May 1983 The Analyst Vol. 108 No. 1286 Platform Atomisation in Carbon Furnace Atomic-em ission Spectrometry Laszlo Bezur," John Marshall John M. Ottawayt and Riad Fakhrul-Aldeen Department of Pure and Applied Chemistry University of Stvathclyde Cathedral Street Glasgow G1 1XL The design and operational characteristics of platforms used in carbon furnace atomic-emission spectrometry are described. Platforms made of graphite and coated with pyrolytic graphite give significant improvements in the sensitivity and detection limits achieved for 23 out of 24 elements investigated. The delay in atomisation allows the production of the atomic vapour at a time when the vapour phase inside the furnace atomiser has reached higher temperatures. Improvements in signal to noise ratio have also been achieved through use of a high-resolution 6chelle spectrometer a novel square-wave wavelength modula-tion system and the better separation of the analyte signals from blank signals.Keywords Carbon furnace atomic-emission spectvometvy ; platform atomisation A number of publications in recent years have demonstrated the utility of the carbon furnace as a source for the generation of atomic-emission signals and a state of the art review has recently been pub1ished.l Much of the previous work has employed atomic-absorption instrumentation imposing severe limitations on the technique as neither the design of the furnaces nor the optical arrangement of these systems has been optimised for the measurement of carbon furnace atomic-emission spectrometric (CFAES) signals.In the measurement of carbon furnace atomic-emission signals a continuum background signal from the tube wall is always present .2 Even under optimum focusing a residual background signal persists which has been shown to be the result of Rayleigh scattering of black-body radiation from the tube wall into the monochromator.3 Correction for such continuum background radiation has been successfully achieved by wavelength modulation generated by an oscillating refractor plate.4 A novel method of square-wave wavelength modulation was employed in the instrument used for the work described in this paper and its construction and operation have been de-scribed Using this system the background signal from the furnace can be auto-matically subtracted obviating the need to obtain a separate background measurement for each analytical signal.Previous work has been based on the optimisation of the signal to background ratio in the measurement of CFAES signals. I t follows that the amount of scattered light entering the monochromator has a considerable influence on the over-all detection limit of the technique. For such a system it is important that the design of the furnace should not contribute an increase in the scatter signal if real improvements in sensitivity are to be made. However with a system incorporating an automatic background correction device the noise rather than the background is the limiting factor on detection power. Selection of the optimum operating temperature will therefore be based on signal to noise ratios rather than signal to background ratios as previously described.8 The temperature experienced by the analyte atomic vapour greatly affects the sensitivity obtained by CFAES.2y9 Improvements in sensitivity have been achieved by modification of the graphite tube itself and hence its thermal properties.Such modifications include the reduction of the thickness of graphite in various sections of the tube2S1* and the addition of a sample cup to the base of the tube.ll These design modifications are essentially permanent as the heating characteristics of the tubes are irreversibly altered. In addition although effective the modifications described require the precision machining of * Present address Institute for General and Analytical Chemistry Technical University of Budapest, Gellert ter.4 1521 Budapest Hungary. t To whom correspondence should be addressed. 55 554 BEZUR et a,!. PLATFORM ATOMISATION Analyst Vol. 108 tubes and tend to be time consuming. In some instances tube designs limit the range of ele-ments that may be determined,g and this is undesirable from the point of view of obtaining compromise conditions for simultaneous multi-element analysis by CFAES. A tube design is required that has specific advantages for furnace emission measurement which at the same time does not result in a degradation in sensitivity for any class of element. The over-riding consideration in the design of a graphite tube for the measurement of sensitive atomic-emission signals is the temperature attained in the tube and which is experienced by the analyte atoms.9 With an exponential temperature dependence of emission intensity at a constant free atom population the desired solution is one that results in the atomisation of the sample after the maximum furnace vapour temperature has been attained.As the vapour temperature depends on the wall temperature it is necessary to delay the sample atomisation until the vapour-phase temperature is maximised. This has been achieved to some extent by the use of the volatile-element tubelo and the cup-tube,ll where selective heating of parts of the atomiser tube heats the vapour prior to the formation of atoms. However it is also important that the maximum temperature that the tube can reach is not reduced as a result of the design. With the cup-tube the cup temperature does not reach the appearance temperature of some elements and cleaning the tube of an involatile matrix is difficult.The effect of diffusion of the atomic vapour from the furnace must also be considered, especially with respect to the temperature obtained in the vapour phase at the time of atomisa-tion as this will affect the sensitivity. Consequently the heating rate of the tube and the efficiency of atomisation are important parameters in evaluating a tube design with respect to the effective excitation temperature and rate of atom loss due to diffusion. Ideally a tube is required in which the sample is atomised almost instantaneously into a maximum temperature furnace atmosphere. L’vov12 has discussed the advantages of atomisation of the analyte into an atmosphere that has reached isothermal conditions for atomic absorption.Platform atomisation has been suggested as a method of obtaining such conditions and the technique has been investigated by a number of workers.13-17 The platform is a small pyrolytic graphite plate that is placed in the centre of the furnace beneath the injection port and the sample is deposited on it. Because the platform is not part of the tube itself thermal contact with the tube is minimal resulting in a delay in the heating of the platform with respect to the tube. As a result the sample is atomised from the platform at some time later than it would be from the tube wall and is introduced into a vapour that has reached or is close to isothermal conditions. This is precisely the situation required for improvement in the sensitivity of CFAES.Indeed increases in sensitivity would be expected to be much greater with atomic emission as opposed to atomic absorption by virtue of the strong temperature dependence of the emission signal. It is also expected that advantages derived from the use of the platform in atomic absorption would be equally applicable to CFAES. In an earlier paper,5 the detection limits for a range of elements achieved by the use of a platform in an HGA 72 furnace used with an dchelle spectrometer system were reported. In this paper more extensive information is given that includes the optimisation of platform design for CFAES measurement and the analytical parameters used in the operation of the furnace - platform for CFAES.Experimental Instrumentation The instrument system used was described briefly in a previous paper.5 The spectrometer is a Spectrametrics SMI I1 I khelle monochromator and a square-wave wavelength-modulation system was incorporated for automatic background correction of carbon furnace atomic-emission signals. Wavelength modulation was achieved by a rotating quartz chopper with four quadrants of different thickness and was analogous to a system recently described in detail for use in continuum source atomic-fluorescence spectrometry.18 In contrast to the earlier work with the dchelle ~ y s t e m ~ in which the wavelength modulation chopper was mounted near the exit slit in this instance it was positioned near the entrance slit. Operation in this mode is clearly preferable if the potential use of the dchelle system.for simultaneous multi-element analysis is to be realised. The other components of the instrument the Perkin-Elmer HGA 7 May 1983 I N CARBON FURNACE AES 555 carbon furnace atomiser and Brookdeal precision lock-in-amplifier were as described previously5 and are listed in Table I. I t should be noted that the figures for dispersion were given incorrectly in the previous paper.5 hlonochromator . . Wavelength modulation Lens . . . . . . Atomiser . . Detector . . . . . . . . Power and reference unit . . Signal processing . . TABLE I INSTRUMENTATION Spectrametrics SMI I11 Cchelle monochromator Focal length 0.75 m F number 13 Optical speed f / l O Reciprocal linear dispersion-a t 200 nm 0.062 nm mm-l a t 400 nm 0.124 nm mm-l a t 600 nm 0.180 nm mm-l Slit widths 0.2 mm Slit heights 0.5 mm Rand pass a t 400 nm 0.025 nm Laboratory constructed18 Chopper rotation speed 20 rev s-l Modulation frequency 40 Hz Modulation interval see text f = 130 mm silica Perkin-Elmer HGA 72 carbon furnace Laboratory-prepared platforms Purge gas argon 1.5 1 min-' Hammamatsu R292 photomultiplier Laboratory constructed 28 V d.c.supply to the EHT d.c. - d.c. converter f 15 V d.c. power supply and 6 V d.c. for the chopper motor Ortec Brookdeal 5002 current pre-amplifier Ortec Brookdeal 9503D precision lock-in-amplifier The square-wave wavelength-modulation system was developed to give a maximum signal to noise ratiolg and has been used to correct both for residual tube-wall background emission and variable background emission caused by various sample matrices.The modulation system consists of a chopper disc made of four quartz segments of different thickness a d.c. motor and a reference signal generated using an infrared light-emitting source and photodiode used to trigger the lock-in-amplifier. The quartz quadrants were each made of radius 30 mm and thickness 2.5 1.0 2.5 and 4.0 mm as described previously.18 A Perspex rim was attached around the chopper disc with a balancing weight opposite the 4.0-mm quadrant. The lock-in-amplifier was used to achieve automatic continuous background correction by subtracting the background signal obtained during observation through the 1 .O- and 4.0-mm quadrants from the analyte plus background signals observed through the 2.5-mm quadrants.The background-corrected signal was recorded using a Servoscribe R 541.20 potentiometric recorder but a digital transient recorder Type DL 901 (Electroplan Ltd. Royston Hertford-shire) was also used to study rapid signals. To measure the uncorrected signal the d.c. output of the current pre-amplifier was directly connected to the chart recorder. A l-m optical bench was fixed to the base of the spectrometer and the focusing lens and HGA 72 furnace head were mounted on the rail using screw adjustable holders. Wavelength adjustment was normally achieved for convenience using a hollow-cathode lamp of the appropriate element. Where necessary this was also achieved by injection of a suitable con-centration of the element required.In this instance care must be taken to avoid memory effects particularly for involatile elements. Problems due to residual atom signals were often particularly troublesome with this instrument owing to the high sensitivity achieved for many elements and it was found preferable to use an old graphite tube for wavelength adjustment when using a high-concentration analyte solution. In preliminary measurements a silica lens of focal length 50 mm and diameter 40 mm was used to produce a 1 1 image of the centre of the HGA 72 tube at the entrance slit of the mono-chromator. Subsequently and for most of the results reported a silica lens of focal lengt 556 BEZUR et al. PLATFORM ATOMISATION Analyst Vol. 108 130 mm and diameter 40 mm was substituted in order to reduce background emission reaching the detector.The furnace operation programme was optimised for each element separately as will be described, but in all instances the gas-stop mode was used during the atomisation stage. The HGA 72 graphite tube has dimensions of length 53 mm and i.d. 8.6 mm and is significantly larger and wider than most current commercial furnace systems. The absence of windows also helps to reduce problems related to scattering of tube-wall radiation from external surfaces into the monochromator. The gas flow and dispersion of atom populations formed in the tube pro-ceeds from the centre to the ends increasing the atom residence time and particularly under gas-stop conditions increasing the temperature experienced by the atom population.2 Standard Perkin-Elmer HGA 72 graphite tubes were used but were coated with pyrolytic graphite in sits prior to use.This was achieved by the introduction of a flow of 50 ml min-1 of methane into the argon purge gas stream for 3-6 min whilst the furnace was set to operate at 2100 “C. Platforms were prepared in the laboratory and were made of different designs as shown in Fig. 1. It was desirable to minimise the contact between the platform and the tube to reduce the rate of heat transfer. Ideally the platform should be wide enough to make contact with the sides of the tube without touching the bottom. Sections of graphite cut from standard HGA 72 tubes were found to be very suitable as their curvature matched that of the tube wall and gave the desired degree of contact when mounted in the centre of the tube.A Perkin-Elmer HGA 72 graphite furnace was used for atomisation and excitation. 1.5 t 4 L1 Fig. 1. Graphite platforms used A curved “rectangular” platform; and B as A but with sections removed from the contact edges. The platforms used were of the two basic designs shown in Fig. 1. The simplest was a curved “rectangular” piece cut from the wall of the HGA 72 tube of dimensions 8 x 11 nim. The mass of this piece was approximately 0.12 g. The second type had similar dimensions but sections were removed from the two contact edges to reduce further the tube to platform contact to that at the four corners of the platform. Platforms of different dimensions were prepared to arrive at this optimum the thickness in each instance being maintained at the thickness of the HGA 72 tube wall.Platforms were coated with pyrolytic graphite under the same conditions as those used for coating the tubes. Platforms were placed in the centre of the HGA 72 tube and were introduced easily via the open ends of the atomiser. Both types could accommodate up to 50 p1 of sample solution, which was transferred through the normal injection port of the atomiser using an Oxford micro pipette. Atomic-absorption measurements were also made with this system using the appropriate hollow-cathode lamp positioned on the optical bench and focused through tlie HGA 72 atomiser and on to the entrance slit of the monochromator. A mechanical chopper was used to achieve intensity modulation of the hollow-cathode lamp radiation and a similar reference signal wa May 1983 IN CARBON FURNACE AES 557 generated to operate the lock-in amplifier.The Wavelength modulation unit was disconnected during absorption measurements. The mechanical chopper was mounted in front of the hollow-cathode lamp and between it and the HGA 72 furnace. Temperature measurements of the graphite tube and platform were made using an Ircon Series 1100 recording optical pyrometer using the appropriate current shunts and the R 541.20 potentiometric recorder. The pyrometer was focused as required on the inside tube wall or platform through the injection port of the HGA 72 atomiser. Temperature measurements were made assuming an emissivity of unity. Reagents All reagents were of the highest available purity and high-purity distilled water was used for the preparation of all solutions.Stock solutions (1 000 pg ml-l) of each element were pre-pared from the metal or appropriate sulphate or nitrate salts and acidified to a final concentra-tion of 10-2 M sulphuric or nitric acid. Working solutions of the appropriate concentrations were prepared when required. Detection Limits signal equal to the noise of the blank signal. The detection limits were calculated as the concentration of analyte element that produced a Results and Discussion Optimisation of Signal to Noise Ratio using commercial atomic-absorption systems to measure furnace atomic-emission signals detection limits were defined in terms of signal to background ratio. Using wavelength modulation for continuous automatic background correction detection limits be-come dependent on the signal to noise ratio.4 Assuming thermal equilibrium in the furnace tube2* and consequently a Boltzmann distribution of energy the measured emission signal or spectral line intensity at any instant will be given by2Ov2l In previous e x p ( g ) .. . . . . * - (1) KhcNtAijgiL 47~B ( T ) hij I = where I is the intensity over the total line width in W m-2 sr-l K is a combined unit conver-sion factor and instrument constant N t m-3 is the total instantaneous concentration of atoms in all states in the source L m is the length of the source in the direction of viewing Aij s-l is the Einstein transition probability for spontaneous emission gi is the statistical weight of the upper energy level B(T) is the partition function over all states Ei J is the energy of the upper state i T K is the temperature of the vapour phase in the furnace hij m is the wavelength of the spectral line and h c and R have their usual meanings and values.The dominating noise component of carbon furnace atomic-emission signals using continuous background correction is in most instances the shot noise associated with the residual black-body radiation from the tube all.^,^ To improve detection limits of the carbon furnace atomic-emission technique it is therefore required to increase I without significantly increasing the residual black-body radiation. Because in optimising the conditions for CFAES measurement the most sensitive spectral line will be chosen and fixed and the optical aperture parameters included in the value of K will be made as large as possible within instrumental constraints the only factors that can be used to increase the value of I are Nt L and T .In electrothermal atomisers an excessive increase in L is generally unacceptable owing to the increased possibility of self-absorption and increased temperature gradients of the furnace leading to condensation etc. at the cooler parts of the tube. Although some modified graphite tubesg?l1 have been designed to vary the heating characteristics with respect to length Nt and T remain the most important parameters. In a graphite furnace operated under gas-stop conditions the vapour-phase temperature T , corresponds closely to the tube-wall temperature.20 The maximum tube-wall temperature is limited by the rapidly diminishing lifetimes of graphite tubes above 3300 K and the power available in most commercial systems is restricted to levels that prevent this temperature being exceeded.The instantaneous total atom concentration Nt is a result of a series of comple 558 BEZUR et at. PLATFORM ATOMISATION Analyst Vol. 108 solid melt and gas phase reactions that depend on the chemical and physical properties of the analyte and matrix as well as instrumental factors such as the heating rate the final equilib-rium temperature set and the surface material of the atomiser. According to L'vov's model of the atomisation process,12 the maximum number of analyte atoms in the furnace Nmax. is related to the total number of analyte atoms introduced into the furnace No and is given by Nmax.= 2N0 2 - - 1 + e-Ti/T,> . . 71 '(" 7 2 where r1 is the atomisation time and r2 is the residence time. The loss of analyte atoms in graphite tube furnaces is expressed by r2 and is due partly to the internal gas flow and partly to diffusion. Under interrupted gas-flow or gas-stop atomisation conditions diffusion and convection caused by the expanding gases are the important factors and will be dependent on the internal dimensions of the tube. Under diffusion-controlled conditions r2 is a function of temperature through the tempera-ture dependence of the diffusion coefficient where - - (3) 7 2 Gcc-" and m varies between 1.5 and 2. The rate of loss therefore increases with increasing tempera-ture and atomic-absorption results often show that Nmax.and hence the absorption maximum ( A max,) occur at an optimum temperature that may be considerably less than the maximum or equilibrium temperature. When atomic-emission measurements are made the maximum signal occurs after Nmax. (or Amax.)2s22 and at a stage when the number of atoms available for excitation is diminishing. This is because IA is also dependent on T which for most elements is still increasing at the time of Nmax The time of peak intensity of the emission signal there-fore depends on the two changing parameters N t and T. The fact that under normal operation the temperature is changing during atomisation is not ideal for atomic-absorption measurements. An improvement in sensitivity and reduction or elimination of matrix interference effects was achieved by increasing the heating rate of the furnace and also by delaying the evaporation or atomisation of the sample.15 In this instance Nmax.occurs at a time when the furnace is closer to the equilibrium or maximum temperature, and molecules that may be formed as a result of interfering reactions are more completely dissociated. The use of higher heating rates and the delay of atomisation to times at which higher temperatures are available is even more advantageous in emission measurement than absorption as the possibility then exists of simultaneously increasing Nma,. and T . Thus maximum power heating using the HGA 2200 carbon furnace gave a significant improvement in the atomic-emission detection limits for involatile elements,23 owing to increases in Nma,.. In this work platform atomisation was used to delay atomisation to give an increase in T during the lifetime of the atom population. Detection limits in CFAES are limited by the noise generated by the intense emission from the graphite tube wall. In a properly aligned and baffled optical system it should not be possible for tube wall radiation to enter the spectrometer slit directly but a residual signal is still ~bserved.~ The main sources of this residual radiation which has the characteristics of black-body radiation may be the following : (a) tube-wall radiation scattered by molecules and particles in the hot part of the furnace (Rayleigh scatter) ; (b) tube-wall radiation scattered by particles formed by condensation at the cooler ends of the tube (Mie scatter); (c) radiation scattered by incandescent carbon particles from the decomposition of the graph-ite tube ; (d) radiation reflected into the spectrometer from furnace windows and other optical com-ponents; and (e) aberrations or dirt particles in the optical system focusing lens furnace windows etc., which might allow tube-wall radiation to enter the spectrometer directly despite correct alignment.The intensity of the emission signal given by any of the above sources is proportional to the intensity of the radiation from the tube wall which is given by Planck's Law which for May 1983 I N CARBON FURNACE AES 559 narrow band pass and in the wavelength region and temperatures of interest in carbon furnace atomic emission has the form3 .. . . ' ' (4) CI E -- A5 exp(C,/AT) where E is the integrated intensity of the black-body radiation C and C2 are constants A is the wavelength and T the absolute temperature. Thus the black-body radiation and the fraction of it scattered are exponentially dependent on the temperature in the same way as the analyte atomic-emission intensity given by equation (1). Thus an increase in temperature (at the same atom concentrations) increases the signal and background in proportion and gives no advantage in signal to background ratio.* The noise component of the background-corrected signal however is proportional to the square root of the background intensity in the shot noise-limited examples and will therefore change less rapidly with temperature.In CFAES with wavelength modulation background correction detection limits can be improved by maximising the temperature as far as possible and also by minimising all sources of residual background emission. The angular distribution of the scattered light intensity I, in the Rayleigh region (A> diameter of particles d) is expressed by T12 where I is the impingement intensity 8 is the angle between the incident and scattered light beams n is the refractive index of the particle V is the volume of the particle h is the wave-length and Y is the distance of the particle from the point of observation. Thus the intensity of Rayleigh scattered light is proportional to the incident intensity and to the sixth power of the particle diameter and inversely proportional to the fourth power of the ~ a v e l e n g t h .~ ~ ~ The intensity is largest at angles near 0 and 180". The intensity of scattered light caused by larger particles (d > A) Mie scatter is not dependent on wavelength but is dependent on the angle 8 giving maximum intensity at small angles and a minimum at 180". The intensity of Mie scatter is also proportional to the number of scattering particles and will thus increase in the presence of a sample matrix that forms condensed vapours at the cooler ends of the tube.24 This is particularly relevant in the open-ended HGA 72 atomiser used in this work but will be minimised in other atomiser designs. Mie scattering will also reduce the analyte atom intensity reaching the monochromator and will thus degrade both the signal to background and signal to noise ratios by simultaneously changing both factors.Modifications to the tube or gas flow could be used to reduce the effect of Mie scattering. Scattering of black-body radiation by carbon particles will increase with the age of the graphite tube. A fine loose carbon dust is gradually formed in most types of graphite tubes, but the effect is reduced by coating with pyrolytic graphite. The latter was used throughout this study. Optical Design Both were used to form a 1 1 image of the central section of the tube at the entrance slit of the monochromator. The apertures of the two lenses were identical producing a signal from a central cylindrical section of the graphite tube of 2 mm diameter. The 50-mm lens gave a more deformed image of the 53-mm long HGA 72 tube and even in the optimum position a proportion of tube-wall radiation entered the monochromator directly.The 130-mm lens gave a much improved image with a distinct dark area in the middle of the slit indicating that no tube radiation entered the monochromator directly. The detection limit for manganese was determined for each system giving 0.14 pg 1-1 for the 50-mm lens and 0.029 pg 1-1 for the 130-mm lens an improvement of 4.8 in signal to noise ratio. In CFAES correct lateral alignment of the furnace and optical components is essential. The emission intensity of the tube-wall radiation (background) and the total emission for 50 pl of 1 pg 1-1 manganese solution were measured at the correct position and with the furnace tilted out of the exact horizontal axis.Several positions were used and the results are shown in Fig. 2 as a function of the angle. The signal to background ratio changed from 4.3 to 0.6 when the angle was changed from 0 to 2". Two projector lenses of focal lengths 50 and 130 mm were compared 560 BEZUR et al. PLATFORM ATOMISATION Analyst Vol. 108 1 .o 0.8 0.2 \.C !I 4 2 h I I 0 0 1 2 a, Fig. 2. Effect of the lateral alignment of the graphite furnace on (A) the emission signal from 50 p1 of 1 pg 1-l manganese (B) the background signal a t 403.08 nm and (C) the signal to back-ground ratio (A/B). Wavelength Modulation The resolution of the SMI bchelle monochromator is about an order of magnitude better than that usually available in commercial atomic-absorption instruments and results in an im-proved signal to background ratio.The square-wave wavelength-modulation system not only provided automatic correction for the residual background signal but also correction for the less reproducible background emission from various matrices. The chopper consisting of four quartz quadrants was mounted at the entrance slit of the monochromator at an incident angle of 24" (at a slit width of 0.2 mm) to the optical axis. The extent to which the light is displaced by refraction on passing through the chopper is dependent on the thickness of the quartz,18 and it was arranged that light at the analyte atom wavelength passed through the two 2.5-mm quadrants. The 1 .O- and 4.0-mm quadrants then allow the background signal to be measured on either side of the atomic line.The chopper rotation speed was 20 s-l which produced a square wave modulated signal of frequency 40 Hz. The incident angle of 24" was chosen to give efficient modulation over the range from 200 to 800 nm at tlie 0.2-mm slit width. The total modulation distance was 0.52 mm at 200 mi 0.50 mm at 400 nm and 0.46 mm at 800 nm and gave modulation intervals of 0.032 0.062 and 0.115 nm at the respective wavelengths. The minimum acceptable angle of incidence was determined by calculation and was also measured using a chromium hollow-cathode lamp as the line source at 425.4 nm. The anode current of the photomultiplier tube was measured on a 1-MR resistance using an oscilloscope in the d.c. mode. If the angle of incidence is large enough the minimum current is equal to the dark current.The measured minimum angle 23" was larger than that calculated 21" owing to the aperture of tlie spectrometer and the slight de-focusing of the slit image caused by the chopper disc. To ensure correct modulation an incident angle of 24" was chosen. Dimensions and Design of Platforms Previous studies of platform atomisation have concerned their use in atomic absorption in which the most important requirement is to minimise the reduction of intensity of the hollow-cathode lamp. Increased background emission reaching tlie detector is rarely important but assumes much greater significance in atomic-emission measurements not only because of the generally higher atomisation temperatures used but also because of the direct relat ionsliip between detection limits and the residual background emission signal.In addition flat or plane platforms appeared to be unsuitable in the HGA 72 as the sample tended to spread to tlie edges of the platform and then wet the tube wall. To minimise both problems curved platforms were designed with tlie shape and dimensions shown in Fig. 1. These curved platforms can be prepared from a tube with dimension May 1983 IN CARBON FURNACE AES 561 identical with those of the tube in which they are to be used. Only the lower edge of the plat-form touches the tube wall satisfactorily reducing heat conduction. The upper edge of the platform is some distance from the tube wall reducing the tendency of the sample to wet the wall and ensuring that samples are more accurately located even compared with injection directly on the tube wall.Normal injection of samples through the injection hole of the HGA 72 tube is straightforward and 5O-pl aliquots are conveniently atomised using platforms of optimum dimensions. Platforms with sections removed from the contact edges were prepared which further reduced the rate of heating and cooling. The platforms indicated in Fig. 1B were found to be significantly better for volatile elements such as gallium and indium. The effect of the different platform configurations in the light beam is indicated by the avail-able angular apertures at the centre of the tube and the resulting free tube diameter shown in Table 11. In the system used the aperture is limited by the optical system but the reduction in free tube diameter is significant as the radiating surface approaches more closely to the optical observation zone resulting in a higher scattered light intensity.TABLE I1 EFFECT OF PLATFORM ON AVAILABLE ANGULAR APERTURE AND FREE INNER DIAMETER OF AN HGA 72 TUBE Available angular Free inner Tube aperture O diameter/mm HGA 72 . . . . . . 18.8 8.5 HGA 72 with single curved platform . . . . 18.8 7.0 HGA 72 with double curved platform . . 18.8 6.0 Temperature Characteristics of Platforms heat to the p1atf0rm.l~~~~ the change of platform temperature with time is therefore given by16 I t is clearly established that the reduced tube - platform contact results in a slow transfer of Heat is transferred to the platform almost entirely by radiation and where E C Q S and V correspond to the platform and E is the emissivity C is the specific heat, Q is the density S is the surface area and V is the volume of the platform 8 is the Boltzmann constant and T and T are the temperatures of the tube wall and platform respectively.To increase the temperature lag of the platform only a few of the parameters in equation (6) are significant. The choice of material is restricted to the different forms of graphite tantalum and tungsten alternative materials being subject to chemical and thermal decomposition. With any specific material the only means of increasing the temperature lag is to increase the thickness of the platform. The effect of platform dimensions and mass is shown in Fig. 3. The use of a longer platform of the same thickness is unsatisfactory as the surface area in-creases almost in proportion to the mass or volume and the temperature lag is not greatly increased (compare curves B and D in Fig.3). Longer platforms also suffer from the dis-advantage of the greater spreading of the sample in the tube. It may then be subject to variable heating rates owing to the temperature gradient existing in the tube. They will also reduce the available angular aperture. Almost the same increased delay was achieved by using the platform with reduced contact edges shown in Fig. 1B (see curve C Fig. 3). This also increased the time required for cooling the platform between injections. Some practical advantage was found in using two platforms mounted one on top of the other.The tempera-ture lag is about the same as achieved by doubling the mass of a single platform. The platforms only touch at the edges and are heated equally by the tube wall on either side. The temperature of the platforms is then the same and there is no heat transfer between them. Although no tube or platform emission was allowed to enter the monochromator directly, background emission signals increased owing to increased scattering of radiation. Fig. 4 illustrates the increase in background from the use of a single platform ( x 1.28) and a double platform ( x 1.83) these signals being taken at 500 nm directly from the pre-amplifier i.e., without background correction 562 BEZUR et al. PLATFORM ATOMISATION Analyst Vol. 108 3000 s !?? t? 2000 tl n I- : 1000 2 4 6 8 10 Time/s I I I I 1 1 I 0 8 6 4 2 0 Ti me/s Fig.3. Temperature veYsus time graphs for an HGA 72 furnace operated a t 2600 K. (A) Standard tube; (B) standard tube plus platform Fig. 4. Background signals taken shown in Fig. 1A; (C) standard tube plus platform from the pre-amplifier a t 500 nm. HGA shown in Fig. 1B; and (D) standard tube plus 72 operated a t maximum power 999 platform of dimensions 8 x 20 mm. units for 10 s. (A) Standard tube; (B) standard tube plus platform; and (C) standard tube plus two platforms placed one on top of the other. Platform Atomisation The effect of the platform on atomisation was investigated for chromium and gallium, representing elements of different volatilities. Both atomic-absorption and atomic-emission signals were observed under identical conditions using a standard HGA 72 tube and a standard tube with platform.Background correction was used in the emission mode but not in absorp-tion where it was unnecessary for the pure analyte solutions used. Tube and platform temperatures were also observed under the same conditions and all signals were synchronised in time from the start of the atomisation stage of the furnace. Platform temperature measure-ments made using an optical pyrometer are likely to be in error owing to the reflection of light from the tube wall scattered off the platform. In this work no account has been taken of this source of error. As indicated platform temperatures will only be too high it will be noted that this error does not affect any conclusions drawn which clearly demonstrates the characteristic properties of platform atomisation.For chromium 50-pl aliquots of a 50 pg 1-1 solution were used for both absorption and emission and the results are illustrated in Fig. 5. At this concentration of chromium the start of the atomic-absorption signal in the standard tube (curve C Fig. 5) occurs at 1920 K (curve A). Atomisation from the platform appears to start at a platform temperature of 1950 K (compare curves E and B). These atomisation temperatures are in good agreement but are higher than the appearance temperature of chromium at 1 740 K,f2 possibly owing to the effect of c~ncentration.~~ When atomisation from the platform starts the tube temperature has reached 2350 K giving a lag of 400 K.I t is clear from curve E that at the time of maximum atom population 5.5 s the temperature of the tube has not reached its maximum. A further delay of about 2 s giving 7.5 s in total would be required to achieve coincidence of Nmax. and the equilibrium temperature Tmax The peak absorbance for chromium was about the same with and without the platform (compare curves C and E) but the emission intensity increased by a factor of 1.6 using the platform (compare curves D and 1;). As at the temperatures used in electrothermal atomisers most atoms remain in the ground energy state the total concentration of atoms in all states at any time N t is proportional to the absorbance at that time A t . Substitution in equation (1) and rearrangement give May 1983 6 -C d -I $ 5 -4 -IN CARBON FURNACE AES 563 I I I 3000 2000 g 2 2 4-n E 1000 + 0 2 4 6 8 10 Ti me/s Fig.5. Absorption and emission curves for 50 p1 of 50 pg 1-' chrom-ium atomised at an HGA 72 furnace setting of 2800 K (A) change of temperature of the graphite tube; (B) change of temperature of the plat-form of type A (Fig. 1) ; (C) chromium atomic absorption following tube-wall atomisation ; (D) chromium atomic emission following tube-wall atomisation ; (E) chromium atomic absorption with platform atomisation ; and (F) chromium atomic emission with platform atomisation. where K' and E are constants for a specific transition and K' includes all the non-exponential terms in equation ( l ) plus the conversion constant relating N t to At.Hence a plot of the In IA/At veysus T-l should be a straight line of slope - E i / k . For the chromium wavelength of 425.43 nm this slope will be -3.38 x 104 K-1. Vapour temperatures were not measured in this study. Therefore tube-wall and where relevant measured platform temperatures were used to examine this relationship. In Fig. 6 In IA/A t is plotted for chromium atomisation in a standard HGA 72 tube without platform against the reciprocal tube-wall temperature, T;:. A linear relationship is obtained but with a greater negative slope than the theoretical value suggesting that the average vapour temperature experienced by the analyte atoms a t any time is slightly less than the tube-wall temperature at that time. In Fig. 7 values 3 ' I I I 3 3.5 4 4.5 W K - ~ x 10-4 Fig.6. Plot of InTA/A versus T-l for 50 pl of 50 pg I-' chromium atomised from the tube wall a t a furnace setting of 2800 K. Broken line indicates theoretical slope -E,/k, of -3.38 x lo4 I<-l for chromium a t 425.43 nm. 7 6 55 -I 4 C B ' A I I I 3 3.5 4 4.5 T V K - ~ x 1 0 - ~ Fig. 7. Plot of 1nI)./At versus T-l for 50 p1 of 50 pg 1-' chromium atomised from the platform a t a furnace setting of 2800 I<. A, T T,; 13 ,T = Tev; and C T = T,. Ihoken line indicates theoretical slope as for Fig. 6 564 BEZUR et al. PLATFORM ATOMISATION Analyst Vol. 108 obtained for chromium atomisation from a platform are plotted against T -4 and against the measured temperature of the platform T-t as well as the average of the two temperatures, T;:.All give linear relationships and the theoretical slope (broken line) falls between T; and T-;. Similar measurements were carried out for gallium using 50 p1 of a 50 pg 1-1 standard solution. Absorbance and emission measurements were made at 403.29 nm and are illustrated in Fig. 8. Both the absorption signals from the tube wall and the platform start at 1400 K, which is close to the average literature value for the appearance temperature of 1350 K.l2 7 - 0.7 6 - 0.6 .- 5 5 - 0.5 m a .- E 4 -; 0.4 S E a $ U 0 2 4 6 8 10 Absorption and emission curves for 50 pl of 50 pg 1-1 gallium at an HGA 72 furnace setting of 2950 I<. (A) Change of temperature of the graphite tube; (B) change of temperature a t the platform of type A (Fig. 1) ; (C) gallium atomic absorption following atomisation froin the tube wall ; (U) gallium atomic emission for tube-wall atomisation; (E) gallium atomic absorption for platform atomisation ; and (F) gallium atomic emission for platform atomisation.Time/s Fig. 8. The tube-wall temperature at the start of the absorption signal from the platform is 2 180K, giving a platform delay in temperature of 780 I< at this time. For gallium an increase in peak absorbance by a factor of 3 is obtained on using the platform (curves C and E) suggesting more complete dissociation of molecular species in the higher vapour-phase temperature available following platform atomisation. The enhancement in the emission signal on the use of the platform is a factor of 5 (curves D and F).The maximum absorption signal for gallium with 3 3.5 4 4.5 5 T ’ / K - ’ x Plot of lnl~,/A vevszfs T-l for 60 pl of 50 p g 1-1 gallium atomised from the platform a t a furnace setting of 2950 I<. A T = T,; B,T = Tav; and C T = T,. Hroken line indicates theoretical slope -E,/k’ of 3.56 x lo4 1i-l for gallium at 403.29 nm. Fig. 9 May 1983 I N CARBON FURNACE AES 565 the platform still occurs about 3 4 s before the tube reaches its maximum or equilibrium temperature. All three are linear but in contrast to chromium the slope of the Ti? line is closer to the theoretical slope of 3.56 x lo4 K-l. The relationship examined in Figs. 7 and 9 provides an interesting method of testing the application of the Boltzmann equation to the excitation of atoms during electrothermal atomisation and of estimating the average vapour-phase temperature experienced by the transient atom population.The differences between chromium and gallium cannot be explained at present. Clearly the platform temperature lags behind the tube-wall tempera-ture during the atomisation of most elements and the vapour-phase temperature may not follow the tube wall exactly either. At least the part of the tube that contains the platform is at a temperature lower than the tube wall and this may help to reduce the over-all vapour-phase temperature. It is clear that the above relationship may be used to predict the emission curve and the time of maximum emission intensity for an element if the absorbance curve of the element and the temperature characteristics of the tube are known.The expected increases in emission intensity as a function of temperature at constant atom concentrations can also be easily calculated from equation (1). Typical values for chromium, gallium and lead are shown in Table I11 and are similar to those values observed experimentally (shown in Table VII and discussed later). The actual increases in peak emission intensity will depend on both the temperature and atom concentrations at the time of each peak and on the energy of the upper state of the transition. The In (IA/At) relationships with T;’ TF’ and T;;;1 are shown in Fig. 9. TABLE I11 THEORETICAL INCREASES IN EMISSION INTENSITY WITH TEMPERATURE (Nt CONSTANT) Element and wavelengthlnm Ei/eV T J K Cr 425.43 . . . . . . . . 2.91 2 800 2 700 2 700 Ga 403.29 .. . . 3.07 2 800 2 700 2 700 Pb 405.78. . . . 4.38 2 800 2 700 2 700 TZIK 2 900 2 900 3 000 2 900 2 900 3 000 2 900 2 900 3 000 I2lIl 1.52 2.38 3.51 1.5 2.5 3.7 1.87 3.66 6.62 Platform Drying Temperatures The reduced heat transfer to the platform means that much longer drying times are required if the tube temperature is set as usual close to the boiling-point of the solvent. Using 373 K for water and a 50-pl aliquot of sample drying times increased to 180-200 s from the normal 30-40 s. Increasing the tube temperature to 473-573 K reduced the drying time to 3040 s without sample losses by spluttering. Heat transfer to the platform at low temperatures is by conduction and energy losses arising from the evaporation of the water balances the plat-form temperature allowing a smooth drying rate even at higher tube temperatures.When the solvent has evaporated the temperature of the platform rises to that of the tube wall, which might then cause losses of very volatile elements. In practice this has only been observed for indium when a lower drying temperature of <450 K must be used. Drying temperatures are best chosen by visual observation of solvent evaporation. Effect of Platform on Emission Peak Shape In our experience the use of platform atomisation results in improvements in signal to background and signal to noise ratios above that due to the increased vapour-phase temperature. This arises from improved peak characterisation and shape and can be illus-trated for a number of elements.Background-corrected signals for a blank sample and 50 pl of appropriate concentrations for silver indium and lead both with and without a platform are shown in Figs. 10-12. Without a platform large and variable blank signals are obtained for silver and lead which appear to b 566 BEZUR et al. PLATFORM ATOMISATION Analyst Vol. 108 -\ \ \ \ \ \ I 1.1 \ A I D +- Time Fig. 11. Emission peak shapes for indium at 4- Time 410.f7 nm. (A) Blank-firing of standard tube; (B) 50 pl of 100 pg 1-1 indium from tube wall; (C) blank firing with platform; and (D) 50 pl of 0.5 pg 1-1 32Ft 1;''. " ~ ~ i ~ p e a ' ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ indium from platform. Temperature settlng, 2900 K. tube; (B) 50 p1 of 50 pg 1-' silver from tube wall; (C) blank firing with platform; and (D) 50 pl of 5 pg 1-l silver from platform.Tem-perature setting of the HGA 72 furnace, 2900 K. due to a memory effect caused by evaporation of analyte from the ends of the tube and contact cones. The analyte peak after the sample injection is ill-defined and superimposed on the variable blank signal. Using the platform a substantial increase in sensitivity is observed, but in addition the peak is more clearly separated from the blank signal and much more sharply defined. At the measurement of peak emission with the platform both sensitivity and signal to noise ratio are improved. With indium (Fig. 11) the blank signal is similar for the tube-wall and platform atomisation but both sensitivity and peak shape are improved I A d Fig.12. Emission peak shapes for lead at 405.78 nm. (A) Blank firing of standard tube; (B) 50 pi of 1000 pg 1r1 lead from tube wall; (C) blank firing with platform; and (D) 50 pl of 100 pg 1-1 lead from plat-form. Temperature setting 2 900 K May 1983 I N CARBON FURNACE AES 567 when using the platform. Several reasons can be given for the improved atomisation obtained with the platform : (a) the sample is localised on the platform and is therefore prevented from spreading away from the centre of the tube. The sample is therefore evaporated in the hottest part of the tube and in a shorter time; (b) evaporation takes place into a much hotter gas which results in faster evaporation and dissociation of gas molecules; and (c) during platform atomisation the temperature at the ends of the tube is higher when the sample is evaporated and thus condensation is reduced.Blank signals were found to be significant for a number of elements and arise in part from the very high sensitivities achieved using this technique. Several causes may be considered. The blank signal may be an analyte signal derived from contamination present in the tube material from slow release of refractory species formed on the graphite surface or from con-densed analyte on the ends of the tube or contact cones. I t may also be due to a failure in the wavelength modulation system due to the presence of a highly structured background signal, or to the partial intensity modulation given by the wavelength modulation system used.26 Characteristic blank signals are shown in Fig.13 together with appropriate analyte signals using concentrations near the detection limits. For manganese gallium lead indium and silver there is no problem from the blank signal which is either negligible or well separated from the analyte peak. With copper and chromium a slowly rising blank signal is obtained, which was shown to be due to analyte atoms present in the form of a memory effect. Thus the magnitude increased after the injection of higher concentrations of analyte and decreased when gas-flow atomisation was used. In many instances the memory effects shown here are com-parable to the "double peaks" observed in the atomic-absorption determination of volatile elements. +Time Emission peak shapes for several elements together with blank signals [EL) at the same wavelength.Platform atomisation. Mn 1 pgl-1 manganese; Cu 1 pg 1-' copper; Cr 1 pg 1-1 chromium; and Ga 1 pg 1-' gallium. Wavelengths as in Table VI; 50-p1 aliquots; temperature setting 2 900 K. Fig. 13. Owing to the sensitivity achieved by furnace atomic emission with platform atomisation it is essential to remove the analyte completely from the tube after each injection. Although gas-stop atomisation is required for maximum sensitivity it tends to increase the condensation effects and hence blank signals. Maximum sensitivity with minimal memory effects is achieved by gas-stop atomisation maintained only until the peak emission intensity has been observed, followed by a clean stage at maximum temperature under gas-flow conditions until the signal returns to zero.Such criteria will also be relevant in the use of platform atomisation in atomic absorption although the effects for each element will vary with the sensitivity given by each technique 568 BEZUR et al. PLATFORM ATOMISATION Analyst Vol. 108 Temperature Programme Mention was made above of the choice of drying temperature which is critical if analysis times are not to be unnecessarily long. The platform also takes longer to cool to ambient temperatures and an extra 20-30 s delay must be allowed at this stage. The selection of the maximum ashing temperature is very similar to that for the standard tube without platform and recommended values were normally used without difficulty. The choice of atomisation temperature however is different from that reported earlier for a system without wavelength modulation.8 Above a specific temperature the signals remained constant with increase in temperature ie.a plateau region was reached. Atomisation temper-atures given in Table IV are either the maximum available (2873 K) or close to it and as TABLE IV OPTIMUM ATOMISATION TEMPERATURES FOR CFAES MEASUREMENT Element . . . . . . . . 3 :: Be . . Cd . . Cr . . c u . . . . . . Fe . . . . Ga . . K . . . . Mg . . Mn . . . . Na . . . . Ba Co Dy Eu Ho In Ni, Pb Sc Sr Yb . . T / K -7 STD tube Platform >2 773 -~ 2 8 2 3 >2 823 - >2 823 >2773 2 873 >2 723 ->2673 >2773 ~ 2 6 2 3 > 2 673 >2773 ->2823 2 873 >2 823 2 873 ~ 2 4 7 3 >2673 >2673 >2 823 2 873 2 873 predicted,8 are higher than for a system without wavelength modulation.The optimum temperatures given are the minimum required to give maximum intensity and little change is observed at higher temperatures. The differences between the temperature found to be optimum with and without the platform are very small and the major difference in conditions is caused by the introduction of wavelength modulation background correction. This is a function of the change in dependence of detective power from signal to background to signal to noise ratio as originally reported by Epstein et aZ.4 The use of the platform and wavelength modulation will clearly facilitate the selection of compromise conditions for a potential simul-taneous multi-element analysis system based on CFAES without significant sacrifice in sensitivity for any element.Improvement factors for the improvement in sensitivity with the use of the gas-stop mode are shown in Table V for both platform and tube atomisation. The improvement with the TABLE V EFFECT OF GAS FLOW ON EMISSION SIGNAL SENSITIVITY MEASURED USING 50-pl ALIQUOTS OF ANALYTE AT 1 pg ml-I Element Mg . . In . . Cr . . Mn . . Fe . . cu . , co . . Ni Na Sensitivity gas stop Sensitivity gas flow 7-Platform Standard tube 6.5 9 6 90 3 4.5 10 50 6 30 4 G 5.8 17 4.4 4.3 19 4 May 1983 IN CARBON FURNACE AES 569 platform is in all instances smaller than with the tube alone. The main advantage of the gas-stop mode in atomic emission is to reduce the rate of loss of analyte and allow a greater vapour-phase temperature to be obtained at gas flow.The delay caused by the platform acts in an analogous manner and hence reduces the effect of the gas-stop mode. However the improve-ment factors with the platform are still significant and the gas-stop mode should be used for the highest sensitivity. Effect of Platform on Reproducibility When large sample volumes are injected directly on to the tube wall of a furnace the sample spreads out either instantaneously or sometimes during the drying stage to an extent that depends on the surface tension of the solvent. With many sample types reproducibility is impaired through variable spreading effects. The platform counteracts this by preventing the sample from extending over its perimeter.The surface tension of the droplet ensures that it does not fall off the top surface of the platform. The reproducibility of sample placement is therefore improved and the reproducibility of analysis is consequently also improved. The results in Table VI indicate this to be so particularly for the more volatile elements. Precision is also improved because of the faster and more complete evaporation and dissociation of the analyte and because of the reduction in condensation effects at the tube ends which results in the sharper peak shapes illustrated in Figs. 10-12. The platforms used in this study could accommodate sample volumes of up to 50 pl. TABLE VI REPRODUCIBILITY OF CARBON FURNACE ATOMIC-EMISSION MEASUREMENT OF SELECTED ELEMENTS Wavelengths given in Table VII ; optimuni conditions as given in Table I V ; 5O-pl injections.Relative standard deviation yo * Concentration/ Element pg 1-1 Cr . . 50 c u . . 50 Fe . . 100 In . . 500 K 1000 Mn . . 10 Na . . 1000 A f \ Standard tube Standard tube with platform 2.6 1.9 4.1 2.3 4.3 3.1 5.5 3.2 2.4 1.9 2.6 2.3 2.6 2.0 * 10 replicates. Effect of Platform Atomisation on Sensitivity and Detection Limits As reported previ~usly,~ the detection limits achieved with this system are extremely low for many elements. The number of elements investigated to date has been considerably extended and the detection limits obtained are shown in Table VII and compared with values obtained using other relevant spectroscopic systems.Table VII includes the wavelengths used for each element and the corresponding energies of the upper level of each transition and also the appearance temperatures of most elements.12 Also included are the improvement factors obtained by use of the platform in terms of both detection limit and sensitivity. Column 5 in Table VII lists the best detection limits achieved without background correction but using the optimum temperature and tube type referred to. Column 6 gives the values obtained with a 0.75-m Ebert monochromator combined with an HGA 2200 furnace and the oscillating refractor plate form of wavelength modulation. Comparison of columns 5 and 6 shows the advantages of the use of wavelength modulation and a higher resolution .monochromator as this system gave significant improvements for all elements except lead.Further significant improvements have been achieved with the present system both with and without platform atomisation. Beryllium and barium are the only elements for which detec-tion limits are not lower using the present system barium undoubtedly owing to the poo 570 BEZUR et aE. PLATFORM ATOMISATION TABLE VII COMPARISON OF CFAES DETECTION LIMITS Detection limit (DL)/yg I-’ Element 2 * * * * Au Ba ,. Be Cd Cr Go c u Fe Ga . . Ho In K Na ._ Ni Pb s c Sr Yb : 2 a . * * - I nm 328.07 396.15 267.59 553.55 234.86 326.11 425.43 345.35 324.75 404.89 459.40 371.99 403.29 405.39 410.17 404.41 285.21 403.08 330.23 341.48 405.78 402.04 460.73 398.79 eii 3.78 3.14 4.63 2.24 5.28 3.80 2.91 4.02 3.78 3.06 2.70 3.33 3.08 3.06 3.02 3.06 4.34 3.0R 3.75 3.11 4.38 3.08 2.69 3.11 ‘ K 1120 2070 1370 2 200 730 1740 1640 1460 (-) (-) 1540 1350 (-4 lkb 1580 1530 1510 1250 1675 1060 2 450 2100 (-) 3 Appearance Without With Present work Wavelenethl Ed temueraturel*/ wavelength wavelength Present work STD tube + modulation mod~lation4.~~ STD tube platform 0.44t 4.17 12s 160t 4601 52s 35a “?7,, 1.45 l.3** 0.66t 2 y 505 217 (-1 38 1.511 (-4 23% 27t 147 1.27 0.367 1.4 0.9 0.4 0.044 2.4 0.1 13.0 1.0 0.3.5 0.1x 2.1 2.6 0.9 0.06 5.7 2.1 0.92 0.31 0.026 360 .5 0 2 8on 26 370 0.051 0.030 0.26 5.0 0.023 1.6 0 .In1 9.1 0.73 0.1 8 0.039 17.4 0.025 2.0 0.3 0.029 3.6 2.1 3.4 0.35 0.25 0.016 25 5 0 Analyst Vol. 108 Improvement in detection Improvement limits in sensitivity, DL(STD) S(PL)* ~ -DL(PL) 27 30 14 10 56 1.5 1.9 1.5 2.9 1.4 I .4 1.4 1.6 1.5 H4 1.3 7 2.6 1.6 1.0 2.6 1.2 1.6 108 I , S (STD) 10 20 3.3 1.3 4.5 2.0 1.4 2.3 I .3 1.5 2 5.8 1.4 1.3 3.5 2.9 1.6 1.0 5 3.3 1.2 1.6 22 89 * Ratio of sensitivities measured under same conditions. t With cup tube.ll $ (-) Not measured.5 Standard tube. 7 H-igh temperature tube.* 11 With tapered tube.D ** With volatile-elements tube.D sensitivity of the photomultiplier in the wavelength region of the barium line. Comparison of columns 7 and 8 in Table VII indicates the improvements obtained by use of a singlecurved platform. Calculated improvement factors for both the detection limits and sensitivities are given in columns 9 and 10. The effectiveness of platform atomisation in CFAES is shown by the improvement in detection limits for all elements varying between 1.5 and 108 and in sensitivity between 2 and 89. For elements such as lead and silver the improvement is partly due to the reduction in noise and for others such as gallium it is almost certainly partly due to improved atomisation efficiency.The main contribution however derives from the increased emission intensity provided by the higher temperatures available during the lifetime of the atom population. The increases in sensitivity are consistent with the calculated values shown in Table 111. The linear range of the method is considerably extended as a result of the improvement in detection limits. Some typical growth curves using platform atomisation are shown in Fig. 14. Curvature begins to recur at roughly the same analyte concentration as with tube-1 I 1 I I I -4 -3 -2 - 1 0 1 2 3 Log c Fig. 14. Analytical curves of growth using plat-form atomisation for chromium lithium manganese, iron and nickel log (relative intensity IR) veYszts log (concentration pg ml-l). Wavelengths as in Table VI; conditions as in Table IV; 5O-pl aliquots May 1983 IN CARBON FURNACE AES 57 1 wall atomisation,2s but the linear part of the graph extends to lower concentrations owing to the improvement in detection limit offered by the platform.The linear dynamic range is about three orders of magnitude for most elements which is adequate for most analytical purposes, but not as extensive as the linear range obtained in inductively coupled plasma emission spectrometry. For many elements alternative wavelengths are available for the higher con-centration regions. I t should be noted that the detection limits for several elements are approaching the ultimate measurement levels where the greatest problems do not involve the sensitivity of the technique but those derived from contamination.The sensitivity of the method using the main lines for sodium and potassium is so great that for most practical purposes weaker lines must be used, as indicated in Table VII. Precise comparison of the two spectrometer wavelength modulation systems shown in columns 6 and 7 of Table VII is difficult. The square-wave wavelength modulation system used here should give an over-all improvement of 1 .8,19 but variations in performance are bound to be contributed by the different spectrometers and photomultiplier tubes used and particu-larly by the different carbon furnaces. The HGA 2200 and HGA 72 atomisers have different tube sizes heating rates and diffusional loss mechanisms and consequently both the atomisa-tion efficiencies and also the average vapoiir temperatures experienced by the atom population will be different.Direct comparisons of the alternative methods of wavelength modulation and alternative furnace types using the same spectrometer system have yet to be carried out and remain of interest. The characteristics of a monochromator optimum for CFAES will be a balance between aperture and band pass. In such a system furnace design will also be an important parameter With automatic background correction better results may possibly be obtained with a spectrometer of wider aperture and band pass than that used in this work. The results presented in this paper suggest that CFAES is one of the most sensitive emission techniques for a wide range of elements and that excellent results may be achieved with at least two different optical systems.Platform atomisation offers significant advantages in terms of sensitivity and combined with the reduction in chemical interferences,15~17 represents the most convenient and efficient means of atomisation so far described. Platforms can be used with any commercial electrothermal atomiser require slight adjustments to the atomisa-tion programme and are simple to prepare. The increase in tube lifetimes which also results, is an attractive by-product in view of the current price of graphite tubes for most instruments. The possibility of simultaneous multi-element analysis by CFAES is made more feasible by the use of platform atomisation as compromise atomisation conditions can be used without signifi-cant reductions in sensitivity for any element.The very low detection limits reported in this paper suggest the application of the technique to many real analytical problems. A specific interest in this laboratory is the analysis of trace elements such as manganese chromium and lead in biological materials.6 The detection limits for these and other elements are well below the currently accepted normal levels in blood serum and urine and indicate that results of good accuracy and precision could be obtained with little or no sample pre-treatment. The wave-length modulation used will also correct for increased background signals from light scattered by matrix particles. Detailed results of these studies will be presented in the near future. This work was made possible by the award of grants (to J.M.O.) by the Royal Society for the purchase of the HGA 72 carbon furnace atomiser and by the SEKC for the purchase of the khelle spectrometer.The award of a visiting fellowship (to L.B.) as part of the cultural agreement between the Ministry of Education of Hungary and the British Council is also gratefully acknowledged as is the leave of absence granted to L.B. from the Technical Uni-versity of Budapest. References 1. 2. 3. 4. 5 . 6. Ottaway J . M. Hutton 13. C. Littlejohn D. and Shaw,.F. Wzss. 2. Karl-Alnrx-Univ. Lezpzzg 1979, Ottaway J. M. and Shaw F. AppZ. Spectrosc. 1977 31 12. Littlejohn D. and Ottaway J. M. Analyst 1977 102 553. Epstein $1. S. Itains T. C. and O’Haver T. C. APPZ. Spectrosc. 1976 30 324. Ottaway J . A I . 13czur L. antl Marshall J. ‘4nalyst 1980 105 1130. Ottaway J . A I . I3czur Id. Fakhrul .4ltlecn I t . Frech W. antl Marshall J . “‘l‘racc 1’:lemcnt Analytical 28 357. Chemistry in Jlcdiciiic and 13iology,” Walter dc Gruytcr Munich 1980 p. 575 572 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. 27. 28. BEZUR MARSHALL OTTAWAY AND FAKHRUL-ALDEEN Marshall J. Bezur L. Fakhrul Aldeen R. and Ottaway J. M. Anal. Proc. 1981 18 10. Littlejohn D. and Ottaway J. M. Anal. Chim. Acta 1979 107 139. Littlejohn D. and Ottaway J. M. Analyst 1979 104 1138. Ottaway J. M. and Hutton R. C . Analyst 1976 101 683. Littlejohn D. and Ottaway J.M. Analyst 1978 103 662. L’vov B. V. Spectrochim. Acta Part B 1978 33 153. Gregoire D. C. and Chakrabarti Anal. Chem. 1977 49 2018. L’vov B. V. Pelieva L. A. and Sharnopolskii A. I. Zh. Prikl. Spektrosk. 1977 27 395. Slavin W. and Manning D. C. Anal. Chem. 1979 51 261. Katskov D. A. and Grinstein I . L. Zh. Prikl. Spektrosk. 1978 28 968. Slavin W. and Manning D. C . Spectrochim. Acta Part B 1980 35 701. Michel R. G. Sneddon J. Hunter J. K. Ottaway J. M. and Fell G. S. Analyst 1981 106 288. O’Haver T. C. Epstein M. S. and Zander A. T. Anal. Chem. 1977 49 458. Littlejohn D. and Ottaway J. M. Analyst 1979 104 208. Littlejohn D. and Ottaway J. M. Analyst 1978 103 595. Ottaway J. M. and Shaw F. Analyst 1975 100 438. Littlejohn D. and Ottaway J . M. Anal. Chim. Acta 1978 98 279. L’vov B. V. “Atomic Absorption Spectrochemical Analysis,” Hilger London 1970. Rowston W. B. and Ottaway J . M. Analyst 1979 104 645. Bezur L. Marshall J. and Ottaway J . M. in the press. Epstein M. S. Rains T. C. Brady T. J. Moody J . R. and Barnes I. L. Anal. Chenz. 1978 50 874. Littlejohn D. and Ottaway J. M. Can. J. Spectrosc. 1979 24 154. Received September 30th 1981 Accepted December 20th 198