摘要:

JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY APRIL 1992 VOL. 7 48 1 Photon Detection Based on Pulsed Laser-enhanced Ionization and Photoionization of Magnesium Vapour Experimental Characterization Giuseppe A. Petrucci Raul G. Badini* and James D. Winefordnert Department of Chemistry University of Florida Gainesville FL 3261 1-2046 USA A resonance ionization detector (RID) based on the two-step laser-enhanced ionization of Mg in a miniature air-acetylene flame is described. The detector utilizes the 285.213 nm resonance absorption of Mg as the signal transition i.e. photons to be measured. Magnesium atoms excited by absorption of photons at 285.2 nm were further excited to higher lying energy states by absorption of photons at 571.2 or 435.2 nm from which collisional ionization could occur or by absorption of photons at 300.9 nm which directly ionized the excited Mg atoms via an autoionizing level above the ionization continuum. The miniature flame used (about 2 mm in diameter) had very low noise characteristics and the minimum detectable number of photons (MDP) was limited in some instances by the noise of the transimpedance amplifier used.The lowest MDP obtained for the excitation scheme 3s2 *S (285.2 nm)-3p IPo (435.2 nm)+ 6d 'D was experimentally determined to be 1 x lo3 (7 x 1 0-l6 J). The quantum efficiency of this excitation scheme defined as the number of electrons created per photon absorbed at the signal transition (285.2 nm) was found to be 0.75. The spectral response bandwidth of the detector in the signal transition was determined to be mostly Lorentzian in nature owing to pressure broadening in the atmospheric pressure flame with a stray light rejection ratio of approximately 1 x 1 0-5 at 100 cm-l displacement from the absorption maximum of the RID at 285.213 nm.An application of the RID is given in the detection of weak Raman scatter of carbon tetrachloride chloroform and dimethyl sulfoxide. Keywords Photon detection; pulsed laser-enhanced ionization; photoionization; magnesium vapour; reson- ance detection The efficient detection of photons within a selected spectral interval is a fundamental measurement problem in optical spectroscopy. Generally photon detectors are inherently broadband in spectral response and must be coupled to some wavelength-selective device. This combination al- though versatile has the disadvantage that increased spectral resolution can only be achieved at the expense of optical collection and throughput efficiency. Thus disper- sive spectrometric systems become insensitive when high resolution of the order of atomic linewidths is required.The flame is an experimentally simple efficient and reproducible environment capable of maintaining a con- tinuous atomic population. Stepwise atomic ionization methods in the flame have proved to be very sensitive analytical methods capable of detecting analyte concentra- tions in the picograms per millilitre range."* This high sensitivity is due primarily to the ability not only to ionize specific atoms efficiently but also detect those ions with near 100% probability.The analytical signal is derived by measuring the enhanced rate of ionization in flames caused by absorption of resonant photons. An interesting variation of the atomic ionization tech- nique is the utilization of the laser-induced ionization process as a sensitive photon detector. This concept first proposed several years has received considerable a t t e n t i ~ n ~ - ~ owing to its potential applications. The basis of this approach is as follows. A partial energy level diagram for a five-level atom is shown in Fig. 1. In the resonance ionization detector (RID) the transition 1-2 of the atom M is the signal transition. When an atom M absorbs a photon with energy equal to hvsig it is excited to level 2 M*. Simultaneously a laser henceforth referred to as the detection laser tuned to the excited-state transition 2-3 and spatially and temporally coincident with the signal photons in the RID promotes M* into a higher energy state yielding a highly excited atom M**.These highly a.i. at- \\ 4 I I I I I 2- 4 I Fig. 1 General energy level diagram for a five-level atom kul collisional rate constant (s-l); A, the Einstein coefficient for spontaneous emission (s-l); oUl transition cross-section (cm2); and ZuI photon irradiance (cm-* s-l). In all instances the subscript ul denotes starting u and ending 1 levels of a transition. The superscript p.i. indicates non-specific photionization and a.i. an autoionizing transition *On leave from Departmento de Fisicoquimica Facultad de Ciencias Quimicas Universidad Nacional de Cdrdoba SUC.16 C.C. 6 1 RA 50 16 Cdrdoba Argentina. ?To whom correspondence should be addressed. excited atoms if promoted to levels close enough to the ionization continuum (IC) are efficiently ionized by colli- sions with flame species. Alternatively M* can be photo-482 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY APRIL 1992 VOL. 7 ionized directly via an autoionizing level (a-i.) or by multiphoton absorption. The resulting ions are collected by measuring the current across an electrically biased flame. If the steady-state (thermal) population of level 2 is negligible then only when M is excited to M* by absorption of a signal photon can an ionization signal be recorded. The magni- tude of the ionization signal is directly proportional to the number of incident signal photons absorbed and hence the RID behaves as a spectrally selective photon detector centred at the signal transition hvsi,.The attractiveness of such an approach is mainly based on its high light throughput very narrow frequency re- sponse bandwidth and subsequently high stray light rejec- tion (SLR). Recently,6 the use of an RID detector has been proposed as a high-resolution detector for Raman spectros- copy. Also the Raman spectrum of carbon tetrachloride has been obtained using an atmospheric pressure flame RID based on the 285.2+470.0 nm two-step excitation scheme with Mg using a pulsed laser.' No experimental attempt however has yet been made to study systematically the fundamental characteristics of such a detector. In addition although basic qualitative assumptions can be made to evaluate different atomic systems as RID candidates except when assuming ideal behaviour of very simple excitation schemes there is a lack of fundamental informa- tion necessary to model and predict the actual behaviour with accuracy.General Considerations The choice of a suitable optical scheme for the RID is very important and depends of course on the atomic element selected. Different general optical schemes have been proposed for the RID.sv8-12 After the photon to be detected has been absorbed ionization of the excited atoms M* and M** can be achieved by direct ionization via an autoioniz- ing level (resonant) multiphoton ionization (non-resonant) field-induced Rydberg state ionization and collisionally assisted ionization following stepwise optical pumping.The last on which the laser-enhanced ionization (LEI) tech- nique is based is attractive because the pumping scheme corresponds generally to bound-bound transitions allow- ing optical saturation at modest laser powers. However this type of scheme requires a collisional environment to bridge the energy gap between the highest optically excited state and the ionization continuum. The flame environment is well suited for such an excitation-ionization scheme. RID Element The RID elefnent must possess several favourable charac- teristics. (i) The number density of atoms in the originating level of the signal transition (usually the ground state) must be high enough to ensure complete absorption of all incident signal photons. Therefore in the flame RID the element must be easily atomized and must have a large free atom fraction.This will provide the needed atom density while minimizing both the amount of analyte solution introduced into the flame and consequently the number of thermal ions created reducing electrical noise in the detection process. (ii) The thermal population of the RID atoms in the first excited state (level 2) of the RID scheme must be minimal. The RID atoms promoted to this first excited state by a non-resonant process e.g. thermal excitation are available for further excitation by absorption of a photon from the detection laser and subsequent ionization thereby contributing to the charges produced by absorption of the second laser alone. However as the first excitation was not brought about by absorption of a signal photon these ions are not proportional to the intensity of the light to be measured and contribute only to the electrical noise of the detector.The energy of the signal transition will therefore be much greater than that available from the atomizer alone i.e. h~,,~>>kT,~,,,,,,. This minimizes the steady-state thermal population of the first excited state. Fctr the acetylene-air flame therefore signal transitions at wavelengths less than about 350 nm should be considered.13 (iii) The oscillator strength of the signal transition should be as high as possible (>0.5) so that again the minimum number density of the RID element is required for 100°/o absorption of signal photons. As discussed above the lower number density will serve to minimize the noise of the detector.Also the lower number density will decrease the absorption half-width resulting in an increased resolving power and SLR (see under Resolution and stray light re-jection). ( i v ) The second step (excited-state transition 2-3) should be of low energy (long wavelength) so that multiphoton ionization of the RID element (from the ground state) or of flame species is not likely. The oscillator strength of this transition should also be relatively high (:+0.05) so that optical saturation can be effected at modest laser powers. The use of modest laser powers is also beneficial in decreasing the extent of background multipho- ton ionization. ( v ) The RID element must possess a possible excitation scheme such that the effective energy defect AE% i.e.energy difference between the uppermost laser- populated level and collisionally coupled levels and the IC be no more than 3kT to ensure a high probability of cctllisional ionization.14 Alternatively the second step could directly photoionize the excited atom either state-specifi- calXy through an autoionizing state or non-specifically by absorption of multiple photons at the excited-state transi- tion wavelength. As will be discussed in a collisionally dominated atmospheric pressure atomizer the latter are not favourable alternatives to collisional ionization owing to their relatively low cross-sections and the higher laser powers required. Optical Saturation of Excited-state Transition For the LEI approach to be used as a photon detector the ionization signal must necessarily be linearly dependent on the number of incident signal photons.Also in order to attain maximum sensitivity it is imperative that once the atoms have been excited to level 2 by the absorption of signal photons they be ionized with 100% efficiency. Omenetto et aL8 stressed the importance of differentiat- ing between the quantum efficiency i.e. the ratio between the detector output event and its photon input and the ion yield i.e. the ratio between the number of electrons created during the interaction and the number of atoms present in the probe volume (see under Quantum efficiency). It is irnportant to emphasize that obtaining a quantum effici- ency of unity for the over-all detection process is contingent upon realizing an ion yield of unity also in the excited-state transition.Here an excited-state ion yield is defined as the number of electrons created divided by the number of al.oms excited into level 2. In other words it is clear that in order for the RID to be successful as a photon detector each signal photon incident on the RID must promote an atom of the RID element to the first excited state M*. At this point in the detection process there is a relatively ]low number density of atoms in level 2 all having reached that state ideally by absorption of signal photons. It is therefore also obvious that in order to achieve a quantum or detection efficiency of 100% each of these excited atoms &I* must be ionized with 100% probability i.e. the ion yiield from level 2 must be 100%.This photon detection approach can therefore be corn- p,ared with excited-state single-step LEI in a flame of M* analyte as the ionization process results from a single laser- induced transition (2-3 in Fig. 1). The ion yield from levelJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY APRIL 1992 VOL. 7 483 2 and hence the sensitivity of the detector are independent of the first step. For single-step excited-state LEI in a flame linear excitation in the transition 2-3 with nanosecond laser pulses is not expected to produce a significant ion yield.15 If however the transition is optically saturated i.e. the rate of induced absorption is equal to the rate of stimulated emission and assuming collisional ionization from level 3 the ion yield per laser pulse is given by15 where n2 (cmW3) is the atomic density of level 2 g2 and g3 are the degeneracies of levels 2 and 3 respectively kJi is the collisional ionization coefficient (s-l) from level 3 and Ati is the interaction time (s).For nanosecond pulses the interac- tion time is given by the duration of the laser pulse At (s). Only collisional ionization is considered in eqn. (I) from which it can be seen that the ion yield is independent of the laser energy when the transition is optically saturated and is dependent solely on the product of k3iAtL. Therefore to approach an ion yield of unity when the laser is tuned to the transition 2-3 the product of the collisional ionization coefficient and the laser pulse duration must be greater than unity. This means of course that for laser pulse widths of 10 ns k3i must be of the order of 1 x lo8 s-l or greater.As stated above this criterion will be met only for atoms excited to within approximately 3kT of the ionization limit. Likewise in a photoionization scheme (from level 2) the product of the photoionization rate and laser pulse duration must be greater than unity. Evaluation Criteria for RID There are several fundamental figures of merit from which one could evaluate a photon detection system including quantum efficiency sensitivity spectral and temporal response resolution and SLR. Quantum eficiency The quantum efficiency 'I of a photon detector is defined as the number of fundamental detectable events per incident signal photon at the peak of the spectral response of the detector.For a photomultiplier tube this is the number of photoelectrons ejected from the photocathode per incident photon. For the RID on the other hand the quantum efficiency corresponds to the number of charges created per incident signal photon. Therefore for a signal source that is spectrally broader than the response bandwidth of the RID only those photons falling within the absorption bandwidth of the signal transition are consi- dered in the calculation of q. The quantum efficiency of the RID is closely dependent on the RID element and the excitation scheme chosen. In order for the quantum efficiency to approach unity the relative rate of ionization of atoms optically excited to level 3 must be much greater than the sum of the rates of all the 'deactivation' pathways of the atoms in level 3 i.e.k3i+ o ~ ~ ~ ~ I ~ ~ > > A ~ ~ + k32+A3rn+ k3rn. The ion branching ratio,9 ti defined as must be equal to unity. All terms are defined in Fig. 1. As discussed above ti can approximate unity only if the atom is optically excited to within approximately 3kT of the ionization limit for collisional ionization or if the atom is directly photoionized especially by excitation through an autoionizing level. The quantum efficiency at the centre of the absorption transition is then given by where Ri ti= - Rrn+Ri ti (3) and Ri=kzi+ k 3 i + ~ 3 ~ i . 1 2 3 R21 =A21 +k21 Rrn=A3rn + k 3 m +A32 + k32 a, is the absorption cross-section of the signal transition (cm2) labs is the absorption pathlength (cm) and n is the number density ( ~ m - ~ ) of atoms of the RID element in the ground or originating state of the signal transition.From eqn. (3) if Ri>>RZ1 then 27 is proportional to ti and therefore directly reflects the probability of ionization from the uppermost laser populated level. For the determination of q the signal photons are provided by a laser tuned to the signal transition 1-2 which is made temporally and spatially coincident in the flame with the detection laser tuned to the excited-state transition 2-3. The energy of the signal laser and hence the number of signal photons are measured with an energy meter. An energy meter is however a broadband detection device therefore if the laser bandwidth is broader than the absorption (response) bandwidth of the RID a correction must be made for the photons incident on the detector that are not within the response bandwidth of the RID.This correction factor is obtained by means of a simple absorp- tion experiment. The fraction of incident photons within the absorption bandwidth that is absorbed is reflected in the factor [ 1 -exp(c~,~~n~l~~,)] which can be made equal to unity by increasing either the number density of RID element atoms or the absorption pathlength. Experimentally is obtained from the slope of a plot of the number of charges created versus the number of signal photons absorbed. Minimum detectable number of photons (MDP) and linear dynamic range (LDR) In a properly designed experiment (i.e. 100% charge collection efficiency and optimum overlap in the RID of the signal photons and the detection laser) the MDP is dependent only on the excitation-ionization scheme used.Under optimum conditions the MDP can be defined as where s is the estimate of the standard deviation of the blank (no photons at hvsig). As stated above is equal to the slope of the calibration graph of number of charge pairs created versus number of signal photons absorbed. In the absence of any electrical (ionization) gain mechanism within the RID,16 this value is ideally equal to 1. Also of importance is the range of number of incident signal photons over which the detector response is linear. The LDR can be obtained from a log-log plot of the number of charges created versus the number of signal photons absorbed and is defined by LDR= UQIMDP where UQ is the upper limit to the number of photons where the deviation from linearity has decreased by 5%.Clearly to optimize RID performance r] must be maximized while decreasing s. As is discussed below however this is not easily accomplished. MDP=slq (4) Resolution and SLR Another important quality of a spectral detection system is its ability to resolve spectrally two closely spaced lines. For484 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY APRIL 1992 VOL. 7 conventional optical detection systems the practical resolu- tion is defined ultimately by the grating dispersion. The maximum resolution however can only be achieved at the diffraction-limited slit-width. This of course limits the amount of light that is measured and a compromise must therefore be reached between acceptable resolving power and light throughput.Stray light rejection on the other hand describes the response of the detection system to wavelengths shifted from the wavelength of interest lo. Often it is reported as the ratio of light intensity falling on the detector at a given shift from ;lo to the intensity at Lo. This ratio (for 100 cm-' displacement from Ao) is of the order of 1 x for the best monochromator-photomultiplier system. The required res- olution is often obtained before adequate SLR is achieved. When one is considering only a small spectral range A& less than one order of magnitude larger than the required resolution the resolution and SLR might be viewed as converging into one entity. This applies to the RID. Both the resolution and SLR of the RID are fixed by choice of RID element wavelength scheme and atom cell and are given directly by the absorption profile of the signal transition in the RID.In a flame-based RID Lorentzian- dominated profiles of the RID element signal transition can yield resolutions of the order of picometres with corre- sponding SLRs of 1 x 1 0-5 at 100 cm-l displacement from lo. By using a low-pressure atom cell where the profile of the signal transition will be more Gaussian dominated in nature sub-picometre resolution with SLRs of < 1 x at 1 cm-' displacement can be achieved at no cost to light throughput of the detector as atomic resolution is inherent in the detection process. Temporal response The question of temporal response of the RID also deserves some comment.The RID is fundamentally an integrating detector which does not preserve temporal information concerning the input light source. Obviously photons that are not in temporal coincidence with the detection laser cannot be detected and those arriving during the more intense portions of the temporal distribution of the detec- tion laser will be detected more efficiently. The temporal response will of course also be dependent on the rate of excitation from level 1 to level 2 and on the collisional ionization rates from the uppermost excited levels. This is true regardless of the duration of the detection laser pulse and so in principle the RID is capable of detecting photons within very short and well defined time intervals. While the RID is a potentially fast (ns) gated detector limited practically by either the time duration of the detection laser pulse At or the time duration of the signal pulse A tslg it is an integrating detector during the detection pulse ( i e .all signal photons falling on the detector within A t d are counted and integrated) and therefore if A td<A tslg temporal information contained within the signal pulse is lost. 'The temporal response of the RID may also be affected by the time evolution of the electron current pulse that is geinerated. Theoretical models' 7-20 describing the time dependence of the LEI signal have shown in good agree- ment with experiment that the current induced in the detection circuit due to electron formation under the inlluence of an electric field is of the order of 10-200 ns depending on the ionization rates of the laser populated levels and experimental parameters such as the magnitude of the electric field electrode spacing and flame compo- sition.Therefore a lower limit on the temporal response of the RID is set. Considerations Specific to the Mg Flame RID Magnesium is nearly an ideal element for an atmospheric pressure flame-based RID. The ground-state transition at 285.2 13 nm is among the strongest existing resonance lines and couples directly to several strong excited-state transi- tions through which the atom can be excited to within a fraction of an electronvolt of the IC. Also the relatively high energy (21kT at 2500 K) of the first excited state ensures that there is a negligible thermal population of this state (about 1 x A detailed description and modelling of the Mg atomic sys'tem has recently been given by Omenetto et a1.,* with special attention being given to the ion yield and quantum efficiency of some possible excitation-ionization schemes.In this paper only the energy levels and schemes directly relevant to the present work will be discussed. that of the ground state). Excitation-Ionization Schemes Considered In ,all cases the 285.213 nm resonance transition of Mg was considered as the first (signal) transition for reasons stated t 571.2 6 285.213 Fig. 2 Three excitation-ionization schemes for Mg studied for the RID; all wavelengths are given in nanometres ~ ~~ Table 1 Transitions and energy levels considered* Wavelength of Energy level of second step/ final state/ Energy defect/ Scheme nm Transition cm-l cm-lj- A 57 1.2 3p 'P0+5s 'S 52 556 -91 13 B 300.9 3p 'PO-+ 68 285 f6616 C 435.2 3p 'P0+6d ID 58023 - 3646 continuum *In all instances the first step was provided by signal photons at 285.213 nm (3s2 l S - 3 ~ IPO).?A -( +) sign indicates a final energy level below (above) the ionization limit of the atom.JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY APRIL 1992 VOL. 7 48 5 above. From this first excited state several different excitation-ionization schemes were studied as shown schematically in Figure 2. Table 1 lists each scheme with corresponding transitions and associated energy levels. Collisionally assisted ionization from level 3 Two possible schemes which rely on collisionally assisted ionization of level 3 are considered (Fig.2). In Scheme A atoms in the first excited state 3p IP0 are further excited to the 5s IS state by absorption of a photon at 571.2 nm bringing them to within 9100 cm-l of the IC or approxi- mately 5.5kT (at 2500 K). This transition has not been much studied and the oscillator strength is not known. The quantum efficiency however is not expected to be signifi- cant when the energy defect is so great. Also if the oscillator strength of this transition is low as is often the case for excited-state transitions optical saturation may not be possible with the laser powers available. To increase q a second high-power laser could be used to photoionize directly the atoms from level 3. The fundamental wave- length output of a Nd:YAG laser as the second ionizing laser is considered in Fig.2 A and C. Of course the atoms in level 3 could also be ionized by absorbing a second photon at A 2 3 ; however because of the high laser powers required this is not considered a viable scheme. Absorption of a photon from level 3 at 1064 nm would photoionize the atom with a resulting energy overshoot (in Scheme A) i.e. energy above the IC to which the atom is promoted of approximately 300 cm-'. In Scheme C the excited-state transition is to the 6d ID state by absorption of a photon at 435.19 1 nm. This places the atom within about 3600 cm-I of the ionization limit. As discussed by Axner and Berglind,14 the collisional ioniza- tion efficiency from such a proximity to the IC is of the order of 50%. The oscillator strength21 of the transition 3p IP0-+6d ID (435.2 nm) is about 0.1.This is reasonably strong for an excited-state transition and should allow (neglecting any atom traps other than the IC) 'saturation' at modest laser powers (about 1 x J in a 10 ns pulse). Again as in Scheme A ionization may also proceed directly from level 3 by absorption of a second photon at 435.191 nm or by absorption of a photon from a second laser imparting enough energy to the excited atom to photoionize it. In both Schemes A and C the ionization efficiency qlon from level 2 (not considering a second ionizing laser) is described by Ri 'lion = Rj+Rm+R2i k2i+ k3i+ o33iP.'. IZ3 + A3 + A2 + k3k + k2 + Azl + k2 where R = A3 + k3,+ A2 + k2,. Other terms are defined in eqn. (3) and the caption of Fig. 1. Several simplifying assumptions can be made in eqn.( 5 ) (i) kzj<<k3i; (ii) 03Pi'123<<k3i at modest laser powers of i 2 3 ; and (iii) the transition 2-3 is optically saturated and therefore the rate of stimulated absorption from level 2 is much greater than the sum of all deactivation rates from the same level i.e. 023123>>A2m+A21+ k2,+ k21. Eqn. (5) then simplifies to Hence a measurement of the ionization efficiency will yield information as to the relative deactivation rates of level 3. When k3p>A3,,,+k3 a value of qion close to unity is expected. Photoionization via an autoionizing state Ionization of magnesium atoms excited to level 2 can also be accomplished by tuning a frequency-doubled dye laser to the peak of a broad autoionizing transition of the Mg atom at 300.9 nm,22723 Scheme B.The autoionization cross- section 02r.i. for this transition is 3 x cm2 allowing optical saturation at modest to high laser powers. Now R,=k2,+AZm Rzl is as defined above and Ri= k2i+~2pi.12(a.l,). From this it can be seen that qion-+l only when the sum of deactivation rates from level 2 to the ground and metastable states i.e. R,+R21 is much smaller than the over-all rate of ionization Ri. Since the rate of collisional ionization from level 2 is at most of the same order of magnitude as other deactivation rates (usually it is orders of magnitude smaller) Ri can be made greater than R,+R21 (for a given excitation-ionization scheme) only by increasing the rate of photoionization 02t.*.12(a,I.). This requires 02pi-Z2(a,i,)> R + R21. For a photoionization cross- section of 3 x cm2 and R,+R2,=10S s-' this requires 12+ cm-2 s-'.In a 2 mm diameter laser beam of 10 ns duration this in turn corresponds to an energy requirement of approximately 2 x J to saturate the transition and approach qlon= 1. In this work an RID based on the two-step LEI of Mg in a miniature air-acetylene flame is discussed. Several excita- tion-ionization schemes were studied and evaluated for quantum efficiency MDP and resolution and SLR. The practical utility of such a detection approach was also demonstrated by the recording of the Raman spectra of several compounds. Experimental The experimental set-up is shown in Fig. 3. The frequency- doubled output of an Nd:YAG laser (532 nm) (Model YG58 1-30; Continuum Santa Clara CA USA) was used to pump two dye lasers (Models TDL-50; Continuum).The n n Fig. 3 Experimental set-up RC = Raman cell A= aperture and L= lens doublet486 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY APRIL 1992 VOL. 7 Table 2 Output characteristics of dye lasers and laser dyes used Wavelength/ Linewidth/ nm Laser dye Pm Remarks 285.2 Rhodamine 590 17+2 Frequency doubled 571.2 Rhodamine 590 20+- 3 Fundamental output 300.9 Rhodamine 640 14+2 Frequency doubled 435.2 LDS 750 (Styryl 7) 39+9 Frequency mixed bandwidth of the first dye laser (used for Raman excitation and to simulate the signal photons at 285.2 nm for the determinatons of quantum efficiency) was 17 -t 2 pm. The fundamental output of the dye lasers could be frequency doubled or frequency mixed with the 1064 nm from the Nd:YAG laser.Both the doubling and mixing processes were controlled by an automatic tracking system thus facilitating wavelength scanning. The bandwidth of the second dye laser was 39+9 14+2 and 2Ot-3 pm for wavelengths of 435.2 300.9 and 57 1.1 nm respectively. Table 2 shows the dyes utilized for each wavelength together with output characteristics. The two laser beams were counterpropagating and were made spatially and temporally coincident in the flame. The temporal coincidence was measured with a fast photodiode (Model ET2000; Electrooptics Technology Fremont CA USA) (rise time t 2 0 0 ps) and a storage oscilloscope (Model 3834; Tektronix Beaverton OR USA). The diameter of the beams taken at the lle point was measured by manually moving a narrow slit (50 pm) across the beam and measuring the intensity passing through with a photodiode. The absorption of the 285.2 nm line was monitored continuously with a hollow cathode lamp (Hamamatzu Bridgewater NJ USA) and double monochromator (Model 1680; Spex Industries Edison NJ USA).The burner (Fig. 4) used to support the air-acetylene flame consisted of a 2 cm x 1.9 mm i.d. stainless-steel tube (1) pressure fitted into a phenolic base (3) and was fitted into a conventional atomic absorption pre-mix spray chamber. The burner was shielded from r.f. pick-up by an aluminium tube-shield (2). As the high back-pressure from 3 f 1 - 1.9 - - 5 0 1 Fig. 4 Design of burner used to support miniature flame viewed from (a) the side and (b) the top 1,2 cm long stainless-steel tube; 2 A1 tube shield; 3 phenolic base; and V signal.All dimensions given in mm such a small flame orifice prevented nebulization with a conventional pneumatic nebulizer Mg solution was intro- duced with a laboratory-constructed ultrasonic nebulizer arid was carried into the flame in the air stream. The concentration of the Mg solution in the nebulizer was 100 ppm and the nebulization rate was increased (by increasing the r.f. power applied to the ultrasonic nebulizer) until >95% absorption of the 285.2 nm line from a Mg hollow caLthode lamp was obtained. A negative high voltage was applied to an immersed water-cooled tubular electrode24 (3 mm 0.d.). The signal (V) was taken from the burner tube which served as the second ellectrode. The laser-induced current change in the flame was a.c. coupled into a current-to-voltage amplifier (Thorn- EM1 Model A l ; Gencom Plainview NY USA) with a high-voltage capacitor (2.2 nF 2 kY).The voltage output from the amplifier was then fed into a boxcar and a personal computer for further processing. The small dimensions of the burner tube in combination with the aluminium shield reduced pick-up of r.f. interfer- ence. The typical operating d.c. current of the miniature flame was t 1 PA one order of magnitude smaller than for a conventional 1 cm diameter flame,' resulting in a small shiot-noise contribution from the flarne itself. Both r.f. pick- up noise and flame-induced shot-noise levels were found to be lower than the transimpedance amplifier noise. Calibration of the Transimpedance Amplifier Tlhe transimpedance amplifier was calibrated for charge integration by injecting a rectangular voltage pulse of known amplitude vn (V) and duration t (s) into a current divider. The input of the Model A1 amplifier was taken at the open end of a 50.1 kC2 resistor. The total charge injected into this amplifier anj was then calculated from Tlhe amplitude and duration of the input current pulse across the 50.1 kQ resistor were measured directly with an ocsilloscope.Experimentally the transimpedance amplifier was cali- brated by setting the width of the boxcar gate equal to the width of the ouput signal (at its base) thereby integrating the entire charge contained within the output signal pulse. Tlhe calibration factor was then calculated by dividing the integrated output signal (V s) by the amount of charge injected (A s).To test the response of the amplifier for diverse input signal it was calibrated with input current pulses ranging from 20 to 2000 ns in width and amplitudes from 0.2 to 1 V. The calibration factor of the amplifier found to be independent of the input pulse amplitude and duration within the range calibrated was determined to be 1.05 ( rt 0.04) x lo5 (V s)/(A s). Determination of Quantum Efficiency To determine the quantum efficiency of the particular excitation scheme under study the first dye laser passing through the flame was used to emulate the signal photons.JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRYy APRIL 1992 VOL. 7 487 This laser beam was expanded 8-fold and limited to approximately 2 mm in diameter with an iris diaphragm in order to obtain a more spatially homogeneous beam.It was then focused in the flame to a spot size of (200 pm with a lens of 25 cm focal length. The energy of this laser was measured with a pyroelectric detector (Model 53-09; Molec- tron Detector Portland OR USA). The beam was systema- tically attenudated with calibrated neutral density filters and the ionization signal measured. The second laser was focused to a spot size of 600 pm with a lens of 125 cm focal length and made temporally and spatially coincident with the first beam. Care was taken to ensure that the ‘signal beam’ was totally encompassed by the detection laser beam. The spectral energy density required for optical saturation of the second step was determined experimentally by means of a saturation curve obtained by observing the two-step LEI signal as a function of laser en erg^.^^^^^ As the RID has a spectral response that is narrower than the ‘signal’ laser bandwidth the fraction of incident photons absorbed at the transition maximum was measured directly by scanning the laser through the transition while monitor- ing the laser energy transmitted through the flame with a photodiode.Care was taken in these measurements to ensure that the signal transition was not optically saturated. As discussed in the previous section for the determina- tion of q width of the boxcar gate was set equal to the base width of the signal to integrate the charge collected in the signal pulse. As is discussed below unless the laser beams and hence the charges formed are at the cathode surface a geometrical correction must be applied to account for the charge collection efficiency of less than unity.RID Wavelength Response The wavelength response of the RID was obtained by placing an aluminium plate at the Raman sample position (RC in Fig. 3) to act as a diffuse scatterer imaging the scatter into the RID. The RID signal was measured at discrete wavelength intervals on the blue wing of the transition maximum in order to increase the signal-to-noise ratio. It was assumed that the profile is symmetrical about the transition maximum. The concentration of Mg in the flame was maintained at a level to ensure >95% absorption of a hollow cathode line to mimic the conditions of the Raman experiment.Care was taken that the transition was not optically saturated at the line centre with scatter from the aluminium plate. To increase the signal-to-noise ratio of these measurements the profile determination was per- formed with the second laser also incident on the flame and tuned to the excited-state transition at 435.2 nm. Raman Spectra The Stokes-shifted Raman spectra were obtained by scan- ning the exciting laser from higher energies (lower wave- lengths) at a fixed rate to the Mg transition maximum at 285.2 nm. The scattered photons were collected and imaged into the flame with a lens doublet (L 5 cm focal length 5 cm diameter). The system was aligned by observing the ultraviolet scatter from an alumina-water mixture placed in the sample cell and imaging it on the flame volume illuminated by the second laser.The second laser was tuned to one of the excited-state transitions of the RID. The energy flux of the detection laser was maintained at about 1 mJ cm-2 to saturate optically the excited-state transition. The energy flux of the exciting (Raman) laser over the wavelength range scanned (279-286 nm) was approxi- mately 10 mJ cm-2. -A12-l where A I is the wavelength of the exciting laser and A12 is the wave- length of the first (signal) transition of the RID element (k 285.2 nm). Since A12 is necessarily fixed by the choice of RID element and signal transition a Raman spectrum is obtained by scanning the exciting laser. When the energy difference between the exciting photons and the energy of the signal transition is equal to a Raman shift of the sample an enhanced rate of ionization is observed in the RID and a Raman signal recorded.The Raman shift is given by Aij Results and Discussion Geometric Correction for Charge Collection The simplest description of the detection process in flame LEI is that of charge induction.iJ8.20 In this model the method of images2’ is often used to describe the amount of charge q induced on an electrode by a charge formed in an electric field E. Consider two electrodes at some distance d apart and at some potential difference V to each other with a conducting medium between them. An electric field will then exist everywhere between the electrodes provided the applied voltage is greater than the saturation voltage. * If the laser produced charge pair is formed in a region of non-zero electric field then the much slower processes of charge collection by convection and diffusion can be neglected.28 Each charge pair induces a total charge e in the external circuit eventually.A fraction of e is induced in the external circuit by the motion of the electron from its place of origin toward the anode and the remainder of e is induced by the motion of the ion from the same point of origin to the cathode. Note that the induced current will also depend on the velocity of the which in turn is dependent on the product of ion mobility and electric field magnitude. To apply this model it must therefore be assumed that the electric field is not significantly disturbed by the laser- produced charges.In conventional LEI measurements the current is usually measured at the anode with a boxcar gate width of 200-500 ns spanning the electron pulse but cutting off the ion pulse owing to the much smaller mobility of the ions. Therefore only the contribution to the charge in the external circuit from the motion of electrons is integrated. From the treatment given by Havrilla et a1.,20 it follows Table 3 Results of MDP qexp and q determinations for Schemes A-C Wavelength MDE#Y LDR of second tlexp* rlt Scheme step/nm (e- per photon) (e- per photon) MDP fJ A 571.1 0.054 0.069 2 . 4 ~ 104 17 1 x 104 B 300.9 0.0039 0.0049 2.7 x 105 190 NDS C 435.2 0.59 0.75 1000 0.7 1 x 105 *qeexp= Quantum efficiency of RID determined experimentally. tq=True quantum efficiency of the RID (see text).$Minimum detectable energy at 285.2 13 nm. §ND =not determined.488 JOURNAL QF ANALYTICAL ATOMIC SPECTROMETRY APRIL 1992 VOL. 7 1x108 U 0 4 1x10’ g 1x106 C a a - r p 1 x 1 0 ~ 0 . 0 . 0 0 0 0 * ~ _ _ _ _ _ _ ~ 1 x 1 0 ~ 1 x 1 0 ~ 1 x 1 0 ~ 1x10’ 1 x 1 0 ~ 1 x 1 0 ~ 1 ~ 1 0 ~ ~ iXioli 1 x 1 0 ~ ~ No. of photons absorbed at h, Fig. 5 Log-log plot of quantum efficiency curve obtained for Scheme C it follows that 100% collection efficiency is obtained only when the laser-produced charges are formed at the cathode surface. In all other cases a geometric correction should be applied. In fact when performing analytical LEI measure- ments the lasers are aligned as close to the cathode as possible. In the present study owing to the difference in beam sizes and space restrictions it was more convenient to align the beams farther from the cathode.In determining the quantum efficiency the electrode spacing d was held at 19 mm and the laser was aligned 4 mm from the cathode surface. With this geometry a collected charge fraction of 0.79 results.20 The true detector quantum efficiency q is then given by 1.27 times the experimentally determined quantum efficiency qexp. The values of MDP (at 285.2 nm) LDR qeXp and q are reported in Table 3. Scheme A The excitation-ionization scheme chosen for the initial evaluation of the flame RID is shown in Fig. 2 A. In this scheme the excited-state transition was to the 5s IS state occurring at 57 1.2 nm. In this scheme (see earlier) the Mg atoms are excited to a final level well below the IC.The quantum efficiency q and MDP obtained were 0.069 and 2 . 4 ~ lo4 photons (17 fJ at 285.2 nm) respectively. The excited-state transition was optically saturated with an incident spectral energy density of 5.3 x J cm-3 nm-l or a photon irradiance of 3 x cm-2 nm-l s-l. As the excited-state transition was optically saturated q is less than unity probably because of the large energy defect and correspondingly low ion branching ratio ti from level 3. The limiting noise for this scheme was determined to be that of the transimpedance amplifier used; that is there was no observable contribution to the noise from the flame high voltage Mg atomic density radiofrequency pick-up or the detection laser. Therefore with a given instrumentation an improvement in the MDP will result only by increasing q.Scheme B A second potential approach of increasing q is by state- specific ionization of Mg atoms in the 3p lPo state via an autoionizing level a.i. An atom excited to a bound level above the IC must necessarily decay into the csntiuum with a probability of unity. This excitation-ionization scheme is shown in Fig. 2 B. The autoionizing transition23 from the 3p ]Po level occurs at 300.9 nm and has a cross-section of 3 x cmz. This is considerably larger than the cross- section for non-specific photoionization. The MDP and q for this scheme were 3 x lo5 photons and 0.0049 respec- tively. Both figures of merit are poor compared with Scheme A. This might be ascribed tcb several causes given below. Owing to the still modest (autoionizing) cross-section the transition could not be saturated with available photon irradiances.As a result as described earlier the ion yield and hence the quantum efficiency are not expected to be significant unless optical saturation is reached. In such a situation q will be linearly dependent on the energy of the ionizing laser. Therefore to maximize q the largest avaLilable photon irradiance at 300.9 nm was used. This however led to a substantial increase in background noise. The ionization potential of Mg lies 61 669 cm-l above the ground state. Calculation readily shows that the simultaneous absorption by ground-state RID atoms of two photons at 300.9 nm imparts 66 467 cm-l to the atom ionizing it. Therefore even with the signal photons absent from the RID there exists a large multiphoton ionization background due to A2 alone.As a result a poor MDP was also observed. Scheme C The third and most successful scheme studied promoted Mg atoms from the 3s lPo state to the 6d lD state by absorption of photons at 435.2 nm. This excited-state transition has a relatively large cross-section of 4 x 10-13 cm;! and promoted the Mg atoms to within 3600 cm-l (about 2kT at 2500 K) of the IC. The high probability of ion.ization from an energy state with such a low energy defect and large ti are reflected in the high q value of 0.75 obtained with this scheme. From the log-log plot of the calibration graph (Fig. 5) an LDR or imore than five orders of magnitude is obtained. Further the large cross-section allows optical saturation at modest laser powers.Also multiphoton ionization is negligible as the simultaneous absorption of three photons at 435.2 nm must take place to ionize ground-state Mg atoms directly. This serves to decrease the noise which together with the high q yielded an MDP of 1000 photons. It should be stressed that Scheme C is a nearly ideal scheme for the flame RID as it is limited by noises not connected with the Mg or the laser. It should be noted however that the estimated shot noise induced by the d.c. current in the miniature flame (about 300 electrons for 1 ,UA)~ falls just below the transimpedance amplifier-limiting noise found. Further improvements in SLR and MDP can be obtained only for RIDs based on differet atom cells e.g. a low- pressure vapour cell or heat pipe oven the low intrinsic nois’es of which will allow the use of’ more sensitive and quieter electronics.Of course in some instances such low- pressure RIDs may rely on alternative ionization mecha- nisms such as field-induced ionization or photoionization. Influence of 1064 nm Laser on Quantum Efficiency in Schemes A and C One means of increasing q might be by non-specific photoionization from the uppermost laser populated state with the fundamental wavelength output of the Nd:YAG laser. This was attempted in both Schemes A and C. the 1064 nm output was made coincident in the RID with the signid photons and the detection laser. Photon irradiances of the 1064 nm beam of up to 4 x cm-* s-l were used. For Scheme A absorption of a photon at 1064 nm from levell 3 promoted the atom to 285 cm-l above the ionization limit.Since as described earlier there is direct competition between ionization and deactivation of level 3 (Fig. 2 A) a quantum efficiency of less than unity implies that the collisional ionization rate is smaller than the deactivationJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY APRIL 1992 VOL. 7 489 ( C) 1x10' I 1 Q) 1x10-' I L 0 B I 1x10-2 0 a U g l X l O S E g 1x10-4 .- - m 1x104 10 20 30 40 50 60 70 80 90 100 Displacementkm-' Fig. 6 Wavelength response of the Mg RID based on the 3s2 IS-+ 3p 'Po transition at 285.2 13 nm measured as described in the text. The curves were obtained by use of eqn. (9); the solid curve was obtained by using A? determined experimentally (2.1 cm-l) and the broken curve was calculated using a value of A? that was lox smaller i.e.0.21 cm-* rate of about 1 x lo9 s-l derived from gas kinetic con- siderations in atmospheric pressure flames. Assuming a photoionization cross-se~tion~~ of about 1 x cm2 using a photon irradiance of about 1 x cm2 s-l should result in a photoionization rate of about 1 x lo8 s-l. Therefore the relative contribution to the total ionization rate from photoionization by the second laser (1 064 nm) is expected to be comparable to the collisional ionization rate. This was indeed observed experimentally by a 2.5-fold increase in the total ionization rate on introduction of the 1064 nm beam into the flame. For Scheme C no significant increase in the total ionization rate was observed on introduction of the 1064 nm laser beam into the flame.The large quantum efficiency observed (0.75) implies that the collisional ionization rate from level 3 (Fig. 2 C) is greater than the deactivation rates [cf eqn. (6)]. Recalling that collisional deactivation rates are of the order of 1 x lo9 s-l a photoionization rate (or laser fluence in the 1064 nm beam) greater than was available experimentally is needed in order to observe a significant contribution to the total rate of ionization. Resolution and Stray Light Rejection The SLR of the RID depends closely on its spectral response which also governs the frequency response of the system. The SLR may be approximated in a general form by 1 - -mjml(t,,At)exp[ -k(to,At)Z]d(AiJ) 1 - -mjml(t,,At)exp[ - k(T~~,AiJ))l]d(At) SLR(ts)= - (9) where I(TS,At) is the intensity of the signal source centred at a wavenumber of ts and having a certain spectral distribu- tion described by a Gaussian function and A t =I to- t i is the wavenumber displacement of the source maximum from the absorption maximum; k(t,,At) is the absorption coefficient for the signal (absorption) transition in the RID centred at a wavenumber of i j o .The denominator serves to normalize the SLR(t,) to the maximum absorption signal observed experimentally. The SLR is necessarily dependent on t i.e. the wavenumber maximum of the source. In other words the SLR will depend on the magnitude of the displacement between the frequency of interest and that of the stray light. Analysis of the SLR equation shows that the spectral response of the RID near the resonant transition frequency (to) is also dependent on both the light source profile i.e.on I(ts,A5) and the actual absorption lineshape i.e. k(to,At) of the resonance transition both of which have a definite spectral dependence. Once the spectral profiles of the source and absorption lines are known eqn. (9) can be solved numerically for discrete values of At which is the wavenumber displace- ment of the source maximum from the absorption maxi- mum. Clearly as eqn. (9) has been normalized to maximum overlap when A t =O (i.e. Fs=DO) the SLR is equal to unity. As was described earlier experiments were carried out to verify the response of the RID about its resonance transi- tion. The profile determined experimentally is shown in Fig. 6. As can be seen the highest signal obtained with the laser tuned to the transition at 285.2 nm was normalized to 3 Q a C 5 J v) .- -0.2 I I I 1 I 1 I I I 0 100 200 300 400 500 600 700 800 0 120 240 360 480 600 720 0.4 0 0 100 200 300 400 500 600 700 Raman shift/cm-' Fig.7 Raman spectrum of (a) CCll obtained with the Mg flame RID using Scheme C; (b) CHC13 obtained with Mg flame RID; and (c) DMSO obtained with M g flame RID. See text for experimental conditions490 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY APRIL 1992 VOL. 7 t 1 I I 1 0 200 400 600 800 I Raman shift/cm-' Fig. 8 Raman spectrum of CC14 obtained with a 0.22 m double monochromator-PMT detector at the RID position. See text for experimental conditions 1. The SLR at 100 cm-l displacement from to is about Also shown in Fig.6 are two curves obtained by solving eqn. (9) numerically for different values of A t . In calculat- ing both curves the absorption profile is assumed to be Lorentzian-dominated owing to the collisionally dominated flame environment. The full width at half-maximum intensity (FWHM) of the resonance absorption line was taken to be 0.43 cm-l (0.0035 nm at 285.2 nm).31 The highest reported value was chosen as some degree of resonance broadening was expected owing to the optically thick conditions of the RID at 285.2 13 nm. The solid curve was obtained by using the experimentally determined (signal) laser FWHM of 2 cm'l (0.017 nm at 285.2 nm) in eqn. (9). Good agreement is seen between the calculated and experimentally observed SLR. The broken line is obtained by using a theoretical (signal) laser FWHM of 0.2 1 cm-l (0.0017 nm at 285.2 nm).It is interesting to note that the spectral response and SLR(F) of the RID is relatively insensitive to the spectral distribution of the signal source and at least with a Lorentzian-dominated RID response is primarily dependent on the absorption profile and FWHM of the signal transition in the RID. 3 x 10-5. Detection of Raman Scatter The Raman spectra of dimethyl sulfoxide (DMSO) carbon tetrachloride and chloroform were recorded with the flame RID operating with Scheme C. The spectra were obtained as described under Experimental. No attempt was made to maximize the solid angle of collection ae, of Raman photons. The simple lens system used allowed for collection of only about 1.5% of the signal photons.Clearly for the RID which has no limiting aperture Qexc could be increased to greater than 2n sr. The Raman spectra shown in Fig. 7(a)-(c) were obtained with a single scan of the excitation laser. A 1 s time constant was used. Each scan took approximately 15 min. Clearly the signal-to-noise ratio could be improved by averaging multiple scans. However owing to the complexity of scanning a frequency-doubled pulsed dye laser this is a difficult and time-consuming task. Even with a single scan however the three spectra obtained show a reasonable signal-to-noise ratio. An interesting advantage of the high resolution afforded by the RID is the direct measurement of Raman line profiles. As both the detector spectral response and the spectral width of the excitation source are narrower than the Raman bands the FWHM and profile of the Raman bands could be obtained directly from the spectra.Finally the performance of the RID can be compared directly with that of a double monochromator-photomulti- plier tube (PMT) system. Fig. 8 shows the Raman spectrum of carbon tetrachloride obtained with a 0.22 m double monochromator-PMT system at the RID position. Appro- priate optics were used to fill the monochromator exactly thus obtaining maximum throughput with minimized stray light introduction. The monochromator-PMT was aligned and optimized on the 313 cm-l Raman band of carbon tetrachloride. The optimized slit-widths were 100 pm. To erriulate the RID the monochromator was held fixed at 285.2 nm and the exciting laser was again scanned.Clearly under the same experimental conditions the performance of the monochromator is inferior to that of the RID. As expected the bandwidths reflect directly the bandpass of the monochromator. Also the signal-to-noise ratio is degraded over that of the RID. Conclusions The Mg-based flame RID has been shown to be capable of detecting about 1000 photons with a resolution dictated by the absorption half-width of the signal transition in the flame of better than 0.5 cm-I. Also as the limiting noise was due to the transimpedance amplifier it can be said wiih confidence that the best detection limit was not obtained and is now dependent on amplifier technology rat her than detection concept. It is believed therefore that these results for a non- optimized Raman system combined with the high quan- turn efficiency good SLR and high potential light through- put.are encouraging and justify fiirther work on such detection schemes. In a flame-based RID the Lorentzian-dominated absorp- tion profiles of the RID element can yield resolutions of the ordler of picometres with corresponding SLR of about 1 x at 100 cm-' displacement from Ao. At present the use of a low-pressure RID to increase the resolution to sub- picometres is being considered. Also a response profile of a lowpressure RID would be Gaussian dominated allowing SLIR capabilities of 1 x at a 1 cm-l displacement from LO- This research was supported by Department of Energy DOE-DEOFGO5-88ER1388 1. The authors gratefully thank Dr. N. Omenetto and Dr.B. W. Smith for many helpful discussions especially with regard to the interpreta- tion of the results. One of the authors (R.G.B.) thanks the CoiiseJo Nacional de Investigaciones Cientificas y Tkcnicas de la Republica Argentina for the fellowship which made possible his stay at the University of Florida. References 1 Travis J. C. Turk G. C. DeVoe J. R. Schenck P. K. and van Dijk C. A. Prog. Anal. At. Spectrosc. 1984 7 199. 2 Zorov N. B. Kuzyakov Y. Y. and Matveev 0. I. J. Anal. (:hem. USSR 1982 37 400. 3 Matveev 0. I. Zorov N. B. and Kuzyakov Y . Y . J. Anal. Chem. USSR 1979,34 654. 4 Ganeev A. A. Matveev 0. I. and Sholupov S. E. J. Anal. Chem. USSR 1987 43 1424. 5 Omenetto N. Smith B. W. and Winefordner J. D. Spectro- chim. Acta Part B 1989 Special Supplement on Proceedings of the Symposium held at Pontifical Academy of Sciences 9 1.6 Smith B. W. Omenetto N. and Winefordner J. D. Spectro- chim. Acta Part B 1989 Special Supplement on Proceedings of the Symposium held at Pontifical Academy of Sciences 10 1. 7 Smith B. W. Farnsworth P. B. Winefordner J. D. and Omenetto O. Opt. Lett. 1990 15 823.JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY APRIL 1992 VOL. 7 49 1 8 Omenetto N. Smith B. W. Farnsworth P. B. and Wineford- ner J. D. J. Anal. At. Spectrom. 1992 7 89. 9 Omenetto N. Smith B. W. and Farnsworth P. B. paper presented at the Institute of Physics Conference Series No. 114 Section 9 Varese Italy 1990. 10 Bloom S. H. Korevaar E. Rivers M. and Liu C. S. Opt. Lett. 1990 15 294. 11 Okada T. Andou H. Moriyama Y. and Maeda M. Opt. Lett. 1989 14 987. 12 Matveev 0. I. and Pribytkov V. A. J. Appl. Spectrosc. USSR 1987 46 16. 13 Fujiwara K. Omenetto N. Bradshaw J. B. Bower 9. N. Nikdel S. and Winefordner J. D. Spectrochim. Acta Part B 1979 34 317. 14 Axner O. and Berglind T. Appl. Spectrosc. 1989 43 940. 15 Omenetto N. Smith B. W. and Hart L. P. Fresenius’ 2. Anal. Chem. 1986 324 683. 16 Petrucci G. A. and Winefordner J. D. Spectrochim. Acta Part B 1992 47 437. 17 Berthoud T. Drin N. Lipinsky J. and Camus P. J. Phys. C7 1983 44 67. 18 Schenck P. Travis J. C. and Turk G. C. J. Phys. C7 1983 44 75. 19 Magnusson I. Spectrochim. Acta Part B 1987 42 1 1 13. 20 Havrilla G. J. Schenck P. K. Travis J. C. andTurk G. C. Anal. Chem. 1984 56 186. 21 Wiese W. L. Smith M. W. and Miles B. M. Atomic Transition Probabilities Sodium Through Calcium US Gov- ernment Printing Office Washington DC 1969. 22 Bradley D. J. Dugan C. H. Ewart P. and Purdie A. F. Phys. Rev. A 1976 13 1416. 23 Petrucci G. A. Stevenson C. L. Smith B. W. Winefordner J. D. and Omenetto N. Spectrochim. Acta Part B 1991,46 975. 24 Turk G. C. Anal. Chem. 1981 53 1187. 25 Alkemade C. Th. J. Spectrochim. Acta Part B 1985,40 133 1. 26 Omenetto N. Benetti P. Hart L. P. Winefordner J. D. and Alkemade C. Th. J. Spectrochim. Acta Part B 1973,28 289. 27 Jackson J. D. Classical Electrodynamics Wiley New York 1975 p. 54. 28 Lawton J. and Weinberg F. ElectricalAspects of Combustion Clarendon Press Oxford 1969. 29 Cobine J. D. Gaseous Conductors Theory and Engineering Applications Dover Publications New York 194 1 p. 56. 30 Saloman E. B. Spectrochim. Acta Part B 1991 46 319. 31 Parsons M. L. Smith B. W. and Bentley G. E. Handbook of Flame Spectroscopy Plenum Press New York 1975 p. 419. Paper 1 /04958E Received September 26 1991 Accepted December 5 1991

ISSN:0267-9477    

DOI:10.1039/JA9920700481

出版商:RSC    

年代:1992

数据来源: RSC

摘要:

JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY APRIL 1992 VOL. 7 493 Effect of Argon Pressure on Spectral Emission of a Plasma Produced by a Laser Microprobe Mikio Kuzuya and Osamu Mikami Department of Electronic Engineering College of Engineering Chubu University 1200 Matsumoto-cho Kasugai-shi Aichi 487 Japan The effect of an argon atmosphere on the self-absorption of emission spectra of a laser-induced plasma was studied with the use of a normal laser microprobe by changing the gas pressure. The degree of self-absorption was determined from the observed profiles of copper resonance lines using the curve-fitting method with the assumption of a Lorentzian profile. According to the results self-absorption was reduced by decreasing the argon pressure and was eliminated at low pressures the range of which depends on the analyte concentration.Studies of the spatially resolved spectrum show that the confining effect of the plasma by the argon atmosphere becomes effective at higher pressures resulting in an increase in the emission intensity. As a result there exists a moderate pressure at which self-absorption can be eliminated without losing too much intensity from reduction of confinement. At a pressure of around 150 Torr (1 Torr= 133.3 Pa) high emission intensity of the spectral line virtually free from self-absorption was obtained about 1.5 mm above the sample surface for the determination of copper (concentration 1-9.8%) in aluminium samples and a linear calibration graph with a slope of unity was obtained. Keywords Laser microprobe; self-absorption; argon pressure; emission intensity; calibration graph The laser microprobe analyser (LMA) which was first demonstrated by Brech and Cross' in 1962 has been used as a convenient instrument for microanalysis because of its capability for microsampling and directly analysing any type of material including electrical non-conductors in air at atmospheric pressure in sit^.^-^ However the emission spectrum of the laser-induced plasma has disadvantageous characteristics such as self-absorption line broadening and an intense background continuum which lead to poor precision of quantitative analyses.It is for this reason that so far the inherent potential of LMA has not been fully exploited and its use has been mainly limited to qualitative analy~is.~.~ In particular self-absorption of a spectral line is a serious problem because it causes a small slope of the calibration graph resulting in poor concentration sensitivity. Further in some instances where elements having high concentra- tions are to be determined self-reversal occurs and quanti- tative analysis becomes impossible. In order to surmount the problem of self-absorption therefore a procedure for correcting self-absorption theoretically is proposed and that metal concentrations can be determined by this method even for spectral lines with self-absorption is shown.* From the viewpoint of practical analysis however it is essential to have a light source free from self-absorption.For this reason several efforts have been devoted to the problem of reducing or eliminating the effect of self- absorption.Kubota and Ishida9 studied the influence of thle analytical conditions on self-absorption using the normal laser microprobe with auxiliary spark excitation. They have shown that the effect of self-absorption is reduced and an improvement in the slope of the calibration graph can be obtained by decreasing the laser energy and by optimizing the spark parameters. Although significant improvements in the slope of the calibration graph could be obtained the slope values were still less than unity. The effects of the various surrounding atmospheres on the emission characteristics of laser-induced plasmas have been investigated by many re~earchers.~-~ Treytl et allo studied various gas media i. e. argon oxygen nitrogen and helium and showed that high emission intensities were obtained in argon.Kagawa et aL1l also compared the effects of different gases such as argon helium nitrogen and carbon dioxide and showed that argon at about 1 Torr (133.3 Pa) is the best for emission spectrometry of laser- induced plasmas. Mohrl* studied the effect of an argon atmosphere on emission spectra and found that self- absorption can be avoided and steeper calibration graphs can be obtained in an argon atmosphere in the pressure range 50-760 Torr. From these studies a reduced argon atmosphere seems to offer an overall improvement in terms of the emission intensity and self-absorption. Recently Niemax and ~ o - w o r k e r s ~ ~ - ~ ~ made detailed studies of laser-produced plasmas in an argon atmosphere and showed that an argon atmosphere reduced to about 100 Torr is optimum for optical emission spectrometry of the laser plasma.Iida16*17 also reported that moderate confine- ment of the plasma and a resultant increase in emission intensity are achieved in argon at pressures of 50- 100 Torr. However these studies were mainly concerned with the Q- switched laser and did not deal with the self-absorption problem of LMA in detail. In this work the effect of the argon pressure on self- absorption of the emission spectra of the plasma produced by a normal laser microprobe was investigated quantita- tively. The spatially resolved emission spectra of resonance lines of copper were measured at various argon gas pressures. The results obtained show that a reduced argon pressure is effective for the elimination of self-absorption and for the improvement of the slope of the calibration graph to unity.Experimental Fig. I shows a schematic diagram of the experimental set- up. The laser microprobe used was a JEOL Model JLM-200 provided with a neodymium-doped glass laser (rod size 100 mmx6.5 mm diameter) and was operated in a normal (free-running) mode. The laser beam was directed down- ward by a 90" prism and was focused on the surface of the sample by an objective lens (focal length 35 mm) through a quartz glass window of the sample chamber. The laser- produced plasma was observed at right-angles to the laser beam by imaging the plasma on an entrace aperture (1 mm diameter) just in front of an optical fibre ( 1.2 mm diameter) in a 1:l ratio by a single lens (focal length 50 mm) and was led to the spectrometer.The spatial distribution of the emission intensity was measured by varying the position of the optical fibre in the image plane. For spectral measure- ments a I .25 m modified Czerny-Turner mounted spectro-494 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY APRIL 1992 VOL. 7 li Vacuum II WE=- -I lens a Sample chamber Gas l - d tank Vacuum pump Fig. 1 Schematic diagram of the experimental set-up Laser beam I [ F T Monochromator P- 250 Digital oscilloscope Y Specrron PDA - I Computer PC-9801 I L _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ J (Optical processing system I Printer used for the measurement of emission spectra of the laser-produced plasma Sample / stage - To vacuum pump Fig. 2 Cross-section of the sample chamber meter (JEOL Model JSG-125) having a grating of 1200 grooves mm-' and a reciprocal linear dispersion of 0.62 nm mm-l was used.The spectra were detected by a silicon- intensified photodiode-array (PDA) detector (Tracor Northern Model TN-6 144) with 7 14 active elements (element size 25 pm x 2.5 mm spectral range 11.0 nm). Spectral data acquisition and processing were carried out with an optical processing system (Seki Technotron Model SK-296). The timing of the operation of the laser and the PDA detector was controlled by a pulse generator that was constructed in our laboratory. The sample was placed in a vacuum-tight sample cham- ber (Fig. 2) which was designed in our laboratory for controlling the ambient atmosphere. The sample chamber is a rectangular box with two windows (36 mm diameter) on the front and back sides and a window (10 mm diameter) on top through which the laser beam enters.The sample chamber is mounted on a precise z-axis transfer stage and the sampIe stage can be rotated to provide a new surface after each laser shot. The sample can be set by viewing through a microscope whose optical axis coincides with the axis of the laser beam when a 90" prism is pulled out. After evacuation argon gas was introduced into the chamber through a conical nozzle in the form of a jet directed against the surface of the sample. The gas jet helps to prevent the top window from being contaminated with ejected sample material. The pressure and flow rate of argon gas were regulated by the valves in the argon line and in the pumping line.The pressure was varied in the range 1-760 Torr and monitored by a semiconductor pressure gauge (Okano Works Model VA-2076s-2) calibrated with a mercury manometer. The following standard samples were used a brass sample (MBH B9E Cu 95.0%; Zn 4.93%; other elements <0.02%) and standard aluminium samples (Alcoa CU- 1-B CU-2-B CU-3-B CU-4-A and SA-1973). The chemical compositions of the latter standards are given in Table 1. The samples were polished using water-proof silicon car- bide paper (No. 1000) and then washed with acetone in an ultrasonic cleaner. In the experiment for obtaining the calibration graph aluminium samples were embedded together in an acrylic resin plate and then polished. This sample plate was set in the sample chamber so that each sample was analysed successively without opening the sample chamber.Results and Discussion Effect of Argon Pressure on Self-absorption Typical emission spectra of a brass sample containing 95% copper in argon at various pressures is shown in Fig. 3. The spectra were measured at a height of 1 mm above the sample surface. The emission lines are copper resonance lines (Cu I 324.7 nm Cu I 327.4 nm). The laser energy was 1.0 J. At pressures higher than 350 Torr both lines were completely self-reversed as is usually observed in air at atmospheric pressure. When the pressure was decreased to Table 1 Chemical composition of standard aluminium samples (Yo) Sample c u Si Fe Mn 0.002 CU- 1 -B 1.01 0.14 0.25 CU-2-B 2.53 0.14 0.25 0.00 1 CU-3-B 4.52 0.15 0.25 0.001 CU-4-A 7.02 0.14 0.25 - SA- I973 9.80 0.17 0.22 0.00 1JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY APRIL 1992 VOL.7 495 - tn Cu 1324.7 nml Cu 1327.4 nm 4- Q Wavelength/mn - c Fig. 3 Emission spectra of the Cu I resonance lines (324.7 and 327.4 nm) observed at a height of 1 mm above the sample surface in argon at various pressures (1 Tom= 133.3 Pa). The sample is a brass containing 95% Cu t I m c I I I -0.4 -0.2 0 0.2 0.4 uln rn Fig. 4 Comparison of experimental and theoretical line profiles. The experimental profile (dotted line) corresponds to that of the Cu I 327.4 nm line at 750 Torr ( I Tom= 133.3 Pa) in Fig. 3. The solid lines show the theoretical profiles calculated for (a) Lorentzian distribution for p=2.39 6=4.4x nm and AS=3.6x nm; and (6) Gaussian distribution for p=2.36 S=5.8 x nm and A,=3.1 x nm 200 Torr it can be seen from the spectral profile of the 327.4 nm line that the effect of self-absorption was reduced considerably although self-reversal of the 324.7 nm line was observed.When the pressure was decreased further ( 1 00 Torr) no self-reversal was observed and a resultant increase in emission intensity was obtained. However at lower pressures the emission intensity decreased. The same tendency of the effect of an argon atmosphere was obtained P 1 5 10 50 100 500 1000 Pressureflorr Fig. 5 Variations of the evaluated values of the absorption parameter p of the Cu I 327.4 nm line with the pressure of argon gas for a Cu concentration of A 95 and B 7.02% ( I Tom= 133.3 Pa) The error bars represent the standard deviation of three runs for aluminium resonance lines (A1 I 308.2 and A1 I 309.2 nm).l* The effect of the argon pressure on self-absorption was evaluated by the absorption parameter p calculated from curve-fitting theory. Assuming that the plasma consists of a high-temperature core surrounded by atomic vapour of low temperature the intensity distribution of the spectral line is given by19 I(u)=I,P(u-&)expi -P[P(~)~P(O)ll (1) with u=A-A where I.is total intensity of the spectral line for no absorption P(u) a spectral line-shape function 1 is the wavelength and A the wavelength at the centre of the line. The wavelength shift As which is the separation between the emission and the absorption peaks was introduced to take account of asymmetry in the spectral profile.* The absorption parameter p indicates the degree of self-absorption. That is for p= 0 self-absorption is absent and with increase in the value of p the self-absorption increases.For p>l the spectral profile has a dip in the centre which means self-reversal. Therefore the degree of self-absorption can be estimated by knowing the value ofp. The absorption parameter p was determined by fitting the theoretical profile [eqn. (l)] to the observed profile by means of the least-squares method. The shape function P(u) was first examined by comparing the experimental profile with the two typical theoretical profiles given by the Lorentzian distribution P(u)=(d/n)/(u*+S2) (2) P(~)=(ln2/nd~)'/~exp[ -ln2(u2/B)] (3) and the Gaussian distribution where S is the half-width at half-maximum intensity.The results for the Lorentzian and Gaussian distributions are shown in Fig. 4(a) and (b) respectively. The experimental data which are indicated by the points were taken from the profile of the copper 327.4 nm line at 750 Torr in Fig. 3. Obviously the experimental profile is well approximated by the Lorentzian distribution so the Lorentzian distribution was subsequently used as the shape function. Curve A in Fig. 5 shows the variation of the absorption parameter p obtained for the copper resonance line (327.4 nm) in Fig. 3 with the pressure of argon. The degree of self- absorption is reduced significantly with decreasing argon pressure down to 100 Torr. However at 100 Torr the value of p obtained is 0.42 which means an intensity reduction of 34% at the line centre and the emission spectrum is still self-absorbed even though a sharp spectrum is observed as496 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY APRIL 1992 VOL.7 z 6 h f! c .- - .- a 4 v) 0 +- .E 2 0) .- Fig. 6 Dependence of the emission intensity of the Cu I 327.4 nm line on the pressure of argon gas (1 Torr=133.3 Pa). The observation height is 1.5 mm above the surface of the sample. The sample is an A1 standard containing 6.58% Cu I I I I 1 5 10 Concentration of Cu (%I Fig. 7 Calibration graph obtained for the analysis of Cu (327.4 nm) in A1 standards in argon at a pressure of 150 Torr (1 Torr= 133.3 Pa). The observation height is 1.5 mm above the sample surface. The error bars represent the standard deviation of five runs shown in Fig.3. At pressures lower than about 10 Torr the effect of self-absorption was much improved and the reduction in peak intensity was only a few percent. An aluminium standard sample with a copper concentra- tion of 7.02% was analysed and the degree of self- absorption of the Cu I 327.4 nm line was evaluated by using the same procedure. The result obtained is shown by curve B in Fig. 5. The degree of self-absorption becomes small compared with the former instance (curve A) as would be expected owing to the lower concentration of copper. A spectral line almost free from self-absorption can be obtained at the pressures up to about 150 Torr. From these results it is concluded that self-absorption can be reduced by decreasing the pressure of argon and that emission spectra free from self-absorption can be obtained in the lower pressure region.This pressure region extends to higher pressures as the concentration of the analyte element becomes lower. The reason for the reduced self-absorption at lower pressures is considered to be less confinement of the plasma as described later. The result in Fig. 4(a) where good agreement was obtained between the experimental profile and the theoreti- cal profile calculated using the Lorentzian distribution also indicates that the main source of line broadening of the plasma produced by the normal laser is Lorentzian broadening i.e. pressure broadening due to collisions with atmospheric species. This is supported by the result that the estimated half-width for the spectra in Fig.3 decreases monotonically from 4.4 x to 2.8 x low2 nm as the argon pressure is decreased from 750 to 1 Torr. The decrease in line broadening at lower pressures could be explained by decreased collisional broadening with argon atoms and by lower collision rates within the plasma caused by weaker confinement of the plasma as shown later From an analytical standpoint on the other hand it is necessary to study the effect of the argon pressure on the emission intensity in addition to self-absorption. Fig. 6 shows how the Cu I 327.4 nm peak intensity varies with the argon pressure. The sample used was an aluminium sample containing 6.58% of copper and the observation height was 1.5 mm above the sample surface. In this experiment no self-reversal was observed over the pressure range 1-760 Torr.The emission intensity increases with decreasing pressure of argon but a further decrease in pressure causes a decrease in the emission intensity. The maximum emission intensity was obtained at a pressure of around 150 Torr. A decrease in emission intensity at higher pressures near atmospheric pressure may be explained by the effect of self-absorption but not at lower pressures. The effect of the argon pressure on the emission intensity will be considered in some detail later on the basis of spatially resolved observations of emission spectra. Calibration Graph On the basis of the foregoing experiments a calibration graph was constructed by plotting the intensity of the 8l 7 6 5 4 3 2 1 E E " X O 2 4 6 8 1 0 1 2 8p- 7 0 2 4 0 2 4 6 8 1 0 1 2 0 2 4 6 Intensity (a r bi t ra ry units) Fig. 8 Spatial distribution of the emission intensity of the Cu resonance line (327.4 nm) at various pressures of argon gas (a) 760 (h) 150 (c) 10 and (4 1 Torr (1 Torr= 133.3 Pa).The positions x and y represent the lateral displacement from the laser beam axis (0 x=O; A x= 1.5; and 0 x=3.0 mm) and the height above the sample surface respectively. The sample [indicated in (a) by the hatched area] is an aluminium standard (CU-4-A) containing '7..02% CuJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY APRIL 1992 VOL. 7 497 copper 327.4 nrn line against the copper concentration as shown in Fig. 7. The samples used were standard alumi- nium samples. The emission intensity was measured at a height of 1.5 mm above the sample surface in argon at a pressure of 150 Torr.The laser energy was set to 0.5 J. The slope of the calibration graph obtained was almost unity indicating a linear relationship between the emission intensity and the concentration. Scott and Strasheim6 determined copper (0.1 - 10%) in aluminium samples in air at atmospheric pressure using a normal mode laser and obtained a calibration graph with a slope of 0.4. They indicated that the cause of this small slope value was the effect of self-absorption. Comparing these results it is concluded that a reduced argon atmosphere is effective for the elimination of self-absorption. Spatial Distributions of Spectral Emission Fig. 8 shows the spatial distribution of the emission intensity of the Cu I 327.4 nm line at different argon gas pressures.The sample used was aluminium (CU-4-A). The positions x and y indicate the lateral displacement from the laser beam axis and the height above the surface of the sample respectively. At 760 Torr the emission intensity along the laser beam axis up to 1 mm above the sample surface was more intense than that of the upper and off-axis regions and the strong emitting region was localized just above the sample surface. At a pressure of 150 Torr the emitting region extends in both vertical and horizontal directions and the intensity of the emission line 3 mm off the laser axis was detected up to a height of 3 mm. The most intense emission was obtained at around 1.5 mm above the sample surface along the laser beam axis. When the pressure was reduced to 10 Torr the emitting region extended in a more spacious region and was not localized.However the emission intensity became weaker. This tendency became greater at 1 Torr. From the results shown in Fig. 8 it is concluded that the spatial distribution of the emission intensity depends strongly on the pressure of argon. At lower pressures near vacuum owing to the weak confining effect of the plasma by the surrounding gas the laser-produced plasma expands rapidly and becomes rare resulting in a decrease in the emission intensity. With increase in pressure the confining effect becomes stronger and hence produces a denser and hotter plasma which gives rise to a high emission intensity. At higher pressures however the confining effect also would cause an increase in concentration of the absorbing species surrounding the hot plasma which would cool the plasma and cause self-absorption.As a result intense emission spectra free from self-absorption are obtained in a moderate pressure range of the argon atmosphere. Conclusions Using a normal laser microprobe the effects of reduced argon pressure not only on self-absorption but also on the emission intensity of the spectral lines were studied. The results showed that self-absorption can be reduced by decreasing the argon pressure and lower pressures eliminate self-absorption. On the other hand the confining effect of the plasma is effective at higher pressures resulting in an increase in the emission intensity. These two effects produce a high emission intensity of the spectral line almost free from self-absorption about 1.5 mm above the sample surface at a pressure of around 150 Torr.Under these conditions a calibration graph with a slope of unity was obtained for the determination of copper in aluminium samples up to a concentration of 9.8%. Therefore a reduced argon atmosphere was found to be useful for the improve- ment of the analytical performance of LMA. Finally this work suggests that it might be possible to determine the major components of the sample and further studies will be undertaken in the near future. The authors thank T. Mikami for technical assistance and M. 1 2 3 4 5 6 7 8 9 10 1 1 12 13 14 15 16 17 18 19 Kato for his help in preparing the samples. References Brech F. and Cross F. Appl. Spectrosc.1962 16 59. Harding-Barlow I. Snetsinger K. G. and Keil K. in Microprobe Analysis ed. Andersen C. A. Wiley New York 1973 ch. 12 p. 423. Laqua K. in Analytical Laser Spectroscopy ed. Omenetto N. Wiley New York 1979 ch 2 p. 47. Piepmeier E. H. Analytical Applications of Lasers Wiley New York 1986 ch. 19 p. 627. Moenke-Blankenburg L. Laser Microanalysis Wiley New York 1989 p. 1. Scott R. H. and Strasheim A. Spectrochim. Acta Part B 1970 25 31 1. Van Deijck W. Balke J. and Maessen F. J. M. J. Spectrochim. Acta Part B 1979 34 359. Kuzuya M. Mikami T. and Mikami O. J. Spectrosc. Suc. Jpn. 1987 36 333. Kubota M. and Ishida R. J. Spectrosc. SOC. Jpn. 1974 23 74. Treytl W. J. Marich K. W. Orenberg J. B. Carr P. W. Miller D. C. and Glick D. Anal. Chem. 1971 43 1452. Kagawa K. Ohatani M. Yokoi S. and Nakajima S. Spectrochim. Acta Part B 1984 39 525. Mohr J. Exp. Tech. Phys. 1974 22 227. Leis F. Sdorra W. KO J. B. and Niemax K. Mikrochim. Acta 1989 11 185. KO J. B. Sdorra W. and Niemax K. Fresenius’ Z. Anal. Chem. 1989,335 648. Sdorra W. and Niemax K. Spectrochim. Acta Part B 1990 45 917. Iida Y. Appl. Spectrosc. 1989 43 229. Iida Y. Spectrochim. Acta Part B 1990 45 1353. Kuzuya M. and Mikami O. Jpn. J. Appl. Phys. 1990 29 1568. Cowan R. D. and Dieke G. H. Rev. Mod. Phys. 1948 20 418. Paper 1 /0464 7K Received September 6 I991 Accepted January 3 I992

ISSN:0267-9477    

DOI:10.1039/JA9920700493

出版商:RSC    

年代:1992

数据来源: RSC

摘要:

JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY APRIL 1992 VOL. 7 499 Palladium Nitrate-Magnesium Nitrate Modifier for Electrothermal Atomic Absorption Spectrometry Part 3.* Determination of Mercury in Environmental Standard Reference Materials Bernhard Welz Gerhard Schlemmer and Jayateerth R. Mudakavit Department of Appjied Research Bodenseewerk Perkin-Elmer GmbH W-7770 Ul~erlingen Germany ~~~~~~ ~ ~~ ~~ ~ A single addition of 15 pg of palladium is sufficient for 40-60 determinations of mercury if the graphite tube is not heated to temperatures higher than 1500 "C in the cleaning step. A pyrolysis temperature of 400 "C can be used without any loss of mercury and a characteristic mass of 0.1 ng of mercury was obtained for an atomization temperature of 1000 "C. Interferences caused by high sodium chloride concentrations could be eliminated by using a mixture of 95% argon and 5% hydrogen as the purge gas.Mercury could be determined accurately in Aquatic Plant Albacore Tuna River Sediment and Coal Fly Ash reference materials. Keywords Electrothermal atomic absorption spectrometry; palladium modifier; mercury determination; sodium chloride interference; environmental samples Mercury is usually determined using cold vapour atomic absorption spectrometry (CVAAS) and with optimized instrumental and experimental conditions absolute masses of less than 50 pg of mercury and concentrations below 1 ng 1-1 can be determined in environmental samples.' Introduction of flow injection techniques with the inherent miniaturization and automation2 has further increased the reliability and attractiveness of CVAAS.Nevertheless this technique is hampered by some notorious interferences due to transition metals such as copper nickel and silver and the volatile hydride-forming elements.2 In addition CVAAS obviously requires a special accessory which might be inconvenient if mercury needs to be determined only infrequently. For these reasons various attempts have been made over the past two decades to determine mercury by electrother- mal atomic absorption spectrometry (ETAAS) with a graphite furnace. Obviously the volatility of this element caused the greatest problems and its stabilization at least until the solvent has been removed completely was the main topic of all the resulting publications. Additives such as hydrogen per~xide,~ ammonium s ~ l f i d e ~ tell~riurn,~ potassium permanganate5 and potassium dichromate6 have been investigated with varying success.One of the problems encountered with some of these reagents was that they could not be applied to solutions containing higher concen- trations of organic compounds because the reagent was consumed at least partly by the organic species. Amalgamation with gold has been studied as an alterna- tive method of retaining mercury in the f ~ r n a c e . ~ The sample was heated and the mercury trapped in a gold-plated graphite tube which was then heated in the conventional fashion. Shan and Ni8 proposed the addition of palladium nitrate to the sample solution for the stabilization of mercury a procedure which has been adapted successfully by ~thers.~J* Ping et aL9 investigated gold silver platinum and selenium as alternatives to palladium but found palladium to be the most effective for retaining mercury.The best sensitivity (peak height) was obtained using a mixture of palladium and platinum. Nevertheless the stabilizing effect on mercury reported by this group was not very impressive. *For Part 2 of this series see ref. 1 1 . ?On leave from the Department of Chemical Engineering Indian Institute of Science Bangalore India. They found increasing losses of mercury with increasing pyrolysis temperature with palladium alone and also with the palladium-platinum mixture and they observed that 'the detection limit was a sensitive function of the pyrolysis temperature' above 200 "C.A similar experience was reported by Grobenski et a1.,l0 who observed significant losses of mercury before atomiza- tion if pyrolysis temperatures higher than 200 "C were applied. They therefore used a two-step drying at 90 and 140 "C followed by an atomization at 1000 "C. Zeeman- effect background correction was used in order to compen- sate for the high background absorbance observed in the analysis of environmental samples. They also found that they could avoid losses of mercury during the drying stage only by introducing the palladium modifier into the cavity of the L'vov platform and heating it at 1000 "C for 10 s before they added the sample solution. They assumed that palladium nitrate was reduced to the metal under these conditions but that mercury was not only stabilized by amalgamation but that oxide formation was involved in the reaction.The purpose of this work was to investigate further the stabilization of mercury by palladium under different conditions. It would undoubtedly be desirable to stabilize mercury to a higher temperature so that at least a 'mild' pyrolysis stage could be introduced in order to remove some volatile matrix components. It would also be desirable to simplify the lengthy and complicated temperature programme used by Grobenski et a1.,lo which included the pre-treatment of the palladium modifier prior to every sample introduction. It was our intention to develop a procedure that could be used routinely for the determina- tion of mercury in environmental samples similar to the ETAAS procedures used for other trace elements.Experimental A Perkin-Elmer Zeeman-3030 atomic absorption spectro- meter equipped with an HGA-600 graphite furnace and an AS-60 furnace autosampler was used throughout. An electrodeless discharge lamp for mercury operated at 5 W from an external power supply was used and a 0.7 nm spectral slit-width was selected to isolate the 253.7 nm resonance line. Signal evaluation was by means of inte- grated absorbance values computed by the AA instrument exclusively. Pyrolytic graphite coated graphite tubes with500 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY APRIL 1992 VOL. 7 Table 1 Temperature programme for the determination of mercury after a single injection of 15 pg of palladium as modifier on to the platform and its reduction to the metal at 1000 "C Step Parameter 1 2 3* 4 5 6 t 7 Temperature/"C 90 120 200 20 1000 1500 20 Ramp time/s 1 10 5 1 0 1 1 Hold timeis 10 10 2 5 9 5 5 5 Read Internal gas flow rate/ - - - On - - - ml min-I 300 300 300 300 0 300 300 *A 400-450 "C pryrolysis temperature may be used if palladium modifier and sample are mixed prior to injection (see text). TPalladium modifier solution must be added to each sample solution if higher cleaning temperatures are used.pyrolytic graphite platforms were used for all experiments. The graphite furnace temperature programme used for most determinations is given in Table 1. The volumes injected into the graphite tube were 20 pl of sample or standard solution and 10 ,ul of modifier solution. All reagents used were of the highest purity available and at least of analytical-reagent grade.Nitric acid was further purified by distillation in a quartz sub-boiling still (Kuerner Analysentechnik Rosenheim Germany). Palladium nitrate solution was prepared by dissolving palladium metal powder (Alfa Products Johnson Matthey Karlsruhe Germany) in the minimum amount of concen- trated nitric acid with the addition of 10 pl of concentrated hydrochloric acid and further dilution with de-ionized water. Further details of the preparation of the palladium modifier solution have been described elsewhere. The palladium nitrate-magnesium nitrate mixed modifier was prepared by mixing equal volumes of solutions containing 3000 mg I-' of palladium and 2000 mg 1-1 of magnesium nitrate. A 10 p1 volume of this solution contained 15 ,ug of palladium and 10 pg of magnesium nitrate.A stock standard solution 1000 mg 1-l of mercury was prepared from Titrisol concentrate (Merck Darmstadt Germany). Standard solutions were prepared daily by further dilution with 0.2 mol 1 - I nitric acid. The following standard reference materials (SRMs) and research materials were used Albacore Tuna (NIST RM 50) and River Sediment (NIST SRM 1645) (US Department of Commerce National Institute of Standards and Techno- logy Gaithersburg MD USA) and Aquatic Plant (BCR 060) and Coal Fly Ash (BCR 038) (Community Bureau of Reference Commission of the European Com- munities Brussels Belgium). Approximately 1 g of sample was weighed accurately into the poly(tetrafluoroethy1ene) beaker of a Perkin-Elmer Autoclave-3 acid digestion bomb 4 ml of water and 4 ml of concentrated nitric acid were added and the digestion bomb was closed tightly heated to 160 "C slowly and kept at this temperature for 1 h.After cooling the autoclave was opened carefully and the con- tents were transferred quantitatively into a 25 ml calibrated flask and diluted to volume. Results and Discussion Optimization of Modifier and Temperature Programme At first a temperature programme that consisted only of a drying stage at 120 "C and atomization at 1000 "C was used in order to minimize low-temperature losses of mercury. Under these conditions a 15-20% greater integrated absorbance was obtained when the standard solution and the modifier were mixed prior to their injection compared with a sequential injection of the two solutions.This might well be due to mercury losses because of incomplete mixing with the modifier solution when the solutions were injected separately. It was also found that there was no difference in the sensitivity for mercury when palladium nitrate alone .was used as the modifier and when it was mixed with magnesium nitrate. In contrast to the observations of Ping et aL9 and Grobenski et a1.,lo it was observed in the present study that the integrated absorbance for mercury remained constant up to a pyrolysis temperature of 450 "C with no indication of any losses when the standard solution and the modifier were mixed prior to their injection. This behaviour was (again identical for the palladium nitrate alone and when it was mixed with magnesium nitrate.In an investigation of the procedure proposed by Groben- ski et a1.,I0 involving pre-treating the modifier at 1000 "C and reducing it to the metal it was observed that the palladium was retained essentially quantitatively on the platform after the atomization of mercury if the tempera- ture was not increased above 1500 "C. This effect is shown in Fig. 1 for 40 determinations of mercury carried out after a single addition of 15 pg of palladium using the tempera- ture programme given in Table 1. The integrated absor- bance signal for mercury decreased by less than 5% over these 40 determinations. This decrease could be slowed further when the maximum temperature used for clean-out was reduced to 1200 "C. More than 60 determinations of mercury could be carried out under these conditions with less than a 5% decrease in the integrated absorbance.The only disadvantage associated with this procedure was a slight decrease in the stabilizing power of the palladium modifier. The losses of mercury became significant at pyrolysis temperatures above 300 "C as can be seen from the pyrolysis curve in Fig. 2. If this should become a problem in the analysis of real samples it could be overcome very easily by adding a small amount of palla- No. of determinations Fig. 1 Forty repetitive determinations of 5 ng of Hg after a single addition of 15 pg of Pd (after reduction to the metal at 1000 "C) 0 5 ng of Hg without modifier; and + 5 ng of Hg mixed with 15 pg of PdJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY APRIL 1992 VOL.7 50 1 v) 0.25 1 0 tu f! 0.20 s n $ 0.15 rn 4- 2 f 0.10 - - - . A BG 0.5 I 1 I J 2000 0.05 0 500 1000 1500 TemperaturePC (a) Fig. 2 Pyrolysis and atomization curves for 5 ng of Hg in the presence of pre-reduced Pd integrated absorbance signal for A increasing pyrolysis temperatures without the addition of Pd (atomization temperature 1000 "C); B increasing pyrolysis temper- atures with 0.6 pg of Pd added to each solution; and C increasing atomization temperatures without the addition of Pd (pyrolysis temperature 200 "C) BG 0.5 dium to each sample and standard solution. It was found that the addition of only 0.6 pg of palladium which is 4% of the usually applied 15 jig restored the thermal stability of mercury. The losses of mercury became significant only at pyrolysis temperatures higher than 500 "C under these conditions as can be seen from curve B in Fig.2. Also shown in Fig. 2 is the atomization curve for mercury using the reduced palladium modifier. The integrated absorbance is essentially constant for atomization tempera- tures between 850 and 1000 "C and decreases almost linearly between 1000 and 2000 "C to less than one third of its original value. An atomization temperature of 1000 "C was used throughout this work because a 'sharp' peak with high sensitivity was obtained under these conditions. The calibration graph for mercury was linear between 100 and 1500 pg 1-l of mercury (integrated absorbance between 0.047 and 0.707 s) with a linear regression equation of y=4.83 x 10-5x-0.0126 and a correlation coefficient of r=0.9998.The characteristic mass was calculated to be m0= 97 pg which is in good agreement with the theoretical value of 69 pg,12 taking into account the 25-30°/o decrease in sensitivity due to the application of the Zeeman effect.12 ( c ) Interference Studies There is very little information in the literature on possible interferences in the determination of mercury when palla- dium is used as the chemical modifier. Cationic interfer- ences are very unlikely to occur because there is probably no other metallic element that could bind mercury more efficiently than palladium or that could significantly influ- ence the stabilizing power of this modifier. Background absorption could well be substantial because of the low pyrolysis and atomization temperatures used for mercury.Zeeman-effect background correction the most powerful technique currently available was therefore used in order to avoid problems due to insufficient background correction. The only type of interference that had to be taken into account seriously was that due to anions such as chloride and sulfate because mercury is known to form stable and volatile chlorides and sulfides in the condensed and/or gas phase. The low pyrolysis temperatures used for this element do not allow the volatilization of significant amounts of the matrix except for acids etc. and the low atomization temperature increases the risk of gas-phase interferences. The influence of increasing concentrations of potassium BG 0.5 Fig. 3 Influence of increasing concentrations of K2S04 on the integrated absorbance signal of 5 ng of Hg (e) BG 0.5 /-- 0 ( f ) AA 0.3 I 8 --- 0 0 2 :::;] _____-- 0 3 6 Time/s Fig.4 Atomization signals for 5 ng of Hg in the presence of increasing concentrations of NaCI (a)-(& cleaning temperature 2000 "C 15 pg of Pd added to each solution; (e)-(g) cleaning temperature 1500 "C one injection and reduction of 15 pg of Pd prior to the determinations. (a) Matrix-free solution; (b) and (e) 0.01 g 1-1 NaCl added; (c) and cf) 1 g 1-I NaCl added; and (d) and (g) 10 g 1-1 NaCl added. Broken lines are background absorbance (BG) sulfate on the integrated absorbance signal of 500 pg 1-l of mercury is shown in Fig. 3. Up to 100 mg 1-* of potassium sulfate there was no effect on the mercury signal and there502 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY APRIL 1392 VOL.7 was a slinht increase in the integrated absorbance for __ concentrations around 1000 mg 1 - i However this effect was within the k 10% margin which is typically considered acceptable in interference studies. The mercury signal began to drop significantly in the presence of potassium sulfate concentrations higher than 5 g l-l which should not be a major limitation in practical analysis. Sodium chloride has a substantial influence on the peak shape and on the integrated absorbance of mercury as is shown in Figs. 4 and 5 and this influence was additionally dependent on the way in which the modifier was applied. When a cleaning step at 2000 "C was used and the palladium modifier was mixed with all solutions sodium chloride up to about 0.1 g 1-l had little influence on the appearance temperature and the signal shape of mercury.There was however some decrease in sensitivity even at very low concentrations of sodium chloride which in- creased to about 40% for concentrations of the latter around 1 g 1-l. In the presence of even higher concentrations of sodium chloride a second early peak appeared for mercury and the main peak was broadened. At the same time the integrated absorbance signal for mercury began to increase again and reached 80% of the original value in the presence of 30 g 1-l of sodium chloride. When the temperature of the cleaning step was reduced to 1500 "C and the palladium modifier was introduced only once onto the platform the appearance time was shifted significantly even in the presence of only 0.01 g 1-l of sodium chloride.Higher sodium chloride concentrations caused no additional shift in the appearance but an increasing distortion of the mercury signal. It is interesting that this shift and distortion had no influence on the integrated absorbance signal of mercury up to sodium chloride concentrations of 0.1 g 1-l. In the presence of about 1 g 1-* of sodium chloride the integrated absorbance signal decreased to approximately the same values that were obtained when palladium was mixed with the solutions even though the signal shape was significantly different. For even higher sodium chloride Concentrations a similar slow increase in the integrated absorbance was observed and the peak shapes showed a comparable distortion.The integrated absorbance decreased without an appar- ent signal distortion when palladium was added to the analyte solution [Fig. 4(a)-(c)]. This can be best interpreted by low-temperature losses of mercury prior to the atomiza- tion stage. Such losses have been reported by others9 and can be explained by the insufficient stabilizing power of palladium at low temperatures when it is added to the solutions as the nitrate. The fact that the integrated absorbance signal for mercury actually increased again for sodium chloride concentrations higher than 1 g 1-l (Fig. 5) can be explained by a trapping effect. The time-resolved atomization signal for mercury in the presence of 10 g 1-* of sodium chloride [Fig. 4(4] showed two changes compared with those with lower sodium chloride concentrations which support this interpretation an early peak and increased tailing of the main atomization pulse.The early peak is probably due to mercury chloride which was lost prior to atomization in the presence of up to 1 g I-' of sodium chloride but was trapped and retained in part in the presence of higher sodium chloride concentrations. The compound was hence volatilized early in the atomization stage i.e. at a temperature high enough for its dissociation into atoms. This interpretation is further supported by the time- resolved atomization pulses for mercury in the presence of pre-reduced palladium [Fig. 4(e)-(g)]. In this instance the early mercury peak appeared even in the presence of the lowest sodium chloride concentration of 0.01 g I-'.It can also be seen in Fig. 5 that there were no pre-atomization losses of mercury up to concentrations of 0.1 g 1-1 of ;P I .- .Q m 60 2 .c !! 20 401 Fig. 5 Influence of increasing concentration of sodium chloride on the integrated absorbance signal of 5 ng of Hg A cleaning temperature 2000 "C 15 pg of Pd added to each solution purge gas argon; B cleaning temperature 1500 "C one injection and reduction of 15 ,ug of Pd prior to the determinations purge gas argon; and C same as B but purge gas 95% argon-5% hydrogen sodium chloride. The mercury chloride which undoubtedly formed in the condensed phase when mercury nitrate was mixed with an excess of sodium chloride was stabilized by the reduced palladium and released only at temperatures at which it could be atomized.In the presence of I g 1-1 of sodium chloride the molar ratio of C1 to Pd is about 4 so that it is not surprising that the stabilizing power of palladium was no longer sufficient to prevent an interfer- ence. Under these conditions the way in which the modifier was applied was apparently no longer signficant as the extent of the interference was the same for pre-reduced ]palladium and palladium added to the solutions. The fact ithat the early and the later peak were reduced in height [Fig. $ 4 0 1 may be interpreted by the presence of pre-atomization losses and of a gas-phase interference. Finally in the 'presence of 10 g 1-l of sodium chloride [Fig. 4(g)] the 'atomization pulse had a shape very similar to that when palladium was added to the solution and also the integrated (absorbance was similar (Fig.5 ) i. e. it was increasing owing to the trapping effect for increasing sodium chloride concentra- tions. Under these conditions sodium chloride and not pal- ladium was the dominating component in the graphite tube so that it was no longer important in which form it was added. A similar minimum in the interference curve due to sodium chloride as was observed here (Fig. 5) has been found also in earlier work on tha1li~m.l~ In this instance a reduction of the palladium modifier on the L'vov platform prior to the introducion of the sample and the use of a mixture of 95% argon and 5% hydrogen as the purge gas were found to remove the sodium chloride interference effectively. As the first measure the reduction of the palladium was already used for mercury determination in this work only the addition of hydrogen to the purge gas had to be investigated.The result of this experiment is shown in Fig. 5 curve C. Only a single atomization signal was obtained for mercury under these conditions and the integrated absorbance remained constant within 96- 102% of the matrix-free signal up to a concentration of 30 g 1-l of sodium chloride in the solution. This indicated that the various influences of sodium chloride on the mercury signal were predominantly caused by reactions in the gas phase and that the releasing action of hydrogen was mainly that of binding chloride in the form of HCl. The fact that hydrogen was less effective when palladium was not reduced prior to sample injection however is an indication that pre- a,tomization losses were involved and possibly also more complex processes such as sorption of hydrogen on the reduced palladium metal. More detailed investigations would certainly be necessary in order to clarify these mechanisms entirely.JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY APRIL 1992 VOL.7 503 Table 2 Determination of Hg in environmental standard reference and research materials BCR 060 (Aquatic Plant) RM 50 (Albacore Tuna) SRM 1645 (River Sediment) and BCR 038 (Coal Fly Ash) certified values are in parentheses Concentration of Hglpg g-I Independent ~ acid BCR060 RM 50 SRM 1645 BCR038 BCR038* digestion (0.34 k 0.04) (0.95 k 0.10) (1.1 & 0.5) (2.10 -I 0.15) (2.10 2 0.15) 1 0.33 1.05 0.81 1.87 1.94 2 0.34 0.95 0.73 1.79 1.88 3 0.32 0.96 0.77 1.81 2.12 4 0.27 0.97 1.08 - - - 1.03 5 - - - *Determination repeated after standing overnight (see text).Analysis of Environmental Samples Three standard reference materials and one research material were selected for this study representing a wide variety of environmental samples such as plant tissue fish tissue sediment and coal fly ash and covering a mercury concentration range between 0.3 and 2.1 pg g-l. The pressure digestion with nitric acid was not expected to result in complete destruction of the organic matter and a residue-free dissolution of the inorganic constituents which is not necessary for an analysis by ETAAS. It was expected however that mercury was leached from the residues and was in solution quantitatively.The results which are summarized in Table 2 for 3-5 independent acid digestions of the four reference materials indicate that these expecta- tions were met by three of the four materials but not by the Coal Fly Ash for which our results were low by 10-1 5%. It was interesting that the mercury content in all three acid digests of the Coal Fly Ash had increased when the samples were analysed one more time on the next day and were now within 10% of the certified value. Apparently the leaching process had continued overnight and was now more complete. A contamination was less likely because no such effect was observed for the other sample digests. The temperature programme in Table 1 was used for these determinations with a single introduction of 15 pg of palladium at the beginning of the series of determinations and pure argon was used as the purge gas because the samples had only low concentrations of sodium chloride.All determinations were carried out using the standard calibration procedure with matrix-free standard solutions. One digestion solution of each sample was spiked with a mercury standard in order to verify the absence of interfer- ences and the recovery was between 91 and 108%. Problems with the lifetime of the graphite tubes were encountered in earlier experiments when a higher acid concentration than that described under Experimental was used i.e. when the acid digests were less diluted. This was probably due to nitric acid which condensed at the cool tube ends and was not removed during the heating stage at 1500 "C.No such problems were found when the tube was heated to 2650 "C for cleaning or as was done in these experiments when the nitric acid concentration did not exceed 4 ml per 25 ml of solution. Conclusion The proposed procedure has been applied successfully to the determination of mercury in a variety of environmental samples ranging from plant and fish tissue to sediment and coal fly ash. It was found that there was no difference in the stabilizing power for mercury and other performance parameters between palladium alone and palladium mixed with magnesium nitrate as the modifiers. Several variations of the procedure have been investigated including mixing of sample and modifier prior to sample injection pre- reduction of the palladium on the platform and a combina- tion of both whereby the last approach permitted the use of the highest pyrolysis temperature of 450 "C.For routine applications however a single injection of 15 pg of palladium and its reduction to the metal on the platform which was sufficient to stabilize mercury for at least 40 determinations if the temperature was not in- creased above 1500 "C was chosen because of simplicity lower cost and the faster temperature programme involved. The stabilizing power of palladium for mercury was found to be significantly higher than previously reported. Conditions under which pyrolysis temperatures of 400 "C or more could be applied include pre-mixing of sample and modifier and injection of the sample with a small amount of added palladium onto pre-reduced palladium. The complex effect of sodium chloride which resulted in the formation of double peaks and depending on the conditions in pre-atomization losses and gas-phase inter- ferences could be removed by the use of a mixture of 95% argon and 5% hydrogen as the purge gas. More work will be required however in order to understand fully the mecha- nisms of this interference and its removal by hydrogen. 1 2 3 4 5 6 7 8 9 10 11 12 13 References Welz B. and Melcher M. At. Spectrosc. 1984 5 37. Welz B. and Schubert-Jacobs M. Fresenius' J. Anal. Chem. submitted for publication. Issaq H. J. and Zielinski W. L. Anal. Chem. 1974,46 1436. Ediger R. At. Absorpt. Newsl. 1975 14 127. Fujiwara K. Sato K. and Fuwa K. Bunseki Kagaku 1977 26 773. Kirkbright G. Shan X.-Q. and Snook R. D. At. Spectrosc. 1980 1 85. Siemer D. and Woodriff R. Anal. Chem. 1974 46 597. Shan X.-Q. and Ni Z.-M. Acta Chim. Sin. 1979 37 261. Ping L. Fuwa K. and Matsumoto K. Anal. Chim. Acta 1985 171 279. Grobenski Z. Erler W. and Voellkopf U. At. Spectrosc. 1985 6 91. Welz B. Schlemmer G. and Mudakavi J. R. J. Anal. At. Spectrom. 1988 3 695. L'vov B. V. Spectrochim. Acta Part B 1990 45 633. Welz B. Schlemmer G. and Mudakavi J. R. Anal. Chem. 1988 60 2567 NOTE-Ref. 11 is to Part 2 of this series. Paper 1/05003F Received September 30 I991 Accepted November 22 1991

ISSN:0267-9477    

DOI:10.1039/JA9920700499

出版商:RSC    

年代:1992

数据来源: RSC

摘要:

505 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY APRIL 1992 VOL. 7 Palladium Nitrate-Magnesium Nitrate Modifier for Electrothermal Atomic Absorption Spectrometry Part 4.* Interference of Sulfate in the Determination of Selenium Bernhard Welz Gabor Bozsait Michael Sperling and Bernard Radziuk Department of Applied Research Bodenseewerk Perkin-Elmer GmbH W-7770 Uberlingen Germany The possible sources of sulfate interferences in the determination of selenium even when stabilized temperature platform furnace conditions were used were investigated using sulfuric acid sodium sulfate and magnesium sulfate as model compounds. The molecular spectrum of CS was observed when sodium sulfate was volatilized. The most likely mechanism of interference in this instance is an expulsion of the analyte element with the violently volatilized matrix early in the atomization stage.Sulfuric acid and magnesium sulfate decompose at elevated temperatures with the formation of SO3 which was reduced at least in part to SO2 in the graphite tube. The most likely interference mechanism in these instances is the formation of SeO in the presence of excess of SO3 and its volatilization during the pyrolysis stage. Addition of barium nitrate to the palladium nitrate-magnesium nitrate modifier reduced the background absorption and also the sulfate interference significantly. In the presence of barium sulfate is bound at least in part as barium sulfate which decomposes in the atomization stage. This means that SeO if it is formed is volatilized at temperatures which are high enough for its atomization.The proposed mixed modifier allows an interference-free determination of selenium in mineral waters with high sulfate content. Keywords Electrothermal atomic absorption spectrometry; selenium determination; sulfate interference; mineral waters; molecular spectra of CS and SO Selenium is still among the elements that are causing more than average problems in electrothermal atomic absorption spectrometry (ETAAS) with a graphite furnace. Numerous publications can therefore be found about atomization and interference mechanisms for this The use of the stabilized temperature platform furnace (STPF) con~ept,~ which allows atomization under near-isothermal condi- tions has brought about a significant improvement in the determination of selenium. The use of chemical modifiers which is part of the STPF concept has been found to be indispensable for the stabilization of selenium during the pyrolysis stage in order to prevent low-temperature volatili- zation losses.The originally proposed nickel modifiers has been shown to be not particularly effective in stabilizing s e l e n i ~ m . ~ ~ ~ It has been found that in the presence of nickel as the chemical modifier the different oxidation states of selenium exhibit significantly different thermal behaviours,10 which may be the reason for various analytical errors. The palladium modifier originally proposed by Shan and Ni," has proved to be much more effective in stabilizing selenium particularly when mixed with magnesium ni- trate.6J2-16 In preliminary experiments,17 however it was found that the previously described sulfate interference on the determination of seleniumlsJ9 could not be removed completely even under STPF conditions and using the palladium nitrate-magnesium nitrate modifier.The aim of this work was to investigate the source of the observed interferences and with a better understanding of the mechanisms to search for means for their elimination. Mineral and medicinal waters with exceptionally high sulfate content which previously did not permit an interfer- ence-free determination served as test materials. Experimental A Perkin-Elmer Zeeman-3030 atomic absorption spectro- meter equipped with an HGA-600 graphite tube atomizer *For Part 3 of this series see J. Anal. At. Spectrom. 1992,7,499.t o n leave from the National Institute of Hygiene Budapest Hungary. and an AS-60 autosampler was used for all atomic absorption measurements. A selenium electrodeless dis- charge lamp operated at a power of 6 W was used for all determinations. The wavelength was set to 196.0 nm and a 0.7 nm slit-width was used throughout. A typical temperature programme for the determination of selen- ium in the presence of high sulfate concentrations is given in Table 1. A cool-down step was used routinely prior to atomization and pyrolytic graphite coated graphite tubes with a pyrolytic graphite platform were used for all experiments. A Perkin-Elmer Model 3030 atomic absorption spectro- meter equipped with an HGA-500 graphite tube atomizer and an AS-40 autosampler was used for molecular absorp- tion measurements. The deuterium arc lamp of the back- ground corrector was used as the radiation source and the measurements were carried out point by point with manual wavelength setting and a spectral slit-width of 0.07 nm.All reagents were of the highest available purity but at least of analytical-reagent grade. Nitric acid was purified by sub-boiling distillation (quartz sub-boiling still Kiirner Analysentqchnik Rosenheim Germany). A mixed palladium nitrate-magnesium nitrate-barium nitrate chemical modifier was prepared by dissolving 450 mg of palladium metal powder (-22 mesh m4N8= 99.998% Alfa Ventron Products Johnson Matthey Karls- ruhe Germany) in 1.5 ml of concentrated nitric acid with the addition of 10 pl of concentrated hydrochloric acid and diluting to 100 ml with de-ionized water resulting in a solution of 4500 mg 1-l of palladium.A 1265 mg amount of magnesium nitrate hexahydrate (Suprapur Merck Darms- tadt Germany) was dissolved in de-ionized water and diluted to 100 ml resulting in a solution containing 1200 mg 1 - I of magnesium. A 285 mg amount of barium nitrate (Suprapur Merck) was dissolved in de-ionized water and diluted to 100 ml resulting in a solution containing 1500 mg 1-l of barium. The final modifier solution was prepared by mixing equal volumes of the three solutions. A 10 pl volume of this mixed solution contained 15 pg of palla- dium 4 pg of magnesium and 5 pg of barium. For some of the experiments mixed modifiers with different ratios of the506 JOURNAL OF ANi4LYTICAL ATOMIC SPECTROMETRY APRIL 1992 VOL.7 BG 0.6 - Table 1 Recommended temperature programme for the determination of selenium in the presence of high sulfate concentrations using the palladium nitrate-magnesium nitrate-barium nitrate modifier sample volume 10 pl Step (a) Parameter 1 2 3 4 5* 6 7 TemperaturePC 80 100 800 20 2300 2500 20 Ramp time/s 1 10 10 1 0 1 1 Hold time/s 10 20 20 20 5 5 5 Internal gas flow rate/ ml min-I 300 300 300 0 0 300 300 BG 0.6 *The read cycle was activated in this step. ( b ) three elements were prepared using different volumes of the same stock solutions. a BG 0.6 Results and Discussion The mixed palladium nitrate-magnesium nitrate modifier which has been used very successfully for the determination of a large number of elements including selenium in a variety of samples could not remove completely the interference of sulfate on selenium determination.This can be seen in Fig. 1 in which the signal for 1 ng of selenium is shown in the presence of 20 pg of sulfate as sulfuric acid sodium sulfate and magnesium sulfate. Sulfuric acid and magnesium sulfate did not cause either significant signal distortion or excessive background absorption. The inte- grated absorbance values in the presence of these concomi- tants were nevertheless 20% and 1 O% respectively lower than the value for the matrix-free solution. Sodium sulfate also reduced the integrated absorbance signal by about 20%. In this instance however a relatively strong background .-. . . . . . . . . ( c ) . - . *. . .. 0 1 - 1 ...... ......-*........I al 0 b n AA 0.4 BG 0.6 (d) 0 2.5 Time/s 5.0 Fig. 1 Atomization signals for 1 ng of selenium in the presence of various sulfate compounds (20 pg of using the palladium nitrate-magnesium nitrate modifier pyrolysis temperature 800 "C; and atomization temperature 2100 "C.The dotted signal is background absorbance (BG) (a) reference solution matrix-free integrated absorbance 0.2 19 s; (b) sulfuric acid integrated absor- bance 0.173 s; (c) sodium sulfate integrated absorbance 0.170 s; and (d) magnesium sulfate integrated absorbance 0.204 s absorbance and a significant change in peak shape were observed. It is interesting that in the presence of sodium sulfate the selenium signal appeared earlier and was clearly distorted at the time when the background signal reached its highest value.Although graphite might influence some of the reactions or the temperatures at which they occur explanations for the observed phenomena can be found when the physical data of the model compounds and their behaviour on volatilization are considered (see Table 2). Sulfuric acid and magnesium sulfate decompose with the formation of SO3 whereas sodium sulfate is volatilized without decomposi- tion. Sodium sulfate could apparently not be volatilized during the pyrolysis stage at 800 "C and was hence vaporized very rapidly and violently early in the atomiza- tion stage and part of the selenium was co-volatilized with the matrix and atomized in part in the hot gas environment which explains the early appearance and rapid rise of the atomization signal.At the same time however the resi- dence time of the selenium in the absorption volume was influenced by the violent vaporization so that complete atomization was not possible. The small background absorbance of about 0.05 that appeared under the selenium signal in Fig. l(a) (b) and (6) was not due to a matrix-induced background but was caused by a slight overlapping of the a-components of the Zeeman-split selenium line with the emission line of the primary light source. The same background signal was observed for a pure standard solution [Fig. l(a)] and in the presence of sulfuric acid [Fig. l(b)] and magnesium sulfate CFig. l(d)]. As both of these compounds caused significant background absorbance when lower pyrolysis temperatures were used the absence of a matrix-induced background under the conditions in Fig.1 indicated that both sulfuric acid and magnesium sulfate decomposed at temperatures below 800" C. This was as expected for sulfuric acid but not for magnesium sulfate which according to the data in 'Table 2 should decompose only at higher temperatures. Carbon did apparently play an active role in the low- temperature decomposition of magnesium sulfate. In order to investigate the processes occurring during decomposition and/or volatilization of the investigated compounds the background absorbance spectrum over the range of interest was measured using a deuterium arc lamp as the radiation source. No chemical modifer was used in Table 2 Melting (m.p.) or boiling point (b.p.) and behaviour of the various sulfate compounds during volatilization Compound M.p.or b.p./"C* Volatilization behaviourl H2S04 B.p. 328 Decomposes to SO3 + H 2 0 Na,SO M.p. 884 Not decomposed MgS04 M.p. 1124 Decomposes to S03+Mg0 BaSO M.p. 1580 Decomposes to S03+Ba0 *Ref. 20. ?Refs. 21 and 22.JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY APRIL 1992 VOL. 7 507 1.2 1 .o 9 0.8 Q z 2 0.6 d 0.4 0.2 280.0 nm I n l I I I I I I 1 Y 200 220 240 260 280 300 320 340 Wavelengthlnm Fig. 2 Molecular absorption spectrum measured during the vaporization of 100 pg of magnesium sulfate without the addition of a chemical modifier and without the pyrolysis stage vaporiza- tion temperature 2100 "C these experiments in order to keep the spectra simple and easy to interpret. Also no pyrolysis stage was used so that all compounds and decomposition products were volatil- ized together at 2100 "C.The molecular absorbance was measured with the smallest spectral slit-width of 0.07 nm and the smallest possible wavelength steps of about 1 nm available at the instrument. It was obvious that the low resolution of the spectrometer and the manual step-by-step registration would result in some distortion of the recorded spectra. Some maxima might not show up at the exact wavelength be of the wrong relative intensity or even be missing. It is nevertheless believed that the recorded spectra made an identification of the volatilized species possible because only one compound was introduced into the graphite tube and volatilized and hence the variety of species that could be expected was limited.The spectrum that was recorded during the vaporization of 100 pg of magnesium sulfate is shown in Fig. 2. The spectrum recorded on vaporization of sulfuric acid had a similar pattern but was much weaker and is not shown here. The spectrum depicted in Fig. 2 is probably that of the SO2 molecule. In the literature it is described as a relatively dense band spectrum in the range between 264 and 3 19 nm which declines towards the red end.23 The most intense of the approximately 40 bands described in the literature are between 278 and 304 nm and some of them could be identified well in the recorded spectrum such as those at 280.0 296.1 and 300.1 nm. Apparently the SO3 generated in the decomposition of magnesium sulfate was reduced at least partly to SO2 in the graphite furnace. No estimation could be made based on this experiment of the extent to which SO3 was reduced to SO2 because the spectrum of SO3 is much weaker than that of SO2 and its unambiguous identification is possible only in the absence of S02.23 In any event however the amount of SO2 formed was in large excess over selenium and as was derived from Fig.l(d) it was formed in the pyrolysis stage. Because of the great similarity of selenium and sulfur in their chemical behav- iour it might be concluded that the analogous selenium compound Se02 was also formed under these conditions. Selenium dioxide sublimes at a temperature of 340-350 O C Z o and might therefore be responsible for the observed analyte losses in the pyrolysis stage in the presence of Oml t O '2;o 2;o 2;o 2k 2;o 280 *;o 3'00 Wavelengthhm Fig.3 Molecular absorption spectrum measured during the vaporization of 30 pg of sodium sulfate without the addition of a chemical modifier and without a pyrolysis stage vaporization temperature 2 100 "C magnesium sulfate and also of sulfuric acid. This is in contrast to the observations of Styris et a1.;6 their measure- ments however were made with pure solutions in the absence of any sulfate matrix. An entirely different pattern was exhibited by the molecular spectrum observed during the evaporation of 30 pg of sodium sulfate (Fig. 3). The spectrum closely resembled that of the CS molecule which is characterized by the prominent band head at 257.6 nm which degrades to the red.23*24 This means that sodium sulphate was not volatilized without decomposition in the graphite furnace but was partly reduced on reaction with the graphite.It should not be excluded that in this instance selenium might have reacted in an analogous manner and that the formation of molecules such as CSe and SSe might be responsible for some analyte losses and interferences. Co- expulsion with the violently volatilizing matrix however is probably the predominant mechanism of interference as discussed earlier. In the presence of all the investigated sulfate model compounds the mechanism of interference involved a loss of the analyte element at different stages of the temperature programme and not a gas-phase interference. With magne- sium sulfate and sulfuric acid these losses were caused by the formation and volatilization in the pyrolysis stage of SeOz in the presence of the oxidizing SO3 matrix that was generated on decomposition of these compounds.In the presence of sodium sulfate the losses were caused predomi- nantly by analyte expulsion with the violently volatilizing matrix early in the atomization stage. In both instances the analyte losses could be reduced significantly by converting the sulfate into a thermally more stable compound such as barium sulfate (see Table 2) which decomposes only at a temperature high enough to make analyte losses in the pyrolysis stage very unlikely. Any Se02 formed in the atomization stage however should be atomized under the STPF conditions used in this work. Analyte element expulsion would also be less likely than in the presence of sodium sulfate because of the higher thermal stability of barium sulfate which should result in a less violent volatilization.The addition of barium would obviously have the expected releasing effect only if the stability of barium sulfate as indicated in Table 2 was not affected too much by carbon as was the case for magnesium sulfate. The final answer about the success of this measure could only be expected from the experiment. The atomization signals for 1 ng of selenium in the presence of sulfuric acid sodium sulfate and magnesium sulfate using the palladium nitrate -magnesium nitrate-barium nitrate modifier are shown in Fig. 4. In contrast to the signals with the palladium nitrate-magnesium nitrate modifier shown in Fig. 1 a508 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY APRIL 1992 VOL.7 Table 3 Recovery of 1 ng of selenium added to various mineral and medicinal waters comparison of palladium nitrate-magnesium nitrate and palladium nitrate-magnesium nitrate-barium nitrate modifiers. The selenium content in all mineral waters was below 3 pg 1-' Matrix concentration/mg I-' Recovery of selenium* (O/O) -- Sample Na SO42- Pd-Mg Pd-Mg-Ba Uberkinger (Germany) 995 1 1 1 1 Imnauer (Germany) 28 650 Weissenberger (Germany) 5.4 265 Paradi (Hungary) 470 313 Poljustrovo (Romania) 100 135 Radenska (Yugoslavia) 5 50 108 Icelandic (Iceland) 4.0 < 5 *Mean 2 SD (n = 5). 83 k 3.0 94k 1.5 89 ~t 2.0 99k 1.2 88 k 3.1 98 k 2.0 89 rt 2.5 97 rt 2.5 87 k 3.0 89 k 2.3 98 k 2.0 9 8 2 1.7 100k2.1 100k2.4 200 "C higher atomization temperature had to be applied in the presence of barium because of the signal broadening caused by this modifier component.The slightly smaller half-width of the signals in Fig. 4 compared with those in Fig. 1 was hence caused by the higher atomization temperature. The appearance temperature however was the same with the two modifiers which means that the atomization mechanism was not altered by the barium. This was actually according to expectations as the compo- nent responsible for analyte stabilization palladium was active in the same manner in both modifier mixtures. The most significant change in signal shape was found in the presence of sodium sulfate where the maximum of the background appeared significantly later and the early selenium peak which was caused by co-volatilization and analyte expulsion had disappeared. The interference due to sulfate disappeared as well; the integrated absorbance 0 L for selenium in the presence of sodium sulfate was only insignificantly lower than that in a matrix-free solution.The interferences in the presence of sulfuric acid and of rrnagnesium sulfate disappeared similarly. The only appar- ent change was that the selenium signal was less symmetric and more noisy in the presence of the palladium nitrate- magnesium nitrate-barium nitrate modifier. This could be explained by the decomposition of the barium sulfate during the atomization stage which also became apparent in the higher background signal. This however did not result in any detectable interference in the atomization of the analyte element as is apparent from the integrated absorbance values measured.Table 3 shows the recovery of selenium added to a series of mineral waters with different sulfate contents. The natural content of selenium in all mineral waters was below the detection limit of 3 pg I-'. The recovery of selenium was significantly better in the presence of the palladium nitra- te-magnesium nitrate-barium nitrate modifier compared with the previously used palladium nitrate-magnesium initrate modifier. The proposed addition of barium there- *fore proved to be useful also in a practical application .which supported the proposed interference and release %mechanisms. It was not possible however to remove the :sulfate interference entirely using the proposed conditions. 'The tolerable amounts of sulfate for a 95% recovery of selenium using the recommended masses of the palladium nitrate-magnesium nitrate-barium nitrate modifier a py- rolysis temperature of 800 "C and an atomization tempera- ture of 2300 "C were 15 pg of sulfate in the form of sodium sulfate 20 pg in the form of sulfuric acid and 60 pg when sulfate was present as the magnesium salt.The character- istic mass for selenium was around 25 pg under these conditions which is in good agreement with published data. The detection limit expressed as three times the standard deviation (SD) of a blank solution was 3 pg 1-l of selenium using 10 pl of sample solution. AA 0.4 BG 0.6 (d) Fig. 4 Atomization signals for 1 ng of selenium in the presence of various sulfate compounds (20 ,ug of S042- each) using the palladium nitrate-magnesium nitrate-barium nitrate modifier pyrolysis temperature 800 "C; and atomization temperature 2300 "C.The dotted signal is background absorbance (BG) (a) reference solution matrix-free integrated absorbance 0.18 1 s; (b) sulfuric acid integrated absorbance 0.17 1 s; (c) sodium sulfate integrated absorbance 0.174 s; and (d) magnesium sulfate integrated absor- bance 0.183 s Conclusion The palladium nitrate-magnesium nitrate mixed modifier is extremely valuable and versatile as it could be applied successfully to a large number of analyte elements in a variety of matrices. Care should be taken however in calling it a universal modifier because it certainly has limitations that then require special action i.e.deviation from the concept of a 'universal' modifier. In this work this special action was the addition of a third compound barium nitrate to the modifier mixture and in earlier work on the determination of thallium in a high-chloride matrix it was the pre-reduction of palladium and the addition of hydrogen to the purge gas.2s In this context it should be considered that it is probably not desirable to have a truly universal modifier ie. a modifier that stabilizes all elements to the same extent. This would essentially lead toJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY APRIL 1992 VOL. 7 509 the situation that matrix components are stabilized to the same extent as the analyte element preventing a separation during the pyrolysis stage which is considered an important requirement of a chemical modifier.Instead of searching for a truly universal modifier and also of investigating more and more sophisticated proce- dures to increase the efficiency of the undoubtedly success- ful palladium nitrate-magnesium nitrate modifier it might be better to look for alternative methods when the mixed modifier does not fulfil the requirements. For the present application the determination of selenium in the presence of high sulfate concentrations such alternative procedures could be the application of hydride generation atomic absorption spectrometry or an on-line preconcentration and separation of the analyte element26 prior to its determination by ETAAS. References 1 Cedergren A. Lindberg I. Lundberg E. Baxter D.C. and Frech W. Anal. Chim. Acta 1986 180 373. 2 Styris D. L. Fresenius’ 2. Anal. Chem. 1986 323 7 10. 3 Droessler M. S. and Holcombe J. A. Spectrochim. Acta Part B 1987 42 98 1. 4 Dedina J. Frech W. Lindberg I. Lundberg E. and Cedergren A. J Anal. At. Spectrom. 1987 2 287. 5 Droessler M. S. and Holcombe J. A. J. Anal. At. Spectrom. 1987 2 785. 6 Styris D. L. Prell L. J. Redfield D. A. Holcombe J. A. Bass D. A. and Majidi V. Anal. Chem. 1991 63 508. 7 Slavin W. Manning D. C. and Carnrick G. R. At. Spectrosc. 1981 2 137. 8 Ediger R . D. At. Absorpt. Newsl. 1975 14 127. 9 Dedina J. Frech W. Cedergren A. Lindberg I. and Lundberg E. J. Anal. At. Spectrom. 1987 2 435. 10 Welz B. Schlemmer G. and Voellkopf U. Spectrochim. Acta Part B 1984 39 501. 11 Shan X.-Q.and Ni Z.-M. Acta. Chim. Sin. 1981 39 575. 12 13 14 15 16 17 18 19 20 Schlemmer G. and Welz B. Spectrochim. Acta Part B 1986 41 1157. Lindberg I. Lundberg E. Arkhammar P. and Berggren P. O. J. Anal. At. Spectrom. 1988 3 497. Teague-Nishimura J. E. Tominaga T. Katsura T. and Matsumoto K. Anal. Chem. 1987 59 1647. Welz B. Schlemmer G. and Mudakavi J. R. J. Anal. At. Spectrom. 1988 3 93. Welz B. Schlemmer G. and Mudakavi J. R. J. Anal. At. Spectrom. 1988 3 695. Bozsai G. Schlemmer G. and Grobenski Z. Talanta 1990 37 545. Carnrick G. R. Manning D. C. and Slavin W. Analyst 1983 108 1297. Esser P. and Durnberger R. Fresenius’Z. Anal. Chem. 1987 328 359. CRC Handbook of Chemistry and Physics ed. Weast R. C. CRC Press Cleveland OH 58th edn. 1978. 21 Schmidt M. and Siebert W. in Comprehensive Inorganic Chemistry eds. Bailor J. C. Emeleus H. J. Nyholm R. and Trotman-Dickenson A. F. Pergamon Press Oxford 1 973 VOl. 2 pp. 795-933. 22 Gmelins Handbook of Inorganic Chemistry Gmelin Verlag Clausthal-Zellerfeld 8th edn. 1949 p. 160. 23 Pearse R. W. B. and Gaydon A. G. The Identifrcation of Molecular Spectra Wiley New York 1963. 24 Martinsen I. and Langmyhr J. F. Anal. Chim. Acta 1982 135 137. 25 Welz B. Schlemmer G. and Mudakavi J. R. Anal. Chem. 1988,60 2567. 26 Sperling M. Yin X. and Welz B. J. Anal. At. Spectrom. 1991 6 295. NOTE-Refs. 15 and 16 are to Parts 1 and 2 of this series respectively. Paper 1 /05002H Received September 30 I991 Accepted November 28 1991

ISSN:0267-9477    

DOI:10.1039/JA9920700505

出版商:RSC    

年代:1992

数据来源: RSC

摘要:

JOURNAL OF ANALYTICAL ATOMIC SPECTRMETRY APRIL 1992 VOL. 7 51 1 Use of Organophosphorus Vapours as Chemical Modifiers for the Determination of Cadmium by Electrothermal Atomic Absorption Spectrometry Les Ebdon Andrew S. Fisher and Steve Hill Plymouth Analytical Chemistry Research Unit Department of Environmental Sciences Polytechnic South West Drake Circus Plymouth PL4 8AA UK The use of the vapours of several organophosphorus liquids as chemical modifiers for the determination of cadmium by electrothermal atomic absorption spectrometry is described. The best results were obtained when triethyl phosphite was used as the modifier. These vapours were found to stabilize cadmium to the same extent as a more traditional modifier but with a reduced possibility of contamination. The method was validated using certified reference materials (NIES Human Hair and Sargasso).Keywords Electrothermal atomization; organophosphorus vapour; chemical modification; cadmium determina- tion; atomic absorption spectrometry Cadmium is known to induce toxic effects in certain organisms at low concentrations. Therefore the determina- tion of cadmium in a variety of matrices by electrothermal atomic absorption spectrometry (ETAAS) has become common. 1-5 Serious problems may be encountered during the analyses because of the very high volatility of cadmium. This restricts the ashing (pyrolysis) temperatures which can be used hence the chemical constituents cannot be readily removed during the ash stage and thus non-specific absorbance interference (smoke) may seriously affect the analytical signal.To overcome this problem a number of chemical modifiers have been used to stabilize thermally the cadmium and allow higher ashing temperatures. These modifiers include alcoholic potassium hydr~xide,~ ammo- nium oxalate which also aids the volatilization of chloride ions,4 lanthanum-nitric acid,6 magnesium nitrate,’ pal- ladium-magnesium nitrate8v9 and palladium-ammonium nitrate.IO The most common modifiers used for cadmium are the mono- and dibasic forms of ammonium phos- phate. Unfortunately the use of many of these modifiers may give rise to several disadvantages. The lanthanum modifiers have been found to have a corrosive effect on the graphite tubes which curtails their useful lifetime,g the palladium modifiers can be expensive and the ammonium phosphate modifiers are often not available in a state of high enough purity9 and can give rise to elevated background signal^.^^^^ Some workers have attempted to purify their modifiers by passing them through a Chelex 100 chelation-exchange column12 or by extraction with ammonium tetramethylene- dithiocarbamate (ammonium pyrrolidinedithiocarbamate) -chloroform.l 3 This however leads to additional time- consuming preparation stages which are undesirable analytically. An inexpensive contamination free gaseous chemical modifier would go some way to ameliorating these disad- vantages while still possessing the thermal stabilizing properties exhibited by the ‘wet chemical’ modifiers. This paper describes the use of the vapours of some organophos- phorus compounds to stabilize cadmium up to tempera- tures comparable to those used with the traditional ‘wet chemical’ modifiers.Experimental Reagents and Standards High-purity water was obtained by using reverse osmosis (Milli-R04; Millipore Harrow UK) followed by adsorp- tion de-ionization and ultrafiltration (Millipore Milli-Q system). A 1000 pg ml-l standard of cadmium as the nitrate (Spectrosol Merck Poole UK) was used as the stock standard solution. Working standards were prepared by serial dilution of the stock standard on a daily basis. Various organic liquids with a relatively high vapour pressure were used as prospective chemical modifiers. These were triethyl phosphite (GPR Merck) triethyl phosphate (GPR Merck) and trimethyl phosphite (97% Aldrich Gillingham Dorset UK).As a comparison am- monium dihydrogenphosphate (AnalaR Merck) was also used for some analyses. Instrumentation All analyses were performed using an atomic absorption spectrometer (PU 9 1 OOx Philips Scientific Cambridge UK) fitted with an electrothermal atomizer (PU939Ox Philips Scientific) and a data station (PU9 178x Philips Scientific). A Dreschel bottle containing the modifier under investigation was placed in the gas line between an argon source and one of the alternative gas inlets of the electroth- ermal atomizer. Electrographite tubes were used through- out and in-house fabricated electrographite platforms were used for some analyses. Sample introduction was by hand- held micropipettes (Gilson Villiers-Le-Bel France). Sample Preparation Certified reference materials NIES No.5 Human Hair and NIES No. 9 Sargasso (National Institute for Environmental Science Yatabe-Machi Tsukuba Ibaraki Japan) were prepared in the following manner the sample (approxi- mately 0.2 g) was weighed accurately into acid-washed poly(tetrafluorethy1ene) bomb receptacles 2 ml of nitric acid (Aristar Merck) were added and the receptacles were fastened securely in stainless-steel bomb cases. These were heated in an oven at 1 10 “C for 2 h and after cooling 0.5 ml of perchloric acid (AnalaR Merck) was added to the receptacles which were then re-fastened in the cases and replaced in the oven for a further 1 h. After cooling the contents were transferred quantitatively into acid-washed 25 ml calibrated flasks and diluted to volume with water.A blank was prepared in a similar fashion but omitting the sample.512 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY APRIL 1992 VOL. 7 ~~ ~~ Table 1 Furnace programme used for the ashing plots for cadmium by wall atomization Set 110 35 20 - 20 50 Stage temperature/"C Hold time/s RampPC s-* Dry Ash Atomize 1600 2.5 Full power Clean 1800 3 Full power Table 2 Analytical operating conditions for cadmium analyses Wavelength 228.8 nm Lamp current 4 mA Band p a s s 0.5 nm Injection volume 20 pl Table 3 Furnace programme used for ashing plots for cadmium using platform atomization Set 180 35 30 - 20 50 Stage temperaturePC Hold time/s Ramp/"C s-' Dry Ash Atomize 2000 3 Full power Clean 2200 4 Full power Table 4 Furnace programme for ashing plots for cadmium using a platform and ammonium phosphate modifier Set 180 35 30 180 35 30 - 20 50 Stage temperature/"C Hold time/s Ramp/"C s-* - Dry Cool 20 20 Dry Ash Atomize 2000 3 Full power Clean 2200 4 Full power Procedure Initially ashing plots were constructed for a 1 ng ml-1 cadmiurn standard with wall atomization in the absence of any modifier in the presence of triethyl phosphite then triethyl phosphate and finally trimethyl phosphite.The vapours were introduced during the dry and ash stages. The furnace programme and the analytical operating conditions used are detailed in Tables 1 and 2 respectively. A similar experiment was then performed with platform atomization using no modifier triethyl phosphite and ammonium phosphate 'wet chemical' modifier. The furnace programme used for platform atomization with the organophosphorus chemical modifiers and for the 'wet chemical' modifier are detailed in Tables 3 and 4 respectively.The ammonium phosphate modifier was introduced into the furnace during a cool stage just after the dry stage. Results The results obtained from wall and platform atomization are shown in Figs. 1 and 2 respectively. It can be seen that all of the modifiers thermally stabilize the cadmium to some extent. For wall atomization the maximum ashing temper- ature without any modifier was only 400-450 "C but the use of triethyl phosphite vapour stabilized the cadmium to temperatures of up to 700 "C. The extent to which the different modifiers stabilized the cadmium was determined by the hydrolytic properties of the individual vapours.Triethyl phosphite hydrolyses in moist air whereas the triethyl phosphate hydrolyses only slowly in the presence of 1 .- c ""250 350 450 550 650 750 850 Ashing temperature/"C Fig. 1 Ashing plots of cadmium with different modifiers and no platform A no modifier; B trimethyl phosphite; C triethyl phosphate; and D triethyl phosphite 0.6 22 $ 0.5 (rJ n 5 0.4 M) n a 0.3 Q) * & 0.2 P) + 350 550 750 950 - 0.1 Ashing temperaturePC Fig. 2 Ashing plots for cadmium from a platform A no modifier; 13 triethyl phosphite; and C ammonium phosphate water.I4 By introducing the vapours during the dry and ash stages of the furnace cycle the triethyl phosphite vapour should readily hydrolyse depositing modifiying species in the graphite tube.These can then interact with the cadmium ions which are consequently thermally stabil- ized. The triethyl phosphate only partially hydrolyses and hence it has a smaller stabilizing effect. Trimethyl phos- phite was used because it could be expected to have a higher vapour pressure than the ethyl analogue. It should therefore be more efficient in transporting modifying species to the furnace. Its powerful smell seemed to confirm that it did have a higher vapour pressure but the results obtained showed no further improvement in the stabilizing effect indicating that either the methyl derivative is less readily Inydrolysed or that sufficient modifying species were being deposited with the ethyl derivative to obtain maximum stabilization. In addition the methyl derivative also re- sulted in a slight depression in sensitivity which could not be readily explained.Inspection of Fig. 2 shows that the triethyl phosphite vapour was as efficient at stabilizing the cadmium as the "wet chemical' counterpart. Additionally the ammonium phosphate led to an increased signal. Analysis of modifier solutions with no added cadmium yielded signals compar- able to the increased signal observed in Fig. 2. This increased signal was therefore attributed to cadmium contamination of the 'wet chemical' modifier. The use of blank measurements indicated that no such impurities were present in the organophosphorus vapour. Impurities such as those found in the ammonium phosphate are obviously undesirable especially when determinations close to the detection limit are to be performed.As the use of triethyl phosphite yielded the best results it was decided that this vapour alone should be used for all further studies. Evaluation of the Method Having established that aqueous cadmium standards may be thermally stabilized by triethyl phosphite vapour it wasJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY APRIL 1992 VOL. 7 513 0.4 e 8 2 (D 0.3 0 0 P 0.2 2 *.' 0 Y u. I ~ 200 300 400 500 600 700 800 900 Ash i ng tern per at u rePC Fig. 3 Ashing plots for certified reference material NIES No. 5 Human Hair A no modifier; and B triethyl phosphite necessary to evaluate the method with real samples. Certified reference materials NIES No. 5 Human Hair and No. 9 Sargasso were chosen to validate the method. Ashing plots of the digests from wall atomization were constructed to ensure that the triethyl phosphite would thermally stabilize cadmium in real samples in addition to aqueous standards. The furnace programme and analytical operating conditions used are detailed in Tables 1 and 2 respectively.The results obtained for the ashing plots of these materials showed that a maximum ashing tempera- ture of 700 "C could be used in the presence of the modifier but only 600 "C in its absence. The results for hair are shown in Fig. 3. As the maximum ashing temperature for cadmium in an aqueous standard was only 450 "C it would appear that the chemical was exerting a stabilizing effect on the cadmium. Presumably this was due to the presence of phosphate ions in the chemical. An ashing temperature of 700 "C was chosen for the analysis of the certified samples.The results based on integrated absorbance measurements of the analyses are shown in Table 5. Good agreement with the certificate values was obtained. As accurate results were obtained from wall atomization it was not deemed neces- sary to repeat the analyses using platform atomization. Conclusions Owing to the extremely volatile nature of many of its compounds chemical modification has become almost a prerequisite for cadmium determinations. The use of 'wet chemical' modifiers are fraught with impurity problems. The use of triethyl phosphite vapour has been shown to stabilize cadmium thermally to the same extent as the traditional modifiers but without the concomitant contam- ination problems.Other organophosphorus vapours were Table 5 Results for the analysis of certified reference materials n= 5; results expressed as rt 20 Value Certified Material valuelpg g-' obtained/pg g-' NIES No. 5 Human Hair 0.2 k0.03 0.2 f0.02 NIES No. 9 Sargasso 0.15 k0.02 0.12 f 0.02 also used but proved to be either less efficient or as efficient but more unpleasant to use than the triethyl phosphite. Finally the method was validated using certified reference materials. Reasonable agreement with the certificate values was obtained. The authors acknowledge the funding of this work by the Science and Engineering Research Council (SERC) and Philips Scientific under the Co-operative Award in Science and Engineering (CASE) scheme. 1 2 3 4 5 6 7 8 9 10 11 12 13 14 References Delves H. T. and Woodward J. At. Spectrosc. 1981 2 65. Alvarado J. and Petrola A. J. Anal. At. Spectrorn. 1989 4 41 1. Calapaj R. Chiricosta S. Saija G. and Bruno E. At. Spectrosc. 1988 9 107. Knowles M. B. J. Anal. At. Spectrom. 1989 4 257. Ebdon L. and Lechotycki A. Michrochem. J. 1987,36,207. Hunt D. T. E. and Winnard D. A. Analyst 1986 111 785. Majidi V. and Holcombe J. A. J. Anal. At. Spectrom. 1989 4 439. Bulska E. Grobenski Z. and Schlemmer G. J. Anal. At. Spectrom. 1990 5 203. Welz B. Schlemmer,G. and Mudakavi J. R. J. Anal. At. Spectrom. 1988 3 695. Smeyers-Verbeke J. Yang Q. Penninckx W. and Vander- voort F. J. Anal. At. Spectrom. 1990 5 393. Yin X. Schlemmer G. and Welz B. Anal. Chem. 1987,59 1462. Foote J. W. and Delves H. T. Analyst 1988 113 91 1. De Benzo Z. A. Fraile R. Carrion N. and Loreto D. J. Anal. At. Spectrom. 1989 4 397. Merck Hazard Data Sheets Merck Poole 1989. Paper I /05 I44J Received October 10 I991 Accepted December 5 1991

ISSN:0267-9477    

DOI:10.1039/JA9920700511

出版商:RSC    

年代:1992

数据来源: RSC

摘要:

JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY APRIL 1992 VOL. 7 515 Study of Palladium-Analyte Binary System in the Graphite Furnace by Surface Analytical Techniques Yang Peng-yuan Department of Chemistry Xiamen University Xiamen 361 005 China Ni Zhe-ming Research Center for Eco-Environmental Science P. 0. Box 934 Beying 100085 China Zhuang Zhi-xia Xu Fu-chun and Jiang An-bei Analytical Testing Center Xiamen University Xiamen 36 1 005 China The mechanisms of the stabilization of Pb Zn and As by a palladium modifier were investigated by scanning electron microscopy X-ray diffraction spectrometry and X-ray photoelectron spectrometry. All three elements were found to form an intermetallic solid solution with excess of Pd. The concentration ratio of Pd to analyte varies with the surface depth of Pd-anafyte species.Chemical shifts measured in the binding energy of Pd and analyte are insignificant. The monophase of the Pd lattice in which some Pd atoms are displaced by the analyte atoms has been identified as being dominant. It is suggested that the reduced Pd and analyte form a stable intermetallic solid solution during the ashing stage and the analyte atoms remain in the Pd lattice until the temperature is high enough to break down the lattice. Keywords Palladium modifier; electrothermal atomic absorption spectrometry; surface analysis; atomization mechanism Since Shan and Nil first reported the use of Pd as a stabilizer for Hg in 1979 Pd has been widely employed as a matrix modifier for many volatile elements in electrother- mal atomic absorption spectrometry (ETAAS).I-lO Re- cently this modifier has been demonstrated to be also a very effective adsorbent for trapping hydrides.I0*I1 How- ever the mechanism of the Pd modification is still not well understood. Early investigations on the formation of Pd-Pb and other binary systems were carried out by Shun12 in 1981. Later Shan and Wang13 reported that the formation of Pb-Pd and Bi-Pd bonds could be determined from the shift of binding energies by X-ray photoelectron spectrometry (XPS). More recently Wend1 and Muller-Vogtl* reported some interest- ing results of X-ray diffraction (XRD) spectrometry show- ing the presence of Pd3Pb and Pd3Pb2 intermetallic com- pounds in the graphite furnace but no further data were available. The reduction of Pd has been testified as a necessary step by several workers.Voth-Beach and ShraderIS applied an additional reducing agent to the Pd modifier such that Pd is reduced to the metallic form early in the temperature programme resulting in a high ashing temperature for analytes. A similar observation was reported by Teague- Nishimura et a1.,16 who observed that Se can be stabilized by Pd effectively in the presence of ascorbic acid. Recently Rettberg and Beach" employed 5% hydrogen as a sheath gas during the Pd pre-treatment step and found that the Pb signal recovery was nearly consistent from 200 to 1000 "C. In order to clarify the mechanism of the stabilization of analytes by a Pd modifier many fundamental studies need to be pursued especially those related to the surface chemistry of Pd.In this work the surface interaction between Pd and analytes on the graphite platform was investigated. Three surface analytical techniques XPS XRD spectrometry and scanning electron microscopy (SEM) were applied to determine the species generated on the graphite platform. Possible mechanisms are proposed and discussed. Experimental Instrumentation The sample pre-treatment was performed on a Perkin- Elmer 3030 atomic absorption spectrometer fitted with an HGA-500 graphite furnace. In all experiments pyrolytic graphite platforms (Perkin-Elmer part No. BO 10-9324) were utilized to hold the samples. The XRD spectra were obtained by a RigakulNew X-ray diffraction system (Rotaflex Dlmax-c Series) equipped with a copper target source. The samples were scanned over 20-120" diffraction angles at a scanning rate of 6" min-l.Computerized data acquisition and data processing systems were employed to collect the data and analyse the results. Scanning electron micrographs of sample surfaces were accomplished using a Hitachi S-520 instrument associated with an EDAX-9 100 X-ray energy dispersion analyser. The XPS was applied to characterize the surface com- pounds of the Pd-analyte system on the graphite platform. The instrument used was an Escalab MK I1 (VG Analytical) equipped with a magnesium target (Ka) and fitted with an argon ion gun ( 5 kV) for sputtering the surface layer. The erosion rate was approximately 1 nm min" at 30 pA current with a beam diameter of -8 mm. The instrument is also capable of performing measurements of accompanied Auger electron spectra (AES).Reagents and Solutions Analytical-reagent grade chemicals were used. A Pd stock solution was prepared by dissolution of Pd powder (99.999%) in a minimum amount of concentrated HN03-HCl diluted to a final concentration of 4 g 1-l. Stock solutions (1 g 1-l) of Pb and Zn were prepared from the chlorides and of As from the arsenite. The sheath gas was pure argon (99.99%). Procedure In the general procedure the sample solution was placed on the platform and dried at 100 "C for 1-2 min and this process was repeated until the required amount of sample had accumulated. The sample was then ashed at a desig- nated temperature for more than 2 min. Particularly when the sample was prepared as a Pd-analyte system an ashing temperature of 1000 "C was applied and when an analyte- only sample (Pb As and Zn) was ashed as a reference system in the absence of Pd a temperature of 300 or 400 "C was used to avoid losses of the analyte due to overheating.After cooling the graphite platform was carefully removed516 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY APRIL 1992 VOL. 7 3.0 (a) 0 8 - 0' I I 20.0 50.0 100.0 120.0 20.0 50.0 100.0 120.0 0.8 0.4 n io.0 50.0 100.0 120.0 26 ("1 Fig. 1 XRD of Pd-analyte system (a) Pd-Pb species; (b) Pd-As species; and (c) Pd-Zn species with surface sputtered by a 5 keV 30 pA argon ion beam for 30 rnin. The sample was ashed on the graphite platform at 1000 "C for 4 min prior to obtaining the spectra. (This sample treatment also applies to the spectra shown in Figs.2-5 unless specified otherwise.) For the marked values above the identified peaks the first number is the experimental d- parameter of the Pd-analyte system and the second number in parentheses is the deviation of this d-parameter relative to the experimental d-parameter for a pure Pd system. The three unmarked wide peaks refer to graphite carbon from the graphite tube and stored in a sealed glass tube purged with argon gas. No further sample treatment was carried out before using the sample for surface testing. The waiting time between sample preparation and testing could last from a few minutes to several hours depending on the instrument loading. No on-line experimental measurements were considered owing to difficulties with the required instrumentation.The amount of analytes and Pd were usually 1000- 10 000 times greater than those applied in a normal routine analysis mainly owing to the less sensitive detection capabilities of the surface analytical instruments employed. The amount of samples were 4 mg for Pd 0.07 mg for As 0.05 mg for Pb and 0.05 mg for Zn in the Pd-As Pd-Pb arid Pd-Zn systems tested respectively. In XRD analysis the XRD spectrum for the pure Pd sample was obtained first and served as a reference stlandard prior to the subsequent identification of other spectra. Results and Discussion XRD Results Palladium can form a number of binary systems (or alloys) with many volatile elements exhibiting some similar feature in their phase diagrams.18 The solubility of a volatile element in Pd is about 10-20% in a Pd monophase and the binary system formed is called an intermetallic solid solution.An intermediate phase would appear only when a large amount of the doped volatile element is present and an intermetallic compound could result. The formation of an intermetallic solid solution or compound can be identi- fied by the XRD technique. Fig. l(a) shows the XRD spectrum of the Pd-Pb binary system. All peaks were carefully identified against the standard powder diffraction d-values for Pd.19 It was surmised that some major peaks of a Pd,Pb (x,y=1-3) compound would appear14 in an intermediate phase. Unfortunately the major peaks illustrated in Fig. l(a) virtually correspond to a Pd monophase with a correspond- ing peak sequence and intensity order.Hence no peak for a Pd,Pb intermetallic compound was observed with the sensitivity of the XRD instrument used in this work. The amount of Pb was purposely increased to 0.3 mg to increase the ratio of Pb to Pd but the peaks of Pd,Pb and of Pb were b'arely visible. It was evident that the XRD spectrum for the Pd-As binary system also does not exhibit peaks for Pd,As and As as depicted in Fig. l(b). In Fig. l(b) all peaks are identified as being for the Pd a-monophase. The XRD spectra were also taken for both the surface and inner layer of the Pd-Zn system and the spectrum for the inner layer is illustrated in Fig. l(c). Again no peak other than for a Pd monophase can be seen. Small shifts in the powder diffraction d-values between the pure Pd and Pd-analyte systems were observed and the order of peak intensities also differs from that for pure Pd as indicated in Fig.l(a)-(c). For instance the correspond- ing shift in the d-value for the Pd-As system is about 0.01. As is known from the phase diagram the presence of an intermediate phase (e.g. a Pd,Pb phase) is hardly possible on the basis of the solubility of Pb in Pd metal when the ratios of Pd to analytes are greater than 1O:l. The experimental results are essentially in good agreement with the phase diagrams of Pd-Pb,20 Pd-Znzl and Pd-As.22 SlEM Results The displacement energy spectrum of the Pd-As system was scanned by the SEM X-ray microanalyser. The As species which could not be directly identified in the XRD spectrum exhibits a strong peak in the spectrum obtained.From the data acquired by the X-ray microanalyser the surface mole ratio of Pd to As was roughly obtained from several speckles on the sample surface having a value close to 3 1 analogous to that reported previously.' Because the electron beam can penetrate 1-3 pm below the surface caution should be exercised when SEM results are applied tlo interpret surface chemical phenomena. Particular consideration was given to acquiring an image map of species on the surface such as those reported in a similar experiment on a Pd-Se system. l6 The image maps of As and Pd La emission scanned ten times over a surfaceJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY APRIL 1992 VOL. 7 517 speckle of the sample by the X-ray microanalyser were obtained. The developed micrographic films displayed no bright spots for either Pd or As.XPS Results Fig. 2 illustrates the binding energy for Pb species in the absence and presence of Pd. The spectra were analysed against the reference value.23 When the Pb species (added as PbC12 solution) exists in the absence of Pd the XPS peak corresponds to that of PbC12 and/or PbO (curve A the 0.7 eV gap in binding energy between PbClz and PbO is not readily distinguishable). In the presence of Pd double peaks appear in the spectrum. The peak of 138.5 eV was identified as a mixture of PbC12 and/or PbO and the other peak of 136.8 eV was Pb metal when the added amount of Pb is large (curve B) or small (curve C). The presence of surface oxygen and chlorine was examined by XPS scanning around 284.7 eV for oxygen and 198.8 eV for chlorine. The peak height for Pb is much greater than that for PbC12 and PbO indicating that the reduction of PbO seems to be sufficient in the presence of Pd.24 An argon ion beam was utilized to sputter the surface layer to determine the variation of the amount of Pb with increasing surface depth.The spectra obtained are also displayed in Fig. 2. Curves C (without sputtering) D (3 rnin sputtering) and E (6 rnin sputtering) have similar profiles indicating that the distribution of Pb atoms is nearly uniform in the tested layer. The mole ratio of Pd to Pb was calculated to be about 14 I using Wagner's approach.25 The measured shift of Pb binding energy in the Pd-Pb system was found to be relatively small within -0.25 eV and the shift of the Pd binding energy is only about 0.1 eV.This small shift is as expected. As is well known empirically in XPS,23 intermetallic species usually exhibit smaller shifts in binding energy than those of chemical compounds 1 I I I I 134 136 138 140 142 144 146 Binding energyfev Fig. 2 XPS of Pb species for the Pd-Pb binary system. The standard binding energies are Pb 136.8 PbO 138.2 and PbC12 138.9 eV. A Surface spectra of PbClz (ashed at 300 "C); B surface spectra of Pd-Pb system with a ratio of 13 1 ; C surface spectra of Pd-Pb system with a ratio of 80 1; D inner-layer spectra of Pd-Pb after sputtering by a 5 keV 30 PA argon ion beam for 3 min; and E inner-layer spectra of Pd-Pb after sputtering for 6 rnin / / I I 1 I I I I 35 37 39 41 43 45 47 49 Binding energyfev Fig. 3 XPS of As species for the Pd-As binary system.The standard binding energies are As 41.3 and 41.9 and As,O 44.2 and 44.8 eV. A Surface spectra of As (As203 ashed at 300 "C); B surface spectra of Pd-As system; C inner-layer spectra of Pd-As after sputtering by a 5 keV 30 pA argon ion beam for 5 min; and D inner-layer spectra of Pd-As after sputtering for 10 rnin probably owing to the free electron mobility in the intermetallic binding. The Pd-As system was investigated in a similar manner and the spectra are illustrated in Fig. 3 for As203 in the absence of Pd (curve A) and the Pd-As system on the surface (curve B) and on the inner layers after 5 min (curve C) and 10 rnin (curve D) sputtering by the argon ion beam. The surface As peaks for the Pd-As system (curve B) were identified as being that of As203 (large peak at 44.2 eV oxygen was checked by XPS but its source was not identified) and that of As (small peaks at 4 1.3 and 4 I .9 eV).However no As203 peak appeared on the inner layers (curves C and D). It can be deduced from Fig. 3 that a large amount of As203 is present on the surface layer and that the amount of As species gradually decreases as the surface depth increases. The mole ratio of Pd to As on the surface is about 2:1 compared with 3:1 calculated from the SEM experi- ment. It is not clear why As203 could remain on the surface resisting a 1000 "C ashing. It is possible that the reduction in surface As203 on the graphite is incomplete even in the presence of Pd. It seems that if As203 could be reduced completely at an earlier time in the ashing stage As would be stabilized by Pd more effectively.16 A shift of 0.1 eV in the Pd binding energy was indicated in the XPS of Pd in the absence and presence of As. A relatively large shift about - 0.6 eV in the binding energy of As203 can be seen in the As spectra in the absence and presence of Pd.However these shifts do not support the suggestion that the Pd-As system is close to a chemical compound. It was found that the behaviour of Zn in the Pd-Zn system exhibits a different trend to that of the Pd-Pb system. Fig. 4 shows the Zn spectra on the surface of the Pd-Zn system (curve A) and on the inner layers after sputtering by the argon ion beam for 20 min (curve B) 35 rnin (curve C) 45 min (curve D) 65 rnin (curve E) and 85 min (curve F).In contrast to the species in the Pd-As system surface Zn or ZnO could barely be detected (curve A). As the surface depth increases the Zn peaks become more and more intense (curves B-D) and finally level off (curves D-F). The absence of ZnO was inferred by XPS owing to blank oxygen signals.518 JOURNAL OF ANPLYTICAL ATOMIC SPECTROMETRY APRIL 1992 VOL. 7 I I 1 I I 1 1015 1017 1019 1021 1023 1025 1027 Binding energyleV Fig. 4 XPS of Zn species for the Pd-Zn binary system. The standard binding energy for Zn is 102 1.45 eV. A Surface spectra of Pd-Zn system; B inner-layer spectra of Pd-Zn system after sputtering by a 5 keV 30 ,uA argon ion beam for 20 min; C inner- layer spectra after sputtering for 35 min; D inner-layer spectra after sputtering for 45 min; E inner-layer spectra of Pd-Zn after sputtering for 65 min; and F inner-layer spectra of Pd-Zn after sputtering for 85 min 2 D 258 262 266 270 "1019 1023 1027 Binding energylev Fig.5 X-ray photoelectron and accompanying Auger electron spectra of Zn species in the absence of Pd. (a) Auger electron spectrum; the standard binding energies are 264.6 eV for Zn and 269.4 eV for Zn2+; and (b) X-ray photoelectron spectrum; the standard binding energy is 1021.45 eV. The ZnClz solution was injected and dried several times to accumulate a total Zn mass of 1 mg; the sample was then ashed at 300 "C for 4 min. A Spectra for the surface Zn sample; B inner-layer spectra after sputtering by a 5 eV 30 pi argon ion beam for 60 s; C inner-layer spectra after sputtering for 180 s; and D inner-layer spectra after sputtering for 360 s The results showed that the shift of Pd binding energy in the Pd-Zn system is again as small as 0.1 eV and the shift of Zn binding energy is about -0.25 eV. This small shift is in agreement with that explicitly depicted in the Pd-Zn XRD spectrum in which only the Pd monophase is present.In the absence of Pd the X-ray photoelectron spectra [Fig. 5(b)] and accompanying Auger electron spectra [Fig. 5(a)] were recorded for Zn which was added as ZnC1 and subjected to a 400 "C ashing temperature. Auger electron spectrometry was used because of its much larger chemical shift in binding energy (4.6 eV) for Zn2+ in Auger electron spectrometry than that in XPS (0.6 eV).26 It can be seen from Fig. 5 that the Zn2+ (as ZnC1 or ZnO) species is dominant on the surface layer [curve A in Fig.5(a) and (b)]. In the tested inner layer the proportion of Zn2+ is notably smaller and the reduction of ZnO seems to be the major process [e.g. curve D Fig. 5(a)] ZnO was confirmed from a Zn-only Auger electron spectrum). Apparently most of the surface Zn species are unreduced ZnO species and would suffer from being lost at a high temperature even in the presence of Pd as can be seen from comparison of the curves A-D in Fig. 5(b) with curves A-F in Fig. 4. It seems that Zn for which the atomization process could be dominated by ZnO dissociation in the gas phase,27 should preferably be reduced early during the ashing stage in order to prevent ZnO from being lost even in the presence of Pd modifier.28 Bond Formation Between Pd and Analyte Although SEM and XPS indicate the co-existence of analyte and Pd XRD showed only the Pd monophase.The lack of an intermediate phase for an intermetallic compound confirmed that Pd and the analyte form an intermetallic solid solution. Both the intermetallic solid solution and compound have the features of an intermetallic bond but they are different in structure. In the Pd-analyte solid solution analyte atoms replace Pd atoms remaining in the monophase rather than forming an intermetallic compound in a new crystal lattice.18 However the intermetallic bond between Pd and the analyte is probably the major force ]in retarding the analyte vaporization to a relatively high ashing temperature. Weak shifts in the binding energies for the analyte and Pd in their binary system were observed in XPS.These observations provide further support for the above deduc- tion. The intermetallic bond between Pd and the analyte resembles that between Pd atoms in the Pd monophase such that the binding energy shift due to substituted analyte (atoms is expected to be small in a Pd-analyte solid solution. Unfortunately the formation of an intermetallic bond cannot account for some of the results observed. The amount of Zn atoms increases with increase in surface depth of the Pd-analyte species; in contrast that of As atoms decreases with increase in depth and that of Pb atoms remains almost constant in different layers. In fact a large amount of As203 exists on the surface of the Pd-As system and a small amount of PbO on that of the Pd-Pb system but no ZnO exists on the surface of the Pd-Zn system.A possible alternative explanation is that in addition to the intermetallic bond between Pd and the analyte the formation of a chemisorbed bond between Pd and Pb or As oxide species might also be chemically feasible," attributable to an additional spare electron pair in the electron orbital of As or Pb which can also form a bond with the Pd d-~rbital.*~ Conclusion This study explored the possible mechanisms of Pd as a stabilizing modifier for some volatile elements. It is concluded that the reduced Pd and analyte probably form an intermetallic solid solution in a Pd monophase during the ashing and pre-atomization stages. Hence the analyte atoms can remain in the Pd lattice until the temperature is high enough to break down the Pd lattice.JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY APRIL 1992 VOL.7 519 The shifts of binding energies for the studied system were found to be small. The mole ratio of analyte to Pd varies with the surface depth depending on the element investi- gated. It is still unclear why at a relatively high tempera- ture AszO3 and PbO can exist stably on the surface of the Pd-analyte system but ZnO cannot. The sample preparation employed in this study suffers from cooling and being exposed to air in addition to the large excess of sample hence caution should be exercised when the proposed reaction mechanism is applied to real analytical situations. This research was supported by the National Postdoctoral Fellowship Foundation of China The authors appreciate valuable suggestions from Professors B.Huang and T. Huang. References 1 Shan X.-q. and Ni Z.-m. Acta Chim. Sin. 1979 37 261. 2 Shan X.-q. and Ni Z.-m. Acta Chim. Sin. 1981 39 575. 3 Ni Z.-m. and Shan X.-q. Spectrochim. Acta Part B 1987,42 937. 4 Shan X.-q. Ni Z.-m. and Zhang L. Talanta 1984,31 150. 5 Ping L. Fuwa K. and Matsumoto K. Anal. Chim. Acta 1985 173 315. 6 Niskavaara H. Virtasalo J. and Lajunen L. Spectrochim. Acta Part B 1985 40 1219. 7 Grobenski Z. Erler W. and Voellkopf U. At. Spectrosc. 1985 6 91. 8 Schlemmer G. and Welz B. Spectrochim. Acta. Part B 1986 41 1157. 9 Welz B. Schlemmer G. and Mudakavi J. R. J. Anal. At. Spectrom. 1988 3 93. 10 Zhang L. Ni Z.-m. and Shan X.-q. Spectrochim.Acta Part B 1989,44 339 751. 11 Sturgeon R. E. Wille S. N. Sproule G. I. Robinson P. T. and Berman S. S. Spectrochim. Acta Part B 1989 44 667. 12 Shun H.-w. Masters Thesis University of Technology and Science Beijing 198 1. 13 Shan X.-q. and Wang D.-x. Anal. Chim. Acta 1985 173 315. 14 Wendl W. and Muller-Vogt G. J. Anal. At. Spectrom. 1988 3 63. 15 Voth-Beach L. M. and Shrader D. E. J. Anal. At. Spectrum. 1987 2 45. 16 Teague-Nishimura J. E. Tominaga T. Katsura T. and Matsumoto K. Anal. Chem. 1987 59 1647. 17 Rettberg T. M. and Beach L. M. J. Anal. At. Spectrom. 1989 4 427. 18 Evans R. C. Crystal Chemistry Cambridge University Press Cambridge 2nd. edn. 1964. 19 Powder Diffraction File Joint Committee on Powder Diffrac- tion Standards Philadelphia 1974. 20 Marcotte V. C. Metall. Trans. B 1977 8 185. 21 Nowotny H. Bauer E. and Stempfl A. Monatsh. Chem. 1951,82 1086. 22 Smithells Metals Reference Book ed. Brandes E. A. Butter- worth Guildford 6th edn. 1983 pp. 11-64. 23 Handbook of X-ray Photoelectron Spectoscopy-a Reference Book of Standard Data for Use in X-Ray Photoelectron Spectroscopy ed. Wagner C . D. Riggs W. M. Davis L. E. Moulder J. F. and Muilenberg G. E. Perkin-Elmer Physical Electronics Division Eden Prairie MN 1985. 24 Volynsky A. Tikhomirov S. and Elagh A. Analyst 1991 116 145. 25 Wagner C. D. J. Electron Spectrosc. 1983 32 99. 26 Snyder H. R. and Kruse C. W. J. Am. Chem. SOC. 1958,80 1942. 27 L'vov B. V. and Ryabnuke G. N. Spectrochim. Acta Part B 1982 37 673. 28 Zhuang Z.-x. Yang P.-y. Jie L. Wang X.-r. and Huang B.- l. J. Can. Spectrosc. 1991 36 9. 29 Griffith B.H. and Marsh J.D.F. Contact Catalysis Oxford University Press London 1957. Paper 1/00162K Received January 14 1991 Accepted October 29 1992

ISSN:0267-9477    

DOI:10.1039/JA9920700515

出版商:RSC    

年代:1992

数据来源: RSC

摘要:

JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY APRIL 1992 VOL. 7 52 1 Direct Determination of Impurities in Powdered Silicon Carbide by Electrothermal Atomic Absorption Spectrometry Using the Slurry Sampling Technique Bohumil Docekal* and Viliam Krivan Sektion Analytik und Hochstreinigung der Universitat Ulm Albert-Einstein-Allee 1 1 W-7900 U/m/Donau Germany A direct method of analysis of powdered silicon carbide for the determination of Al Cd Cr Cu Fe Mg Mn Ni Ti V and Zn based on electrothermal atomic absorption spectrometry (ETAAS) using the slurry sampling technique is described. Possible spectral interferences caused by the refractory matrix components were studied. The technique was optimized with regard to sample preparation dispensing thermal pre-treatment and atomization parameters.The accuracy was checked by comparison of the results with those obtained by ETAAS and inductively coupled plasma atomic emission spectrometry involving decomposition of the sample and by instrumental neutron activation analysis. For most of the elements investigated the achievable limits of detection are at the sub-microgram per gram level. Keywords Silicon carbide direct analysis; slurry sampling technique; electrothermal atomic absorption spectrometry Powdered silicon carbide has increasingly been used as a basic material for the production of advanced ceramics. The importance of trace element impurities for the quality of products and methods used for their determination was recently reviewed by Broekaert and co-workers. 1*2 Silicon carbide has been found to be a refractory and chemically resistant s~bstance.~ Its complete decomposition necessi- tates extreme chemical treatments which from a trace analytical point of view can have several disadvantageous consequences.The decomposition of silicon carbide by fusion with alkali metal ~ a l t s ~ - ~ has proved to be rapid and efficient and therefore attractive for routine analysis. However the introduction of blank and high salt contents considerably limits its application especially to samples of higher purity. Therefore decomposition with high-purity mineral acids is in general preferred. A mixture of sub- boiled concentrated acids HN03-HF-fuming H2S04 seems to be the most suitable reagent for this purpo~e.~-~ However the decomposition even when performed in a high-pressure autoclave lined with poly(tetrafluor0ethy- lene) (PTFE) is time consuming and the resulting medium is inconvenient for some methods.These dissolution difficulties make simpler analytical methods desirable. Thus in spite of some methodological limitations it was possible to develop a technique for the direct analysis of nebulized slurries of powdered silicon carbide by induc- tively coupled plasma atomic emission spectrometry (ICP- AES).7*9 Direct determination of 54 elements by instrumen- tal neutron activation analysis (INAA) has also been reported.8 It was shown by Slovak and DocekallO that a very fine powder of aluminium oxide can be analysed for trace impurities up to the parts per million level by electrother- mal atomic absorption spectrometry (ETAAS) using direct sampling of the aqueous sample suspensions.This tech- nique was subsequently not adopted in the field of analysis of ceramic materials but was taken up by us and investi- gated for its applicability to the determination of impurities in silicon carbide. However in a direct analysis of silicon carbide by ETAAS problems associated with the thermal stability of silicon carbide are expected as a result of which *Permanent address Czechoslovak Academy of Sciences Institute of Analytical Chemistry Veveri 96 CS-6 1 142 Bmo Czechoslovakia. the matrix cannot be removed by thermal pre-treatment. Consequently spectral interferences caused by the matrix and other particles formed during the atomization might be the reason for the substantial limitations of this method. Accurate calibration in this solid-sampling technique is another critical aspect.Finally depending on the homo- geneity representative sampling of silicon carbide for slurry preparation and for dispensing can also be a problem. In this paper an optimized procedure for the direct determination of impurities in powdered silicon carbide by ETAAS using slurry sampling is presented. The technique is demonstrated by the determination of 1 1 elements in silicon carbide samples analysed recently by other analyti- cal methods. Experimental Samples Commercially available silicon carbide powders were sup- plied by H. C . Starck (Goslar Germany type A 10 and B lo) Lonza (Waldshut Germany type UF 15) and Elektro- schmelzwerke (Kempten Germany type S 933); for simpler identification they are henceforth denoted I-V.After ultrasonic pre-treatment the typical average particle dia- meter and standard deviation of the grain size distribution were 0.41-0.48 pm and 0.36-0.42 pm respectively. The particle size did not exceed 5 pm. Except for the treatment in an ultrasonic bath no further pre-treatment such as grinding or fractionation was applied. Apparatus A Perkin-Elmer Model 5000 Zeeman-effect corrected atomic absorption spectrometer (inverse transversal AC system) equipped with an HGA-500 graphite furnace an AS-40 autosampler and a Model 3600 Data Station was used for the determination of Al Cd Cr Cu Fe Mg Mn Ni Ti V and Zn. In addition the applicability of a con t in u um-source compensated spectrometer for this pur- pose was tested using a Perkin-Elmer Model l 100 B atomic absorption spectrometer equipped with a deuterium-arc correction system an HGA-400 graphite furnace and an AS-40 autosampler. In order to minimize possible filtration effects by particles collected on the end of the capillary522 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY APRIL 1992 VOL.7 Table 1 Experimental conditions used for the ETAAS slurry technique Element Experimental conditions Instrumental parameters- Wavelength1 nm Spectral bandwidth/ nm Source/ mA W Temperature program me$- Drying9 Charring Temperat ure/"C Atomization temperature II Read/s Internal gas flow (argon)/ Cleaning Cooling Modifier"*/mg Linear working range?- Up to concentration ml min-I (PPm) Atomic line/nm Slurry composition (% d v ) A1 Cd Cr c u FC Mn 256.8 228.8* 357.9 324.7 305.9 279.5* 396.1 372.0* 373.7 25:2.71.248.3* 0.7 0.7 0.7 0.7 0.2 0.2 HCU EDL HCL HCL HCL HCL 25 4 25 15 30 20 320 "C ramp 10 s hold 20 s for all elements Ramp 20 s hold 20 s for all elements 1700 300 1500 1200 1400 900 2700 "C maximum power heating hold 5 s for all elements 5 1 5 5 5 5 300 300 300 100 300 300 2700 "C hold 2 s for all elements 20 "C hold 20 s for all elements 0.2 no 0.2 0.2 no no 350 - 20 10 400 10 256.8 - 357.9 324.7 305.9 279.5 (60) (3i2.0) 0.5 2.5 0.5 0.5 0.5 0.5 Mg 285.2 202.5 383.81. 0.7 HCL 4 900 5 300 no 35 285.2 0.01 Ni 352.5 232.01. 0.2 HCL 20 1400 5 300 no 30 352.5 2.5 Ti 364.3 0.2 HCL 30 1400 5 300 no 150 364.3 0.5 V 3 18.4 0.7 HCL 40 1700 5 50 no 90 3 18.4 0.5 Zn 21 3.9* 0.7 HCL 10 300 1.5 100 no 0.8 2 13.9 2.5 *Possible spectral interference see text and Table 5.?Not recommended because of spectral interference. $Hollow cathode lamp. $Internal gas flow 300 ml min-l Ar except in atomization step. ?Sampling volume 20 pl. IlRecorder and magnet and READ commands are on for 5 and 1 s respectively before starting the atomization step. **Mg(NO)3)2-6H20 Suprapur (Merck). ??Linear working range up to concentration in ppm at the specified wavelength and slurry composition assuming dispensing of 20 pl aliquots. during pipetting the common sampling capillary of the autosampler (Part No. 101 161 i.d. 0.5 mm) was replaced with another one of i.d. 0.8 mm. Pyrolytic graphite coated graphite ringed tubes with fork-shaped platformsll (see Fig.1) supplied by Ringsdorffwerke (Bonn Germany) were used. For a few comparative measure- ments ordinary electrographite (Part No. 070699) and pyrolytic graphite coated graphite (Part No. 09 1504) tubes without platforms were also utilized. The instrumental parameters used are summarized in Table 1. Procedure For the preparation of sample slurries 0.1-0.5 g of silicon carbide powder was mixed in a cleaned plastic beaker (30 ml) with 20 ml of doubly distilled water previously checked for the blank value. Suspensions were pre-treated in a Sonorex RK 255 H ultrasonic bath (Bandelin Electronic Berlin Germany) for 15 min in order to disintegrate particle agglomerates. Under continuous stirring with a PTFE-covered magnetic bar from the position usually assigned to a beaker containing the modifier solution 20 pl aliquots of the homogenized slurries were automatically dispensed into the cavity of the platform.Stirring was performed by a remote-controlled rotating magnet situated (because of steric hindrance) over the covered sample beaker. l2 A concentrated solution of Suprapur-grade mag- nesium nitrate as chemical modifier was added to the suspension and the mixture was introduced into the furnace or it was separately dispensed into the cavity of the platform via the sample volume mode of an autosampler prior to the sample suspension. For standardization by the standard additions technique the suspensions were spiked with aqueous standard solutions. For the analysis of the silicon carbide powders by ICP- AES and ETAAS the samples were decomposed in an autoclave with a mixture of concentrated HN03 HF and fuming H2S04.A more detailed description of the decom- position procedure and of the INAA used as an additional independent method can be found Scanning Electron Microscopy Scanning electron micrographs of the platform surface were obtained with a Philips SEM 500 microscope with magnifi- cation ranging from 10 to 2500 times at an accelerating voltage of 20 kV. The same microscope equipped with aJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY APRIL 1992 VOL. 7 523 Philips EDAX 9800 microprobe system was used for element-specific surface analysis of the platform. Results and Discussion Thermal Decomposition of Silicon Carbide According to the thermodynamic data,3 the total decompo- sition of silicon carbide requires thermal treatment at temperatures above 2500 "C whereby the decomposition mechanism can be affected by impurities.During the decomposition process elemental silicon and carbon are produced. Owing to its higher volatility silicon vaporizes from the mixture whereas carbon is expected to form mainly a solid residue. From the practical point of view this phenomenon can limit the maximum applicable sam- ple amount and consequently can also affect the limits of detection. In order to ensure a fast and complete decomposition and vaporization of the sample in the graphite furnace the maximum available atomization temperature of 2700 "C and maximum power heating mode were chosen in all experiments with consideration of the lifetime of graphite tube or the stabilized temperature platform furnace (STPF).Dispensing a sample portion of 0.1 mg caused no significant interference during the over-all tube lifetime reaching approximately 400-500 atomization shots of 5s duration. After this number of shots a small residue identified as the carbon matrix was observed in the centre of the platform as can be seen in the scanning electron micrograph shown in Fig. 1. Repeated introduction of 0.5 mg of sample however gave rise to considerable accumulation of carbonaceous matter on the platform which had to be removed mechani- cally after approximately 60-80 determinations. The use of full internal gas flow (300 ml min-l) during the atomization stage reduced the amount of deposited matter on the platform. Neither silicon nor other main impurity elements were detected by the EDAX microprobe in the residue remaining on the surface of the platform.The conclusion can be drawn that the decomposition of silicon carbide is complete and that trace elements trapped in the bulk are released during the atomization stage. The tube and/or platform should not be significantly corroded by interaction with silicon carbide as was manifested by the long lifetime of both components. Spectral Interferences Owing to its refractory properties the sample matrix cannot be removed during the charring treatment. Therefore considerable interferences by non-specific attenuation and/ or emission13 caused by particles and molecules produced during the atomization stage can be expected. The levels of the background attenuation shown in Fig.2 were measured over the wavelength range 200-450 nm with an atomic absorption spectrometer equipped with a deuterium-arc lamp in the background mode and by the two-line method at the following non-sensitive lines Cu 249.2 282.4 and 323.1 nm A1 307.0 nm Zn 212.5 nm Fe 358.1 and 421.6 nm and Ni 231.4 and 362.5 nm. More pronounced or even negative absorbance background values were observed if the two-line method was used. The background increased proportionately only if small sample portions up to 50 pg were injected followed by a slight increase for larger sample amounts. These differences and the shape of the non- specific spectra indicate that the attenuation is caused by molecules with fine-structured molecular band spectra.According to literature data,14J5 C2 C3 and Si2 seem to be the most probable interfering species but other molecules such as Sic2 and Si2C can also be pr~duced.~ For example the Mulliken system for a C2 molecule shows a headless Fig. 1 Fork-shaped platform with the carbonaceous residue in the central part of the cavity after 400 repeated atomizations of 0.1 mg of silicon carbide (sample IV); final magnification of 6.2 200 300 400 Wavelengthlnm Fig. 2 Spectra of non-selective attenuation measured with deuterium-arc continuum for a 2.0 nm spectral bandwidth at an argon flow rate of A 0; B 50; C 100; and D 300 ml min-' and for 0.1 mg of dispensed sample IV band with a maximum at 232.5 nm and the comet-head group of the C3 molecule appears at 404.9 nm as a red degraded band.14 Owing to the fine structure of the bands over- and under-correction errors especially for continuum- source compensated spectrometers are to be expected.In fact these interferences were observed at several lines of interest for a number of analyte elements (see Fig. 3). For example double peaks were observed for the lines of Zn at 213.9 nm and Fe at 252.7 nm and over-corrections occurred at the lines of Cd at 228.8 nm Ni at 23 1.1 nm and Ni at 232.0 nm if deuterium-arc corrected signals were measured. Second peaks (Zn Fe) and negative signals (Cd Ni) coinciding in time with the background attenuation do not occur (see Fig. 4) if other analytical lines are used in the continuum-source compensated measurement [see Figs. 3(a) and 4(a) 3(h) and 4(g) and 3(2) and 4(h)] or if Zeeman- effect background correction is applied at the same line [see Figs.3 0 and 4(e)]. The volatilization of Zn and Cd prior to the matrix enables this problem to be overcome using time-separated measurements [see Figs. 3 0 and (g)]. However this procedure is not applicable to the measure- ment of Ni and Fe at the lines given above. Therefore for these elements other analytical lines have to be chosen (see Table 1).524 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY APRIL 1992 VOL. 7 AA-BG 0.1 BG 2.01 ( C ) AA-BG 0.2 1 ( f ) I BG 0.3 A AA-BG 0.05 BG 0.3 0 AA-BG 0.2 BG 0.5 0 / .. ,--. AA-BG 0.05 ( i ) BG 0.4 t Time/s Fig. 3 Under- and over-correction errors caused by silicon carbide decomposition products. Signal traces AA - BG continuum source corrected; BG (dotted line) background attenuation; AA uncorrected; and ZAA Zeeman-effect corrected.Internal gas flow rate 300 ml min-' Ar except for Cu (50 ml min-I). Amount of sample injected 0.5 mg for Cd Ni and Zn; 0.1 mg for Ca (AA) Cu Fe and Mg; no sample for Ca (ZAA AA-BG). (a) Mg 383.3; (b) and (e) Ca 239.9; (c) and (d) Cay 422.7; U Zn 213.8; (g) Cd 228.8; (h) Fe 252.7; (i) Ni 232.0; and (J) Cu 324.7 nm A large interference (disturbed signals) was observed with and without dispensed samples (new tube) for the main Ca lines at 422.7 and 239.9 nm using both the continuum- source and the Zeeman-effect background correction sys- tems [see Fig. 3(b)-(e)]. This interference might be due to contamination of the tube and/or platform material with diffusively distributed thermally stable calcium carbide or emission13 of carbonaceous species ejected from the sur- face.This interference made the determination of Ca impossible. At internal gas flow rates of lower than 200 mi min-l of argon especially on the tailing part of the Zeeman-effect corrected signals for Cu Fe Mn Ni Ti and V an interfering effect manifested in the spike-like form of the absorbance record [see Fig. 3(j)] occurred. This inter- ference is probably caused by a discontinuous vaporization or release of carbonaceous residue and its emission. Therefore atomization with the full gas flow rate (300 nil min-') proved to be the best for the analysis of this material. Standardization and Analysis of Samples 'The direct analysis of solid samples by ETAAS requires suitable matrix-containing solid standards with known (concentrations of the trace elements of interest.However reference standard materials for silicon carbide with certified trace element contents are not available and the thermochemical aspects make a standardization using synthetic standard materials as is applied to the analysis of aluminium oxide,l* impossible. The formation of silicon carbide requires treatment of the starting substrates in an lelectric oven at very high temperatures. It can be expected that several of the spiked elements will be lost during this procedure as can also be deduced from the low content ofJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY APRIL 1992 VOL. 7 525 0 5 Time/s Fig. 4 Typical signal shapes for some elements. For absorbance specification see Fig.3. Internal gas flow rate 300 ml min-l. Amount of sample injected 0.5 mg for Cr Ni Ti V and Zn; 0.1 mg for Al Fe and Mn; and 0.025 mg for Mg. (a) Mg 202.5; (b) Mn 279.5; (c) V 31 8.4; (d) Ti 364.3; (e) Zn 213.8; CI) Al 256.8; (g) Fe 248.3; ( h ) Ni 352.5; and (i) Cr 357.9 nm the volatile elements (Cd Zn Cu etc.) generally found in this material (see Table 2). Hence the standard additions technique based on spiking the suspensions with aqueous standard solutions is considered as a very useful standardi- zation method. However this standardization technique has to be verified for accuracy.16 The results obtained by the slurry technique were compared with those of ETAAS and ICP-AES after sample decomposition and with those of INAA. In general an accurate standardization requires that the analyte element added as a standard to the sample and the indigenous analyte element in the sample behave similarly during the charring and atomization stages.According to this stipulation a ‘generalized standard additions method’ reported recently” could not be applied to this slurry technique. In this work the trace impurities might be trapped in the sample bulk they might be physically sorbed or chemically bound as a result of the production processes applied (smelting zone melting grinding) on the surface of the particles or they might be present as discrete particles disseminated throughout the silicon carbide powder. Hence the kinetics and efficiency of vaporization and atomization are in general influenced by the physical and chemical form of the analyte element which consequently affects the analyte signal.An equalization of the behaviour of different forms of the analyte element can be achieved by a suitable matrix-analyte modification. Preliminary experiments using wall atomization from ordinary and pyrolytic graphite coated graphite tubes showed large discrepancies between the expected and found contents even if the signals were evaluated from integrated absorbances (peak areas). These experiences and the above considerations suggested the application of the STPF concept to this problem. Therefore in subsequent experi- ments a fork-shaped platform in a ringed tube and integrated absorbance (s) evaluation were used. Suspensions of powdered silicon carbide were prepared in water in order to achieve hydrolysis of the spiked analytes followed by retention of the hydrolysis products on the very large particle surface (approximate specific surface area 15 m2 g-l).The sorption of the analyte elements in the added form should eliminate the physical separation of the spiked analyte present originally in the aqueous phase from the sample particles by being sucked into the graphite or by drying processes during the first stage of the temperature programme. In special experiments the leaching behaviour of the elements and their distribution between the two phases of the suspension were studied. For this purpose after ultrasonication the phases were separated and the liquid phase was analysed. The results of these experiments showed that neither the analytes contained in the samples nor the analyte spikes were significantly leached from the silicon carbide particles into the solution.The modifiers used for the equalization of the behaviour526 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY APRIL 1992 VOL. 7 Table 2 Element contents determined in different silicon carbide samples by the slurry technique (A) and comparison with results obtained by ETAAS or ICP-AES after sample decomposition (B) and INAA (C) Contentlpg g-l Element A1 Cd Cr c u Fe Mg Mn Ni Ti V Zn Method* A B A B C A B C A B A B C A B A B C A B C A B A B A B C I UF 15 Lonza 251 k 14 267 f 3 (0.02 0.23 k 0.04 (0.25 3.1 k0.3 3.0k0.1 2.6 k 0.2 3.4 k 0.3 5.3 k 0.4 260k 10 251 k 4 239 f 5 30.8 k0.5 33.6 k 0.6 6.9 k 0.6 7.0 k 0.2 7.9 k 0.2 21.4f0.8 24kO.l 2 1.7 k 0.3 10723 102 +2 34k 1 36+1 0.68 k 0.02 0.8+0.15 <0.5 I1 B 10 Starck 338k 15 347 k 2 (0.02 (0.07 <2 18.5 k 1.5 18.2 k 0.2 15.1 k 1.0 8.9 k 0.4 8.0 2 0.8 650 k 30 656 k 16 646 k 39 17.6k 1.0 12.6 k 0.3 3.7 2 0.2 3.8 k 0.2 3.9 k0.2 12.4 + 0.5 12.9 k 8.1 13.0 2 0.3 82+ 1 82k2 29k 1 26k1 0.12 k 0.02 (0.15 (0.5 I11 A 10 Starck 480 k 40 461 k 2 (0.02 (0.07 (0.1 2.6 k 0.3 2.4 k 0.1 1.9 k 0.15 1.920.1 15.3 + 0.2 2.3k0.1 17.0 k 0.7 16.1 k 1.2 5.6 f 0.5 5.0kO.l 0.14 k 0.08 -0.2 0.16 k 0.01 3.0 k 0.4 -2 0.81 k 0.04 6 1 f l 65+ 1 29a2 282 1 0.43 k 0.04 0.11 kO.01 (0.1 IV A 10 Starck 224-1 15 198k5 (0.02 n.d:t (0.2 5.5 f 0.3 5.6 k 0.2 4.4 k 0.8 1.3k0.1 - 1 150k 10 134k 1 13222 4.0kl.l 1.7k0.1 0.78 k 0.06 0.69 k 0.02 0.74 k 0.0 1 4.6 k 0.5 w2 1.6k0.2 40k3 41 2 2 9 k 2 9.6 k 3.3 0.18 k 0.02 n.d.(0.6 V ESK S 933 177 & 19 178 k 4 <0.02 <0.5 (0.09 7.1 k0.5 7.1 k 0.2 6.9 k 0.5 3.2 2 0.6 2.9 & 0.3 320k 1 322 k 14 5.2 +_ 0.8 3.7 k 0.2 0.72 f 0.08 0.73 f 0.14 4.8 2 0.7 (3 3.8 k 0.2 340 k 20 0.70k0.01 130k 13 158k 1 8 4 f 1 84k2 0.31 k0.03 n.d. (0.5 *A ETAAS using slurry sampling technique; n= 5. B ETAAS or ICP-AES after high-pressure decomposition of the sample (refs. 8 and 9); 7n.d. =Not determined. n=6. C INAA (ref. 8). of all analyte forms should react with silicon carbide at sufficiently low temperatures. Palladium and magnesium nitrate seemed to be well suited for this p~rpose,~ when applied in sufficiently large amounts. For example 0.6 mg of magnesium nitrate is needed for complete decomposi- tion of 0.1 mg of silicon carbide.Nevertheless utilization of larger amounts of these modifiers is mainly limited by their purity (affecting the blank) and by their chemical interac- tion with graphite. Proceeding from the relatively well known properties of magnesium nitrate as a universal modifier,'* the modifying efficiency for this matrix was studied in more detail. Although silicon carbide cannot be removed during the charring stage the determination of a suitable charring temperature plays a decisive role in the standardization. Thermal treatment at too high temperatures can give rise to losses of the spiked analytes from the unmodified suspen- sion and consequently higher results would be obtained. On the other hand at too low pre-treatment temperatures giving rise to no loss of the analyte the reaction of the modifier with the matrix might be insufficient or it might not take place at all.In this instance there would be a risk of obtaining lower results owing to easier atomization of the spiked analyte. The optimum experimental conditions for the determina- tion of the main impurity elements Al Cr Fe Mn Ti and V the content of which in various samples was verified by other methods are summarized in Table 1. The contents determined in silicon carbide powders of different origin by this slurry technique and the other independent methods V Ti 1 100 200 300 400 Contentfpg g-' Fig. 5 Standardization by using samples analysed after decompo- sition; 0.1 mg of sample introduced. Samples 0 I; M 11; 0 111; 0 IV; and A V are compared in Table 2.Surprisingly for most of the elements investigated no addition of modification agents was necessary. This is obviously due to the utilization of the STPF concept and the evaluation of integrated absorbance data and probably also to the convenient form of the analyte elements in the sample. Excluding some few instances where the analyte element contents were low the results obtained by the slurry technique are in good agreement with those obtained by INAA and by ICP-AES or ETAAS after sample decomposi- tion.JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRYy APRIL 1992 VOL. 7 527 Table 3 Reproducibility of sampling and dispensing procedure in the determination of Cr in silicon carbide (sample IV) Sampling Relative standard deviation* (Oh) sequence No.Sampling Dispensing 3.3 2.2 3.8 3.4 4.0 3.9 Average 6.4 3.4 *Based on dispensing 20 pl aliquots of 0.5% m/v sample suspensions and using integrated absorbance evaluation; n = 6. Table 4 Reproducibility of dispensing in the determination of Fe for various silicon carbide samples Sample Relative standard Concentration No. deviation* (%) level/pg g-I I 3.7 250 I1 5.1 650 IV 8.3 130 V 6.8 320 *See Table 3. The wide range of the element contents in the silicon carbide samples analysed by other methods permits the use of these materials for standardization purposes. An example is shown in Fig. 5 in which the integrated absorbances of Al Fe Ti and V measured for 0.1 mg of each of the five samples are plotted against the content deter- mined by ETAAS after decomposition of the sample.Another possible approach is to apply various amounts of one well characterized silicon carbide material either by dispensing suspensions of various concentrations or by repeating the introduction and drying step. If solid silicon carbide standards are used the analysis can be performed also by the wall atomization technique. In this manner tailing (memory) effects might be decreased for carbide- forming elements and integrated absorbance can be used for the evaluation. Sampling Errors Sampling errors with the slurry technique are generally associated with variance of the analyte mass sampled resulting from the variances of the particle mass number of particles present in the volume injected concentration of the analyte element (homogeneity) and volume pipetted.l9 Ceramic powders obtained by grinding can be expected to have good homogeneity. Therefore sample portions of 0.1 g of silicon carbide for analyses should normally be sufficient for the preparation of the suspension. However the dispensing of relatively small volumes of slurries might cause considerable fluctuations of signals owing to the irregular distribution of particles and impurities as discussed under Standardization and Analysis of Samples. From this point of view suspensions as concentrated as possible should be introduced into the atomizer. However the maximum dispensable amount of silicon carbide is limited by the accumulation of carbon-aceous residue on the platform and by spectral interferences as mentioned above. The standard deviation of the sampling procedure including the preparation of the slurry and its dispensing with the autosampler was determined for some elements by analysing 0.5 and 2.5% m/v (0.01% m/v for Mg) aqueous suspensions. Results for the reproducibility for Cr at the 5 ppm level in silicon carbide (sample IV) and for Fe in various samples are given in Tables 3 and 4 respectively. It can be seen that good reproducibilities for both sampling and dispensing are achieved for 20 pl aliquots of 0.5% m/v sample slurry (for an average number of particles analysed of about 1 x lo9).These reproducibilities are similar to those usually found in the analysis of homogeneous solu- tions. In addition they proved that the stirring technique with the remote-controlled magnetic bar was sufficient for homogenization of the suspensions also with respect to the dispensing procedure used being superior to dispensing by hand.The good agreement of the results obtained directly by the slurry technique and by other methods (see Table 2) Table 5 Detection limits achieved in the analysis of silicon carbide powders by ETAAS using the slurry sampling technique Element A1 Ca Cd Cr c u Fe Mg Mn Ni Ti V Zn Wavelength/ nm 256.8 422.7 228.8 357.9 324.7 372.0 248.3 285.2 279.5 352.5 364.3 3 18.4 2 13.9 Internal gas flow1 ml min-I 300 300 50 50 0 300 100 300 50 50 100 50 0 Detection limit*/ Pg g-' - - 0.02 0.2 0.05 2 0.4 0.02 0.8 5 2 0.01 - Comment sf Not determined owing to high content Impossible to determine Determined for a spiked slurry m V S S Not determined owing to high content m m m S V *Based on three times the standard deviation of the blank fluctuation evaluated from integrated absorbance applying 20 pl aliquots of a ?Detection limit determined by my memory effect; v maximum vaporizable amount of sample; and s spectral interference high 2.5% m/v suspension of a silicon carbide sample with low trace element contents.background level.528 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY APRIL 1992 VOL. 7 is evidence for a high degree of homogeneity of the samples and it also proves that no significant separation processes take place during the slurry agitation or dispensing (e.g. sedimentation or filtration effects). Detection Limits In most practical instances the detection limits are deter- mined by the fluctuation of the blank values.In our experiments chemical contamination could be minimized by the proposed batch technique. The laboratory ware (beakers and stirrer bars) could be cleaned effectively. Water of extremely high purity was used for the preparation of suspensions and modifier solutions. A further advantage of this technique is that the actual blank level can be easily controlled by analysis of the dispensing medium prior to the addition of the sample. If owing to fluctuations too large blanks are determined a new dispensing liquid can be prepared and checked again. The main factors determining the limits of detection include memory effects especially for carbide-forming elements spectral interferences fluctuation of the baseline and the maximum applicable sample amount.As can be seen from Table 5 the detection limits for many elements are at the sub-microgram per gram level. Conclusions In spite of its refractory properties powdered silicon carbide can be directly analysed by ETAAS for Al Cd Cr Cu Fe Mg Mn Ni Ti V and Zn using the slurry sampling technique. In this technique the risk of introducing contaminants is much lower and can be controlled and reduced more easily than in solution ETAAS. Thus detection limits being at the sub-microgram per gram level are determined not by the blank but by spectral interfer- ences and by the maximum vaporizable sample amount dispensed for atomization. Using the STPF concept and if necessary suitable modification standardization can be performed by spiking the slurries with aqueous standard solutions.The method developed is adequate for routine analysis. We thank B. Hutsch Ringsdorffwerke (Bonn-Bad Godes- berg Germany) for providing the new graphite fork- designed platforms before they were commercially avail- alble W. H. Fritz Sektion Elektronenmikroskopie Uni- versity of Ulm for scanning the micrographs and M. Franek Sektion Analytik und Hochstreinigung University of‘ Ulm for making available the INAA results. This work was supported by financial assistance from the Deutsche Forschungsmeinschaft Bonn-Bad Godesberg. References I Broekaert J. A. C. Graule T. Jenett H. Tolg G. and Tschopel P. Fresenius’ 2. Anal. Chem. 1989 332 825. 2 Broekaert J. A. C. and Tolg G. Mikrochim. Acta 1990 11 173. 3 Gmelins’ Handbook of Inorganic Chemistry System No.15 Springer Berlin Silicium Part B 1959 p. 816; Silicon Carbide suppl. vol. B2 Part 1 1984 pp. 105-107 163 2 17; suppl. vol. B3 Part 2 1986 pp. 1 163 371-381. 4 Harada Y. Kurata N. and Furuno G. Bunseki Kagaku 1987 36 526. 5 Dornemann A. Kolten K.-H. and Rudan D. Fresenius’ Z. Anal. Chem. 1987 326 232. 6 Xu Z. Lihua Jianyan Huaxue Fence 1987 23 43. 7 Graule T. van Bohlen A. Broekaert J. A. C. Grallath E. Klockenkamper R. Tschopel P. and Tolg G. Fresenius’ 2. Anal. Chem. 1989 335 637. 8 Franek M. and Kvan V. Fresenius’ J. Anal. Chem. 1992 342 118. 9 Docekal B. Broekaert J. A. C. Graule T. Tschopel P. and Tolg G. Fresenius’ Z. Anal. Chem. 1992 342 1 13. 110 Slovak Z. and Docekal B. Anal. Chim. Acta 1981 129 263. !I I Shuttler I. L. Schlemmer G. Carnrick G. R. and Slavin W. Spectrochim. Acta Part B 1991 46 583. 12 Slovak Z. and Docekal B. paper presented at the 5th Czechoslovak Conference on Atomic Spectrometry Nitra September 1980. 13 De Loos-Vollebregt M. T. C. and van Ochten P. J. J. Anal. At. Spectrom. 1990 5 183. 14 Pearce W. B. and Gaydon A. G. The Identification of Molecular Spectra Individual Bands Chapman and Hall London 1965 pp. 97 and 277. 15 Ohlsson K. E. A. and Frech W. Spectrochim. Ada Part B 1991 46 559. 16 Welz B. Fresenius’ Z. Anal. Chem. 1986 325 95. 17 Baxter D. C. J. Anal. At. Spectrom. 1989 4 415. 18 Docekalova H. Docekal B. Komarek J. and Novotny I. J. Anal. At. Spectrom. 1991 6 661. 19 Holcombe J. A. and Majidi V. J. Anal. At. Spectrom. 1989 4 423. Paper I /05 1 701 Received October I I 1991 Accepted January 13 1992

ISSN:0267-9477    

DOI:10.1039/JA9920700521

出版商:RSC    

年代:1992

数据来源: RSC

摘要:

JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY APRIL 1992 VOL. 7 529 Slurry Procedure for the Determination of Titanium in Plant Materials Using Electrothermal Atomic Absorption Spectrometry Ignacio Lopez Garcia Pilaf Vi Aas and Manuel Hernandez Cordoba* Department of Analytical Chemistry Faculty of Chemistry University of Murcia 30071 -Murcia Spain The use of slurries to determine the titanium content of vegetables by electrothermal atomic absorption spectrometry is discussed. As slurries from ground plant material produce serious deterioration in the pyrolytic graphite coating of the tube the samples are first submitted to a process of mild calcination. Suspensions are prepared in the 0.1 -1 0% range in water containing 0.03% hexametaphosphate. Calibration can be performed with aqueous standards although the standard additions method is recommended. The detection limit for titanium using 5% suspensions is 0.03 pg g-I and the relative standard deviation is a4.670 for 1.3 ng of the element introduced into the furnace.Data for the titanium content of lettuce spinach pea and commercially available paprika samples are in good agreement with those obtained by an acid dissolution procedure. Keywords Electrothermal atomization atomic absorption spectrometry; solid sample; slurry; titanium; plant material The use of slurries as a method of introducing solid samples into an electrothermal atomizer seems attractive and has been the object of much research. A recent review' on the use of solid samples in electrothermal atomic absorption spectrometry (ETAAS) reflected this interest and gave details of slurry procedures for more than 30 determinands in a wide range of materials.The slurry-ETAAS approach is not of general application because an essential condition i.e. a sufficiently low particle size must be fulfilled although in many instances a simple treatment of the sample can produce this conditi~n.~-~ In this paper the determination of titanium in vegetables using the slurry-ETAAS approach is discussed. The litera- ture on the determination of titanium using ETAAS is scarce and the references are mainly related to the need for using good quality coatings in the atomizer.6 As far as we know there are no reports dealing with the determination of titanium using slurry-ETAAS procedures. As titanium has been to have beneficial effects on yield and quality of fruit in several crops a simple fast procedure as is discussed here can be of practical use.Experimental Apparatus A Perkin-Elmer Model 1 lOOB atomic absorption spectro- meter with deuterium-arc background correction and an HGA-400 electrothermal atomizer were used. In an effort to obtain more reliable background correction the mea- surements were performed at the 320.0 nm line of titanium instead of the more usual 364.3 nm line. The hollow cathode lamp was operated at 30 mA and a spectrometer bandwidth of 0.2 nm was used. Previous experiments showed that the best repeatability was obtained when the peak height instead of integrated absorbance was measured. For this reason background-corrected peak heights were used as the analytical signal.Commercially available pyrolytic graphite coated graphite tubes (Perkin-Elmer) were used. Reagents A stock solution of titanium (1 000 mg dm-3) was obtained from Fisher Scientific. Working solutions were prepared daily by appropriate dilution. Hexametaphosphate (HMP) was obtained from Fluka Chemie and used without further purification. *To whom correspondence should be addressed. Table 1 Recommended furnace programme purge gas argon; and flow rate 300 cm3 min-' (stopped during atomization steps). All temperatures quoted are values set on the HGA-400 power supply Stage TemperaturePC Ramp time/s Hold time/s Dry 150 10 10 Ashing 1200 5 15 Cooling 200 0 15 Atomize 2700 0 3 Clean 2700 1 4 Cooling 200 0 15 Atomize 2700 0 2 Clean 2700 1 4 Procedures Commercially available paprika samples were used as received.Lettuce spinach and pea samples were washed with doubly distilled water and then oven dried at 105 "C to a constant mass. Special care was taken to prevent contami- nation of the samples during the long period of time needed to carry out this step. The samples were then ground using a domestic coffee mill and stored in polyethylene bags until the determinations were carried out. Slurry procedure A 10 g portion of dried sample was weighed and ashed at 350 "C in a porcelain crucible for 1 h. The ashes were weighed ground in an agate ball mill for 15 min and sieved with a 325 mesh sieve. The small fraction which did not pass through the sieve was discarded. Amounts of sieved sample in the range 0.025-2.5 g were taken and 25 cm3 of water containing 0.03% HMP added.The suspension was stirred magnetically for a minimum of 10 min. While the slurry was being stirred 25 mm3 aliquots were taken and injected into the electrothermal atomizer. The furnace programme given in Table 1 was run and the corrected peak height values of signals at the first atomization step were obtained. Calibration graphs for up to 10 ng of titanium were prepared using aqueous standards containing 0.03% HMP or the standard additions calibration method was applied. Acid dissolution procedure For comparative purposes the procedure recommended by Anton et ~ 1 . ~ was followed. A 1 g portion of sample was ashed at 600 "C until the ash appeared white (about 24 h).5 30 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY APRIL 1992 VOL.7 The ash was dissolved using 5 cm3 of a nitric acid solution (6 mol dm-3) heated at 90 "C for 60 min and finally the volume made up to 25 can3. Aliquots (25 mm3) of this solution were taken and the peak height due to titanium was measured using a temperature of 1100 "C for ashing (8 s) and 2700 "C for atomizing (3 s). Calibration was performed using the standard additions calibration method. Results and Discussion It has been reported1°-14 that the use of a cool-down step before the atomization step can improve the analytical performance of ETAAS procedures. With this in mind several preliminary experiments were carried out using pyrolytic graphite coated graphite tubes and the furnace programme shown in Table 1 was selected as the most suitable.Because the determination of titanium by ETAAS poses a problem due to the low volatility of the element the recommended programme includes two cool-down steps each followed by a firing obtained by applying the maxi- mum heating rate in the atomizer. The peak height of the signal obtained at the first firing of each run is used to quantify titanium. The second firing is important because it provides information on the amount of titanium retained in the atomizer and thus on the alteration of the pyrolytic coating. A number of experiments were devoted to studying the possible effect of cross-contamination due to the retention of titanium in the tube and the ageing of the tube. For this slurries of calcined paprika samples prepared as indicated under Experimental were used.A 25 mm3 aliquot of a 0.2Oh slurry from a sample containing 21.8 pg g-l of titanium were injected into a new tube and the programme was run repeatedly. A similar set of experiments were performed on another new tube using a different slurry prepared from a sample with a lower content of titanium. As can be seen from Fig. 1 where the results are summarized in both cases the signal obtained from the second firing was about 6% of that obtained from the first. It is important to note that when using the recommended programme the signal from the third firing (this would be the first atomization step of a new sample) was reduced to about 1.7 or 1.3% of that obtained in the first atomization of 9.05 or 2.8 ng of titanium respectively.It was concluded that under the recommended experimental conditions the effect of cross- contamination between the samples to be analysed was reduced to tolerable levels. The ashing temperature was studied both for aqueous standards and for slurries. As can be seen in Fig. 2 where the peak heights obtained in the first atomization step have been normalized for purposes of comparison the peak height was dependent on the volume introduced into the furnace. Thus for 10 mm3 of an aqueous titanium solution the maximum ashing temperature to be used was about 1600 "C while when 25 mm3 were used the ashing could be performed at 1200 "C without any loss of determinand. This effect was attributed to sputtering due to the short ramp time used in reaching the ashing temperature.A similar effect was noted for the suspensions although higher temperatures could be used. Finally in order to decrease the time necessary to carry out the furnace programme a temperature of 1200 "C was chosen for the ashing of the slurries because the peak height of the background signal at the maximum signal from titanium was low (0.0 12-0.040) depending on the slurry concentra- tion and the use of higher ashing temperatures did not produce significant decreases in the background. The slurry-ETAAS approach requires a low particle size and so a grinding step is necessary. The use of slurries prepared directly from the ground plant material proved to be inadequate because of a rapid deterioration of the pyrolytic coating which led to a decrease in sensitivity and an increase in the background.The height of the peak due to titanium when successive injections of 25 mm3 of a 10% lettuce slurry were analysed is shown in Fig. 3. As is illustrated by curve A there was a near continuous decrease in the signal when the number of injections increased. Simultaneously the peak height measured at the second 1 I I I I 900 1200 1500 1800 2100 2400 Ashing temperature/ 4 Fig. 2 Effect of ashing temperature on the relative signal for A and B 10 and 25 mm3 respectively of an aqueous titanium solution; and C and D 10 and 25 mm 3 respectively of a 0.2% slurry prepared from calcined paprika 5 0.06 .- Q1 .c A L 0.03 1 2 3 4 5 No. of firings Fig. 1 Variation of the peak height when consecutive runs were performed after the injection of A 25 mm3 of a 0.2% slurry prepared from calcined paprika; and B the same volume of a 4% slurry prepared from another calcined sample with low content of titanium 0 15 30 45 60 75 No.of firings Fig. 3 Effect of consecutive injections on peak height A and B signals at the first and second firing respectively obtained for a 10% lettuce slurry without prior calcination; and C and D as above using a 3% slurry prepared from the same lettuce sample with prior calcinationJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY APRIL 1992 VOL. 7 53 1 Titaniumhg per 25 mm3 0 10.0 22.5 0.26 2.6 Titaniumhg per 10 mm3 0 5.0 9.0 0.1 1 .o l o t \ 8 2t I 8 1 // I I 0 0.2 0.4 2 6 10 Slurry (%I Fig. 4 Relative standard deviations for ten injections of A 10 and B 25 mm3 of a slurry prepared from calcined paprika; E 10 and F 25 mm3 of a slurry prepared from calcined pea; C and G 10 and D and H 25 mm3 injections of aqueous titanium.Numbers on the upper axes indicate the amount of titanium introduced into the tube as a slurry or as an aqueous solution firing of each run increased (curve B) revealing greater absorption of titanium into the graphite tube owing to the damage caused to the coating. Furthermore several stan- dard additions graphs were obtained from slurries prepared with different percentages of lettuce and the slopes of these graphs were dependent on the percentage. In an effort to overcome the difficulties described above the samples were submitted to a mild calcination step in a muffle furnace and the slurries prepared from the ground sieved ashes as indicated under Experimental.As can be seen in curve C of Fig. 3 when this simple approach was followed the peak height at the first firing of a 3.0% slurry prepared from lettuce remained nearly constant for 60 injections and a decrease of only about 8% in the signal was noted after 75 injections. Furthermore the signal from the second firing of each run (curve D) revealed no significant damage to the coating of the tube. It has already been reported that this simple approach based on a gentle calcination overcomes problems caused by the physical characteristics of the ~ a r n p l e ~ ~ ~ and makes diminution of the particle size possible. In a study of the effect of the different percentages of sample in the slurries several suspensions in the 0.02-0.48% range were prepared from a commercially available paprika sample with a high amount of titanium and other suspensions in the 1-10% range were prepared from peas with a low titanium content.Both 10 and 25 mm3 were used as the injection volumes in the recommended programme. The results indicated a linear relationship (values of the correlation coefficient r were in the 0.9989-0.9999 range) between the peak height obtained at the first firing of each run and the percentage of sample in the slurry for the range studied. An additional study was made using these suspensions to compare the repeatability of the measurements as a function of the mass of titanium introduced into the furnace with the repeatability shown Table 2 Titanium content found in several plant materials.All the values were obtained using the standard additions method (four points) for four replicates Sample Paprika It Paprika 2 Paprika 3 Ground red pepper 1 Ground red pepper 2 Ground red pepper 3 Pea Spinach Lettuce Mean value ( & SD)/,ug g-I Proposed Acid digestion method and ETAAS* 21.8k0.9 0.33 k 0.02 16.4 k 0.8 12.3k0.6 0.44 k 0.02 0.41 k 0.03 0.32 k 0.02 0.46 +_ 0.03 0.72 2 0.03 21.7k 1.3 16.22 1.0 12.1 k0.9 0.35 k 0.05 0.43 k 0.04 0.29 k 0.02 0.74k0.04 0.41 k 0.03 0.47 + 0.03 *Ref. 9. tAll paprika samples are commercially available. shown by aqueous standards under the same experimental conditions. The results are summarized in Fig. 4. As could be expected the relative standard deviation (RSD) ob- tained for slurries was higher than that found for aqueous standards containing the same mass of titanium.The high RSD found for very dilute slurries should be noted. This is attributable both to the low mass of titanium introduced into the tube and to the error involved in sampling slurries. Calibration Under the recommended experimental conditions a cali- bration graph obtained using aqueous standards of titanium gave a slope of 0.0685-+0.0005 absorbance per ng and a value for the characteristic mass of 63 pg. To prove the absence of any matrix effect four slurries were prepared from calcined lettuce covering the 2-10% range and standard additions calibration graphs were obtained. The slopes of these graphs were in the 0.067-0.069 range proving that direct calibration with aqueous standards is valid.The decrease in sensitivity due to damage of the pyrolytic coating does not appear to be excessive because after the recommended programme was run about 200 times a calibration graph for aqueous titanium was again obtained and the slope was 0.059 -+ 0.001 (characteristic mass 74 pg). This small effect due to the ageing of the coating can easily be overcome if analytical results for titanium are obtained using the standard additions method. The detection limit (20) for titanium when using 25 mm3 of a 5% slurry prepared from mildly calcined lettuce was calculated to be 0.03 ,ug g-l. The results obtained for the determination of titanium in several plant materials using the proposed procedure and another procedure reported in the literature9 are presented in Table 2.The data show that there was agreement between the two procedures and that the precision attained with the slurry approach was similar to or even better than that obtained with the procedure used as a reference. Note that the titanium content of the commercially available paprika samples analysed was about 20-40 times higher than those found in lettuce spinach and peas. In order to ascertain whether such a high level of titanium was due to an unusual accumulation of the element in red peppers (Capsicum annuum) from which commercially available paprika is obtained three different batches of red pepper fruits were obtained. The fruits were carefully washed dried and ground in the laboratory and the titanium content was determined. As can be seen in the Table 2 the results were very similar to those found in other532 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY APRIL 1992 VOL.7 plant materials. The level for the commercially available samples analysed appears to be a consequence of the industrial treatment of the original plant material. Financial support from the Spanish Direccion General de Investigacion Cientifica y Tecnica (Project 90-0302) and from Consejeria de Cultura Education y Turismo de la Comunidad Autonoma de la Region de Murcia Spain (Project PCT90127) is gratefully acknowledged. References Bendicho C. and de Loos-Vollebregt M. T. C. J. Anal. At. Spectrom. 1991 6 353. Ebdon L. Fisher A. S. Parry H. G. M. and Brown A. A. J. Anal. At. Spectrom. 1990 5 321. Fagioli F. Landi S. Locatelli C. and Bighi C. At. Spectrosc. 1986 7 49. Bendicho C. and de Loos-Vollebregt M. T. C. Spectrochim. Acta Part B 1990 45 695. Hernandez Cbrdoba M. and L6pez Garcia I. Talanta 1991 38 1247. 4 '7 8 9 10 1 1 12 13 14 15 Slavin W. Manning D. C. and Carnrick G. R. Anal. Chem. 1981,53 1504. Pais I. J. Plant Nutr. 1983 6 3. Dumon J. C. and Ernst W. H. O. J. Plant Physiol. 1988,133 203. Antbn E. Jara J. R. Martinez F. GimCnez J. L. and Alcaraz C. F. 111 Symposium Nacional sobre Nutricidn Mineral de las Plantas ed. Servei de Publicacions i Intercanvi Cientific de la U.I.B. Palma de Mallorca Spain 1990 p. 109. Manning D. C. and Slavin W. Spectrochim. Acta Part B 1985 40 461. Falk H. and Glismann A. Fresnius' 2. Anal. Chem. 1986 323 748. Hinds M. W. Katyal M. and Jackson K. W. J. Anal. At. Spectrom. 1988 3 83. Shuttler I. L. Delves H. T. and Hutsch B. J. Anal. At. Spectrom. 1989 4 137. Lynch S. and Littlejohn D. J. Anal. At. Spectrom. 1989 4 157. Holcombe J. A. and Majidi V. J. Anal. At. Spectrom. 1989 4 423. Paper I /058 74F Received November 19 1991 Accepted January 20 1992

ISSN:0267-9477    

DOI:10.1039/JA9920700529

出版商:RSC    

年代:1992

数据来源: RSC

摘要:

JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY APRIL 1992 VOL. 7 533 Comparison of Sodium Chloride and Magnesium Chloride Interferences in Continuum Source Atomic Absorption Spectrometry With Wall Platform and Probe Electrothermal Atomization John Carroll,* Nancy J. Miller-lhli and James M. Harnly US Department of Agriculture Nutrient Composition Laboratory BA RC- East Beltsville MD 20705 USA Thomas C. O'Haver Department of Chemistry University of Maryland College Park MD 20742 USA David Littlejohn? Department of Pure and Applied Chemistry University of Strathclyde Cathedral Street Glasgow G 7 IXL UK Various masses between 10 and 500 pg of NaCl or MgCI were added to 1 ng of Cr Cu and Mn 2 ng of Cd and Pb and 4 ng of Co Mo Ni and V to study the interferences encountered when wall platform and probe atomization are applied in continuum source electrothermal atomic absorption spectrometry.No char step was used and a compromise atomization temperature of 2700 "C was selected for simultaneous multi-element measurements. No great difference was observed in the analyte signal recovery values obtained with the three atomization modes when NaCl was the interferent. However with MgCI greater interferences were observed with wall atomization. Overall probe atomization proved as good as if not better than platform atomization for the elements considered. With both procedures freedom from NaCl or MgClp interferences was achieved at chloride sa1t:analyte mole ratios of 1 x 1 03-l x 1 05. Although it was not possible to make a definitive assessment of the procedures responsible for the interferences observed there was some evidence that vapour-phase chemical effects are more important for MgCI than for NaCI occlusion of Cd Mn and Pb occurs in NaCl and expulsion of Co Cr Cu Mn and Ni occurs due to co-vaporization with NaCI.Keywords Continuum source atomic absorption spectrometry; electrothermal atomization; simultaneous multi- element measurement; sodium chloride and magnesium chloride interference effect The emergence of continuum source atomic absorption spectrometry (CSAAS) has for the first time enabled the analytical sensitivity afforded by an electrothermal atom- izer to be exploited on a truly simultaneous multi-element The simultaneous multi-element AAS with con- tinuum source (SIMAAC) instrument developed by Harnly et aZ.,4 employed a 300 W xenon-arc continuum lamp as the source with a high resolution echelle polychromator for spectral detection.Computer controlled wavelength modu- lation allowed the acquisition and calculation of back- ground corrected absorbances on up to 16 channels. The detection limits obtained were comparable to those of conventional line source AAS for elements with resonance lines above 280 nm. Below this wavelength however detection limits were poorer owing to the reduction in intensity of the continuum source and the lower spectral efficiency of the echelle spectrometer in this wavelength region. With electrothermal SIMAAC a restriction is normally placed on the choice of char and atomization temperatures that can be applied especially when elements of widely differing volatility are determined.In particular if compar- atively volatile elements are present in the sample the maximum char temperature may be restricted to a value that for many matrices does not allow the bulk of the sample to be removed prior to the atomization step. The use of chemical modifier^^^^ can improve the situation but care must be taken to avoid the occurrence of spectral overlap interferences by high concentrations of magne- sium,' added in the form of Mg(N03)z and the introduction of high blank contamination levels of some elements. *On study leave from the University of Strathclyde. Present address ICI Wilton Materials Research Centre P.O. Box 90 Wilton Middlesborough Cleveland TS6 8JE UK. ?To whom correspondence should be addressed. Thus under the conditions required for simultaneous multi-element analysis the residual matrix levels present during the atomization step may be significantly higher than would normally be observed under optimized condi- tions for single element determinations.Consequently any electrothermal atomization procedure employed for this purpose should be capable of limiting matrix interference effects in addition to providing satisfactory sensitivity for all the elements to be determined. Recent developments in electrothermal atomization tech- nology have centred on procedures which establish high temperature conditions in the atomizer prior to the release of the analyte into the vapour phase. This leads to improved dissociation of any molecules containing the analyte and reduces interferences caused by interaction with matrix salts (e.g.chlorides). Procedures such as p l a t f ~ r m ~ ~ ~ probelOJ and t w o - ~ t e p ~ ~ J ~ atomization have been shown to be effective in reducing matrix interferences. Harnly and Kanel have shown that for wall and platform atomization in CSAAS the atomization temperature must have a high bias in favour of efficient atomization of the more refrac- tory elements. Under such conditions release of volatile and medium volatile elements may occur before the atomizer temperature has stabilized and this can reduce the effectiveness of the platform procedure in minimizing chemical interferences. The problem can be avoided if the sample is introduced into the atomizer after constant temperature conditions have been established.This can be achieved with the two-step constant temperature atom- izerl2J3 and probe atomization.lOJ1 Lundberg et al. l 4 re- ported that it was difficult to select compromise atomiza- tion conditions for a two-step atomizer which gave maximum sensitivity in CSAAS for elements of different volatility. However lower background absorption levels carry-over contamination and halide interferences were obtained compared with conventional atomization proce- dures. At present the two-step atomizer seems to be the534 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY APRIL 1992 VOL. 7 most attractive system for multi-element analysis by elec- trothermal CSAAS (ET-CSAAS) although the need for a special atomizer and two power supplies might limit its general use.Previous studies using probe atomizati~n~~ for simulta- neous multi-element analysis by CSAAS have shown that at the compromise atomization temperature selected (2 700 "C) similar detection limits (within a factor of four) were achieved for 12 elements with wall and probe atomization. However the probe gave better detection limits than the platform for eight of the elements studied and the performance was similar for the four other analytes. It was suggested that in comparison with the other procedures used with a conventional atomizer design probe atomiza- tion might offer some advantages in CSAAS over tube wall and platform atomization as it gives satisfactory sensitivity and might offer better control of chemical interferences.No evidence was given to substantiate the latter point so in the present study the performance of wall platform and probe atomization has been compared with respect to the extent of chemical interferences on nine elements caused by NaCl and MgCl? added at concentrations of up to 5% m/v. The same compromise conditions were applied for each mode of atomization and the nine elements were measured simultaneously using the SIMAAC system. No char step was used in the furnace programme so the mass of interferent present during atomization was the same for all elements and with each procedure. Several different processes can cause a decrease in the magnitude of the signal when the analyte elements are vaporized in the presence of NaCl or MgCl?. (i) A true gas- phase interference can occur whereby analyte atoms and chlorine-containing species from the matrix salts react to form gaseous metal chlorides thereby reducing the free metal concentration in the atomizer volume.(ii) A con- densed-phase interference can occur resulting in the vola- tilization of the analyte as a chloride molecule which is removed from the absorption volume without being atom- ized to the same extent as for an interferent-free solution. (iii) A combination of (i) and (ii) can occur. The analyte and matrix are volatilized as chloride molecules and dissociation of the analyte chloride is suppressed owing to the mass action effect of the excess of chlorine-containing species produced by the interferent matrix. At any tempera- ture and chloride salt concentration the extent of the interference for different elements is likely to depend on the dissociation energy of the monochloride.Elements with a low dissociation energy will probably suffer the least interference. (iv) The matrix salts may cause occlusion of the analyte which can reduce the atomization efficiency and cause loss of the analyte by a carrier mechanism. (v) IRapid evolution of vapours from the comparatively large mass of matrix salts may cause expulsion of the analyte species from the absorption volume and so give a reduction iin the atomic absorption signal of the analyte. This study compares the interferences caused by NaCl i%nd MgClz when wall platform and probe atomization are used for ET-CSAAS. The results obtained suggest that more than one of the above processes occur and that there are some differences in the mechanisms of the NaCl and MgC12 interferences.Experimental Spectrometer 'The SIMAAC system used for simultaneous multi-element imeasurements has been described previously. 1-4 The instru- ment consists of a 300 W xenon lamp as the primary source ian Cchelle polychromator modified for wavelength modula- tion and a PDP 11/34 mini-computer. The mini-computer generated the modulation waveform and performed data acquisition for up to 16 elements. In the present study absorbance measurements were made simultaneously for {Cd (228.8 nm) Co (240.7 nm) Cr (357.9 nm) Cu (324.8 nm) Mn (279.5 nm) Mo (313.3 nm) Ni (232.0 nm) Pb (283.3 nm) and V (318.4 nm). Electrothermal Atomizer '4 Perkin-Elmer HGA-500 electrothermal atomizer was used with an automatic graphite probe assembly which has been documented previously.l5 For probe atomization :studies pyrolytic graphite coated microporous glassy car- bon probes were employed for all measurements. Details of ithe analytical performance of this probe material have been published. l6 Platforms were fabricated from pyrolytic graphite tubes using the method outlined by Koirtyohann i2nd Kaiser." Although the platforms were slightly different iin dimensions and design from commercial equivalents the platforms were used as per current convention in that the edges were in minimal contact with the tube surface. IPyrolytic graphite coated electrographite tubes were used for all wall platform and probe measurements. With probe atomization a small slot was made in the graphite tube wall beneath the injection hole to allow entry and exit of the probe head.Sample aliquots (10 mm3) were deposted onto the atomization surface (wall platform or probe) using a IPerkin-Elmer AS-I autosampler. Table 1 Atomizer temperature programmes Temperature/ Ramp time/ Hold time/ Step "C S S Wall 120 5 45 Platform 250 Probe 450 Dry - - - - Pre-atomize Probe 2700 0 Atomize 2700 0 Clean 2700 0 Conditions RC* Rt GSS *RC=recorder function selected in this instance to initiate removal of the probe. t R = spectrometer read. IGS-argon internal gas flow stopped; 300 cm3 min-' at all other steps.JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY APRIL 1992 VOL. 7 535 ~~~~~ Table 2 Comparison of NaCl interferences for wall platform and probe atomization using SIMAAC In t e rferen t levellpg Analyte Element masdng 10 50 100 200 500 Atomization Signal recovery* (O/O) Cd 2 c o 4 Cr 1 c u 1 Mn 1 Mo 4 Ni 4 Pb 2 V 4 Wall Platform Probe Wall Platform Probe Wall Platform Probe Wall Platform Probe Wall Platform Probe Wall Platform Probe Wall Platform Probe Wall$ Platform$ Probe Wall Platform Probe 100 90 100 95 95 100 85 95 100 85 95 100 65 100 100 50 50 105 105 90 100 82 90 100 95 95 105 80 65 65 75 70 55 65 65 65 50 65 85 30 60 65 70 40 105 85 65 65 85 65 100 95 70 65 70 65 65 40 50 45 35 50 50 40 55 75 15 40 50 80 45 95 65 60 60 92 40 100 70 55 40 55 60 60 20 30 30 25 40 40 35 55 70 15 30 40 85 45 75 50 45 40 86 58 100 30 30 25 NDt 64 60 ND 25 25 ND 25 25 ND 50 55 ND 10 40 ND 45 50 ND 30 30 ND 48 65 ND 25 5 *Based on integrated absorbance signals.tND = not determined. $Recoveries of 95 and l0OI for wall and platform atomization respectively with 5 pg of NaC1. Electrothermal Atomizer Operating Conditions The operating parameters for each mode of atomization are given in Table 1. Since the nature of simultaneous multi- element measurements precludes the use of a high char temperature when volatile elements are present it was decided to omit the char stage for the purposes of the present study rather than remove the volatile elements from the list of test analytes. Similarly a compromise atomization temperature of 2700 "C was employed for all modes of atomization as this had previously been shown to provide satisfactory analytical performance when elements were determined on a simultaneous basis.lJ5 Calibration Solutions Multi-element test solutions were prepared from multi- element stock solutions (Spex Industries).Stock solutions containing 10°/o m/v NaCl and 10% m/v MgC12 were prepared for the interference studies by dissolving the appropriate amount of pure salt in ultra-pure distilled water. For the MgC12 solution spectroscopic-grade MgO was reacted with an appropriate volume of Ultrex HCl. This procedure was necessary owing to the unacceptably high analyte blank levels present in the AnalaR grade MgC12 salt. When analyte blank levels were measured in the prepared interferent salt solutions it was discovered that it would not be possible to conduct interference studies for all 16 elements investigated previously1J5 because of the high blank levels for some elements.For this reason the number of test analytes in the present study was restricted to nine. Appropriate solutions of these elements were prepared by serial dilution from multi-element stock solutions to give final elemental concentrations of 100 pg dm-3 for Cr Cu and Mn 200 pug dm-3 for Cd and Pb and 400 pg dm-3 for Co Mo Ni and V. The different analyte concentrations were required because of differences in the SIMAAC sensitivity for the elements. The NaCl and MgCl solutions used in the interference 'study were 0.1 0.5 1 .O 2.0 and 5.0%. Results and Discussion The effects of NaCl or MgC12 on the integrated absorbance values of the nine test elements were measured simultane- ously using the SIMAAC system.Recoveries of the analyte signals obtained in the presence of the interferent salts were calculated relative to the integrated absorbance signals produced by the same mass of analyte injected in an interferent-free solution (all signals were corrected for the blank). The peak height absorbance values were in the range 0.05-0.3. Recoveries were calculated from the mean result of three replicate measuremens and it was assumed that values in the range 90-1 10% indicated freedom from interference. Tables 2 and 3 summarize the results obtained with NaCl and MgC12 respectively as the interferent. With NaCl there was no great difference in the signal recovery values for Cd Co Ni and V when wall platform or probe atomization was used and the wall atomization results were only slightly poorer than those of the other two procedures for Cr Cu and Pb.The recovery values for Mn and Mo were best with probe atomization although similar performance was obtained for Mn with the platform. Different trends were noted when MgCI2 was used as the interferent (Table 3). Only the more refractory elements Cr,536 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY APRIL 1992 VOL. 7 Table 3 Comparison of MgC12 interferences for wall platform and probe atomization using SIMAAC Interferent levellpg Analyte 5 10 50 100 200 500 Element masshg Atomization Signal recovery* (%o) Cd 2 c o 4 Cr 1 c u 1 Mn 1 Mo 4 Ni 4 Pb 2 V 4 Wall Platform Probe Wall Platform Probe Wall Platform Probe Wall Platform Probe Wall Platform Probe Wall Platform Probe Wall Platform Probe Wall Platform Probe Wall Platform Probe 55 75 100 45 110 105 95 100 100 40 100 105 35 95 95 90 75 115 70 110 110 5 ND-F 100 100 105 110 55 45 90 30 105 110 95 100 95 25 100 100 20 80 90 90 80 110 55 110 110 0 100 100 95 105 110 45 60 80 15 105 100 95 70 65 20 85 80 10 40 80 85 95 105 30 100 95 0 ND 85 80 80 85 40 50 50 5 85 75 80 55 45 20 65 70 10 30 50 70 70 90 25 75 75 0 75 75 75 55 50 35 40 45 5 65 70 75 35 40 25 50 70 3 20 45 75 65 95 25 55 75 0 50 15 70 50 55 20 45 30 0 60 60 75 35 25 15 40 60 5 15 35 75 55 75 25 50 50 0 35 60 75 40 25 *Based on integrated absorbance signals. TND = Not determined.Table 4 Maximum chloride salt analyte mole ratios without interference using SIMAAC with platform and probe atomization Maximum interferent:analyte mole ratilo without interference? M-C1 MgC12 dissociation __ energy$/ Element temperature*/'C Platform Probe Platform Probe kJ mol-1 NaCl Atom appearance - Cd Pb c u Mn c o Ni Cr V Mo 460 790 1080 1240 1370 1400 1470 2000 1970 9.6 x 103 1 .8 ~ 104 1.1 x 104 9.4 x 103 2.5 x 103 2.5 x 103 8.9 x 103 2.2 x 103 t 4 x 103 9.6 x 1103 3.5 x 1105 1.1 x ]LO4 9.4 x 11 03 2.5 x !LO3 2.5 x Ii O3 8.9 x 1103 2.2 x 1103 4.1 x 1104 *Ref. 18. tBased on results in Tables 2 and 3. $Refs. 18 and 19; errors not available for Cd-Cl and Co-Cl values. §NA =Not available. ti x 103 1.1 x 104 6.6 x lo3 5.9 x 103 7.7 x 103 7.7 x 103 5.5 x 103 1 . 4 ~ 103 1 . 3 ~ 104 5.9 x 103 1.1 x 104 5.9 x 103 7.7 x 103 7.7 x 103 5.5 x 103 1 . 4 ~ 103 LOX 104 6.6 x lo4 208 301 k 2 9 383 +- 5 361 & 10 389 372k21 366 & 24 477 +- 63 NA§ Mo and V exhibited similar signal recovery values for each mode of atomization.The interferences observed for Cd Co Cu Mn Ni and Pb were greater in general with wall atomization than for the platform or probe methods. For most of the analytes the signal recovery values obtained with the platform and probe were similar although the probe method was superior for Cd and Mn. The differences in the NaCl and MgC12 results indicate that the interference mechanisms of the two salts might be different. With platform and probe atomization the tem- perature of the gas at the time of analyte vaporization is usually higher than for wall atomization. This appears to be an important factor in determining the extent of MgClz interferences on the more volatile elements but is appar- ently less important with NaC1. This suggests that vapour- phase interference effects such as those implicit in mecha- nisms (i) and (iii) mentioned earlier may be more impor- tant for MgC12. The extent of vapour-phase chemical interferences de- pends on the dissociation energies of the analyte mono-JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY APRIL 1992 VOL.7 537 chlorides as well as the gas temperature. If vapour-phase effects are important some correlation should exist be- tween the order of interferences caused by NaCl and MgC12 and the dissociation energies of the analyte monochlorides. The platform and probe results given in Tables 2 and 3 were used to obtain the mole ratios of chloride to analyte at which interferences were first observed.The data are presented in Table 4 along with the analyte atom appear- ance temperatures1* and the monochloride dissociation energies.18J9 The trends exhibited by the probe and plat- form results are similar. With NaC1 the order of interfer- ence for probe atomization is (V = Ni = Co)>(Cr = Mn = Cd= Cu)>Mo>Pb. With MgC12 the corresponding order is V>(Cr = Cd = Mn =Cu= Co=Ni)>Pb>Mo. The dis- sociation energy values given in Table 4 suggest that the expected order of interference should be V>(Co=Cu= Ni=Cr=Mn)>Pb>Cd. Neither of the observed trends correlate well with the order suggested from a consideration of the monochloride dissociation energies but the agree- ment is better for MgC12 than for NaCl. Cadmium is possibly less affected by the chloride salts owing to its volatility which may cause the element to be vaporized in advance of the interferent such that the temporal overlap of the analyte and chloride vapours in the atomizer tube is less than for the other elements.Expulsion of the analyte vapours with the matrix gases is also a possible cause of chloride salt interferences. HolcombeZ0 has argued that vapour expulsion becomes significant as an interference mechanism only if there is near coincidence in the volatilization temperature of the analyte and matrix species and when more than 1 x moles of matrix gas are evolved. It is known that NaCl can be vaporized from a platform above 800 OC21 and that the volatilization of the salt is almost complete by 1000 "C.Sodium atoms are produced at about 980 "C and Welz et a1.22 have suggested that the flow of Na atoms generated by 100 pg of NaCl can cause rapid expulsion of Pb from an electrothermal atomizer tube. It is possible therefore that elements with appearance tem- peratures similar to or slightly greater than that of Na may also be removed by the excess of flow of the Na atoms (or NaCl molecules) which is likely to persist for 1 or 2 s at high NaCl masses. Of the elements included in this study Co Cr Cu Mn and Ni have appearance temperatures slightly greater than Na and so the on-set of interference should occur at the same NaCl mass for these elements. The results in Table 2 indicate that this occurs for platform and probe atomization and with the exception of Mn also for wall atomization. Shifts in the time of the peak AAS signal were observed for some elements with probe atomization in the presence of NaCl suggesting that suppression of analyte vaporiza- tion might also occur by occlusion of the analyte in the NaCl matrix.For Cd the peak time increased from 100 (no NaCl) to 140 ms with 10 pg of NaCl and to 360 ms with 100 pg of NaC1. For Pb the peak time changed from 140 to 180 to 250 ms and for Mn the change was from 180 to 210 to 290 ms as the NaCl mass was increased from 0 to 10 pg and then to 100 pg. Cadmium Mn and Pb are the most volatile of the elements studied and so are more likely to be affected by occlusion in the NaCl matrix. This effect was not observed with MgC12 as the interferent. The result for Pb is different to that observed by Welz et aLZ2 who reported that the Pb peak occurred earlier in the presence of NaCl when vaporized from a platform.It is possible that the peak shifts could also arise owing to complete temporal overlap in the vaporiza- tion of the analyte and NaCl which for a short period caused total suppression of atom formation. However MgC12 would also be expected to exhibit this effect and cause peak shifts. Conclusions The NaCl interference study indicated that the platform and probe atomization procedures had only a slight advantage over wall atomization for the majority of the elements considered. With MgC12 however the interfer- ence effects observed with wall atomization were generally more severe than for the other two procedures which were similar in performance.Overall the analyte signal recovery values obtained with probe atomization were similar to if not better than those obtained with platform atomization. As a char step was not included in the atomizer programme the interference processes which caused a reduction in the analyte signals must have occurred during the atomization stage. It is probable that a number of phenomena contributed to the interference observed. A comparison of the MgC12 results for the three modes of atomization suggests that under the conditions used for the SIMAAC measurements vapour-phase chemical interfer- ences could be important with this salt. There was some evidence that occlusion of Cd Mn and Pb in NaCl might have contributed to the interferences experienced by these elements.Also expulsion of Co Cr Cu Mn and Ni by rapid production of NaCl or Na vapours may have reduced the maximum analyte atom concentration achieved in the furnace. It is concluded therefore that gas expulsion analyte occlusion and vapour-phase effects all contribute to the interferences observed. However it is not possible from this study to determine the relative contribution of each process to the interferences suffered by the various analyte elements. Without a char step freedom from interference was achieved with probe or platform atomization for interfer- ent:analyte mole ratios of up to 1 x 103-1 x los. Greater freedom from interference may be obtained with chemical modification. A recent study with Pd as the modifier has indicated that a compromise char temperature of 800 "C can be used for simultaneous determination of Cd Cr Cu Mn Mo and Pb by ET-CSAAS with wall atomizati~n,~~ although contamination of the sample by the modifier can be a problem. The authors thank Philips Scientific (now Unicam Limited Cambridge UK) and the Philips Research Laboratories Aachen Germany for provision of graphite probes. They also acknowledge support provided to D.L.by the Pye Foundation and to J.C. by the Science and Engineering Research Council (SERC) and Philips Scientific. References 1 Harnly J. M. and Kane J. S. Anal. Chem. 1984 56 48. 2 Harnly J. M. Miller-Ihli N. J. and O'Haver T. C. Spectro- chim. Acta Part B 1984 39 305. 3 O'Haver T. C. Analyst 1984 109 21 1 . 4 Harnly J. M. Miller-lhli N. J. and O'Haver T.C. Appl. Spectrosc. 1982 36 637. 5 Ediger R. D. At. Absorpt. Newsl. 1974 14 127. 6 Slavin W. Manning D. C. and Carnrick G. R. At. Spectrosc. 1981 2 137. 7 MacDonald L. R. O'Haver T. C. Ottaway B. J. and Ottaway J. M. J. Anal. At. Spectrom. 1986 1 485. 8 Slavin W. Carnrick G. R. Manning D. C. and Pruzkowska E. At. Spectrosc. 1983 4 69. 9 Slavin W. and Manning D. C. Spectrochim. Acta Part B 1980 35 701. 10 Ajayi 0. O. Littlejohn D. and Boss C. B. Anal. Proc. 1988 25 75. 1 1 Littlejohn D. Lab. Pract. 1987 36( lo) 126. 12 Frech W. and Jonsson S. Spectrochim. Acta Part B 1982 37 1021. 13 Lundberg E. Frech W. Baxter D. C. and Cedergren A. Spectrochim. Acta Part B 1988 43 451.538 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY APRIL 1992 VOL. 7 14 Lundberg E. Frech W. and Harnly J. M. J. Anal. At. Spectrom. 1988 3 11 15. 15 Carroll J. Miller-Ihli N. J. Harnly J. M. Littlejohn D. Ottaway J. M. and O'Haver T. C. Analyst 1985,110 1153. 16 Littlejohn D. Cook S. Durie D. and Ottaway J. M. Spectrochim. Acta Part By 1984 39 295. 17 Koirtyohann S. R. and Kaiser M. L. Anal. Chem. 1982,54 1515A. 18 L'vov B. V. Spectrochim. Acta Part B 1978 33 153. 19 CRC Handbook of Chemistry and Physics ed. Weast R. C. CRC Press Boca Raton FL 69th edn. 1988. 20 Holcombe J. A. Spectrochim. Acta Part B 1983 38 609. 2 1 Chaudhry M. M. and Littlejohn D. Analyst 1992 117 7 13. 22 Welz B. Akman S. and Schlemmer G. J. Anal. At. Spectrom. 1987 2 793. 23 Littlejohn D. Egila J. N. Gosland R. M. Kunwar U. K. Smith C. and Shan X.-Q. Anal. Chim. Acta 1991 250 71. Paper 2/00533F Received January 30 I992 Accepted February 4 I992

ISSN:0267-9477    

DOI:10.1039/JA9920700533

出版商:RSC    

年代:1992

数据来源: RSC

摘要:

JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY APRIL 1992 VOL. 7 539 Radiofrequency Atomization and Excitation With a Hot Graphite Cup Electrode for Trace Element Determination by Atomic Emission Spectrometry Kuniyuki Kitagawa and Takashi Katoh Department of Applied Chemistry School of Engineering Nagoya University Furo-cho Chikusa-ku Nagoya 464-01 Japan A discharge lamp was constructed as an excitation source for the determination of trace elements in small samples by atomic emission spectrometry. A graphite cup in which an aliquot (1 0 mg) of sample solution or powder was loaded was located in a stainless-steel cylinder. An r.f. discharge was formed between the graphite cup and the grounded cylinder under a reduced pressure of argon. Concurrently with the formation of the plasma the graphite cup was heated to about 1900 "C owing to the r.f.power dissipation. Asa result the sample was thermally vaporized and/or atomized and subsequently excited to emit radiation in the discharge plasma surrounding thegraphitecup. Basic characteristics were studied in conjunction with the analytical performance. The matrix effect of sodium was tested and found to be negligible in the determination of copper in the presence of up to a 1000-fold excess of sodium. A linear dynamic range of thecalibrationgraphwasobtainedoveraboutfourordersof magnitudeof analyte mass. The discharge lamp was applied to thedirect determination of copper zincandchlorine in National lnstituteof Standards and Technology Standard Reference Materials of biological samples. Keywords Atomic emission spectrometry radio frequency discharge lamp hot cup electrode; atomization; small sample size As exemplified by the inductively coupled plasma (ICP) which has been widely used as a commercially available excitation source for trace element analysis atomic emis- sion spectrometry (AES) has advantageous features such as the capability for multi-element determinations and a wide dynamic range of analyte concentration.Among the most recent requirements in analytical atomic spectrometry is the capability for multi-element determi- nations in small samples in a variety of matrices. In particular this capability is important in clinical medicine biochemistry etc. where large samples are rarely available. Electrothermal atomization atomic absorption spectrome- try (ETAAS) is a well established technique that allows analyses of small samples. 1-3 However multi-element AAS requires complex optical systems.The development of a stable excitation source to permit multi-element determinations in small samples by analyti- cal atomic spectrometry is important. Several techniques to satisfy these requirements have been developed for the ICP e.g. an electrothermal ~aporizer,~ direct insertion of a graphite sample cup5v6 and a discrete nebulization tech- nique where a small volume of sample solution was nebulized batchwise.' In previous studiess-10 several types of discharge lamps with a hot graphite cup electrode which served as a sample vaporizer-atomizer were developed. The atomized sample was subsequently excited in a plasma generated around the hot graphite cup.The use of radiofrequency (r.f.) power was found to be effective for high excitation of analyte atoms. In this work the basic characteristics of a newly con- structed capacitively coupled r.f. discharge lamp were studied for trace analyses of small samples by AES. The r.f. discharge lamp was successfully applied to direct determi- nations of Cu Zn and C1 in National Institute of Standards and Technology (NIST) Standard Reference Materials (SRMs) of biological samples. Experimental Radiofrequency Discharge Lamp and Instrumentation Fig. 1 shows a schematic diagram of the capacitively coupled r.f. discharge lamp with a hot graphite cup 9 Fig. 1 Schematic diagram of the r.f. discharge lamp with a hot graphite cup electrode 1 graphite cup; 2 S i c cup; 3 Mo screw; 4 alumina tube; 5 Macor base; 6 cylindrical stainless-steel enclosure (grounded); 7 quartz windows; 8 to spectrometer; 9 quartz lid; 10 to r.f.matching box; 1 1 to vacuum pump; and 12 r.f. plasma. A graphite cup with plasma region (1-3 mm above cup) viewed laterally; and B graphite cup with emission observed through aperture on lateral face electrode. The graphite cup of 3.5 mm i.d. and 5 mm o.d. 5 mm in height and 3.5 mm in depth ( l ) is fixed in a silicon carbide (Sic) cup of 5 mm i.d. 6 mm 0.d. and 10 mm in height (2) which serves as a thermal resistor against the hot graphite cup. The Sic cup is attached to a molybdenum screw 3 mm in diameter and 53 mm in length (3). The Sic cup and the molybdenum screw are covered with an alumina tube of 6 mm i.d.10 mm 0.d. and 35 mm in length (4) (SSA-S Nippon Kagaku Kogyo). The molybdenum screw is supported by a machinable glass ceramic (Macor) base (9 which serves as an electrical and thermal insulator. These components are incorporated in a cylindrical stain- less-steel enclosure of 38 mm id. 60 mm 0.d. and 77 mm in height (6) with quartz plate windows (7) for spectrometric observation (8). A quartz lid (9) serves as a window for540 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY APRIL 1992 VOL. 7 Table 1 Instrumentation Rf power supply system- Generator Power meter 1 Amplifier 1 Power meter 2 Amplifier 2 Matching box Dummy load Spectrometer- Monochromator 1 Monochromator 2 Detector PMT preamplifier PMT high-voltage supply Recording system- Recorder Computer AID card Program Vacuum system- Vacuum meter Vacuum pump Additional items- Optical pyrometer Microdispenser Ar gas Laboratory-made crystal oscillator Trio function power meter PF-8 I0 Tokyo HyPower HL-400J Taniguchi Engineering Trades P2200 Tokyo HyPower HG3K Daiwa Industry CNW-5 18 Kuranishi Keisoku Kenkyuusho RL-I 200D Nippon Jarrell Ash JE-5OE with (a grating of 1200 grooves mm-l and slits of 25 pm in width Jobin Yvon H.20UV with a grating of 1200 grooves mm-' and slits of 25 pm in width (spectral Hamamatsu Photonics R666 photomultiplier tube (PMT) Laboratory made Laboratory made (spectral bandwidth = 0.04 nm)i bandwidth=O. 1 nm) Toa strip-chart recorder FBR-25 2A NEC 980 1 UV 1 1 Designed by the Chubu Branch of the Japan Society for Analytical Chemistry Written in C-language (Turbo C Version 2.0) Okano Pirani gauge Hitachi rotary pump C3A-150 Minolta TR-630 Drummond Digital Microdispenser Nihon Sanso (99.999Oh) measuring the cup temperature by means of an optical pyrometer.At a flow rate of 50-240 ml min-l STP argon is forced to flow under a reduced pressure of 400-1 866 Pa. The pressure was monitored with a Pirani gauge whose sensor was attached to the vacuum line on the outlet side of the discharge lamp. An r.f. power up to 400 W generated by an oscillator tuned at 13.56 MHz and two r.f. linear amplifiers was applied to the molybdenum screw through an impedance- matching circuit. As a result a plasma discharge was formed between the graphite cup electrode and the grounded cylindrical enclosure (6).Thus the graphite cup surrounded by the plasma was heated by the ref. power dissipation. The time required for the graphite cup to reach an equilibrium temperature was 3-4 s. The r.f. power levels at the linear amplifiers were monitored by in-line transmis- sion-type power meters. An alumina tube (4) prevented an undesirable discharge from being formed on the molyb- denum screw surface and confined the effective discharge around the graphite cup. Table 1 lists the components used for this study. Two spectrometers were used for the simultaneous detection of two analyte emissions or analyte and background emis- sions. The spectrometric signal acquired by a micro- computer through an analogue-to-digital (A/D) converter was processed to calculate the integrated emission after baseline correction.Procedure for Measurement of Liquid Samples An aliquot (10 pl) of a sample solution was placed in the graphite cup using a microdispenser (Drummond). Subse- quently pumping was started with the argon flow stopped. A reduced pressure of < 133.3 Pa was reached when the solvent vaporization was considered to be completed. Then the argon flow rate was adjusted to give a discharge pressure (400-1 866 Pa at a flow rate of 50-240 ml min-l STP). About 2 min later the r.f. power was switched on for the sample atomization-excitation. The next run was made after cooling for about 3 min. Two types of graphite cup 'were used (Fig. 1 A and B). With type A the plasma region .from 1 to 3 mm above the top of the cup was viewed 'laterally by the spectrometer.With type B the emission was observed through one of two apertures drilled on the lateral face of the graphite cup. The counter aperture prevents the black-body emission of the inner face from being viewed by the spectrometer. Type A was used throughout except for measuring the detection limits. Procedure for Measurement of NIST Powder Samples In order to attain stable excitation in the r.f. discharge plasma after the removal of biological matrices electrother- mal charring was applied to the NIST SRMs (SRM 157 1 Orchard Leaves and SRM 1577 Bovine Liver). The same stainless-steel enclosure as that illustrated in Fig. 1 was used for the charring chamber. The discharge electrode was replaced with a cup-type graphite atomizer which served as a heating element for charring or a charring cup.An aliquot (0.15-2 mg) of the powdered sample was weighed into a graphite cup. For the standard additions method 20 pl of a standard solution in 0.05 mol dm-3 sulfuric acid were added to the powdered sample. In order to prevent the powdered sample from being blown by a pressure shock on firing the discharge the sample was covered with a graphite lid 3.2 mm in diameter and 1 mrn in thickness; with the use of the lid the emission peaks became broader but there was no decrease in the integrated emission which was even slightly enhanced probably owing to the confinement of the resulting vapours. The graphite cup was then mounted in the charring cup. Under an argon purge the sample was dried at about 100 "C for 1 min and then charred at 500 "C for 25 s by passing a current through the charring cup.Finally the sample cup was transferred into the Sic cup of the discharge lamp and the discharge process described above was carried out. Reagents Stock standard solutions were prepared by dissolving metal sulfates or nitrates of analytical-reagent grade in distilled,JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY APRIL 1992 VOL. 7 54 1 f 30 4- .- C .g 20 v) W .- E 10 0 Time -+ Fig. 2 Typical emission peaks for (a) Cu and (b) Na simultane- ously measured by two separate spectrometers A peak of background emission (see text). The sample contains 10 ng of Cu (10 pl of 1 ,ug ml-* CuSO,) and Na ( 100 times the amount of Cu in moles) as Na2S04. R.f. power 300 W; Ar pressure 10.664 x lo2 Pa; Acu=324.75 nm and ;INa=589.99 nm ioo 200 300 400 500 R.f powerMl 2000 v E 2 \ 4- E c 1500 Fig.3 Dependences of the graphite cup temperature and Cu emission intensity (integrated signal) on the r.f. power A tempera- tureofcup; B lOOngofCu(10,ulof lOpgml-');andC 10ngof Cu (10 pl of 1 pg ml-l). Ar pressure 7.198 x 1 O2 Pa. For B y-axis values should be multiplied by a factor of 4 de-ionized water and adjusting the acidity to 0.1 mol dm-3 with sulfuric acid or nitric acid of analytical-reagent grade. A stock chloride solution was prepared from magnesium chloride in the same manner. Argon of 99.999% purity and graphite of spectrometric grade were used throughout. Carbon tetrachloride vapour was mixed in the plasma gas for the wavelength setting of C1. In order to enhance the emission intensity of C1 20% v/v helium in argon was used as the plasma gas.Results and Discussion Emission Peak Profile A typical emission peak trace for copper is shown in Fig. 2. The first small peak (A) is the background emission which was confirmed with a blank solution displacing the spectro- meter wavelength by about 0.1 nm from the 324.745 nm analytical line of Cu I. The most likely species responsible for the background emission are diatomic species such as 0 6.66~102 13.33~1 O2 20x1 d Pressure/ Pa Fig. 4 Dependence of the Cu emission intensity (integrated signal) on the Ar pressure. The sample contains 10 ng of Cu; and r.f. power 300 W OH N2+ and NH. The molecular emission from OH species (A0-0=306.4 nm) N2+ (A0-,,=391.1 nm) and NH (Ao-o= 335.8 nm) is known to be very strong in the 300-400 nm region.The species OH can be generated from the residual water of the sample solution. The species N2+ NH and NO can be evolved from the nitric acid. The strong band of NO (A0-0=226.2 nm) in the 200-300 nm region has been reported as a typical background s o ~ r c e . ~ * ~ For the lines in a shorter wavelength region Cd I 228.802 nm Zn I 213.856 nm and Mg I 285.231 nm however the first peaks were found experimentally to be less intense than those for the lines in a longer wavelength region beyond 300 nm Cu I 324.745 nm Fe I 37 1.994 nm Ag I 328.068 nm Co I 346.580 nm Cr I 357.869 nm A1 I 309.271 nm and V I 318.398 nm. Hence the emissions from the species OH N2+ and NH seem to be stronger than that from NO.A high-speed scanning technique such as a refractor plate-a.c. amplifier combination and chemical modifiers would be effective in removing the background emission peaks. Cup Temperature Fig. 3 shows the dependence of the temperature of the graphite cup on the r.f. power. As the r.f. power increases the cup temperature becomes higher but its increment decreases. The heat loss through radiation according to the Stefan-Boltzmann law (cc T4) becomes larger at higher temperatures resulting in the low increment of tempera- ture. The integrated emission for two different amounts of copper introduced has an analogous dependence on the r.f. power to that of the cup temperature. The temperature of the graphite cup determined by the r,f. power as described above was almost independent of the discharge pressure. On the other hand the emission intensity (integrated signal) has a significant dependence on the pressure.As shown in Fig. 4 the integrated emission increases as the pressure becomes higher. Two possible causes are that the exciting species such as Ar* Ar+ and free electrons (which species is the most relevant is not clear) increase in number density and that the diffusion of the resulting atoms is confined or becomes slower at higher pressures. The former is likely because the emission peak profiles were broadened as the pressure became higher. At pressures > 133.33 Pa however the discharge became unstable forming a streamer which moved irregularly around the graphite cup.542 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY APRIL 1992 VOL.7 *Oo0 5 - In C 3 c .- 1000 h .- In .- 0 1 1 2 3 4 5 Height above cup top/mm Fig. 5 Vertical distribution profile of the Cu emission intensity (integrated signal). The sample contains 100 ng of Cu; r.f. power 400 W; and Ar pressure 10.664 x lo2 Pa. The width of the bar indicates the height of the vertical observation window which is given by a horizontal slit located in front of the entrance slit of the spectrometer Spatial Profile of Emission The profile of the copper emission intensity (area) above the top of the type A graphite cup which was obtained by moving the discharge lamp vertically is shown in Fig. 5. The emission intensity becomes lower at a higher position above the top of the graphite. This is mainly attributable to the decreased number density of atoms caused by a rapid diffusion of the atomic vapour from the confined space in the cup to the free space around the cup.With the type B graphite cup the emission from atoms at a higher number density is observable through the aperture. As a result the detection limit [signal-to-noise ratio (SIN) = 21 was about ten times lower (0.02 ng of copper) than that obtained with the type A graphite cup. Calibration Graph and Detection Limit The calibration graph for copper was linear from the detection limit up to 1 x lo3 ng (about four orders of z CT v) = 8 CI - 0 10 20 30 40 50 Mass of CVpg Fig. 6 Calibration graph for chlorine A integrated emission and B square of integrated emission. Wavelength of C1 I 460.10 nm; plasma gas 20% v/v He in Ar; r.f.power 300 W; and graphite cup temperature 1600 "C. The sample contains MgC12 magnitude) and curved beyond 1 x lo4 ng whereas that of Cl was not linear. As shown in Fig. 6 however the square of tlhe integrated emission is proportional to the mass of C1. 'This suggests that the dissociation of C12 molecules plays a dlominant role in the mechanisms involved the equilibrium equation for Cl and Clz explains the square dependence. In practice green-yellowish radiation was visually observed around the graphite cup on firing the plasma. The radiation can probably be attributed to the molecular band of C12 whose head lies at 516.3 nm.11J2 Detection limits obtained with the developed excitation source are listed in Table 2 together with those given by an ICP with an electrothermal vaporizeI-4 and direct insertion of a graphite CUP.^^^ The detection limit obtained with the discharge lamp developed here is comparably low for copper compared with those attained by the other techi- clues but higher for other elements.The use of helium or neon as the discharge gas possibly improves the detection limits through more effective excitation owing to its higher internal energy. ]Effect of Easily Ionizable Elements Another analytically important factor for an excitation source is freedom from matrix effects. An easily ionizable element (EIE) sodium which is commonly present at Table 2 Detection limts (DL) Dung* Dung* R.f. R.f. Element Unm method Ref. Element llnm method Ref. 0.03$ c u 324.75 0.02 0.02-f Cd 228.80 0.27 0.02-f Fe 37 1.99 0.12 0.04-f Ag 328.07 0.09 0.001$ 285.21 0.02 Mn 403.08 0.38$ 0.02-f Mg Zn 2 13.86 1.4s 0.02$ c o 346.58 2.7 0.02-f 0.006$ Cr 357.87 1.2 A1 309.27 0.9 Pb 405.78 3.8 0.035-f V 3 18.40 24 0.01$ *DL=detection limit based on 9N=2 (S=s-b and N=q where s is the mean value of the peak height for the sample b that for the ?From ref.5 . $From ref. 4. §S is the peak height for the sample and N the baseline noise (range). blank and 0 is the standard deviation of the peak heights for the blanlk). The type B graphite cup was used.JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY APRIL 1992 VOL. 7 543 Table 3 Direct determinations of Zn Cu and Cl in NIST SRMs Foundpg g-'* Element Wavelength/ Certificate Sample and line nm CMt SAM4 value/pg g-I SRM 1571 Orchard Leaves Zn I 2 13.86 31 + 5 26k3 25+3 SRM 1577 Bovine Liver c u I 324.75 179 + 65 171 +55 193+ 10 Zn I 2 13.86 187 +40 129+ 15 130k 10 c1 I 460.10 3.3 + 0.75 2.6§ I1 *The standard deviation is based on the measurements repeated at least three times.fCM =calibration method. @AM= standard additions method. §Values in mg g-l. )/Not certificate but reference value. relatively high concentrations in most practical samples was tested as a typical interferent as its coexistence often caused a notable enhancing effect on the analyte emission intensity in excitation sources such as a microwave-induced plasma13 and a capacitively coupled microwave plasma14 and in the reaction zone near the r.f. load coil of an ICP.lS The dependence of the copper emission intensity (integrated signal) on the sodium concentration is shown in Fig.7. The effect due to sodium is almost negligible up to a molar ratio of 1000 (mass of Na= 3622 ng). At a molar ratio of 10000 (mass of Na=36.22 pg) a suppressing effect is observed. This is mainly due to the plasma instability induced by the change in the discharge impedance which is probably caused by the ionization of the large amount of sodium. Plasma fluctuation was visually observed and consequently induced impedance mismatching was indi- cated on the r.f. power meter. Use of an automatic impedance-matching circuit with a high response or a free- running r.f. generator could overcome the problem. The reason why the effect due to sodium is small in the present discharge lamp is not completely clear. A possible cause is fractional vaporization which causes a separation of the residence time between copper and sodium atoms.Consequently copper atoms emit under conditions free from sodium species. However it was confirmed by experiment that this was not the case. In Fig. 2 the peak profile of sodium emission at 589.99 nm which was simultaneously measured by another spectrometer is shown together with that of emission for copper. This illustrates that the copper and sodium atomic vapours co- existed during the majority of the residence time. Another possible cause of the less significant EIE effect is as follows. Usually the higher number density of free electrons in the plasma has been considered to play an Molar ratio [Nal:[Cul Fig. 7 Dependence of the Cu emission intensity (integrated signal) on the co-existing concentration of Na mass of Cu 10 ng; r.f.power 300 W; and Ar pressure 7.1 98 x 1 O2 Pa. The abscissa is the relative integrated emission with respect to the integrated emission in the absence of Na important role in lessening the EIE effects in the analytical zone of the ICP.16 As the number density of free electrons is not high under reduced pressure it may not be a major factor responsible for the small effect due to sodium. However spatial localization of free electrons due to a hollow effect in which the electric field is condensed in the hollow graphite cup resulting in a higher density of free electrons might contribute to some extent to forming a high number density of free electrons around the graphite cup. Experimental observation of the number density of free electrons using the Hp line,17 requires detailed discus- sion.Application to NIST Biological Standard Reference Materials The optimization of charring is important in the direct determination of trace elements in solid or powder samples without applying chemical digestion. In a preliminary experiment charring of the NIST powder samples was effected in a pre-discharge which was obtained at an r.f. power lower than 100 W before the main discharge for atomization-excitation. However intensive atomic emis- sion was observed in the pre-discharge process which implies analyte loss. This is probably attributable to plasma sputtering. A conventional charring method used in ETAAS was therefore employed as described under Experimental.The discharge was occasionally unstable after charring at temperatures lower than 400 "C whereas the loss of copper and zinc was found after charring above 600 "C. Accord- ingly 500 "C was chosen as the optimum charring tempera- ture. After applying optimized charring the peak profiles of copper zinc and chlorine emissions observed with the NIST sample were almost the same as those obtained with the standard solutions. The results for the NIST SRMs are given in Table 3. Fairly good agreement between the values obtained with the standard additions method and those certified by NIST with relative errors within lloh was obtained for copper and zinc. Positive errors were found in the zinc determina- tions with the calibration method which reflects some chemical interferences.The relative error of 27Oh for chlorine is much larger than that for copper or zinc. In practice the blank signal was much higher (about ten times) in magnitude and fluctuated more than those for other elements. Insufficient purification of water or some back- ground-emitting species might be responsible. Conclusions An r.f. discharge lamp with a hot graphite cup electrode was constructed for trace element analyses of small samples by AES. Its basic characteristics and analytical performance were examined. The atomization seems to be principally544 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY APRIL 1992 VOL. 7 thermal (see the high detection limit for vanadium). As seen from the aluminium loss in the pre-discharge described above however sputtering caused by the plasma species such as Ar+ can also play a role.A wide linear dynamic range of about four orders of magnitude of concentration was obtained. The EIE effect caused by an excess amount of co-existing sodium was negligibly small up to an [Na]:[Cu] relative molar ratio of 1000. Application to the trace determination of copper zinc and chlorine in NIST Standard Reference Materials was made with fairly good agreement with the certificate values. The authors thank S. Takahashi for manufacturing the discharge tube and T. Watanabe and T. Imura for fine processing of the ceramic tubes. This research was partly supported by a Grant-in-Aid from the Ministry of Educa- tion Science and Culture Japan. References 1 Wennrich R. Bonitz U. Brauer H. Niebegall K. and Dittrich K. Talanta 1985 32 1035. 2 Chakrabarti C. L. Delgado A. H. Chang S. B. Falk H. Hutton T. J. Runde G. Sychra V. and Dolezal J. Spectrochim. Acta Part B 1986 41 1075. 3 Itoh K. and Atsuya I. Bunseki Kagaku 1986 35 530. 4 Gunn A. M. Millard D. L. and Kirkbright G. F. Analyst 1978 103 1066. 5 Kirkbright G. F. and Walton S. J. Analyst 1982 107 276. 6 Abdullah M. Fuwa K. and Haraguchi H. Appl. Spectrosc. 1987 41 715. 7 Uchida T. Kojima I. Iida C. and Goto K. Analyst 1986 111 791. 8 Kitagawa K. Narita R. and Takeuchi T. Anal. Chim. Acta 1979 110 291. 9 Kitagawa K. Kanoh S. Ohta K. and Yanagisawa M. Anal. Sci. 1988 4 153. 10 Kitagawa K. Katoh S. and Yanagisawa M. J. Spectrosc. SOC. Jpn. 1989 38 209. 1 I Huber K. P. and Herzberg G. Diatomic Molecules and Spectra Van Nostrand Reinhold New York 1979. I 2 Clyne M. A. A. and Coxon J. A. Proc. R. SOC. London Ser. A 1967,298,424. 1,3 Matousek J. P. Orr B. J. and Selby M. Spectrochim. Acta Part B 1986 41 415. 114 Kitagawa K. Koyama T. and Takeuchi T. Anal. Chim. Acta 1979 109 241. 15 Kawaguchi H. Ito T. Ohta K. and Mizuike A. Spectro- chim. Acta Part B 1980 35 204. 16 Haraguchi H. ICP Hakkobunseki no Kiso to Ouyou Koudan- sha Tokyo 1986. I 7 Blades M. W. and Caughlin B. L. Spectrochim. Acta Part B 1985 40 579. Paper I /0445 7E Received August 27 1991 Accepted November 26 1991

ISSN:0267-9477    

DOI:10.1039/JA9920700539

出版商:RSC    

年代:1992

数据来源: RSC