JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY MARCH 1994 VOL. 9 427 Equilibrium and Mass Spectrometry of Nitrate Decomposition in Electrothermal Atomic Absorption Spectrometry* Trevor M c Al I iste r CSlRO Materials Science and Technology Locked Bag 33 Clayton Victoria 3 768 Australia The decomposition of metal nitrates to oxides is assumed to occur in electrothermal atomic absorption spectrometry during the pre-atomization drying or thermal pre-treatment of nitrate samples. The mechanism of thermal decomposition of nitrates to solid oxides and gaseous NO is well established but some evidence from recent electrothermal mass spectrometry (ET-MS) experiments has been used to support an alternative mechanism that of gaseous metal oxide formation e.g. MNO,-+MO(g) + NO,(g) The ET-MS of nickel cobalt copper magnesium and lead nitrate samples have been re-examined using a quadrupole mass spectrometer with cross-beam rather than axial sampling of gas from the furnace.With this new sampling geometry no gaseous metal oxides were detected over a range of heating rates from slow drying to rapid atomization. These results are in keeping with thermochemical equilibrium calculations as is the observation of gaseous phosphorus oxides in the thermal decomposition of NH,H,PO,. Earlier observations of gaseous metal oxides of the above elements by ET-MS are attributed to the formation of molecular clusters of oxides and/or nitrates during rapid decomposition of the metal nitrates. Keywords Electrothermal atomic absorption Spectrometry; mass spectrometry; equilibrium; nitrate decomposition; gaseous oxide It has generally been assumed in electrothermal atomic absorp- tion spectrometry (ETAAS) that in the analysis of nitric acid matrices or nitrates of metals the metal nitrate precipitated in the furnace by drying will decompose on further heating to give the solid metal oxide and gaseous nitrogen dioxide and oxygen M( NO~)~+MO(S) + 2NO,(g) + O.5O2(g) (1) The production of gaseous metal atoms from nitrates in ETAAS should then occur by the reduction of MO(s) by graphite at higher temperatures or via gas-phase decomposition of a stable gaseous oxide MO(g).The mechanism of the isothermal decomposition of solid metal nitrates has been well established' and more recent has tended to confirm the accepted features of nucleation and growth of oxide in the nitrate and the diffusion of the NO,(g) produced to the solid-gas interface.In ETAAS N02(g) has been observed by mass spectrometry (MS) during pre-atomization heating of nitrates of c o ~ p e r ~ n i ~ k e l ~ lead6*'*' and ~ o b a l t . ~ Also reported in these studies were MO+ ions having been observed during rapid pre- atomization heating and attributed to oxide species in the gas phase arising from the rapid decomposition of nitrate.7 L'vov'' has suggested that the precursors of the MO+ ions are genuine gaseous diatomic oxides MO(g) and that a different mechan- ism of oxide formation is operating in the rapidly heated furnace of the ETAAS system to that in the isothermal kinetic studies M(NO,),(s)~MO(g)+2NO,(g)+0.502(g) (2) or M(N03)2(s)-,M0(g)+2N0(g)f 1.502(g) (3) Although these reactions are not thermochemically favoured compared with reaction (1) in the temperature region 400-700 K L'vov has argued that in ETAAS the kinetic rates of reactions (2) or (3) are substantially faster than that of ( 1 ) owing to solid-state factors such as the rate of diffusion of NO,(g) or NO(g) through the solid reactant and pro- duct.L'vov proposed that the product MO(g) condenses in the l3TAAS furnace at 1 atm (1 atm = 101 325 Pa) so that * Presented at the XXVIII Colloquium Spectroscopicum Internationale (CSI) Post-Symposium on Graphite Atomizer Techniques in Analytical Spectroscopy Durham UK July 4-7 1993. reduction of MO(s) can then proceed without loss of analyte. Electrothermal mass spectrometry (ET-MS) experiments con- ducted in ZIUCUO would however detect the product MO(g) because the gaseous oxide should not condense at the low pressures of those experiments.As this novel mechanism contravenes the accepted view of solid nitrate decomposition and relies on incidental reports of MO' ions in work which had other objectives and therefore neglected the phenomenon the experiments on rapid pre- atomization heating of cobalt copper nickel and lead nitrates have been repeated using a recently modified ET-MS instru- ment of cross-beam geometry," and the thermochemical equi- librium in the decomposition of nitrate was calculated by the free energy minimization m e t h ~ d . ' ~ . ' ~ Experimental Several designs of ET-MS apparatus have been reported. Bass and Holcombe7 used a graphite rod atomizer whereas Styris,14 Sturgeon et aL6 and Ham and McAllisterg used tubular graph- ite furnace atomizers.All systems used quadrupole mass spec- trometry (QPMS) instruments and a variety of sample loading methods were employed involving loading the furnace at 1 atm and transferring it to a pumping region or injecting a sample into the furnace in UQCUO through a carefully aligned capillary. The design of Styris14 was elaborated to permit the sampling of the furnace gases during atomization from 1 atm through a skimmer cone into the QPMS instrument at low pressure. The common feature of all these designs was the placement of the atomizer in line with the axis of the QPMS instrument. The design of Ham and McAllisterg has been modified recently'' to permit cross-beam sampling of the furnace gases with the atomizer situated off the axis of the QPMS system (see Fig.1). Such a geometry is commonly used in MS to prevent particles generated in experiments being translated down the axis of the QPMS instrument and creating inter- ference in the electron multiplier detector. This design has the additional benefit of enabling the furnace temperature to be measured by an infrared IR pyrometer (Ircon Modline 11) and was used recently to determine the appearance temperatures of GeO(g) from alkaline and acid matrices of Ge.I5 The sensitivities of the ET-MS systems designed so far are low especially when QPMS is being used to scan over a mass spectrum rather than monitoring a single peak. A typical sample in ET-MS is therefore of the order of several hundred428 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY MARCH 1994 VOL.9 Ion source Viewing Graphite furnace Twin Cu rod feedthrough Fig. 1 Schematic diagram of the cross beam sampling geometry of the ET-MS instrumentation viewed from above nanograms. In this work 5 pl of 0.1 g 1-l solutions of the metal nitrates in de-ionized water were used. Thermochemical equilibrium in the thermal decomposition of the metal nitrates was calculated as p r e v i o ~ s l y ~ * ~ ~ using the CSIRO Thermochemistry System of Turnbull and Wadsley,12 which employs the free energy minimization technique of Eri ksson. l3 Equilibrium Calculations L'vov calculated the equilibrium partial pressure of MO(g) in reaction (2) for magnesium cobalt and copper nitrate hexahy- drates for reaction (3) for anhydrous nickel nitrate and for reactions (2) and (3) for anhydrous lead nitrate.In view of the uncertainty in the degree of hydration of the deposited salts in vacuum ET-MS and of the lack of reliable thermochemical data especially entropies and specific heats of the nitrates and their hydrates such calculations can only be taken as a very approximate indication of the equilibrium in thermal decompo- sition of these nitrates. When matched with experimental observations of appearance temperatures which are also very approximate only the most tentative conclusions can be drawn. Therefore unhydrated nickel and lead nitrates were taken as examples as used in L'vov's calculations," for the estimation of equilibrium by free-energy minimization allowing for both MO(s) and MO(g) as possible products along with N02(g) NO(g) and 02(g).For Ni(N03)2 the following values were used AH,(298 K)= -415 kJ mol-I from Wagman et ~ 1 ; ' ~ S0(298 K)= 183 J mo1-l K-l calculated by the method of Mills and Latimer in the CSIRO Thermochemistry System;I2 and C,= 13.9+0.067T J mol-' K-l by analogy with the data for Mg(N03)2.12 For Pb(N03)2 the values used were AHf( 298 K) = - 45 1.9 kJ mol from Wagman et ~ 1 . ; ' ~ So = 220 J rno1-l K-l again calculated by the method of Mills and Latimer;12 and C as for Ni(N03)2. The data for NiO(g) and PbO(g) came from the CSIRO Thermochemistry System database.12 The results show that either NiO(s) or PbO(s) should be the overwhelmingly dominant metal oxide product of the thermal decomposition of the anhydrous metal nitrates between 400 and 700 K and that the gas phase is composed of NO2 and O as in reaction (1).The over-all pressures in both systems are given in Table 1. By repeating L'vov's method," but noting that the exper- imental data2-g indicated that NO,(g) rather than NO(g) is produced in the decomposition a restricted equilibrium based on reaction (2) only can be calculated. The partial pressures found for gaseous nickel and lead oxides shown in Table 2 vary from 1 x atm between 300 and 700 K. to 1 x Table 1 Variation with temperature of equilibrium pressure of YO2( g j + O,( g j in the reactions ! i ) Ni(N03),(s)=NiO(s)+ 2NO,(g) +O.5O2(g) (ii) Pb(NO,),(s)=PbO(s)+2NO,(g)+0.5O2(g) ((iiij pressure of N02(g)+02(gj produced in 1 x the rate data of Criado et aL3 for reaction (i) s according to Temperature/K 300 350 400 450 500 550 600 650 700 p(i)/atm* 4 x 9 x lo+ 6 x 1 x lo- 0.18 1.4 7.9 32.7 - p (ii)/atm 2 x lo-'* 2 x 10-9 4 x 10-7 2 x 10-5 4 x 10-4 6 x 5 x 0.3 1.3 p(iii)/atm I 10-19 1 10-15 2 10-9 4 10-7 1 10-5 4 10-4 7 10-3 1 x 10-l' - * 1 atm= 101 325 Pa.Table 2 Variation with temperature of pressure of MO(g) in the reactions (i) Ni( N03),(s) =NiO(gj + 2N02(g)+0.50,(g) (ii) Pb(NO,),(s)=PbO(g)+ 2NO,(g)+0.50,(g) Temperature/K 300 3 50 400 450 500 550 600 650 700 p NiO (g)/atm* 1 x 10-39 1 x 10-25 1 x 10-15 1 x 10-13 1 x 1 x 10-l8 1 x lo-" 1 x 10-'O 1 x P PbO(g)/atm 1 x 10-17 1 10-14 2 10-9 2 10-5 1 10-4 1 x 1 x lo-" 7 x 1 x lo+ * 1 atm= 101 325 Pa.L'vovfo defined the detection limit for the ET-MS as z 1 mPa or 1 x lo-* atm. In comparing experimental observations of appearance temperatures of MO' a lower level probably 1 x atm is more realistic in the ET-MS system used in the present experiments as this tends to be the background pressure in the apparatus. This pressure should be reached by NiO(g) between 650 and 700 K and by PbO(g) at ~ 5 0 0 K. These calculations do correlate approximately with the experimental observations. However at these temperatures the pressure of N02(g) and 02(g) from reaction (1) would be much higher than that of MO(g) from reaction (2). It could be expected that the pressure of these gases expanding in the vacuum of the ET-MS appar- atus would break up the sample crystals unless reaction (1) is retarded for reasons undiscovered by previous investigat~rs.l-~ Indeed in the case of nickel the kinetic data for the thermal decomposition of Ni(N03)2 can be used to estimate the pressure of NO2 and O2 gases generated from the decompo- sition of Ni(N03)2 in 1 x s which is an appropriate time scale for earlier experimental observations.' The results of this calculation using the data of Criado et u Z .~ for a simple zero- order mechanism with A (frequency factor) = 1 x lo5 s-' and E (activation energy) = 84 kJ mol-' are given in the third column of Table 1 [P (iii)]. The data of Mu and Perlmutter2 yielded higher pressures at every temperature in this calcu- lation. The results still suggest substantial pressure of these gases above the background in the vacuum of the ET-MS instrument at = 600 K.Moreover in the case of hydrated salts dehydration reactions2 will yield substantial pressures of H20(g) which will disrupt the sample at temperatures even lower than that of the nitrate decomposition. It is noteworthy that in all cases specified by L'vov as hydrated salts the appearance temperature of the MO' ions was lower than with the unhydrated salts.429 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY MARCH 1994 VOL. 9 Results A common feature of the reports on MO(g) in ET-MS is the use of a high heating rate (z 500 K s-I). Nickel can be taken as a typical example of this. In our earliest experiments on ET-MS it was found that an atomization heating pulse applied to the furnace containing a nitric acid solution of nickel without previous drying or thermal pre-treatment gave a transient signal due to NiO' at the beginning of the heating pulse (see Fig.2).17 In subsequent work on cobalt,' the practice of pre-atomization treatment in uucuo at very low heating rates was adopted which eliminated the COO' signal from the early part of the atomization heating cycle. Droessler and Holcombe' observed NiO+ in rapid heating to 625 K accompanied by strong signals for NO2+ and 02+ and a much weaker signal for Ni(N03),' from 500 ng of Ni(N03)2. A repeat of the nickel experiment in the modified cross- beam ET-MS instrument with the ionizing energy at 15eV and using rapid heating (500 K s-') without pre-atomization treatment showed no NiO' although NO,' and 02+ were observed. An important consideration in comparing this experi- ment with the earlier work is the ionizing energy used.Since the earliest work the practice of using low ionizing energies ( z 15 eV) has been adopted in order to emphasize the signals owing to the ionization of atoms or to molecular ions over those of fragment ions from molecules and also to suppress background signals owing to the mass spectrum of the diffusion pump oil. At 15 eV any significant amount of NiO(g) emitted from the furnace should be detected as has been seen in other cases such as that of GeO(g) from Ge.I5 On the other hand at the ionizing energies used in our early work on nickel (50eV) and in that of Droessler and Holcombe' (70eV) substantial fragmentation of molecules such as Ni(NO,),(g) should occur and the weak Ni(NO,),+ observed by Droessler and Holcombe could be the parent ion in a 70eV mass spectrum containing a prominent fragment NiO+.In the present work Ni(N03)2+ at 15 eV was not observed. In view of the simultaneous detection of NO,(g) and 02(g) the earlier work could be interpreted as having detected not NiO(g) but a nickel nitrate species [which will be called NiO(s) for convenience] introduced into the vacuum during the decompo- sition Ni(N03)2-,NiO(s)+ 2N02(g)+0.50,(g) The most investigated example of this metal oxide-metal nitrate phenomenon is that of lead,6y7g8 where PbO' was observed at z 600 K during rapid heating. Bass and Holcombe7 Ni' also observed that no PbO+ was produced when PbO(s) was heated in the furnace. They concluded therefore that the PbO' was associated with the nitrate decomposition and was due to PbO being released from the surface during the associated crystal rearrangement and generation of gaseous products.They did not specify the state of the PbO thus released. In a similar experiment on a lead nitrate sample in the cross-beam ET-MS system used in the present work at 15 eV ionizing energy no PbO+ was observed. Wang et ul. made very similar observations for Cu(N03) and in addition noted that the CuO + signal was substantially reduced by charring the deposited nitrate at 523 K under 1 atm pressure. This last observation is in keeping with the result of the effect of a low pre-atomization heating on the appearance of COO +.' In the cross-beam ET-MS instrument when rapid heating of samples of CU(NO,)~ and Co(N03) was carried out without pre-atomization treatment neither CuO + or Cu,O + nor COO+ were detected at 15 eV ionizing energy in contrast to the earlier work4y9 from apparatus in which the atomizer was situated on the axis of the QPMS system.The non- detection of these oxide ions in cross-beam ET-MS stands in contrast to the case of ions from volatile oxides of known thermochemical stability such as Ga,O(g) In,O(g) and As,O,(g) which have been detected in the 'on-axis' ET-MS instrument.18 A repeat of these earlier experiments on gallium indium and arsenic using the modified cross-beam ET-MS system found all the oxide species associated with these systems which were predicted by free energy minimization calculations. Experiments were also carried out in the cross-beam ET-MS instrument on the remaining two systems specified by L'vov Mg(N0,)2 and NH,H,PO,.In the case of magnesium no MgO+ or Mg(OH),+ ions were detected. This is however a less relevant case than the other nitrates as the original experiments of Styris and Redfieldlg were carried out on their atmospheric pressure sampling ET-MS instrument and clus- tering of species could have occurred in the expansion inside the sampling cones Mg(OH),+ for example could be an ionized cluster MgO-H20. The case of NH,H,PO is particu- larly interesting because it is the only species investigated by L'vov for which a straightforward calculation of equilibrium composition without restriction of products predicts the gener- ation of a substantial amount of gaseous oxide P4010(g) during thermal decomposition.The original work of Bass and Holcombe,' however found not P4010+ but PO2+ and a range of unspecified heavier oxyphosphorus ions. The mass spectrum of vapours from phosphorus oxides is known to be complex." NiO+ Fig. 2 of Ni(NO,) deposited in a graphite furnace Mass spectrum howing Ni' and NiO+ peaks at m/z 58 60 74 and 76 taken at 50 eV ionizing energy during rapid heating of a sample430 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY MARCH 1994 VOL. 9 In the cross-beam ET-MS system for NH4H2P04 PO2+ was detected at the beginning of heating and P2+ at ~1300°C much later in the heating pulse. This latter ion is presumably due to P2(g) or P,(g) from reduction of the phosphorus oxides by C analogous to the As2+ observed in the ET-MS of arsenic." Discussion The new evidence from the cross-beam ET-MS experiments casts serious doubt on the proposition that reaction (2) is occuring in ETAAS of nitrate solutions.Genuine gaseous diatomic oxides such as NiO(g) CoO(g) CuO(g) or PbO(g) should be detected by the cross-beam apparatus as have been GeO(g),15 the triatomic oxides Ga20(g) and In20(g) and the polyatomic oxide As,06(g).18 What then is the source of the MO+ ions observed in the experiments with axial placement of the atomizer? In the case of nickel the work of Droessler and Holcombe' suggested that the precursor could be a nickel nitrate species but if it is Ni(NO,),(g) the polyatomic ion Ni(N03)2f must be relatively unstable as it was not detected by cross-beam ET-MS.In their work on lead Bass and Holcombe suggested that PbO (in an unspecified state) was released from the surface of the atomizer during the nitrate decomposition whereas heating a sample of PbO(s) in the atomizer showed no PbO'. It could be that disruption of the crystal structure of Pb(N03)2 by the rapid evolution of NO,(g) during rapid heating of the atomizer is spraying PbO off the atomizer surface in the form of large clusters. At a pressure of 1 atm in ETAAS these would rapidly condense again on the atomizer surface but in the vacuum of the axial ET-MS system they might be projected into the ion source by the force of the expanding NO,(g) and having trajectories close to the axis of the QPMS instrument their product ions would also move along the axis of the QPMS and be detected.The observation by Wang et aL4 and in earlier work in this laboratoryg that pre-treatment by low heating rates reduces or eliminates the oxide signal suggests that such pre-treatment gives a less violent decomposition of the nitrate in which substantially less oxide is sprayed into the vacuum of the ET-MS system. In cross-beam ET-MS the clusters produced in the rapid decomposition that entered the ion source would have been accelerated at right angles to the axis of the QPMS instrument by the force of exothermic decomposition of the nitrate. The velocity of the product NiO' ion again at right angles to the quadrupole axis would be substantially higher than that of an ion originating from a gaseous oxide molecule with a normal gas kinetic velocity produced by an equilibrium process at 500-600 K.Once this velocity reaches about 1 x 10' cm s-' it is of the same order as that required by an ion of rn/z=74 moving in an arc of radius 0.3 cm under the influence of a draw-out field of lOVcm-' values which are reasonable criteria for the successful extraction of an ion into the quadru- pole analyser in this apparatus. These clusters need not be solely of oxide molecules but could be a mixture of reactant nitrate and product oxide for which the dominant ion in the ET-MS instrument is MO'. There is scope here for further experiments with more sophisticated ion sources offering higher draw-out potentials than were available here in order to detect these product MO + ions and estimate their velocities.1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 References Stern K. H. J. Phys. Chem. Ref. Data 1972 1 747. Mu J. and Perlmutter D. D. Thermochim. Acta 1982 56 253. Criado J. M. Ortega A. and Real C. React. Solids 1987 4 93. Wang P. Majidi V. and Holcombe J. A. Anal. Chem. 1989 59 974. Droessler M. S. and Holcombe J. A. Spectrochim. Acta Part B 1987 42 981. Sturgeon R. E. Mitchell D. F. and Berman S. S. Anal. Chem. 1983,55 1059. Bass D. A. and Holcombe J. A. Anal. Chem. 1987 59 974. Bass D. A. and Holcombe J. A Anal. Chem. 1988 60 578. Ham N. S. and McAllister T. Spectrochim. Acta Part B 1988 43 789. L'vov B. V. Mikrochim. Acta 1991 11 299. McAllister T. Int. J. Mass Spectrom. Ion Proc. 1990 101 127. Turnbull A. G. and Wadsley M. W. The CSIRO Thermochemistry System Version V CSIRO Division of Mineral Products Port Melbourne 1986. Eriksson G. Chem. Scr. 1975 8 100. Styris D. L. Fresenius' 2. Anal. Chem. 1986 323 710. Doidge P. S. and McAllister T. J. Anal. At. Spectrom. 1993 8 409. Wagman D. D. Evans W. H. Parker V. B. Halow I. Bailey S. M. and Schumm R. H. NBS Tech. Note 270-3 and 270-4 US Department of Commerce 1968 and 1969. Ham N. S. and McAllister T. unpublished data. McAllister T. J. Anal. At. Spectrom. 1990 5 171. Styris D. L. and Redfield D. A. Anal. Chem. 1987 59 2891. Muenow D. W. Uy 0. M. and Margrave J. L. J. Inorg. Nucl. Chem. 1970,32 3459. Paper 3/04059C Received July 5 1993 Accepted September 28 1993