Source of Analysis of Standard Materials Sample
Source
Minerals
Sample
Source
Minerals
1. Rhodenite, sanidine 2. Albite 3. Rock glass (R&S) 4 . Anorthite, olivine, orthopyroxene (camperdown) 5. Andesine, clinopyroxene. orthopyroxene (AC-362) 6. Scapolite
7. Acmite, jadeite, oligociase k-feldspar, chromite, magnetite, ilmenite
A . Albee, California Institute of Technology Deer et ai. (57) Flanagan and Fleischer (52) K. Fredriksson and B. Mason Smithsonian Institution (53)
K. Kiel, University of New Mexico
J. V. Smith, University of Chicago ( 5 4 ) K. Snetsinger, Ames Research Center, NASA
ACKNOWLEDGMENT W e would l i k e t o t h a n k t h e following people for providi n g u s w i t h samples o f well-analyzed standard materials: A. Albee, California I n s t i t u t e of Technology; W. Angelotti, A p p l i e d Research Laboratories; W. Elkington, Teledyne; R. D a h l q u i s t a n d H. Dryer, A p p l i e d Research L a b o (51) W. A. Deer, R. A. Howie, and J. Zussman, "An introduction to the Rock-Forming Minerals," Wiley, New York, N.Y., 1966, p 324. (52) F. J. Flanagan, Geochim. Cosmochim. Acta, 33, 81 (1969); M. Fleischer. ibid, 33, 65 (1969); J. C. Ruckiidge, F. G. F. Gibb, J. J. Fawcett, and E. L. Gasparrini. ibid, 34, 247 (1970). (53) B . Mason and A . L. Graham, Smithson. Contrib. Earth Sci., 3 (1970). (54) 6. W. Evans. Contrib. Mineral. Petrol., 24, 293 (1969); D . M. Shaw. J. Petroiogy. 1 , 261 (1960).
8. Amphibole, garnet
J. Stout, University of
9. CulTe
Minnesota W. Wise, University of California, Santa Barbara
Metals 10. Brass
11. Zircaloy 12. Monels, inconels 13. All SRM samples 14. "T" stainless steel 15. "B" low alloy steel, waspalloy, haspal-
Applied Research Laboratories W. Elkington, Teledyne International Nickel Company National Bureau of Standards (55) A. P. von Rosentiel, TNO Source of analysis restricted
l0Y
ratories; K. Fredriksson, Smithsonian I n s t i t u t i o n ; S. H a r t , Carnegie I n s t i t u t i o n ; K. Heinrich, N a t i o n a l B u r e a u of Standards; K. K e i l , University of N e w Mexico; A. Reid, NASA, Houston; A. P. v o n Rosentiel, Metaalinstituut TNO; J. Rucklidge, University of Toronto; J. V. Smith, University of Chicago; K. Snetsinger, NASA, Ames; J. Stout, University of Minnesota; a n d W. Wise, University of California, Santa Barbara. Received for review October 5, 1972. Accepted February
16, 1973. (55) Nat. Bur. Stand. (U.S.) Spec. Publ., 260, U.S. Dept. Commerce. 1968 and 1970 ed.
Optimization of Mass Spectrometer Ion Yields by Electron Energy Selection and Cryogenic Trapping of Contaminants for the Pulse-Heated Vaporization of Metal Atoms J. M. Freese, A. W. Lynch. and R . T. Meyer Sandia Laborafories. Albuquerque. N.M. 871 15
The detection and measurement by time resolved mass spectrometry of vaporized metal atoms produced in the pulse heating of pure metals and refractory metal c o m pounds has been improved by t w o new modifications to the standard operations of a time-of-flight mass spectrometer which can increase the instrument response for metal vapor species by as m u c h as three orders of magnitude. The improvements are: cryogenic trapping of evolved contaminants both in the d c and pulsed-preheat outgassing of samples and in the pulse-heated vaporization experiments: and selection of the energy of the ionizing electrons to maximize the ratio of metal vapor ions to contaminant ions. The new techniques were demonstrated with the vaporization of magnesium. 1438
ANALYTICAL CHEMISTRY, VOL. 45, NO. 8, JULY 1973
The vaporization of metals a n d m e t a l compounds under pulse-heating conditions is currently of i n t e r e s t in order t o compare t h e vapor compositions w i t h those measured under conditions of steady-state heating a n d thermodyn a m i c equilibrium. A comprehensive review of these investigations has b e e n presented b y K n o x ( 1 ) . In some studies, however, the quantitative mass spectrometric measurements of vapor species generated b y pulsed-resist i v e heating or pulsed-laser heating of refractory materials have been hindered b y t h e simultaneous release of organic contaminants (2-5). The q u a n t i t y of organic vapor was E. E. Knox in "Dynamic Mass Spectrometry,' Vol. 2: Heydon and Son, Ltd.. London, 1971, pp 61-96. (2) P. D . Zavitsanos. Carbon i 0 x f o r d i . 6, 731 (1968). (1)
D.Price. Ed..
often sufficient to saturate the mass spectrometer ion source, to cause electrical discharges to occur at the specimen surface, and to attentuate the thermal radiation transmitted to temperature sensors (6). Two new modifications to the standard operations of a time-of-flight mass spectrometer have been made in order to eliminate or reduce the amount of contamination and to increase the precision of vapor-pressure measurements for vaporized metal atoms. The two improvements are: cryogenic trapping of evolved contaminants both in the dc and pulsedpreheat outgassing of samples and in the pulse-heated vaporization experiments; and selection of the energy of the ionizing electrons to maximize the ratio of metal vapor ions to contaminant ions. This paper discusses the principles, implementation, and success of the techniques in improving instrument response by three orders of magnitude.
PRINCIPLES The ion yield of neutral atomic or molecular species in a mass spectrometer is dependent in part on its partial pressure in the ion source. However, Meyer has shown that a maximum number of ions (-105) are produced, transmitted, and collected during any one cycle in the pulsed mode of a time-of-flight mass spectrometer (7). Therefore, it does not necessarily follow that increasing the partial pressure of a species will increase its ion yield. For example, if the average ionization efficiency is 1%and if the pulse heating of a metal filament evolves 107 contaminant molecules and metal atoms ( i e . , 1105 ions are produced), then an increase in the evolution of metal neutrals will not increase the ion yield of metal atoms unless the evolution of contaminant molecules is substantially reduced. A reduction of volatile impurities in solid samples, including metals, has commonly been accomplished by either dc or pulse heai;ing of the material under vacuum conditions. Experience has shown, however, that an outgassing temperature is required which is in excess of the temperature achieved in the actual experiment in order to reduce the impurities to an acceptable level (6). In most pulse-heated vaporization experiments, one is interested in obtaining very high temperatures, higher than can be readily obtained by steady-state heating. Thus, pulse heating itself must be utilized to outgas the sample. This is a slow, inefficient process because the duration of heating is short, the diffusion of impurities out of the bulk is slow, and both gas-phase collisions and surface collisions of the evolved species lead to recondensation on the sample. Alternately, permanent removal of the evolved species can be accomplished with cryogenic pumping in the immediate region where the impurities are evolved. Low-temperature trapping is particularly successful for the removal of volatiles which have high-vapor pressures. in the free state at normal laboratory temperatures, e . g . , C2H4, C2H2, CHI, Con, H20, etc. Cryogenic pumping of contaminants and substrate vapors also aids in obtaining precise time-resolved measurements of the abundances of the vaporized species by preventing multiple and noncoherent traversals through the ionization zone. Because of the long mean free paths a t the operating pressures in an ion source to 10-7
(3) R. T. Meyer and L. L. Ames in "Mass Spectrometry in Inorganic Chemistry." J . L. Margrave, E d . , American Chemical Society, Washington, D . C . . 1968. p 301 (4)' K . A . Lincoln, lnt. J . Mass Spectrom. /on Phys., 2, 75 (1969). (51 K. A. Lincoln, Anal. Chem.. 37, 541 (1965). (6) R . T. Meyer, paper presented at the 1 7 t h Annual Conference on Mass Spectrometry and Allied Topics, Dallas, Texas, May 1969. (7) R . T. Meyer. J. Sci. lnstrum.. 44, 422 (1967).
Torr), it is possible for a vapor atom or molecule to collide with a surface and pass through the ionization volume more than once during a time-resolved experiment. If a given neutral is not ionized on its first pass, it may be ionized on a subsequent pass. In addition, a neutral which travels completely outside of the ionization volume during its first traversal of the ion source may be reflected through the ionization volume after some surface collision. Therefore, the spectrum measured during a later pass would be a summation of neutrals (and ions) originating from the sample a t varying points in time. Cryogenic pumping prevents multiple traversal of the ionization region by gaseous neutrals by trapping neutrals after their first pass through the ion source. This greatly diminishes the presence of those species which are nontemporally coherent with the ion-injection pulse. Most mass spectrometry has been performed using an ionizing electron energy of 70 eV, because most inorganic gases and hydrocarbon molecules have their maximum ionization efficiency between 50 and 100 eV and because most mass spectral identification tables are based upon 70-eV spectra (8-10). However, most vaporized metal atoms ( e . g . , Na, Mg, Ca, Rb, Sr, Ag, Pb, T1) give maximum ionization efficiencies in the range of 10-50 eV (11, 12). In addition, the ionization cross sections of the metal atoms exhibit sharp decays in magnitude at electron energies above the 10-50 eV range of their peak values. (For a detailed description of the electron-molecule or electronatom interaction mechanisms that account for these effects, see ref. I O and 13.)It should be possible to discriminate against contaminant ions by operation of the electron source at any energy which gives maximum ionization efficiency for the metal but reduced efficiency for the hydrocarbon contaminants.
EXPERIMENTAL The vaporization experiments were conducted with a Bendix Model 14-107 time-of-flight mass spectrometer (14). Operation of the spectrometer in the time-resolved mode for both flash photolysis and capacitor discharge resistive heating experiments has been previously described (3, 7, 15). The trapping of evolved contaminants was accomplished with a new ion source housing designed specifically for the time-of-flight mass spectrometer, which is shown in Figure 1. The housing contains a liquid-nitrogen cryostat which surrounds the sample probe and the electron and ion guns of the spectrometer. The sample is cleaned in situ by preheating (dc or pulsed) to a temperature near but below its melting point. The contaminant vapors evolved during the outgassing and preparative pulse-heating experiments are condensed on the walls of the cryostat housing, thus eliminating their effect on the vaporization experiments. This approach was first demonstrated and reported (6) for carbon samples which were outgassed in a separate, all glass, high-vacuu m chamber prior to their use in mass spectrometry vaporization experiments. During the outgassing treatment by dc heating, localized electrical arcing a t the surface often occurred. The electrical arcing was eliminated by high-temperature bakeout with a n (8) American Petroleum Institute Research Project 44, Catalog of Mass
Spectral Data, Carnegie Institute of Technology, Pittsburgh, Pa.. 1953. (9) D. J . Dunbar and L . A . Harrah, Air Force Materials Laboratory Report No. AFML-TR-64-381,February 1965. (10) M . Krauss and V. H . Dibeler in "Mass Spectrometry of Organic Ions." F. W. McLafferty. E d . , Academic Press, New York, N . Y . , 1963, p 118. (111 L. J. Kieffer, "Compilation of Low Energy Electron Collision Cross Section Data," JlLA Information Center Report, University of Colorado, Boulder, Colorado, Part I , January 10, 1969; Part I I , September 22, 1969. (12) Y. Kaneko,J. Phys. Soc. Jap.. 16, 2288 (1961). (13) F. H . Fieid and J . L. Franklin, "Electron Impact Phenomena and the Properties of Gaseous ions," Academic Press, New York, N . Y . , 1957. (14) D. C. Damoth. Advan. Ana/. Chem. lnstrum.. 4 , 371 (1964). (15) R. T . Meyer, J . Chem. Phys., 46, 967 (1966). A N A L Y T I C A L CHEMISTRY, VOL. 45, NO. 8, J U L Y 1973
1439
LASER BEAM BENDIX TOTMS I O N SOURCE
OBJECTIVE LENS
77 ELECTRON BEAM
S A M P L E - W L D I N G PROBE L I Q U I D (COLD
TRAPS
ELECTRICAL FEEDTHROUCHS FOR DC- OR PULSE-HEATING
New ion source housing for Bendix Model 14-101 time-of-flight mass spectrometer showing cutaway views of liquid-nitrogen cryostat and electron/ion guns
Figure 1.
adjacent liquid-nitrogen trapping surface; however, when these outgassed samples were transferred to the mass spectrometer ion source, they were recontaminated by the laboratory air and the spectrometer background vapors and arced again when heated, Use of the new ion source housing with the liquid-nitrogen cryostat permits a n efficient in situ outgassing and the elimination of electrical arcing. The ratio of metal-ion intensity to contaminant-ion intensity was studied by varying the electron energy from 10 to 80 eV. This change was accomplished on the Bendix spectrometer by changing the potentiometer setting, which controlled the bias voltage applied to the filament of the electron gun. The electron beam trap current was held constant a t 25 p A for those experiments in which the intensity of the mass spectral signals was studied as a function of electron energy for the range 30 to 80 eV. For operation below 30 eV, the trap current was adjusted to 10 pA because 25 p A could not be achieved a t 30 eV. The time-resolved mass spectral records were obtained using Lincoln's three-dimensional intensity-modulated raster technique (16, 17).
RESULTS Figure 26 shows the time-resolved mass spectrum at 70 eV for a resistively pulse-heated magnesium ribbon that had previously been both dc heated and pulse heated without liquid nitrogen in the cryostat. Many high-intensity mass signals due to hydrocarbons were recorded, but mass 24 corresponding to vaporized magnesium atoms was not observed. Figure 2c shows a spectrum containing magnesium vapor (mass 24) and background nitrogen gas (mass 28) when the same magnesium ribbon was dc heated with liquid nitrogen in the cryostat to trap outgassed contaminants. T h e rate of c o n t a m i n a n t release was so low during d c heating t h a t t h e c o n t a m i n a n t signals u e r e not observed. Figures 2d and 2e are the time resolved spectra from the first and fourth pulse heatings after dc outgassing, respectively. A comparison of Figures 26 and 2d with Figure 2e indicated a 10- to 100-fold reduction in the contaminant level due to the procedure of dc and pulse heating with liquid-nitrogen trapping. Furthermore, the mass (16) K. A. Lincoln, Rev. S c i instrum.. 35, 1688 (1964). (17) R. T. Meyer in "Time-of-Flight Mass Spectrometry," D. Price and J. E. Williams, E d . , Pergamon Press, Oxford, 1969. P 61
1440
ANALYTICAL CHEMISTRY, VOL. 45, NO. 8, JULY 1973
24 signal due to magnesium vapor became moderately intense. Thus, an increase in the peak height or ion yield of the metal vapor was achieved by reducing the amount of contaminant released during the pulse-heated vaporization experiment. The effective increase in ion yield for metal-vapor detection permitted a determination of the ionization efficiency as a function of electron energy. Evidence for a significant effect of electron energy on relative ion yields is contained in Figures 2e and 2f which are time-resolved mass-spectra recorded at two different electron energies for the same magnesium sample. Figure 2e was recorded at 70 eV/25 pA and Figure 2f at 22 eV/20 pA (trap current). Although the mass 24 intensity was slightly diminished in Figure 2f, all the other mass signals were greatly reduced. A plot of the measured relative ionization efficiency of Mg (Mg e Mg+ + 2e) is shown in Figure 3. For comparison, a typical ionization efficiency curve for a representative organic molecule, n-butane (9), is reproduced in Figure 4. The Mg+ ion yield increased sharply from 10 to 30 eV and then decreased as the electron energy was further increased to 80 eV; the shape of this curve is in general agreement with that reported by Kaneko for Mg (12). However, the total ionization efficiency for the organic was two times greater at 80 eV than at 30 eV. These data clearly correlate with the results recorded in Figures 2e and 2f, which indicate a large increase at 30 eV in the ratio of Mg+ to Z,I,J+, where I,]+ is ion i of charge j . This effect is demonstrated more graphically in Figure 5 , which shows the mass spectral records for a Mg wire being dc heated at a constant temperature. All mass signals I,]+ in the spectrum except for Mg+ decreased as the electron energy was reduced from 70 to 30 eV; the Mg+ signal increased.
+
-
DISCUSSION Previous studies indicated that the principal source of organic contamination in pulse-heating experiments is the surface and bulk phases of the refractory material itself and not the surfaces of components in the ion source, (3,
MASS lamul
Figure 2. Three-dimensional. time-resolved mass spectra .of a pulse-heated magnesium ribbon Time increases vertically from bottom to top of each Polaroid record at 1 msec/cm (large division): the mass increasesfrom left to right: the signal intensity is approximately proportional lo the brightness Of the vertical traces. (see ref. 16 and 1 7 ) . Ail data recorded at 70 eV except for ( 1 ) . (a) The mass spectrometer background gas. N2 (mass 28). ( b ) The resistive pulse heating 01 an Mg ribbon Which had been previously dc and pulse heated without iiqUid-N? trapping; the start of bright traces corresponds approximately to the start Of the 400-rsec duration electrical pulse: outgassing continues after termination of the pulse since the sample loses energy primarily by radiative cooling. ( c ) DC heating of Mg ribbon (mass 24) to provide Outgassing in the presence 01 the iiquid-Nlr trap. (d) First pulse heating after dc heating: mass 24 is Strong and other masses are diminished in intensity. (e) Fourth pulse heating of Mg ribbon after dc heating. (I) Fifth pulse heating of Mg ribbon using reduced electron energy of 22 ev
6),which is confirmed by the present study. The major improvement in relative ion yield shown here is related to the density profile of the ionizations of the vaporized species which occur in the spectrometer ion source. Specifically, the observed effect is a function of the number of ionizations which occur in that portion of the total ionization volume from which ions are actually extracted into the ion gun for acceleration to the detector. The number of ionizations is a function of both the molecular density and the electron density. Normally, both densities are comparable in magnitude and the total number of ions generated is a linear function of each density. In the present case, the densities are strong functions of position in the ionization volume. The ionizing electrons traverse a line-of-sight path of about 3 cm from filament to trap, hut ions are extracted from only the center 1-cm length of this path. The vaporized metal and outgassed contaminants from the pulse-heated wire expand into the entire ionization volume with a molecular density distribution that is an inverse function of the radial distance (3). At the very high molecular densities molecules/cm3) associated with the transient cloud of contaminant vapors, the number of ionizations per unit length
I
10
20
M
40 50 6(1 ELECTRON WERGY l e V i
70
80
Figure 3. Relative ionization efficiency (intensity in arbitrary units) of Mg vapor as a function of electron energy (ev) for the process Mg + e Mg+ + 2e
-
Data from 10 to 30 eV were recorded at 10 PA trap current, whereas data from 30 to 70 BV were recorded at 25 @A: both Sets were normalized at 30 eV. Data were obtained from a dc-heated magnesium ribbon
I
Figure 4. ionization efticiency curves lor n-butane Parent ion is mess 58; fragment ions are produced by dissociative ionization. Data are those reported by Dunbar and Harrah (ref.9)
ANALYTICAL CHEMISTRY, VOL. 45, NO. 8 . J U L Y 1973
1441
\ ' I
:
MASS iarnul
Mass spectral records of vapor released from a Mg ribbon during dc heating and outgassing as a function of electron energy Figure 5.
Each record S h o w signal intensity (vertical scale) vs. mass (horizontal scale). Masses 24. 25. and 26 are the isotopes of Mg: most Other masses are due to organic-contaminant parent and fragment ions: ( a ) 70, ( b ) 6 0 , ( c ) 50,( d ) 40, ( e ) 35, ( 1 ) 30 eV
along the electron path is not constant as is true for the lower molecular densities (-10'0 molecules/cm3) typical of normal mass spectrometer operations. Rather, the effective electron density falls off along the electron path length because of depletion by molecular ionizations and scattering. Thus, the electron density within the ion-extraction zone (the middle third of the electron beam path length) is reduced and the total number of ionizations for that volume is reduced. The preheating and liquid-nitrogen trapping procedure results in a significantly reduced release of contaminant vapor in the subsequent vaporization experiments and, therefore, a lower total molecular density throughout the ionization volume in those experiments. For the same average electron current, the effective electron density
1442
ANALYTICAL CHEMISTRY, VOL. 45, NO. 8 . JULY 1973
would he expected to be higher in the extraction volume. This permits a higher number of ionizations of the metal vapor, whose molecular density is the same as or possibly higher than for a sample that is not outgassed. Thus, the ratio of metal-vapor ions to all other ions, Le., M+/X,l/+, is greatly increased by virtue of both an increase in M+ and a decrease io Z J z J + . In general, the probability of ionization increases linearly with excess energy above the ionization potential of any species with a maximum occurring a t about 4 to 5 times the ionization potential. The ionization potential itself is determined by the energies associated with the electron orbkals for the atoms and molecules and is most often lower for free atoms than for molecules. A t electron energies higher than the ionization potential, the ionization cross section is determined for molecules by the number of normal modes of vibration, by the density of vibrational levels, and by the thermal distribution among states (10). Since a polyatomic molecule has many more vibrational and electronic states to which the impacting electron energy can be distributed, the maximum in the ionization efficiency curve is typically a t a higher energy than for atoms or simple molecules. Hence, use can he made of the different electron interaction kinetics to optimize the mass spectrometer response for the specific chemical system being studied. By the appropriate selection of the electron energy, the Iatio of magnesium ions to total organic ions has been varied from -10-1 (Figure 2e) to -10 (Figure 2n; this gives an effective increase in relative ion yield of -100fold. When this ion-yield increase is coupled with the 10fold reduction in residual sample contamination, a total improvement factor of 1000-fold is obtained. The result is that quantitative measurements of elemental vapor density as a function of temperature can he obtained in pulseheated vaporization experiments. Significant findings are now being obtained, for example, in studies of the laser vaporization of graphite and various metals (18-20). While the procedures reported herein have been optimized for the detection of metal vapor atoms in time-resolved experiments, the basic concepts of optimizing ion yields by electron-energy selection and by cryogenic trapping can he applied to other chemical systems and to other mass spectrometers. Received for review October 26, 1972. Accepted January 22, 1973. This work was supported by the US. Atomir Energy Commission. ( 1 8 ) R. T.Meyerand A. W. Lynch. High Temp. Sci.. 4,283 (19.72). (191 R. T. Meyer. A. W. Lynch, and J. M. Freese. J . Phys. Chem.. 7 7 , ,lo711 1083 I.".l,. (20) R. T. Meyer. J. M. Freese. and A. W. Lynch. Paper presented at lhe 2 0th Annual Conference on Mass Spectrometry and Allied Topics, C)allas. Texas. June 1972.
