Detection limits and surface interactions in quantitative negative chemical-ionization mass spectrometry

0039-9140/84$3.00+ 0.00 Copyright 0 1984Pergamon Press Ltd

Talanra, Vol. 31, No. 1, pp. 5560, 1984 Printed in Great Britain. All rights reserved

DETECTION LIMITS AND SURFACE INTERACTIONS IN QUANTITATIVE NEGATIVE CHEMICAL-IONIZATION MASS SPECTROMETRY ULTRATRACE DETERMINATION OF METALS AND ORGANIC COMPOUNDS BY MEANS OF THEIR COMPLEXES I. K. GREGOR and M. GUILIIAUS School of Chemistry, The University of New South Wales, Kensington, N.S.W., Australia (Received 6 May 1983. Accepted 25 July 1983)

Summary-Details are given of a selective negative-ion mass-spectrometric method appropriate for the ultratrace determination of metals and organic compounds by means of their complexes. Direct introduction of the sample into the ion-source, attachment of low-energy electrons, and selected-ion monitoring are described, and comparative data are given relating to surface effects at the tips of

insertion-probes on detection limits. Detection limits for chromium and cobalt, determined as their tris(2,2,6,6-tetramethylheptane-3,5dione) chelates, were respectively 1.Oand 0.16 pg, and that for nickel [as its bis(N,Ndiethyldith@arbamate) complex] was 1.0 pg. Detection limits of 2.0 and 1.0 ng are attainable for malathion and ethion by measurement of the nickel(H) complexes of their O,O’-dialkyldithiophosphate

hydrolysis products.

We have already drawn attention to some of the advantages and the potential afforded by the method of negative-ion mass spectrometry for the determination of organometallic compounds.’ Since then there have been significant developments in the use of electron-attachment as a “soft” ionization technique under so-called negative chemical-ionization (NCI) conditions.“’ For molecules possessing positive electron-affinity, attachment of low-energy electrons can result in the formation of long-lived negative molecular ions capable of being detected and selectively monitored for quantitative purposes.7-‘0 Electronic and structural criteria for gas-phase electron capture by organometallic substrates, as well as the function of various gases as electron-energy moderators in such reactions, have recently been more precisely defined.“-I8 Moreover, there have been considerable advances made in methods for quantitative introduction of solid samples of low volatility or thermal lability into mass-spectrometer ionization sources.‘e22 Previous mass-spectrometric determinations of metals (as various metal chelates) have been based largely on either electron-impact (EI) or positive-ion chemical ionization.2s27 Under these conditions very low instrumental background contributions are essential if the analytes are to be detected at ultratrace levels. However, because the normal mass-spectrometer background species are relatively immune to electron attachment, such ioncurrent contributions are usually minimal under NC1 conditions.6s28s29Introduction of gas-chromatography fractions and attachment of electrons under NC1 ion-source conditions has been reported for certain classes of metal chelates.30,3’ In general, this method

has several drawbacks, including the nonquantitative introduction of sample into the ionsource, the thermal lability of various metal chelates, and interaction of the chelates with active sites and surfaces prior to entry into the ionization chamber.32 In this paper we present details and results of a selective method for ultratrace determination of metal chelates introduced directly into the ion-source under standardized electron-capture NC1 conditions. The utility of the technique is demonstrated for the biologically significant metals chromium, cobalt and nickel, and by its extension to analysis for malathion and ethion by formation and determination of the nickel derivatives of their (RO)2PS; hydrolysis products. The analytes tested are listed in Table 1. EXPERIMENTAL Instrumentation

Negative-ion mass spectra were obtained and quantitative measurements made on a VG MM- 16F single-focusing mass spectrometer fitted with a combined EIjCI source, under conditions described previously”-‘4,‘8 or specified below (Tables 24). Methane (Matheson UHP, 99.97% pure) was used as the electron-energy moderator under measured and controlled ion-source pressure (MKS Baratron system, type

170/315BHS-10) and temperature conditions to achieve optimum sensitivity for molecular-ion signals at the collector.18Detection-limit measurements were made by performing repetitive 0.1~setintegrating scans of a selected ion (usually the molecular ion). Thus the signal derived from the collector by the d.c. amplifier and a sample-and-hold circuit could be recorded during the complete evaporation of a quantity of sample within the ion-source, usually within about 30sec. The signal recorded was Gaussian in shape and corresponded to an evaporation profile.13 Samples were introduced quantitatively into the ion-source either in fused silica cups (in some cases silaned and sometimes also coated 55

56

K. GRECORand M.

I.

with SE-30) or on dished Vespel@ rods positioned in tapered Vespel@ probe tips which ensured a tight seal in, and electrical insulation from, the ion-source block.2’,22The desorption chemical ionization (DCI) method of sample introduction was also used;i9 this involved rapid evaporation of sample within the methane plasma from a platinum wire, heated by the passage of a ramp-programmed current supplied by a VG type MR609 sourceheater/emitter-current control unit. Details of our DC1 filament, Vespel@ probe tip and the application of this technique to metal chelates have already been reported.i4 Procedures Analyte solutions of known concentration were delivered by microsyringe in precise volumes of 1-3 ~1 to the solid insertion-probe cups, Vespelm rod or DC1 filament wire, and the solvent was evaporated in the vacuum lock for the solid insertion-probe. In this way, small accurately known amounts of samples deposited on fused silica, silaned or SE-30 coated silica,22Vespel@ or platinum, were introduced quantitatively into the ionization-source methane plasma by the “in-beam” method. I9 Details of typical plasma temperatures and sample heating-rate data have already been given.14 The evaporation profile data acquired during the evaporation and ionization of the sample were used to determine the total monitored-ion charge produced by a given mass of sample introduced into the ion source (G/r g).” Materials and reagents The derivatives analysed were prepared and purified by established methods, and their purity was checked by positive-ion and negative-ion mass spectrometry.‘“r’ Solutions of the analytes were prepared by dilution of stock solutions of accurately known concentration, made with n-hexane, chloroform or methanol, as appropriate. Glassware was cleaned by procedures similar to those already described.‘* RESULTS AND

DISCUSSION

Ultratrace metal determination by chelation and NCI mass spectrometry As speciation of metals in the environment is highly important, analytical methods are needed that not only have high sensitivity, but can also indicate the metal oxidation state and co-ordination number.3’4’ An analytical procedure based on a “soft” ionization method is well suited for this purpose. Both proton-transfer CI and NC1 give less fragmentation of quasimolecular, [M + l] + , and molecular ions, [M] -, than electron-impact ionization does. However, many metal chelates can be regarded as Lewis acids with the metal functioning as the active Table

1. List of analytes

Lewis acid site42 and inhibiting proton attachment. Negative ionization, involving attachment of lowenergy electrons in metal orbitals of suitable energy, can be seen as a more favourable ionization process, particularly as many metal chelates have been observed to undergo electron-attachment reactions leading to the formation of stable long-lived negative molecular ions.“-‘5*3h3’ Though it is known that fluorinated metal chelates may yield between 10 and 5000 times more negative-ion current when present as [Mlthan positive-ion current when present as has been made [M+ II+, no systematic comparison of the sensitivity given by the three most common methods of ionization modes (EI, CI and NCI).30.3’ Table 1 lists the metal derivatives which we have determined quantitatively by direct sample introduction, electron-capture NCI, and mass spectrometry. The biologically significant metals, chromium and cobalt, in their tervalent states, were determined in the form of their tris(2,2,6,6tetramethylheptane-3,5-dione) chelates, M(dpm),, which give long-lived molecular anions on electron attachment.43 Nickel is best determined as its bis(N,N-diethyldithiocarbamate) complex, Ni(Et2dtc),.14

Ultratrace analysis by metal complexation and NCI mass spectrometry Formation of complexes of metals and various organic analytes, coupled with solvent extraction and spectrophotometric analysis, is a commonplace procedure. The potential specificity and sensitivity afforded by negative chemical-ionization mass spectrometry has not so far been utilized in this way, and one of our aims was to evaluate the combination of metal derivative formation and NC1 mass spectrometry. Ethion is commonly determined spectrophotometrically by hydrolysis and subsequent formation of the bis(O,O’-diethyldithiophosphate)copper(II) complex, with a practical limit of detection of ca. 25 pg of analyte.“x4 A similar method is used for malathion, the corresponding O,O’-dimethyldithiophosphate complex being formed. These spectrophotometric methods are not specific for the individual O,O’-dialkyldithiophosphate ligands and thus cannot be specific for the parent pesticides. 37 Negative-ion mass spectra have and their derivatives

Analvte Chromium(III) Cobalt(II1) Nickel(I1) Malathion* Ethiont

Derivative Cr[(CH3),C~CO~CH~COC(CH,),I,; Co[(CH,),C.CO.CH.CO.C(CH,),I,; Ni[S,~CN~(C,H&],; Ni(Et,dtc), Ni[S2.P(OCH,)J,; Ni(Me,dtp)& Ni[S,.P(OC,H,)J,; Ni(Et,dtp)&

CH2.C02C2H, *Malathion: (CH,O),. i

GUILHAUS

’ .CGrCrH,. S.CH

tEthion: [(C,H,O),. i .S] .CH SDerivatives of the hydrolysis iroduct.

Cr(dpm), Co(dpm),

Negative chemical-ionization mass spectrometry Table 2. Effects of sample-surface

interactions on evaporation profiles

Surface

Charge, C*

Vespel” Silaned silica/SE-30 Untreated fused silica

5.5 x lo- I0 2.0 x lo- I0 4.0 x lo-”

Heated olatinum wiret

4x 10-12

Precision @SD), % 8 8 20 60

Profile shape Symmetrical Symmetrical degular with tailing Irreeular

*Charge obtained as negative molecular ions from a bisfN.N-dietbvldithiocarbamato~nickel~I1) containing 1 ng of nickel. tDesor$on chemical ionization. ’

been obtained for many organophosphorus pesticides28345as well as for metal chelates formed from analogous O-O’-dialkyldithiophosphate ligands,15 those for the former being characterized by a degree of fragmentation which is excessive for precise selective-ion monitoring, whereas those for the latter display little fragmentation under the optimum ion-source conditions. With such metal derivatives the specificity afforded by negative-ionization mass spectrometry can be used to identify (and determine) unequivocally the metal chelate, the chelating ligand and hence the original pesticide. In this work, the metal complexes Ni(Me,dtp), and Ni(Et,dtp), were selected as models for determination of malathion and ethion because for these compounds, under controlled conditons, ca. 90% of the total ion-current is carried by [Ml- (the ion selected for monitoring).46 These two metal complexes are known to be Lewis acids,47*48 and this property serves to make their cross-sections for electron capture larger than those of the corresponding parent organic analytes.

sample

of

This derivative was chosen because it has vacant co-ordination sites which could interact with active sites on the various surfaces. The results are summarized in Table 2, and show that the most suitable surface for this chelate is provided by Vespel@. The SE-30 coated surface is also effective, but is known to have an upper temperature limitation.22 A considerable loss of sensitivity and precision, as well as tailing of the evaporation profiles, was apparent in the results for uncoated silica surfaces. The low sensitivity and poor reproducibility given by the platinum DC1 filament wires is consistent with previously reported results obtained with organic analytes at the ng level.‘9s50 Detection limits

Detection limits were determined for the analytes listed in Table 1, by introducing into the ion-source successively smaller amounts of their derivatives, on a dished VespeP’ rod. The detection limit was taken to be the smallest amount of analyte giving an evaporation profile distinguishable from that obtained when only pure solvent was delivered to the Surface efects and their influence on the sensitivity rod. The results are given in Table 3 and the caliWhen the amount of sample introduced on the bration graphs are given in Fig. 1. solid insertion-probe approaches that required to A major limitation to the realization of lower form a monolayer on the surface used (typically 0.1 detection limits is the intensity of the background nmole), the sensitivity can be severely affected by spectrum. Although the negative-ion background is interaction between the surface and the sample.22s49 of lower intensity than that for positive ions, it is Significant improvements in sensitivity as well as less significant when the gain of the detection circuitry is decomposition of thermally labile compounds have very high. The abundance of background ions in the been reported to result from the use of inert surfaces VG mass spectrometer used in this work decreases at for direct sample introduction.2’.22 Such methods m/z > 300 and therefore the degree of background involve silaning glass surfaces (and sometimes then coating the surface with the non-polar gaschromatography stationary phase SE-30). Vespel@ Table 3. Detection limits determined by [Ml- monitoring under has been used to an increasing degree as a chemically NC1 conditions inert surface for direct introduction of thermally Source Detection labile compounds. 21The DC1 method, whereby samtemperature,5 limit, Analytet “C ‘Z Pg ples are desorbed within the ion-source from heated Chromium(III) 601 1.0 170 wires, has also been used for compounds that are Cobalt(II1) 608 0.2 170 thermally labile or difficult to volatilize.‘9*49However, 354 1.0 Nickel(I1) 240 Malathion 372 2000 160 no systematic evaluation of these surfaces has been Ethion 428 160 1000 made for metal chelate analysis. *Methane as electron-energy moderating gas (0.10 mmHg), primary To assess the suitability of the surfaces chosen in electron energy 50 eV, emission current 1.00 mA (total), accelerthis work for ultratrace analysis, the average charge ating voltage 4.0 kV, repeller voltage 0 V. carried by negative molecular ions resulting from tAnalyte derivatives introduced on a Vespel@rod into the ionization chamber. of 1 ng of nickel [as #Plasma replicate introduction temperatures were approximately 90% of the source temNi(Et,dtc),] on the various surfaces was measured. perature.

I. K. GRFSOR and M. GUILHAUS

58

I -13 ,o-12

,o-ll

I ,o-lo

I

Icq

,o-9

,

,

,

,

(

,o-13 ,o-12,o-11,o-Io,o-9 ANALYTE

(g)

‘,

-IO ,o-9 IO

I

,d6

I

,o-7

I

,66

-

Fig. 1. Calibration graphs. A: Co(III) (0) and Cr(II1) (0) as their M[dpm], derivatives. B: Ni(I1) (0) as Ni[S,CN. Et,],, C:Malathion (A) and ethion (A) as the Ni[(RO),PSJ, derivatives; each point represents the mean charge detected for triplicate sample introductions, for which the worst r.s.d. was 20%.

interference in the detection of very low levels of analytes is decreased when the derivatives have high masses. It is therefore advantageous to use metal chelates with molecular weights > 400. Background interference becomes apparent when the probe is inserted. When the probe makes contact with and seals the ionization chamber, the pressure in the chamber rises from about 0.01 to 0.10 mmHg (with methane as the electron-energy moderating gas) and this is accompanied by a rise in sensitivity, resulting from the corresponding increase in the thermalelectron concentration in the ionization chamber. If the gain of the detection system is high, then the rise in sensitivity is apparent as a rise in the background ion-current to a plateau representing equilibration of conditions within the ionization chamber. This is illustrated in Fig. 2 which reproduces the evaporation profile obtained on delivery of 1 ~1 of pure solvent (blank) to the Vespel@ rod, with the mass spectrometer tuned to m/z = 60 1. Also shown are profiles for duplicate introduction of 3.3 and 1.0 pg of chromium [as the Cr(dpm), chelate] in the same volume of solvent. In these evaporation profiles the current due to the analyte is superimposed on the rising background signal. The drop in ion-current at the end of the evaporation profile corresponds to the withdrawal of the probe. The detection limit of 1.0 pg obtained here for chromium is two orders of mag-

LL

nitude better than that reported for a similar technique. 3o The use (in that work) of untreated glass capillaries for sample insertion could have contributed significantly to lowering the sensitivity. Further development of our technique, to minimize rising background signals and thereby produce an improvement in detection limits, could involve either temperature-programming of the probe tip*’ or the use of polyimide-coated DC1 filament wires to eliminate direct interaction between the sample and the wire.5’ At higher resolution some reduction of background interference can be accomplished at the expense of sensitivity. 33 A single-focusing magneticsector mass spectrometer with a maximum resolution of 2000 can accomplish this to a limited extent, and only in the low m/z range. In these experiments, a resolution of 1800 (10% valley definition) was used for all measurements except those for the Ni(Mezdtp)*, Ni(Et*dtp), and Co(dpm), derivatives, for which a resolution of 400 was used. Use of the higher resolution reduced the overall sensitivity by 90% and necessitated more frequent tuning of the mass spectrometer to the top of the narrower peaks. In some cases the detection limit was governed by carry-over of small traces of the sample derivative in the mass-spectrometer probe inlet-lock, with atten-

I

I TIME

Fig. 2. Recorder responses for the introduction trace amounts of Cr(II1) as Cr[dpm],.

TIME

of ultra-

Fig. 3. Recorder responses for the introduction of subpicogram

amounts

of Co(II1)

as Co[dpm],.

Negative chemical-ionization mass spectrometry Table 4. Relative sensitivitv of the EI. CI and NC1 modes. Relative sensitivity,t Cl@ Analyte* Chromium(II1) Cobalt(III) Nickel(I1) Malathion Ethion

CI§

EIf

1 1 1 1

6

1

ii

7

IO9

NCIll 350 750 3 x IO’ 124 71

‘AnaIyte derivatives as in Table 1. tInstrumenta1 conditions adjusted to optimize sensitivity in each mode. BReagent gas methane (0.05 mmHg) primary electron energy 200 eV, total emission current 0.500 mA, accelerating voltage 4.00 kV, repeller voltage 0 V. SPrimary electron energy 70 eV, trap current 0.050 mA, accelerating voltage 4.00 kV, repeller voltage 0 V. (ITotal electron emission current 0.500 mA, all other instrumental conditions as given in Table 3.

dant superimposition on the background signal for the blank. This is illustrated in Fig. 3 for cobalt at the sub-picogram level, and is a potentially limiting factor for significantly volatile analytes at these ultratrace levels. Nevertheless the extremely low detection limits obtained here for cobalt illustrate the high sensitivity of the NCI/direct sample-introduction technique. The detection limits for the nickel(U) derivatives of the ethion and malathion hydrolysis products were higher than those for the other analytes in this work, yet they still represent a significant improvement (by a factor of about 1000) on those attainable by the method.37*” Our spectrophotometric standard method offers a sensitive procedure for a series of similar organophosphorus4’ pesticides capable of similarly forming metal derivatives. Further refinement of this approach, coupled with use of ion-counting methods for negative-ion detection, should result in still lower detection limits. Comparative

sensitivity

of the EI, CI and NCI modes

Comparative measurements were made by separate introduction of 100 ng of each derivative under optimum conditions for each ionization mode. The total charge detected from molecular or quasimolecular ions was measured and the relative sensitivities determined (Table 4). The NC1 mode was found to offer the greatest sensitivity for all the analytes tested, except Ni(Et2dtp)2. The sensitivity is better than that obtained by the CI mode, by factors in the range 100-1000, and this is consistent with predictions from a kinetic model for positive and negative chemicalionization.52 The high enhancement for the nickel chelate may be a reflection of a very low tendency for the complex to undergo a proton-attachment reaction, which is consistent with the known chemistry for such a square planar complex with a transition metal as the Lewis acid centre.42 It is noteworthy that the relative sensitivities determined for the EI mode are all higher than the corresponding positive-ion CI sensitivities. This is

59

probably due, at least in part, to the larger ion-source exit-slit area used. Acknowledgement-Support of this work by the Australian Research Grants Scheme is gratefully acknowledged. REFERENCFS

1. D. R. Daktemieks, I. W. Fraser, J. L. Gamett and I. K. Gregor, Tulantu, 1976, 23, 701. 2. K. R. Jennings, Specialist Periodical Reports, Vol. 4, Mass Spectrometry, R. A. W. Johnstone, Senior Reporter, p. 203. The Chemical Society, London, 1977. 3. D. F. Hunt and F. W. Crow, Anal. Chem.. 1978. SO, 1781. 4. I&m, U.S. Natl. Bur. St&. Spec. Publ., 1978, 519,601. 5. J. G. Dillard, Biomedical AuDlications of Mass Spectrometry, Fir.& Suppl. Vol.,-& R. Wall& and 6. C. Dermer (eds.), p. 927. Wiley-Interscience, New York, 1980. 6. R. C. Dougherty, Anal. Chem., 1981, 53, 625A. 7. H. Budzikiewicz, Angew. Chem. Intern. Ed. Engl., 1981, 20, 624. 8. J. H. Bowie and B. D. Williams, in MTP International Review Science, Physical Chemistry, Series 2. Vol. 5. Mass Spectrometry, A. Macwll (ed.), p. 89. Butter. _ worths,-London, i975. 9. L. G. Christonhorou. Adv. Electron. Electron Phvs.. . , 1978, 46, 55. 10. Idem, Environ. Health Perspect., 1980, 36, 3. 11. J. L. Game& I. K. Gregor, M. Guilhaus and D. R. Daktemieks, Inorg. Chim. Acta, 1980, 44, L121. 12. D. R. Daktemieks, I. W. Fraser, J. L. Gamett and I. K. Gregor, Org. Mass Spectrom., 1980, 15, 556. 13. P. L. Beaumont, J. L. Gamett and I. K. Gregor, Znorg. Chim. Acta, 1980, 45, L99. 14. I. K. Gregor and M. Guilhaus, Org. Mass Spectrom., 1982, 17, 575. 15. N. B. H. Henis, K. L. Busch and M. M. Bursey, fnorg. Chim. Acta, 1981, 53, L31. 16. J. E. Szulejko, I. Howe, J. H. Beynon and U. P. Schlunegger, Org. Mass Spectrom., 1980, 15, 263. 17. P. M. George and J. L. Beauchamp, J. Chem. Phvs. 1982, 76, 2959. 18. I. K. Gregor and M. Guilhaus, Abstr. Papers, 3Oth,Ann. Conf. Am. Sot. Mass Spectrom. Allied Topics, Honolulu, 1982, 141. 19. A. P. Bruins, Anal. Chem., 1980, 52, 605. 20. U. Rapp, G. Dielmann, D. E. Games, J. L. Gower and E. Lewis. Adv. Mass Soectrom., 1980, 8. 1660. 21. R. J. Goiter, Anal. Chem., 1980, 52, i589A. 22. J. P. Thenot, J. Nowlin, D. I. Carroll, F. E. Montgomery and E. C. Homing, ibid., 1979, 51, 1101. 23. A. E. Jenkins and J. R. Majer, Talanta, 1967, 14, 777. 24. M. G. Alcock, R. Belcher, J. R. Majer and R. Perry, Anal. Chem., 1970, 42, 776. 25. J. R. Majer and A. A. Boulton, Methods Biochem. Anal., 1973, 21, 467. 26. J. R. Majer, Taluntu, 1972, 19, 589. 27. T. H. Risby, P. C. Jurs, F. W. Lampe and A. L. Yergy, Anal. Chem.. 1974, 46. 726. 28. R. C. Dougherty hd.E. A. Hett, Environ. Sci. Res., 1978, 12, 339. 29. R. C. Dougherty, Biomed. Mass Spectrom., 1981,8,283. 30. S. Prescott, J. E. Campana and T. H. Risby, Anal. Chem., 1977, 49, 1501. 31. T. H. Risby, Environ. Health Perspect., 1980, 36, 39. 32 T. H. Risby, L. R. Field, F. J. Yang and S. P. Cram, Anal. Chem., 1982, 54,410R. 33. B. J. Millard, Quantitative Mass Spectrometry, p. 100. Heyden, London, 1978. 34. G. S. Hammond, D. C. Nonbebel and C. S. Wu, Inorg. Chem., 1973, 2, 73.

I. K. GREC~R and M. G~~LHAUS

60

35. G. D. Thorn and R. A. Ludwig, The Dithiocarbamates and Related Compounds, Elsevier, New York, 1962. 36. L. Malatesta and R. Pizzotti, Chim. Ind. Milan, 1945,

21, 6. 37. J. R. Wasson, G. M. Woltermann and H. J. Stoklosa, Fortschr. Chem. Forsch., 1973, 35, 65. 38. B. A. Davis, K. S. Hui and A. A. Boulton, Advances in Mass Spectrometry in Biochemistry and Medicine, Vol.

II, p. 405. Spectrum Publications, London, 1976. 39. J. J. Dulka and T. H. Risby, Anal. Chem., 1976, 48, 640A. 40. G. B. Morgan and E. W. Bretthauer, ibid., 1977, 49, 1210A. 41. T. H. Risby, Natl. Bur. Stand. U.S. Spec. Publ., 1981, 618, 120. 42. D. P. Graddon, Coord. Chem. Rev., 1969, 4, 1. 43. J. L. Garnett, I. K. Gregor and M. Guilhaus, Org. Mass Spectrom., 1978, 13, 59.

44. J. R. Graham, Analytical Methocis for Pesticides, Plant Growth Regulators and Food Additives, G. Zweig (ed.), Vol. II, p. 223. Academic Press, New York, 1964. 45. H.-J. Stan and G. Kellner, Biomed. Mass Spectrom., 1982, 9, 483. 46. P. L. Beaumont and I. K. Gregor, unpublished results. 47. D. R. Daktemieks and D. P. Graddon, Aust. I. Chem.,

1971, 24, 2509. 48. S. E. Livingstone and A. E. Mihkelson, Inorg. Chem., 1970, 9, 2545. 49. K. S. Webb, B. J. Wood and R. Davis, Adv. Mass Spectrom., 1980, 8B, 1921. 50. A. P. Bruins, Biomed. Mass Spectrom., 1981, 8, 31. 51. V. N. Reinhold and S. A. Carr. Anal. Chem.. 1982. 54. 499. 52. M. W. Siegel in Practical Spectroscopy, Vol. 3, Mass Spectrometry, Part B, C. Merritt, Jr. and C. N. McEwen (eds.), p. 297. Dekker, New York, 1980.