JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY OCTOBER 1991 VOL. 6 505 Shadow Spectral Filming A Method of Investigating Electrothermal Atomization Part 1. Dynamics of Formation and Structure of the Absorption Layer of Thallium Indium Gallium and Aluminium Atoms Albert Kh. Gilmutdinov Yu. A. Zakharov V. P. lvanov and A. V. Voloshin Department of Physics University of Kazan 18 Lenin Str. Kazan 420008 USSR A technique for investigating electrothermal atomization using shadow spectral filming is presented. This consists of obtaining an image of the interior of the furnace by backlighting with the analytical line of the element under study and recording the image of the shadow produced by the absorbing vapour by use of a cine camera. It is then possible to visualize the dynamics of the atomic vapour filling the furnace and simultaneously giving information on the temperature change of the furnace walls and platform.The method is used to investigate the dynamics of formation and the structure of the absorption layers of TI In Ga and Al atoms in different atomization regimes. The results obtained show strong inhomogeneity of the distribution of atoms within a graphite furnace. The extent of non-uniformity is different for different elements and is largely determined by the atomizer platform. A number of anomalous effects have been established 'inverse' atomization of the elements (the formation of atoms begins not near the platform where the sample is deposited but near the opposite wall); the anisotropy of the velocity of longitudinal and transverse propagation of the atomic vapours within a furnace; and the localization of Al under the platform.Keywords Electrothermal atomic absorption spectrometry; visualization of atomic vapour distribution; shadow spectral filming All versions of atomic spectrometry emission fluorescence absorption and photoionization require prior transforma- tion of the substance from a condensed state into an atomic vapour i.e. atomization of a sample. High-temperature flames inductively coupled plasmas various types of electric discharges and laser emission are normally used as atomizers. One of the most common techniques is elec- trothermal atomization (ETA) of a sample in a graphite furnace. Although the proposed method was originally suggested for use with atomic absorption spectrometry (AAS),' the technique owing to its simplicity and versatility has also found wide application in emission2y3 and atomic fluores- ~ e n c e ~ - ~ spectrometry and laser photoioni~ation.~J' Hence it can be said that the problems presented by ETA are characteristic of atomic spectrometry as a whole.The urgency of the investigation has considerably increased in recent years because workers in this area have come close to achieving controlled9 atomization. Whereas originally ETA was passive i.e. after a sample had been deposited in the atomizer the analyst had virtually no influence on the process of atomization more recently persistent attempts have been made to optimize this process in order to achieve predictable and controllable atomization.Numerous approaches to the solution of this problem have been proposed optimal temperature pro- grammes of atomizer heating;1° introduction of a plat- form;' ' construction of an isothermal atomizer;12 atomiza- tion in an atmosphere which is variable in its compo- sition13 or pressure;14 atomizer surface treatment;lS and finally the most radical intervention in the ETA process chemical modification of a sample.16 Despite the numerous inve~tigationsl~ of each of these techniques it would still be premature to refer to fully controlled atomization. Realiza- tion of this is certainly impossible without detailed studies of atom formation processes. In the 1980s) ETA studies reached a qualitatively new stage that of fundamental investigations characterized by attempts to describe the atomization process at a micro- scopic level.Such an approach became possible due to the use of various independent including non-optical methods these included electronic microscopy X-ray diffraction mass spectrometry and Auger and photoelectron spectros- copy among others. Nevertheless the most common and most appropriate for use in the analysis of real samples is AAS.18 However this technique confronts the user with two problems. Firstly the absorption signal gives information on the atomization which has been distorted to a certain extent by the recording system of the spectr~meter'~ and by atomic transfer in the atomizer gas phase.2o Secondly and more importantly the recorded signal is a complex integration of the atom distribution volume of the furnace being viewed.In order to improve the quality and amount of information obtained using AAS a recording technique that simultane- ously registers the signal from nine zones has been pro- posed.21 This technique was subsequently used for studying the atomization dynamics for a number of elements,21-26 and for the study of stationary absorbing layers.27 It was established as a result of these experiments that contrary to earlier ideas the distribution of atoms across the atomizer might not be homogeneous and the spatial gradient depends on the element under study the wall material and the composition of the atmosphere within the atomizer. The proposed method however has one limita- tion the information is supplied only from a narrow region along the diameter of the furnace which is seen through the slit of the monochromator.This drawback was overcome in other s t ~ d i e s ~ ~ ~ ~ where the data were obtained on the distribution of Na atoms over the entire cross-sections of the CRA-90 and HGA-2100 atomizers. In one instance a resonance Schlieren method was used,28 while in another recording of the attenuation of laser emission by a vidicon recorder with subsequent computer processing of the signal was employed.29 The results of these experiments show that the distribution of Na vapour across the furnace is definitely non-homogeneous with a pronounced decrease in concentration near to the graphite walls. In order to optimize the utility of these approaches these techniques require the use of tunable lasers in the ultraviolet (UV)506 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY OCTOBER 199 1 VOL.6 region where most of the analytical lines are located. Another drawback of these techniques is the requirement of using amounts of metal that exceed those employed in real analyses 200 ng28 and 53 ng29 respectively. Thus the methods described make it impossible to investigate the atomic absorption layers of a wide range of elements. In the present paper a method known as shadow spectral filming (SSF) is proposed.30 It consists of obtaining the image of the atomizer volume by backlighting with the analytical line of the element of interest and recording by use of a cine-camera the shadow produced by the absorbing vapour. The possibilities of the method are illustrated by an investigation of the dynamics of the formation of the absorption layer of the Group I11 elements (Tl In Ga and Al) .The atomization of these elements exhibits a number of anomalies the nature of which is subject to different interpretations. When atomizing salts containing Al double peaks3' and anomalously high residence times of the atoms within the graphite furnace have been recorded.32 Mass ~pectrometry~~ has shown the complex composition of the gas phase where aluminium sub-oxides and carbides are present along with the atomic species. Finally at low rates of heating for microgram amounts of A1 explosive bursts of atomic absorption were o b ~ e r v e d ~ ~ ? ~ ~ against the back- ground of a smooth increase in the signal. The atomization of In Ga and TI in a graphite furnace is also complicated and not fully understood. It is for example accompanied by the formation of molecules in the gas phase which have been classed as hitherto unknown carbides.36 On the other hand mass spectrometric investi- gations of the atomization of In and Ga3' showed an important role for the sub-oxides.The depressive effect of certain halides on the absorption of these elements is well known.38 A characteristic feature of the absorption signals of the elements is their complex profile consisting of two peaks.39 The dynamics of formation and the structure of the atomic layers of these elements is investigated in the present work. The results of a study of the distribution of Al In and Ga molecules across the graphite furnace will be presented in Part 2 of this Experimental SSF Unit Image formation A schematic diagram of an SSF unit is shown in Fig.1. The main purpose is to obtain a monochrome image of the interior of a tube atomizer (3) in the cine-camera frame (13). The principal optical component is a fast MDR-2 monochromator (LOMO USSR) assembled according to the Czerny-Turner design. Two spherical mirrors (7) and (9) with focal lengths of 400 mm and two flat rotating mirrors (6) and (10) serve as the objectives. The monochro- mator is equipped with three 140 x 150 mm replaceable diffraction gratings (8) used in the first order. The gratings have 300 lines mm-l for the 1-2.5 pm spectral region 600 lines mm-l for 0.5-1.25 pm and 1200 lines mm-l for 200-600 nm. In the experiments described a slitless monochromator was used the entrance and exit slits were removed and the light passed through a 30 mm aperture in the casing.The linear spectrum light source (l) the emission from which illuminated the atomizer cavity (3) was placed at the focal point of the lens (2). Numerical estimates of the unit show that the light passing into the monochromator through the diaphragm (4) and the lens (5) forms an image of the tube cavity on the diffraction grating surface with an enlargement factor of 1.14. Subsequently the dispersed emission was focused by the lens (1 1) with an 1 . 2 3 Fig. 1 Schematic diagram of the shadow spectral filming unit. Lower set of figures are distances in mm between the elements of the unit. Focal lengths for lenses 2 5 and 11 are 60 614 and 112 mm respectively.(For ,details see text) aperture diaphragm (1 2) onto the cine-camera frame (1 3). By rotating the grating the image of the furnace cavity can be obtained from the light of the required spectral line. Thus at the distances indicated in Fig. 1 the monochrome image of the tube cavity (reduced 3.5 times) is obtained on the film. The cine-camera was used without its lens. In this situation the entire optical system of the unit (4)-(12) served as an objective. Lenses (2) ( 5 ) and (1 1) were made of quartz which made filming in the UV region possible. As the unit is designed for use over a wide spectral range of extended objects all types of optical aberrations are characteristic. According to estimate^,^' when the wave- length of light in the UV range is changed by 100 nm the focal length of the lens changes by 0.5 cm owing to chromatic aberration.In the present experiments when filming at different wavelengths this type of distortion could be compensated for by a corresponding displacement of the camera along the optical axis. The role of astigmatism is even more significant. calculation^^^ have shown that in the Czerny-Turner system even a point light source is repre- sented by two 0.3 mm long sections separated by a distance of 2.8 mm. The simplest way of reducing these distortions is the use of paraxial beams obtained by decreasing the diameter of the 'pencil' of rays. In an SSF unit this can be achieved by introducing a 3 mm diaphragm (1 2) into the system the aperture angle in object space (relative to the atomizer) becomes 4.8 x rad whereas in image space it is 16 x rad.This technique also decreases spherical aberration. The use of a smaller aperture however results in a decrease in the sensitivity of the unit. The resolution of the diffraction grating in this sytem is low owing to the small operational area thus with a 1200 lines mm-' grating the resolution is only 8200. This nevertheless is acceptable when electrodeless discharge lamps (EDLs) are used as the prime sources as they emit a relatively small number of atomic lines the intensities of which are sufficient for obtaining a bright monochrome image. In the experiments described below conventional electrodeless VSB-2 lamps were used. A typical picture obtained by the unit is presented in Fig.2. Frame (a) shows the interior of a cold furnace from the light of the Ga resonance line; the same picture but with the atomizer heated to 2000 "C is shown in frame (b). It can be seen that the luminous furnace walls produce a bright halo due to the light of the hot central part. The intermedi- ate dark ring in Fig. 2(b) is an image of the cool far end of the furnace and the holder. The near end of the tube also has a lower temperature and therefore remains invisible.JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY OCTOBER 199 1 VOL. 6 507 Fig. 2 Image of the inner tube of the furnace in the light of the Ga 4 17.2 nm line. (a) Cold furnace; (b) furnace temperature 2000 "C and; (c) as for (6) but without the prime source The image of a furnace heated to the same temperature but without the translucent emission from the prime source is shown in Fig.2(c). It can be seen that the light from the furnace walls does not introduce any distortions into the image of the interior of the furnace. Spectral spatial and temporal resolution The spectral resolution of the unit is defined as its ability to produce an image of the atomizer cavity from the light at the required wavelength. This depends on the disper- sion of the spectrometer and the diaphragm of the field of sight (4). All the experiments in this work were carried out using a 1200 lines mm-I grating providing an angu- lar dispersion of 0.044 degrees nm-I. Calculation of the spectral resolution of the unit at the diaphragm (4) with a diameter of 6 mm yielded 39.5 nm. With a 1 mm diaphragm the resolution is 12.5 nm.Further decrease of the diameter leads to vignetting of the atomizer cavity image. The effect of the size of the diameter entrance diaphragm (4) is illustrated in Fig. 3 with an example of the resolution of the Ga lines at 392.4 403.3 and 4 17.2 nm. The light flux passing through the unit was recorded by a photomultiplier which was installed in place of the cine-camera. It is evident from curve A that using a diaphragm (4) with a diameter of 6 mm these spectral lines are not resolved. The use of a 2 mm diameter diaphragm results in an almost complete separation of the light fluxes (curve B) their magnitude decreasing insignificantly. Further decreasing the diameter of the diaphragm (4) to 1 mm somewhat improves the spectral resolution (curve C) however this is accompanied by a considerable loss in image brightness.Therefore in the experiments described below the diameter of the dia- phragm (4) is 2 mm which ensured a spectral resolution of about 15 nm. The location of the spectral lines of the elements under study at a distance of < 15 nm is generally not convenient for SSF as it results in the superimposition of images. However when the difference in line sensitivity is small the proximity of the spectral lines can be used to advantage. For instance the 394.4 and 396.1 nm A1 doublet cannot be resolved by the unit and a sufficiently sharp image of the furnace cavity is obtained simultaneously in the light of these two wavelengths. As they differ in sensitivity by a factor of only 1.5 the absorption picture is qualitatively retained the gain in brightness making filming at a higher speed possible.Images of the furnace cavity obtained at characteristic wavelengths of the Ga spectrum are presented in Fig. 4. The cross-hairs are formed by 0.4 mm W wires (numbers 1 and 3 are located 1 mm from the ends of the furnace number 2 is located at the centre). It can be seen from Fig. 4 (frames 2 and 4) that when the spectral lines of Ga are isolated the image of the entire furnace volume is sufficiently sharp and undistorted. In order to evaluate the spatial resolution of the unit the thickness of the wires in the pictures was " 390 400 410 420 430 Wavelengthlnm Fig. 3 Dependence of intensity of the light beam striking the recording system on the wavelength.Diameter of the diaphragm (Fig. 1,4) A 6 B 2; and C 1 mm. Diaphragm (Fig. 1 7) diameter equals 3 mm measured and compared with their true thickness taking into account the magnification of the photograph. These evaluations have shown that the spatial resolution was approximately equal to 0.15 mm which was sufficient for investigation of the 6 mm furnace. Similar filming of the cross-hairs was carried out with a furnace heated to 2000 "C. No image distortions were detected. This suggests that a change in the gas refraction coefficient due to heating has little effect on the formation of the images. Temporal resolution of the unit is determined by the speed of filming. Filming was performed with a 16 mm cine-camera at speeds of 12 16 and 24 frames s-* half the time being taken by exposure the other half by frame transport.Thus the minimum temporal resolution is equal to 0.04 0.03 and 0.02 s respectively. Recording and information treatment In order to record non-stationary absorbing layers A-2 negative cine-film (TASMA USSR) with a spectral sensitiv- ity range of from 230 to 650 nm was used. The speed was 450 ASA and the contrast coefficient 1.5. Density S produced on the film by light that had passed through the absorbing layer was measured by a microphoto- meter. A calibration graph of A=f(S) where f is a mathematical function was plotted in order to translate these densities into more informative absorbance values (A). For this purpose the neutral filters accounting for certain absorbance values were filmed before the experi- ment.As an example Fig. 5(a) presents the results of filter filming on the T1 377.5 nm resonance line. A current of 140 mA was supplied to the EDL which produces a moderate brightness of the line. The figures over the frames are the photometric values of the film densities the lower figures being the absorbance produced by the filter. The calibration graph plotted on the basis of these values is given in Fig. 5(b) curve A. It can be seen from the figures presented that the density can be visualized roughly within the range 0.2-1.1. As follows from curve A Fig. 5(b) these densities correspond to the absorbance range 0.2<A< 1.2. Absor- bance variations beyond this range are virtually indistin- guishable visually. This visual range may be considerably broadened however if the fact that densities depend not only on the absorbance of the absorbing layer but also the brightness of the prime source is taken into account.Fig. 5(b) curve B shows a calibration graph obtained under the same conditions as for curve A but with a brighter light source (the current supplied was 175 mA). Much larger absorbance values correspond to the visual range508 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY OCTOBER 199 1 VOL. 6 Fig. 4 Images of furnace interior and W cross-hairs obtained at various wavelengths of the Ga spectrum. Wire 2 is located in the centre of the furnace wires 1 and 3 are located 1 mm from the ends of the furnace L 0 1 2 Density Fig. 5 (a) Filming of neutral filters at the T1 377.5 nm line and (b) calibration graph of A=f(S) plotted from the results obtained from (a).Current supplied to EDL A 140 mA; and B 175 mA 0.2 <S< 1.1 in this instance. Absorbing layers producing absorbances of less than 0.8 are practically indistinguish- able against the background of bright penetrating light and only optically thick layers can be observed on the film. Thus by varying the brightness of the light source on the SSF unit the absorbing layers can be seen'roughly within the range 0.2-2 A (Fig. 5). (Note that photometry also provides reliable information in the range of densities that are visually indistinguishable up to A = 0.03. This mini- mum value is determined by the non-uniformity of furnace lighting and inhomogeneity of the film emulsion.) When considering the features of SSF it should be noted that the correct data on atomic distribution are obtained only in the absence of non-selective absorption.This limits the possibilities of the method for the investigation of the atomization of samples with complex matrices. Thus before using SSF it is necessary to ensure that there is no background scatter or molecular absorption; in the present study such .measurements were made using a standard spectrometer with background correction. Temperature measurement The temperatures of the wall and platform were measured by using a photodiode which had been calibrated prior to use using a disappearing filament pyrometer (OPPIR- 17 USSR). The emission from the platform was recorded through the dosing hole. The contribution of reflection from the hotter tube was taken into account by a procedure described previo~sly.~~ The wall temperature was estimated by recording the emission from the outside of the tube near the dosing hole.This simplified procedure is not very reliable but the accuracy (-t 50 "C) is sufficient for subse- quent qualitative analysis. Procedures A 10 pl volume of the solution was injected into the furnace with a microlitre pipette. The instrumental parameters of atomizer operation when the sample was deposited onto the wall and the platform are presented in Tables 1 and 2 respectively. It can be seen from the tables that compara- tively low heating rates were used in order to separate in time the different atomization mechanisms. The following lines were used for SSF T1 377.5 nm In 410.1 and 451.1 nm Ga 403.3 nm A1394.4 and 396.1 nm (superimposed).The choice of these lines makes it possible to demonstrate the operation of the unit over a wide range of the spectrum. The selected lines do not coincide with the molecular absorption spectra formed in the UV region during the atomization of these elements.36 Reagents and Materials All of the experiments in this paper were carried out using the atomizing system Graphite- 1 ,44 supplied with aJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY OCTOBER 1991 VOL. 6 509 Table 1 Instrumental parameters for the Graphite-1 system with samples injected on the wall Time/s Internal gas flow/ Step TemperaturePC Ramp Hold ml min-l 1 90 5 50 150 2 300 (Tl) 1 20 150 1000 (Al) 10 10 150 1700 (Al)* 20 1 150 3 1800 (Tl) 5 5 0 2400 (Al) 4 4 0 2300 (AI)* 30 0 150 * Step for explosive atomization.Table 2 Instrumental parameters for the Graphite-1 system with samples injected on the platform Time/s Internal gas flow/ Step Temperature/"C Ramp Hold ml min-l 1 260 1 60 150 2 400 (Tl In Ga) 1 20 150 1300 (Al) 1 20 15 3 1800 (Tl) 5 5 0 2000 (In Ga) 5 5 0 2400 (Al) 4 4 0 Saturn-3 spectrometer (OKBA Severodonezk USSR). The atomizer was mounted in the optical train without any change in the design of the unit. Perkin-Elmer pyrolytic graphite coated graphite tubes (28 mm in length 8 mm o.d. 6 mm id.) and platforms were used in this study. Stock solutions at a concentration of 1000 mg 1-1 were prepared by dissolving TlN03 A1(N03)3 A12(SOJ3. 1 8H20 and In(N03)-3H20 in doubly distilled water.The stock solution of Ga with a concentration of 50 mg 1-l was prepared by dissolving 3.36 mg of Ga203 in 1 ml of concentrated HCl and subsequently diluting with water to a volume of 50 ml. A suspension of Ga203 was prepared by carefully mixing finely dispersed Ga203 powder with water in order to obtain a metal concentration of 50 mg 1-l. Test solutions were prepared by dilution of the stock solutions. All of the reagents were of analytical-reagent grade. Argon gas of the highest quality was used as the internal purge gas and as the sheath gas. Results and Discussion Thallium Owing to the high volatility of T1 the use of a platform and low ashing temperatures not exceeding 400 "C are rec- ommendedl7 for the determination of T1 in a graphite furnace.However under these conditions the atomization impulse has a complex shape especially noticeable at comparatively low rates of furnace heating. In Fig. 6(a) curve A represents the absorbance profile for 10 ng of T1 at a heating rate of about 200 "C s-' (lines T and Tp are images of the wall and platform respectively). This consists of two peaks which demonstrates the complexity of the atomiza- tion mechanism. In order to clarify the situation the same process was recorded by using the SSF method. The visual record of the dynamics of the formation of the T1 absorbing layer is given in Fig. 6(b). Recording began simultaneously with the atomization cycle and thus the first frame in the figure refers to a furnace temperature of 400 "C (see Table 2). Filming was carried out at 24 frame s-l with a moderately bright EDL which is described by curve A in Fig.5(b). As under these conditions absorbance values of less than 0.1 are visually indistinguishable only the second (high-temperature) peak of T1 atomization can be seen on the film. The atoms vaporizing directly from the platform subsequently fill the entire cavity of the furnace. Initially (frames 1 63- 172) weak non-uniformities are observed in the distribution of atoms across the furnace with a greater localization occurring near the platform. Subsequently (frames 175-1 95) these non-uniformities level out. In order to image a weaker low-temperature peak it is necessary to increase the amount of metal deposited. Fig. 7(a) shows the onset of the atomization of 500 ng of T1 which is the appearance of the first peak of atomization. [Note that the peak appears earlier for such an excessive amount of the metal; see Fig.8(6).] The results exhibit a clear anomaly the atoms enter the gas phase not from the platform where the sample was deposited but from the upper furnace wall. This leads to an inverse distribution of T1 concentration in the cross-section of the furnace with a maximum not above the atom source (the platform) but at the opposite wall. A possible explanation of this phenome- non is that the sample first vaporized as molecules which are invisible in the light of the T1 377.5 nm atomic line. Subsequently the molecules dissociate (or are reduced) on the higher temperature graphite wall and re-enter the furnace as atoms.Fig. 7(a) might be the image of this secondary process. In order to avoid the occasional deposi- tion of the sample in the region of the dosing hole when the pipette is inserted these experiments were repeated with the sample being deposited and dried on the platform outside the furnace. The results coincided completely with those presented in Fig. 7(a). The assumption regarding the molecular vaporization of T1 is also confirmed by the following experiment. A sample containing 500 ng of T1 was deposited on the platform and was ashed for 20 s at a temperature of 500 "C. Later the platform was removed while the furnace was re-heated following the same pro- gramme as before (Table 1). The absorbance profile appearing in this instance is presented in Fig. 6(a) curve B.This secondary signal increases with an increase in temper- ature and ashing time. This confirms the assumption that the sample first vaporizes as molecules which are subse- quently atomized on the furnace wall. The use of excessive amounts of metal leads to rapid saturation of density therefore the second peak of T1 atomization in Fig. 7(a) (final frames) is impossible to distinguish. In order to be able to observe this peak it is necessary to use a brighter light source. As shown by Fig. 5(b) an increase in the power supply to the EDL to 175 mA makes it possible to see the absorption layer in the region 0.3tAt1.7 (curve B). The results of filming the same atomization of 500 ng of T1 in this regime are shown in Fig. 7(b). It is a direct continuation of Fig.7(a) and it shows the high-temperature peak of TI atomization. It can be seen that in this instance atomic vapour forms directly at the platform and this process repeats all the features presented in Fig. 6(b). In addition to the initial inverse atomization the results given in Figs. 6 and 7 exhibit two further pronounced features. (2) It can be seen from Fig. 7 that no T1 atoms are present under the platform long after the complete filling of the upper half of the furnace (frames 76-94). This means that in the initial stages the length of the absorbing layer does not exceed that of the platform (1 5 mm). Thus at the beginning of atomization the distribution of T1 atoms is definitely non-homogeneous not only in the transverse but also in the longitudinal direction.In other words there is a pronounced anisotropy in the rate of propagation of T1510 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY OCTOBER 199 1 VOL. 6 1600 1400 800 3 5 7 9 1 1 Time/s Fig. 6 Absorbance versus time profile for the atomization of 10 ng of T1 curve A. Curve B is a result of the secondary atomization of T1 from the wall (see text). Curves T and T present the image of wall and platform temperatures respectively. (b) Imaging the process by SSF. Figures over the frames are their numbers beginning at the onset of atomization. The images were recorded at 24 frames s-* vapour in the graphite furnace i.e. the atoms propagate much faster in the transverse than in the longitudinal direction. (ii) Frames 67-79 in Fig. 7(a) clearly show a sharp decrease in the concentration of Tl atoms on the surface of the platform.As the temperature of the plat- form at this time is approximately 300 "C lower than that of the walls [see Fig. 6(a)] it can be assumed that the T1 vapour is adsorbed (or condensed) intensely at the mark- edly cooler platform. Subsequently as the platform is heated they re-enter the gas phase in the furnace. Thus there are at least three ways in which free T1 atoms can be formed as a result of dissociation (or reduction) of molecules at the wall; by direct atomization from the platform; and through the secondary desorption of atoms from the platform with heating. When the absorbance is recorded using a conventional spectrometer [curve A Fig. 6(a)] the monochromator entrance slit generally isolates a narrow zone along the diameter of the furnace.Therefore the signal A(?) (where ? represents time) is largely formed by the atoms located in this zone. As can be seen from Figs. 6 and 7 the distribution of the TI atoms across the furnace is non-homogeneous. At the same time as has been shown by calculation^,^^ the magnitude of measured absorbance depends not only on the number of absorhlng atoms but also to a great extent on their distribution along the diameter of the furnace. In order to obtain quantitative information on the non- uniformities of the atomic layer photometric measure- ments of the pictures should be carried out. An example of such photometry is given in Fig. 8 for the atomization of 500 ng of TI as described above. The probing beam of the microphotometer had a section with a 0.5 mm side.Data presentation is similar to that described previously:21-26 the absorbance vemm height profiles at time intervals of 0. I7 s (starting at 2.4 s) prior to peak absorbance is shown in Fig. 8(b) and a similar representation for times after peak absorbance is shown in Fig. 8(c). Fig. 8(d) shows the cross- section of the furnace and the region subjected to photome- try (shaded area). (Note that photometry considerably widens the range of densities that can be distinguished.) The microphotometer can reliably record non-uniformities of density within the absorbance limits of from 0.05 [the onset of atomization in Fig. 8(b)] to 1.5. The figure confirms all of the conditions for Tl atomization formulated earlier initially atoms enter the gas phase from the upper wall of the furnace leading to the inverse distribution of concentration [lower curves in Fig.8(b)]; later the flow of atoms from the platform begins to dominate which results in a normal distribution of atomic concentration with a maximum near the platform (upper curves); and finally at atomization end [Fig. 8(c)] the distribution of atoms proves to be uniform across the entire furnace. This shows that at this stage T1 vapours diffuse through the open ends of the furnace without noticeable interaction with the graphite wall. The results given above refer to platform atomization. Fig. 9(4 presents the dynamics of the formation of the absorption layer of the TI atoms for 20 ng of the metal using atomization from the wall. In this instance there are no anomalies in the distribution of the atoms.Initially [frames 44-52 and Fig. 9(b)] the distribution of atoms is as expected with a pronounced concentration maximum near the bottom of the furnace where the sample was deposited. Subsequently as the sample volatilizes the atomic distribu- tion becomes practically uniform [Fig. 9(4 frame 72 and Fig. 9(c)] which verifies the absence of any appreciable interaction between the T1 vapour and the wall. Such behaviour of the atoms being atomized from the wall is in complete agreement with the earlier investigations by Holcombe and co-workers.21-2* Indium Fig. 10(a) presents a record of the atomization of 100 ng of In from the platform. Recording was carried out at theJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY OCTOBER 1991 VOL.6 51 1 Fig. 7 Dynamics of the formation of the absorption layer of the T1 atom using atomization from the platform for 500 ng of the metal (24 frames s-l). Fig. (b) is a continuation of (a) 1.2 1 .o 0.8 0.6 0.4 0.2 0 0 3 6 2 6 10 0 3 6 Height /rn m Time/s Heig ht/mm Fig. 8 (a) Absorbance versus time profile for 500 ng of TI using atomization from the platform. Absorbance versus height contours at 0.17 s time intervals ( h ) prior to and (c) after the peak absorbance. ( d ) Cross-section of the atomizer. The shaded area in ( h ) and (c) indicates the platform 4 10.18 nm line as an intensive band of absorption by the In molecule is superimposed on the normally recommended 303.9 nm line.36 It can be seen that in spite of the atomization of a large amount of metal the distribution of the atoms across the furnace is fairly non-uniform.The main features of the process however are generally the same as for T1 as described above. In fact initially (frames 97-104) there is an inverse distribution of atoms with a sharp decrease in concentration near the platform. This effect is more evident during recording at the non- resonance wavelength of 45 1.1 nm. In this instance Fig. 10(b) gives an image of the behaviour of excited In atoms. The increase in the non-uniformity might be owing to the greater sensitivity of the excited atoms to temperature gradients. As can be seen from frames 76-82 Fig. 10(b) the temperature difference between the platform and the upper wall is 250 "C. It can be easily estimated that as the temperature increases from 1200 to 1 500 "C the concentra- tion of In atoms at the excited 2P3,2 level from which a transition with R=451.1 nm begins increases by about 20%.The dynamics of the formation of the absorbing layer of In atoms during sample atomization from the furnace wall512 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY OCTOBER 199 1 VOL. 6 0 3 6 Heig ht/m m 2 4 6 Timeis 1400 9 F +- 1200 g k I- F 'c--i 1000 0 3 6 HeighVmm Fig. 9 (a) Absorbance versus time profile for 20 ng of TI using atomization from the wall (broken line indicates change of wall temperature). Absorbance versus height contours at time intervals of 0.12 s (b) prior to and (c) after the peak absorbance. ( d ) Imaging of the process by SSF (1 6 frames s-') is virtually the same as for a similar situation using T1 (Fig.9). At the start of atomization there are weakly defined non-uniformities with a greater localization of the atoms near the bottom of the furnace where the sample was deposited. At the point of decay of the impulse the absorbing layer is virtually homogeneous. Gallium Atomization of 50 ng of Ga from the graphite platform recorded at the resonance wavelength of 403.3 nm is shown in Fig. 1 l(a). It can be seen that the formation of the absorbing layer proceeds in the same way as with the atomization of TI (Fig. 7). In this experiment hydrochloric acid was used to prepare the sample. As it is known that chlorine has a strong effect on the formation of the Ga signal,17 the atomization of a suspension of Ga203 was recorded.The results given in Fig. 1 l(b) almost coincide with the preceding record Fig. 1 l(a). This means that the effect of chlorine on the atom distributions presented can be ignored. It is noteworthy that frames 104-1 10 Fig. 1 l(b) indicate a sharp decrease in the concentration of Ga atoms near the lower surface of the platform. This shows that adsorption (or condensation) occurs only on the upper surface of the platform (frames 99-104) but also in the lower part. Thus it may be assumed that the considerably cooler platform acts in the graphite furnace as a 'sponge' first absorbing the atoms and then returning them to the gas phase in the course of subsequent heating. The imaging of Ga atomization from the furnace wall again yields a record that almost coincides with Fig.9 for a similar situation using T1 therefore it is not presented here. In summary it can be said that the atomization of T1 In and Ga proceeds via a fairly similar mechanism. The differences in the dynamics of the formation of their absorbing layers is of a quantitative rather than a funda- mental nature. A vital role is played by the molecules of these elements particularly in the initial stages of the atomization. The formation of molecular layers will be dealt with in Part 2 of this which gives a more detailed description of the atomization mechanism of these elements. Another feature common to all of the elements in question is a pronounced anisotropy of the propagation of the vapours in the furnace volume. The most likely cause for this is the considerable longitudinal non-isothermality of the graphite tube.Indeed the temperature at the ends of a 28 mm HGA-type furnace does not exceed 1000 "C when the centre of the furnace is at 2000 0C.46 Thus longitudinal atomic diffusion occurs in a medium with a rapidly decreasing temperature. This leads to absorption and/or vapour condensation at the cooler ends of the furnace. Another cause for the anisotropy could be the thermal expansion of the inert gas within the furnace as a result of its being heated. As the furnace has open ends and a dosing hole in the centre a picture of the gas flow during itsJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY OCTOBER 199 1 VOL. 6 5135 14 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY OCTOBER 1991 VOL. 6JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY OCTOBER 199 1 VOL.6 515 expansion will be very complicated. For the extreme situation in the absence of*€he dosing hole simple expres- sions for the velocity of gas expansion through the ends of the furnace have been ~btained.~’~~* However under gas stop conditions the ends of an HGA-type furnace are to a certain extent closed to the expanding gas. Therefore it will partially escape through the sample aperture. The velocity of this flow v can be estimated on the basis of another idealized assumption i.e. that the ends of the furnace are completely closed (Fig. 12). The mass of gas inside the furnace will diminish with heating by the law rn(t)=p(t)V where pmconstantlir is the gas density and V the furnace volume varying in reverse proportion to the temperature V= nD2 L/4.Hence the rate of decrease in the mass equals 7TD2 constant dT X L X - X- V x - =- dm dt dt 4 P dt dP -= On the other hand the mass of gas dm escaping during the time dt through the dosing hole of diameter d is dm = v dt nd2p(T)/4 Assuming that in the course of the expansion all of the gas escapes through the sample aperture these values may be equated. This yields an expression for the velocity of gas flow through the dosing hole L dT (s T x d t v = - x - For HGA-type furnaces L=2.8 cm D=0.6 cm d=O.1 cm. As can be estimated from the temperature data presented in Figs. 6 10 and 11 for the onset of atomization of the elements being investigated (l/T) x (dT/dt)=O. 15 s-l which gives ve= 17 cm s-l. At the same time the velocity (vd) of longitudinal diffusion of vapour may be estimated by the formula vd%L/rd where 7 d = L 2 / 8 D * is the ‘diffusion’ residence time of atoms.The diffusion coefficients D in argon for the atoms of the elements in question are listed in ref. 49. Numerical estimates give v,=5-6 cm s-l. Under real conditions the gas will also escape through the ends of the furnace. However even if only 20% of the argon escapes through the dosing hole the values v and v prove to be comparable. The relative role of the thermal expansion is likely to decrease sharply as furnace temperature increases. How- ever at the commencement of atomization of elements of low and medium volatility such as TI In and Ga the contribution of this effect to the formation of the absorbing layer may be considerable.Aluminium Atomization in the analytical regime presupposes con- tinued heating of the atomizer to a given temperature when the internal sheath gas has been stopped. Fig. 13 shows film images of vapour filling the graphite furnace volume during Fig. 12 Idealized pattern of the thermal expansion of inert gas within the furnace. (See text for details) the atomization process. The images were recorded at 16 frames s-l. Atomization of this amount of metal (10 ng) produced sufficiently high densities of atomic vapour for direct observation of the process. It can be seen from the film that despite the absence of an internal argon flow there are pronounced gradients of A1 atom concentration in the cross-section of the atomizer with the atoms being local- ized near the walls of the graphite furnace. The sharp decrease in vapour density near the sample aperture is also noteworthy. This loss may be due to diffusion and thermal expansion in addition to the formation of stable oxides and cyanides resulting from the diffusion of air through the dosing hole.50 In this instance atomization occurs against the back- ground of a rapid rise in the temperature of the furnace.This is indicated on the film by the expanding light ring which is due to emission by the central part of the atomizer walls. The inner dark ring shows the more distant cooler end of the furnace which testifies to the presence of considerable longitudinal temperature gradients during heating of the furnace. Fig. 14(a) shows the record of atomization of 5 ng of A1 under conditions similar to that for stabilized temperature platform furnace (STPF) (Table 2).It can be seen that with atomization from the platform the general tendencies in the distribution of A1 vapour are the same as in the preceding example there are pronounced concentration gradients with a greater concentration of atoms being located near the furnace walls. In order to obtain quantitative data from the film photometric measurements were carried out using a microphotometer which enabled the construction of absor- bance contours by means of a calibration graph such as that shown in Fig. 5(b). The results are presented in Fig. 14(b) for four characteristic stages of atomization i. e. onset (frame 73) maximum (78) and drop (102 and 118) in the vaporization pulse.Again the concentration distributions prove to be in- verse with the maximum amount of the A1 atoms being located not above the platform but a considerable distance from it. This effect is not as pronounced as in the preceding example of TI and Ga but it is fairly distinguishable. This testifies to the considerable role of the A1 molecular species in the formation of atoms. Fig. 14 also shows an interesting vapour re-distribution during sample atomization at the Fig. 13 Dynamics of the formation of the absorption layer of the A1 atom using atomization from the wall for 10 ng of the metal. The images were recorded at 16 frames s-’516 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY OCTOBER 1991 VOL. 6 73 78 102 118 Fig. 14 (a) Dynamics of the formation of absorption layer of the A1 atom using atomization from the platform for 5 ng of the metal.The images were recorded at 24 frames s-'. (b) Structure of the absorption layer at characteristic moments. The figures indicate absorbance values onset [frame 73 Fig. 14(b)] most of the A1 atoms are distributed in the furnace volume above the platform whereas they are largely localized under the platform at the end of the cycle (frames 102 and 1 18). For comparison recording of the atomization impulse in question by the traditional photoelectric method is shown 0.9 Q) 0.6 u m G s n a 0.3 0 3 5 7 Ti m e/s 2500 u 9 2 w 2000 g P I- 1500 1000 Fig. 15 Absorbance versus time profile for 5 ng of A1 using atomization from the platform. Curves T and Tp indicate the change of wall and platform temperature respectively in Fig.15 in which the stages represented in Fig. 14(b) are indicated. Besides imaging the dynamics of the formation of the atomic absorption layer the data presented afford addi- tional information on the character of platform heating. The images in Fig. 14(a) vividly show the lag in heating of the platform relative to that of the walls of the atom- izer when sample evaporation commences the temperature of the wall of the furnace will have almost reached the maximum value which is demonstrated by the bright halo that does not change with time (see also curve T in Fig. 15). In contrast the platform temperature Tp is constantly rising to reach its maximum value only at the end of the atomization cycle (frames 98 and 118).A marked inhomo- geneity in platform luminescence in the cross-section can be seen the emission of the surface layers of the platform is noticeably brighter than that of the interior (frames 83-87). This might be because of the reflection of light emitted by the walls but might also indicate non-uniformity in platform heating in the cross-section. Spikes Imaging This effect first described by L'vov and c o - w ~ r k e r s ~ ~ ~ ~ ~ has been the subject of numerous investigations and reviews on the subject have been p ~ b l i s h e d . ~ ~ - ~ ~ The effect consists of a rapid release of A1 against the background of a slow increase in the signal [Fig. 16(a)]. Spikes are regularly observed at slow heating (5-20 "C s-l) of microgram amounts of aluminium oxide.The SSF method gives a unique opportu-JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY OCTOBER 199 1 VOL. 6 2 0 10 20 Time/s 2300 1700 517 Fig. 16 (a) Explosive atomization of 1 pg of A1 and ( h ) imaging of a single spike by the SSF method. The images were recorded at 24 frames s-l 0.5 al c m e 5 3 0.25 2 6 10 14 Ti me/s 2000 9 f! 2 \ 4- i! 1500 ? 0 Fig. 17 (a) Impulse atomization of 8 ng of A1 with an imposed explosive effect and (b) its imaging with the use of SSF. The images were recorded at 24 frames s-'518 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY OCTOBER 1 99 1 VOL. 6 nity to record the dynamics of development of each individual spike. As an example Fig. 16(b) shows an image of the second impulse of Fig. 16(a) occurring at 19 10 "C. It can be seen that the A1 atoms completely fill the cross- section of the atomizer in less than 0.04 s (frames 2 and 3) i e .the atomization has an explosive character. Subsequent dissipation of the vapour occurs near the location of the dosing hole within about 0.2 s (frames 4- 10). The explosive process in question is initiated and terminated at the side wall of the atomizer (frames 2 and 8) rather than at the bottom of the atomizer where the sample was initially deposited. It is evident that the distribution of A1 in the cross-section of the furnace is distinctly non-uniform at any moment in time. However in this instance the distribution is largely affected by internal gas flow (150 ml min-l). The spikes may also occur with the atomization of A1 using a more typical heating rate and amounts of sample.Under these conditions the spikes are randomly superim- posed on the principal absorption impulse thus introducing large errors in the results. A similar situation arose in the determination under STPF conditions of Vs4 and P.ss Fig. 17(a) shows the absorption signal obtained when atomizing 8 ng of A1 with a superimposed spike. The relevant record of the image is presented in Fig. 17(b) where the second frame shows an explosive impulse. It can be seen that the signal is initiated at the furnace walls rather than at the platform where the sample was deposited. This confirms the earlier assumptions2 that the explosive effect is due to the secondary atomization of A1 vapours adsorbed at the walls of the graphite furnace. It is interesting to note that in this instance also a picture similar to that shown in Fig.14 is observed by the end of the atomization cycle the atomic vapour is localized under the platform (frames 70-72). As this phenomenon is observed under gas stop conditions the following qualitative explanation can be suggested. The platform divides the atomizer into two parts unequal not only in size but also in terms of the composition of the gas phase. Evidently the concentration of oxygen under the platform is minimal whereas the concentration of mole- cules containing carbon is likely to be at a maximum. Thus at the bottom of the furnace conditions are created that are considerably more reducing than those above the platform. This should bring about an increase in the dissociation of the A1 molecules.40 The results of the investigation of different atomization regimes for A1 have been presented. A common feature of all the examples considered is a marked inhomogeneity of the distribution of atoms across the furnace with their pronounced localization near the furnace walls.In order to make unequivocal conclusions concerning the atomization mechanisms of the elements involved the distribution of the molecules in the furnace volume must be investigated. These results will be presented in Part 2 of this The authors are grateful to V. L. Makarov for useful comments on the results obtained in this work. References L'vov B. V. Atomic Absorption Spectrochemical Anaiysis Adam Hilger London 1970. Littlejohn D. and Ottaway J. M. Analyst 1979 104 1 138.Baxter D. C. Frech W. and Lundberg E. Analyst 1985 110,475. Dittrich K. and Stark H.-J. J. Anal. At. Spectrom. 1986,1,37. Dittrich. K. and Stark H.-J. J. Anal. At. Spectrom. 1987 2 63. Bolshov M. A. Zybin A. V. Koloshnikov V. G. and Smirenkina I. I. Spectrochim. Acta Part B 1988 43 519. 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 Bekov G. I. Radaev V. N. andLetokhov V. S. Spectrochim. Acta Part B 1988 43 491. Laser Analytical Spectroscopy ed. Letokhov V. S. Nauka Moscow 1986. Gilmutdinov A. Kh. and Zakharov Yu. A. Izvestiya Akad. Nauk SSSR Ser. Fiz. 1989 53 1820. Slavin W. Carnrick G. R. Manning D. C. and Pruszkovska E. At. Spectrosc. 1983 4 69. L'vov B. V. Pelieva L. A.and Sharnopolskii A. M. Zh. Prikl. Spektrosk. 1977 27 395. Frech W. Baxter D. C. and Hutsch B. Anal. Chem. 1986 58 1973. Frech W. Lundberg E. and Cedergren A. Prog. Anal. At. Spectrosc. 1985 8 257. Hassel D. C. Rettberg T. M. Fort F. A. and Holcombe J. A. Anal. Chem. 1988 60 2680. Ortner H. M. and Kantusher E. Talanta 1975 22 581. Ediger R. D. At. Absorpt. Newsl. 1975 14 127. Slavin W. Graphite Furnace AAS. A Source Book Perkin- Elmer Norwalk CT USA 1984. Katskov D. A. Zh. Prikl. Spektrosk. 1983 38 181. Lundberg E. and Frech W. Anal. Chem. 1981 53 1437. Gilmutdinov A. Kh. and Shlyakhtina 0. M. Spectrochim. Acta Part B 1991 46 1121. Holcombe J. A. Rayson G. D. and Akerlind N. Jr. Spectrochim. Acta Part B 1982 37 3 19. Rayson G. D. and Holcombe J. A. Spectrochim.Acta Part B 1983 38 987. Holcombe J. A. and Rayson G. D. Prog. Anal. At. Spectrosc. 1983 6 225. McNally J. and Holcombe J. A. Anal. Chem. 1987,59,1105. Droessler M. S. and Holcombe J. A. J. Anal. At. Spectrom. 1987 2 785. Wang P. X. Majidi V. and Holcombe J. A. Anal. Chem. 1989,61 2652. Katskov D. A. Savel'eva G. O. Kopeikin V. A. and Grinshtein I. L. Zh. Prikl. Spektrosk. 1988 49 7. Stafford D. J. and Holcombe J. A. J. Anal. At. Spectrom. 1988 3 35. Huie C. W. and Curran C. J. Appl. Spectrosc. 1988 42 1307. USSR Pat. 1330520 MKI G 01 N 21/67. Katskov D. A. and Grinshtein I. L. Zh. Prikl. Spektrosk. 1980,33 1004. Zheng Y. Woodriff R. and Nichols J. A. Anal. Chem. 1984 56 1388. Styris D. L. and Redfield D. A. Anal. Chem. 1987,59,2891. L'vov B. V. and Savin A. S. Zh. Anal. Khim. 1982,37,2110. L'vov B. V. Dokl. Akad. Nauk. SSSR 1983 271 119. L'vov B. V. Norman E. A. and Polzik L. K. Zh. Prikl. Spektrosk. 1987 47 711. McAllister T. J. Anal. At. Spectrom. 1990 5 17 1. Dittrich K. Schneider S. Spiwakow B. Ya. Suchowejewa L. N. and Zolotov Yu. A. Spectrochim. Acta Part B 1979 34 257. Katskov D. A. Grinshtein I. L. and Krug!ikova L. P. Zh. Prikl. Spektrosk. 1980 33 804. Gilmutdinov A. K. Zakharov Yu. A. Ivanov V. P. and Voloshin A. V. J. Anal. At. Spectrom. submitted for publica- tion (0/4000B). Salmon S. G. and Holcombe J. A. Anal. Chem. 1978 50 1714. Goldstein S. A. and Walters J. P. Spectrochim. Acta Part B 1976 31 295. Welz B. Sperling M. and Schlemmer G. Spectrochim. Acta Part B 1988 43 1187. Burakov V. S. Esilevski B. A. Misakov P. Ya. and Pelieva L. H. Zh. Prikl. Spektrosk. 1985 42 336. Gilmutdinov A. Kh. Abdullina T. M. Gorbachev S. F. and Markarov V. L. Zh. Anal. Khim. 1991 46 1481. Wu. S. Chakrabarti C. L. and Rogers J. T. Prog. Anal. Spectrosc. 1987 10 11 5. Holcombe J. A. Spectrochim. Acta Part B 1983 38 609. Gilmutdinov A. Kh. and Fishman I. S. Spectrochim. Acta Part B 1984 39 171. L'vov B. V. Spectrochim. Acta Part B 1990 45 633. Sturgeon R. E. and Falk H. Spectrochim. Acta Part B 1988 43 421.JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY OCTOBER 199 1 VOL. 6 519 5 1 L'vov B. V. Spectrochim. Acta Part B. 1989 44 1257. 52 Gilmutdinov A. Kh. Zakharov Yu. A. and Ivanov V. P. Zavod. Lab. 1989. 11 31. 53 Bendicho C. and de Loos-Vollebregt M. T. C.. Spectrochirn. Acta Part B 1990 45 547. 54 Manning D. C. and Slavin W. Spectrochim. .4cta Part B 1985 40 461. 55 Welz. B. Curtius A. T. Schlemmer G . Ortner H. M. and Birzer W. Spectrochim. Acta Part B 1986 41 1175. Paper 0/0289 7E Received J m e 2 7th I990 Accepted May 17th 1991