. 8. Comparison of the vapour temperature (+) with the wall temperature a t the central (C) and end sections (E) of a volatile-elements HGA-72 atomiser tube operated a t 999 units. Detection limits for nine of the sample elements of varying volatility atomised at maxi- mum power (999 units) in the modified volatile-elements tube are given in Table IV. When the volatile-elements tube was pyrolysed to enhance atomisation of molybdenum, titanium and gadolinium, the wall temperature at both the end and central sections was observed to increase slightly.However, as expected, detection limits for these elements are poor and owing to severe between-run memory effects, it was not possible to calculate a detection limit for molybdenum, although atomic-emission signals could be measured. I n contrast, the detection limits for gallium to chromium are adequate for many analytical purposes. However, the values for lead and gallium are higher than those reported for the initial volatile-elements tube design.13 This is undoubtably related to the observation that at the atom-appearance temperatures of these elements, the tube temperature a t the end sections is still increasing rapidly. Consequently, the analyte-emission signal tends to appear as a shoulder on the rising background signal and is difficult to detect at low atom concentrations.Volatile-elements HGA-72 Tube Fitted with a Sample Cup measurement of atomic-emission signals. further it is apparently necessary to fulfil two conditions. The principle of the volatile-elements tube design appears to be particularly suited to the To extend the range of application of this tube Firstly, the end sections must beDecember, I979 DESIGN FOR CARBON FURNACE ATOMIC-EMISSION SPECTROMETRY 1147 heated far faster than the central atomisation section and should reach an approximately constant temperature before atoms of even the most volatile elements are generated. This implies that the difference between the end and central sections should be as great as possible during the initial seconds of the atomisation sequence.Secondly, the temperature attained by both sections should be as high as possible. In an attempt to achieve these conditions, a new type of atomiser tube was prepared that is substantially different from the others described above in that it incorporates a sample cup placed in the centre of the graphite tube, as well as end sections of reduced wall thickness.14 TABLE I V CFAES DETECTION LIMITS FOR ELEMENTS OF VARYING VOLATILITY ATOMISED AT MAXIMUM POWER IN A VOLATILE-ELEMENTS HGA-72 ATOMISER TUBE Tube-wall temperature at time of maximum atomic Time of maximum emission/K Detection limit using atomic-emission -7 a 50-p1 aliquot/ Element* intensityls Centre End pg ml-I Lead .. . . .. . . Gallium . . .. . . Silver . . . . . . Tin . . . . . . . . Manganese . . . . . . Iron . . . . . . . . Chromium . . . . .. Titanium? . . .. .. Gadolini um t . . . . 4.2 2.3 2.4 3.0 3.5 6.0 7.0 --$ t -+ 2 563 2 373 2 353 2 393 2 323 2413 2 373 2 793 2 793 2 693 2 593 2 613 2 2 3 2 893 2 893 .F -+ t -+ 5.4 0.026 0.008 9 0.063 0.001 5 0.007 3 0.001 3 0.37 18.85 * Wavelengths and spectral band widths as in Table I. t Pyrolytically coated tube. $ Not measured. Optical pyrometer temperature measurements of the cup and reduced wall sections at maximum applied power were given in the original Communication.14 The incorporation of a sample cup as part of the tube wall at the centre enhances the temperature difference between the atomisation (cup) and excitation (tube vapour) sections compared with the volatile- elements tube described in the previous section. The cup is resistively heated with the rest of the tube but because of its greater mass, the temperature rises more slowly and to a lower final value than the sections of reduced wall thickness.Comparison of the temperature profiles in Fig. 8 with those in Fig. 2 of reference 14 indicates that the cup attains a similar equilibrium temperature to the central section of the original volatile-elements tube design. However, the end sections of the cup - tube assembly are almost 300 K higher than those of the tube operated without the sample cup attachment. Fig. 2 of reference 14 illustrates that during the initial few seconds of atomisation, differences of at least 1000 K exist between the cup and the end sections of reduced wall thickness.I t appears, however, that the cup - tube will be restricted in application to elements with atom-appearance temperatures below 2 200 K. Excluding molybdenum, titanium and gadolinium, detection limits for the remaining seven of the sample elements are given in Table V, along with the values obtained for a number of other volatile elements.14 Aliquots of 20 p1 were found to be the maximum suitable for the cup - tube, and comparison of concentration detection limits with the other tubes where 50 pl can be used is at a significant disadvantage. A larger cup size may allow larger aliquots to be used with this type of atomiser but this has not been attempted to date. The tempera- tures in Table V correspond to those of the cup and reduced wall sections at the time of the maximum atomic-emission signal, which was between 2.5 and 60 s after the start of atomisa- tion depending on the volatility of the element.By following the procedure described above and elsewhere14 it is possible to create reproducibly cup - tubes with th. described heating parameters and a tube lifetime similar to that of the standard HGA-72 design.1248 General Comparison The detection limits obtained for the ten sample elements chosen at the start of this study are collated in Table VI for the standard and modified HGA-72 tube designs. The optimum detection limit for each element is in italics. As expected, the high-temperature tube is most suited to the determination of carbide-forming or involatile elements, while the most sensitive atomic-emission signals of volatile elements are obtained with the cup - tube.The volatile-elements and tapered tubes give lowest detection limits for medium-volatile elements and it is interesting to note that of the ten elements investigated only molybdenum achieves optimum sensitivity in the standard tube when operated at maximum power. The detection limits of volatile elements are poor in the high-temperature, standard and tapered tubes as the maximum atom concentrations of these elements are present at low tube vapour tempera- tures and the weak emission signals cannot be easily detected from the rapidly increasing background signals. However, the detection limit for lead in the volatile-elements tube is higher than expected in comparison with the values obtained with the other tube designs.LITTLE JOHN AND OTTAWAY : INVESTIGATION OF ATOMISER TUBE Analyst, vol. 104 TABLE 'v' CFAES DETECTION LIMITS FOR ELEMENTS OF VARYING VOLATILITY ATOMISED AT MAXIMUM POWER IN A VOLATILE-ELEMENTS HGA-72 ATOMISER TUBE FITTED WITH A SAMPLE CUP Element Wavelength/nni* Cadmium . . . . . . 326.1 1 Lead . . . . . . . . 405.78 Zinc . . . . . . . . 307.59 Bismuth . . . . . . 306.77 Galliilm . . . . . . 403.30 Silver . . .. . . 328.07 Gold . . . . . . . . 267.60 Tin . . . . , . . . 286.33 Manganese . . . . . . 403.08 Copper . . . . . . 324.75 Iron . . . . . . . . 371.99 Chromium . . . . . . 425.43 Tube-wall temperature a t time of maximum atomic emission/I< r-----hp 7 c u p Tube wall <: 1573 2 443 1823 2 523 2 003 2 613 1698 2 773 1773 2813 1698 2613 2 253 2 693 2 223 2 683 2 223 3 683 2 093 2 733 - - - - Detection limit using a 20-pl aliquot! pg ml-l 0.06s 0.027 1.57 0.033 0.000 66 0.00044 0.16 0.030 0.003 1 0.002 0 0.10 0.006 3 * Spectral band width 0.2 nm (Pb and Bi, 0.07 nm).With the exception of molybdenum17 the optimum detection limits in Table VI represent the best achieved to date with a standard commeircial furnace - spectrometer system. The values also compare favourably with detection limits presented recently using wavelength modulation background correction to compensate automatically for the tube wall back- ground emission using a Perkin-Elmer HGA-2 100 a t ~ m i s e r . ~ Conclusions The design and temperature gradient of a graphite tube greatly influence the intensity of atomic emission and, consequently, the CFAES detection limits that are obtainable with a Perkin-Elmer HGA-72 atomiser.When the carbon wall temperature decreases towards the ends of the tube from the central atomisation section, as in the standard and high-temperature tube designs, the average vapour temperature is always lower at any time than the tempera- ture of the carbon wall from which atoms are produced. This deviation is more severe in the high-temperature tube due to the increased temperature gradient. Consequently, volatile elements experience low excitation temperatures and exhibit poor detection limits in standard and high-temperature tubes. By reversing the temperature gradient it is possible to pre-heat the vap2ur before atomisation occurs by increasing the temperature of the ends of the tube faster and to a higher equilibrium value than the central atomisation section.Therefore, when atoms are produced either directly from the wall surface or by dissociation of a molecular vapour, they exist in a gas that has a greater temperature than the vaporising surface. The intensity of atomic emission for any element will obviously beDecember, 1979 DESIGN FOR CARBON FURNACE ATOMIC-EMISSION SPECTROMETRY 1149 improved as this temperature difference is increased. The development of a tube operated with a sample cup has allowed this concept to be applied to elements ranging in volatility from cadmium and zinc to copper and chromium. This allows the HGA-72 to be operated in a similar manner to the L’vovlO and Woodriffll atomisers in that atoms and molecules of a number of elements can be introduced into a tube where the vapour is already heated close to its maximum temperature.Further modifications are required, however, to allow efficient atomisation and excitation of involatile elements and elements of high excitation energy, such as arsenic and selenium. At present the CFAES determination of molybdenum, titanium, etc., is best performed with the standard and high-temperature HGA-72 atomiser tubes or by application of rapid furnace heating1’ TABLE VI COMPARISON OF CFAES DETECTION LIMITS FOR ELEMENTS OF VARYING FOR AN HGA-72 ATOMISER VOLATILITY ATOMISED AT MAXIMUM POWER I N DIFFREENT TUBES DESIGNED Values in italics are optimum detection limits.Detection limit/p.g ml-’ Volatile-elements Volatile-elements Tapered Standard High-temperature I 7 -A-pp Element tube with cup tube tube tube tube Lead . . . . . . Gallium . . . . Silver . . . . Tin . . . . . . Manganese . . . . Iron . . . . . . Chromium . . . . Molybdenum . . Titanium . . . . Gadolinium . . . . 0.027 0.000 66 0.000 44 0.030 0.003 1 0.10 0.0063 5.4 0.026 0.008 9 0.063 0.001 5 0.007 3 0.001 3 0.37 18.85 - 2.3 0.036 0.014 0.195 0.001.5 0.009 6 0.000 7 0.084 0.047 9.5 5.37 0.049 0.016 0.24 0.0074 0.025 0.001 6 0.051 0.020 1.32 16.32 1.30 0.033 0.67 0.080 0.16 0.005 0.052 0.017 0.S6 The results presented in this paper indicate that significant improvements in CFAES detection limits can be achieved by alteration of the design and heating characteristics of a standard HGA-72 atomiser tube.Although parts per billion detection limits can now be achieved for a wide range of elements, no single design of commercial atomiser tube is likely to be suitable at present for the optimum atomisation and excitation of all elements. The use of present carbon furnaces with multi-channel spectrometers for simultaneous multi- element analysis would necessitate the adoption of compromise conditions of tube design and temperat~re.~ This may not be a disadvantage in many real analytical situations but would limit the general analytical value of CFAES. The carbon tubes used for all previous published research on CFAES were designed for atomic-absorption measurements. The results indicate that when tubes designed specifically for atomic emission are manufactured, considerable advantages will result if the maximum atom population of all elements can be generated in a vapour already existing a t its maximum temperature. As this is rarely a requirement for atomic-absorption measurements, it is likely that the optimum tube designs for atomic emission and atomic absorption will be different. The optimum tube design for emission may require complete separation of the atomisation and excitation steps in the production of the signal from the analyte. The authors thank The Royal Society for the award of a research grant to J.M.O. for the purchase of the HGA-72 atomiser and the Salters’ Company for the award of a scholarship to D.L.1150 LITTLE JOHN AND OTTAWAY References 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. Shaw, F., and Ottaway, J . M., Analyt. Lett., 1975, 8, 911. Littlejohn, D., and Ottaway, J . M., Analyst, 1977, 102, 393. Hutton, R. C., Ottaway, J. M., Rains, T. C., and Epstein, M. S., Analyst, 1977, 102, 429. Ebdon, L., Hutton, R. C., and Ottaway, J . M., Analytica Chim. Acta, 1978, 96, 63. Epstein, M. S., Moody, J. R., Brady, T. J., Rains. T. C., and Barnes, I. L., Analyt. Chem., 1978, 50, Ottaway, J. M., and Shaw, F., Analytica Chim. Acta. 1978, 99, 217. Alder, J . F., Samuel, A. J., and Snook, R. D., Lab. Pract., 1977, 26, 22. Ottaway, J. M., and Shaw, F., Appl. Spectrosc., 1977, 31, 12. Littlejohn, D., and Ottaway, J . M., Analytica Chim. Acta, 1979, 107, 139. L’vov, B. V., “Atomic Absorption Spectrochemical Analysis,” translated by J. H. Dixon, Adam Woodriff, R., Stone, R. W., and Held, A. M., Apibl. Spectrosc., 1968, 22, 408. Robinson, J . W., and Wolcott, D. K., Analytica Chim. Acta, 1975, 74, 43. Ottaway, J. M., and Hutton, R. C., Analyst, 1976,, 101, 683. Littlejohn, D., and Ottaway, J . M., Analyst, 1978, 103, 662. Littlejohn, D., and Ottaway, J. M., Analyst, 1978, 103, 595. Littlejohn, D., and Ottaway, J. M., Can. J . Spectrosc., in the press. Littlejohn, D., and Ottaway, J. M., Analytica Chim. Acta, 1978, 98, 279. 874. Hilger, London, 1970. Received November 22nd, 1978 Accepted July 9th, 1979