Comparison of polyatomic and atomic primary ... - ACS Publications

Analytical Chemistry 2013 85 (2), 610-639 ... The Journal of Physical Chemistry C 0 (proofing), .... Structure, Energetics, Electronic, and Hydration ...
0 downloads 0 Views 1MB Size
Anal. Chem. 1989, 61, 1087-1093

1087

Comparison of Polyatomic and Atomic Primary Beams for Secondary Ion Mass Spectrometry of Organics Anthony D. Appelhans* and James E. Delmore*

Idaho National Engineering Laboratory, EG&G Idaho, P.O. Box 1625, Idaho Falls, Idaho 83415

The secondary ion sputtering efficiency of a beam of 8-keV SF$- molecules was compared with that of an 8-keV Cs' atomic ion beam for samples of common organic pharmaceuticals. Both beams were operated in the static SIMS mode (picoampere currents) and samples ranged from microgram lo subnanogram quantnies on metal substrates. The efficiency (secondary ion peak Intenslty/primary beam intensity) for the molecular ion peaks (M H)' of the pharmaceutlcals was a factor of 9 to 24 higher for the molecular primary beam than for the atomic primary beam. This efficiency dlfference was reflected in the minimum detectable sample mass, which was a factor of 2 to 20 lower for the molecular beam.

+

INTRODUCTION In the course of development of a fast, highly focused neutral molecular primary beam probe ( I ) for secondary ion mass spectrometry (SIMS), the question as to the relative efficiency of a molecule versus an atom for producing secondary ions of organic molecules from nonliquid surfaces was raised. There have been many studies of secondary ion yields for a variety of atomic ion beams (cf. Benninghoven et al. (2)), and a few comparing relative yields for atomic and diatomic ions (3, 4 ) and for very high energy (5 MeV) atomic and polyatomic ions (5). These previous studies have clearly shown that hcreased energy (up to a point) and increased mass of the primary beam result in higher secondary ion yields. The same general conclusion was reached for static SIMS of polymers (6). Wong, Stoll, and Rollgen (7), in a preliminary study comparing a 40-pA Hg+ primary beam with a 3-pA mixed-mass molecular primary beam for liquid SIMS (Le. specifically not static-SIMS conditions), found up to a factor of 10 enhancement in the secondary ion signal intensity with the molecular primary beam. Their results, while consistent with the results of the present study, cannot be directly compared due to the significant differences between beam intensity, purity, and sample conformation. All of these results apparently have not significantly encouraged further investigation of the sputtering efficiency of molecules since there appears to have been no critical evaluation for a polyatomic molecular primary beam at flux levels and energies commonly used in static SIMS (-PA currents a t -and Cs+ as a function of primary beam accelerating voltage for the codeine (M H) peak. The data were normalized (tothe 8 kV-t of each beam) to illustrate the relative slope of the two data sets; the absolute efficiency of the SF60*-is actually 19 times higher than that of the Cs+ (Table I). The dependence of t on acceleration voltage is approximately the same for SF6O9-and Cs+ over the entire voltage range, although e(SFG0'-)increases slightly faster than e(Cs+). The dependence of secondary ion yield on primary beam energy shown in Figure 2 is consistent with previous results seen with noble gas ions sputtering clean Si ( 4 ) and polymers (6). Primary beam intensity (within the range 5-25 PA) showed no significant affect on secondary ion sputtering efficiency in our measurements. The second indicator of the relative efficiency of each type of beam is the minimum mass detectable for each primary beam. These measurements were performed by preparing samples with submonolayer concentrations on the surface and systematically reducing the sample loading until the peak of interest was no longer detectable. The samples consisted of a single compound, both acetaminophen and codeine were measured, and the parent peak was used to determine detectability. A criteria of a signal-to-noise ratio greater than 2 was used to judge whether a peak was detected, with the signal-to-noise ratio defined as the main peak height divided by the average of the peak heights within f5 amu of the main peak. In all cases the measurements consisted of 10 scans collected over a 2-min period and averaged. Primary beam energy was 15 keV and primary beam current was 20 pA for the SF6Oi-and Cs+ beams, and was 5 pA for the SF60beam. When the samples were run with SF: only and SF6- only, there was no detectable difference in the quality of the spectra

+

1

MASS (M/Z) 120

,,o ,M

,

1

Cs' PRlMARY BEAM ( 19 PA ) 6 , 3 ng CODEINE

I

(M+H)*

I

I

I

w2

MASS (M/Z) Figure 3. Typical secondary ion spectra for nanogram levels of codeine produced with the SF,'*- (top) and Cs+ (bottom) beams for the mass region around the protonated molecular ion peak.

and the peak heights were proportional to the relative beam intensities. With these light sample loadings the codeine signals were more intense than the acetaminophen signals for both the SF609-and Cs+ beams. For codeine the lower detection limit with SF609-was 0.3 ng (-1 pmol) and with cs+ was 6.3 ng, while for acetaminophen the lower limit with SF,$ was 75 ng and with Cs+ was 150 ng. These sample loadings correspond to the equivalent of 1 monolayer for the acetaminophen (-3 X 1014molecules/cm2) and much less than a monolayer for codeine ( 1x 1012molecdes/cm2),assuming the sample was evenly distributed over the surface (it is doubtful that this was the case for these low sample loadings). Figure 3 shows representative spectra of codeine for each beam. These results were reporducible with several samples and clearly indicate that the SF8- molecular beam was a more sensitive probe than the Cs+ for these types of samples and conditions. The third type of measurement reflecting the utility of each beam was of damage cross sections. The lifetime of a monolayer sample of tetrahexylammonium bromide (deposited on a Mo substrate) subjected to a constant primary beam flux for -5 h was measured to obtain the damage cross section u (18) for each type of primary beam. The damage cross section and the sputtering yield both contribute to the transformation probability P (18) (the probability that a molecule disappearing from the surface due to a sputtering event results in formation of a specific secondary ion). It has been shown (18)that P is directly proportional to the sputter yield and inversely proportional to the damage cross section

-

-

-

where Y, is the sputter yield and $8is a surface coverage term. Assuming that O08 is constarit for the two different beams (i.e. that sample preparation is relatively constant), the ratio of

ANALYTICAL CHEMISTRY, VOL. 61, NO. 10, MAY 15, 1989

1091

P for SFsOt-to P for Cs' helps to further define the relative utility of the two primary beam types. P reflects the useful yield (what is measured) relative to the damage incurred (at what cost to the sample integrity). The lifetime measurement requires that the sample be stable in the vacuum system (i.e. not evaporating or decomposing at a significant rate) during the measurement. Unfortunately we found that all of the pharmaceutical samples were not adequately stable over the time period ( - 5 h) required to make a good measurement (the loss due to evaporation, decomposition, or changes in the chemistry of the sample/substrate system while in the vacuum system in the absence of a primary beam was on the order of the loss due to sputtering). Thus we resorted to use of the tetrahexylammonium bromide compound, which we had found to be very stable in vacuum after a preconditioning period. Despite the fact that the ratio of the relative sputtering efficiency for the tetrahexylammonium bromide was only 2.5 (SFso9-/Cs+), making it less than fully representative of the pharmaceutical samples, it did offer a chance at providing further insight into the differences observed between the two beam types. For these measurements 1 pL of a solution of 4 ng/wL tetrahexylammonium bromide in methanol was dried onto the Mo substrate, providing approximately monolayer coverage (assuming even distribution). The samples were held under Torr) for 12 h prior to making the lifetime vacuum (I X measurements to assure stability. In addition, a second sample attached to the backside of the sample probe was used as a control to check for loss of sample due to evaporation or decomposition. The control was monitored at the beginning and end of each run and for brief periods (-120 s) at several intervals during each run. The primary beam energy was 8 kV at a current of 20 pA over a 0.075 cm2area. Figure 4 shows the measured data, the best fit line, and CI based on the best fit line for the SFs0--and Cs' beams. The data are normalized to the initial intensity. The fluence for the SF,O?-beam was calculated both as molecular fluence (top plot) and as atomic fluence (each SF, molecule was counted as seven incoming particles, bottom plot). The units of fluence were chosen for ease of comparison with other data in the literature and should be considered as amp-equivalent for the neutral particles. The intensity as a function of fluence showed a characteristic exponential decay, indicative of depletion of a monolayer sample. Displayed on a semilog plot the slope of the decay curve is proportional to the damage cross section (18) and indicative of the sputtering rate. The results show CI with the SFso>-(1.4 X This is -1.8 times the Cs' CI (8 X indicates the SFso3-is producing more damage per incident particle than the Cs'. However, the efficiency of SF,'*-, c (directly proportional to Y J ,was 2.5 times higher than for Cs+ for tetrahexylammonium bromide. Thus the transformation is 1.4 times higher than for Cs', indiprobability, P(SFso8-), cating that the SF,',- beam is producing more measurable secondary parent ions per unit of damage than the Cs+ beam.

-

CONCLUSIONS Determining the mechanisms responsible for the observed differences in the secondary ion sputtering efficiency of the polyatomic-molecular and atomic primary beams is complicated by the difficulty in measuring or calculating the physical and chemical processes that occur during sputtering, particularly the interactions that occur in the sample following the impact of a polyatomic molecule. Typical Monte Carlo based computer models [cf. TRIM (19)] for predicting collision cascades assume that all atoms within the sample are at rest with respect to the recoiling atoms; this assumption is most probably not valid for collision cascades initiated by polyatomic molecules. The relaxation time for a typical atomic s, particle induced collision cascade is on the order of

=

060.

c \

SF,

0 50 0 40 0 0

p- i

4

x io-"

cm' )

MOLECULAR FLUENCE

'qQ

kcc

-

2 0

r

~

----

-

~

60

4 0

FLUENCE (A-s/cm

*

80

10 0

X lom6)

0 50 0 40 k--0 0

I

1 0

2 0

FLUENCE (A-s/cm

*

3 0

40

X 10-5 )

Figure 4. Exponential decrease in signal as a function of primary ion

dose during bombardmentof ca.a monolayer of tetrahexylammonium bromide on a Mo substrate. Signals have been normalized (to the starting signal value) for ease of comparison and the solid line is the least-squares best-fit for each data set. In the top plot each SF, molecule was considered a single particle in determining the fluence, while in the lower plot each SF, molecule was consideredas seven particles in determiningthe fluence. The units for fluence were chosen to permit comparison with the literature.

which, for typical static SIMS flux, results in each atom-induced cascade being essentially an independent event. However for an incoming molecule, such as SF6,accelerated to energies typical in SIMS, the maximum time difference between when the first atom and the last atom of the molecule s. Thus the collision "hits" the surface is on the order of cascades induced by each of the atoms of the molecule will overlap in time. This, of course, makes it extremely difficult to model, although one can hypothesize on the probable effects of this overlap. It is generally accepted that sputtering of large organic molecules from a surface occurs when energy deposited into the subsurface lattice by the bombarding particle is directed back to the surface and absorbed by the surface molecules, making it possible for them to break the surface bonds and desorb into the vacuum. Several models describing the transfer of energy back to the surface molecules have been proposed (cf. ref 20, 21, 22), and though they differ in describing how the energy is transferred, all acknowledge the importance of the energy density. The mechanism of ionization is less well agreed upon, both preformed ions (23) and formation in the selvedge (24) are possibilities. Independent of the ionization mechanism, it is critical that sufficient energy be transferred to the surface molecules to permit desorption; thus a critical factor is the energy distribution within the sample lattice resulting from the primary particle impact. There have been many studies (e.g., ref 2,6) showing that for atomic primary particles sputtering efficiency goes up with primary particle mass at constant energy. A higher mass particle imparts more of its energy near the surface than an equal energy, lower mass particle. A molecule would be ex-

1092

ANALYTICAL CHEMISTRY, VOL. 61, NO. 10, MAY 15, 1989

pected to deposit more energy near the surface than an atom of equal mass and kinetic energy since the momentum of the individual atoms within the molecule is low and the resulting penetration of these atoms into the lattice will be less than that of the single atom. Assuming then that the spatial energy distribution is important to sputtering yield (the more energy transmitted to the surface the higher the yield), it would follow that a molecule would sputter more secondary ions (on the average) than an equivalent energy atom (assuming sufficiently high energy of each). The model system suggested by Benninghoven (23) illustrated this argument. In the vicinity of the primary particle impact point there is an annular region in which the energy that has been directed back to the surface is of the appropriate form and magnitude that it can be absorbed by a surface molecule and result in desorption of the surface molecule (adequate energy to break the surface bonds without tearing the molecule apart). The inner and outer radii of this region are determined by the primary particle momentum and the resulting collision cascade in the lattice (and the energy requirements of the molecules to be desorbed). Surface molecules residing a t R < R(in) absorb too much energy and are fragmented, while at R > R(out) the energy is insufficient for desorption. Using results from bombardment of small organic molecules (e.g. C2H4)with 3-keV Arc, Benninghoven (23) calculated that, on the average, these radii, R, are of the order of a few angstroms (-5 A). Extending this model to a molecule striking the surface, if the energy distributions due to each atom of the impacting molecule coupled (overlapped), the effective area for desorption could increase and result in an effective area for desorption greater than that of the sum of the individual annuli. In the case of SF,', a symmetric molecule, the S-F bond lengths are 1.7 b, and the momentum of the individual F atoms when the molecule is accelerated to 10 kV is very close to that of 3-keV Ar. If one assumes that the critical sputtering radius for the individual F atoms is of the same order as for the 3 keV Ar, i.e. - 5 A, then the surface energy distributions for each of the F atoms, and the S atom, of the SF, molecule would overlap in space. The typical relaxation time of a collision cascade is s, and the time differential between the leading and following atoms of a 10-keV SF, molecule striking the surface is -3 X s. Thus the collision cascades of the individual atoms of the SF, molecule could overlap in space and in time, possibly resulting in an increase in the effective area for sputtering over the sum of the individual areas for each atom within the molecule, i.e. a "molecular" enhancement. While this conceptualization is certainly less than rigorous and to test it would require knowledge not currently available (the energy distribution around the area of impact and the energy coupling mechanisms controlling sputtering), it does offer an intuitively satisfying picture consistent with the observed experimental results. The experimental results are also consistent with the thermal spike model calculations of Beuhler and Friedman (22), which predict conditions resulting from impact of a cluster molecule that should lead to increased sputter yield. While the SF,',- molecular beam results in slightly more damage to the surface, the number of measurable secondary ions produced is higher (per unit of damage) than that observed for the Cs+ beam. These results suggest further research with even larger primary molecules, or perhaps clusters, although there most probably is some optimum size (and/or shape) primary molecule beyond which the gain in sputtering efficiency is offset by the damage induced. I t may also be that the optimum size and geometry of the primary beam molecules are dependent upon the size and type of molecule to be sputtered.

-

Since the chemistry a t the surface is expected to play a major role in governing the production of ions during sputtering, the possible effects of the different chemical nature of Cs+ and SF6O3-must be considered. It is generally held that the chemical nature of the bombarding species has little or no effect at fluences typical of static SIMS (25)and van Ooij (26) has found that there are no measurable chemical effects and Ar) for caused by the primary beam species (Cs+, 02+, static SIMS of organic polymers. The low primary flux and fluence used in the measurements reported herein would result in an extremely low concentration of the primary beam constituents a t the surface (on the order of mole fraction assuming all primary beam atoms reside in the top monolayer, an extremely unlikely situation). Thus chemical differences between SF,O*-and Cs+ are not expected to be a significant factor influencing the relative sputtering yields in this study. However the energy coupling and surface bond strength are dependent upon the surface chemistry, which is dominated by the sample and substrate materials in these measurements, and so the differences in the relative efficiencies from one sample compound to another are not surprising. Previous measurements ( I ) with this same experimental apparatus showed the secondary ion sputtering efficiency of range, significantly the SF$- beam for Teflon was in the higher than that measured for the pharmaceutical samples. The efficiency for Teflon was remeasured a t the conclusion of this study and confirmed this result. The reasons for the large difference are not apparent, and currently there is not enough data to support a plausible hypothesis. However, we feel that the differences are real since the efficiencies were measured under the same conditions in the same instrument. Further investigations will be necessary to elucidate the mechanisms controlling the sputter yields for these types of materials. We conclude that there may be significant advantages in using molecular primary beams for SIMS of organic molecules, particularly where high sensitivity is critical. In addition the neutral molecular beam easily overcomes sample charging problems without the complication of electron flood guns or other intrusive charge-compensation techniques, making it possible to analyze bulk samples with minimum sample preparation. The experimental results suggest that primary beams consisting of large molecules, and possibly clusters, should be further explored for secondary ion mass spectrometry of large organic molecules.

-

LITERATURE CITED (1) Appelhans, A. D.; Delmore. J. E. Anal. Chem. 1987, 59, 1685-1691. (2) Benninghoven. A,; Rudenauer, F. G.; Werner, H. W. I n Secondary Ion Mass Spectrometry; John Wiiey 8 Sons: New York, 1987; Chapter 2.2.5. (3) Andersen, H. H.; Bay, H. L. Radiat. Eff. 1973, 79, 139-146. (4) Wiilrnaack, K. Surf. Sci. 1979, 9 0 , 557-563. (5) Salehpour, M.; Fishei, D : Hunt, J. Rapid Commun. Mass SDectrom. 1988, 2(3). 59-60. (6) Briggs, D.; Hearn, M. J. Int J . Mass Spectrom. Ion Processes 1985, 67 .. , 47-56 . . ._ (7) Wong, S. s.; Stoll, R.; Rollgen, F. W. 2 . Naturforsch.. A : Phys., Phvs. Chem.. 1982. 3 7 A . 718-719. (8) Thbmpson, D. A. Radiat. €ff. 1981, 56, 105-150. (9) Benninghoven. A.; Rudenauer, F. G.; Werner, H. W. I n Secondary Ion Mass Spectrometry; John Wiley & Sons: New York, 1987; Chapter 2. (10) Delmore. J. E. I n t . J . Mass Spectrom. Ion Processes 1983, 51, 191-205. (11) Delmore. J. E.: Aooelhans. A. D. J . Chem. Phvs. 1988. 88. 9. .. 5561-5570. (12) Delmore, J. E.; Appelhans, A. D. J . Chem. Phys. 1986, 84. 11, 246 6238-6.~ (13) Heo, N. H.; Dejsupa, C.; Seff, K. J . Phys. Chem. 1987, 97, 3943-3944. (14) Delrnore, J. E.; Appelhans, A. D. Int. J . Mass Spectrom. Ion Processes 1986, 6 8 , 327-336. (15) Appelhans, A. D. I n t . J . Mass Spectrom. Ion Processes, in press. (16) Benninghoven, A,; Rudenauer. F. G.; Werner, H. W. I n Secondary Ion Mass Spectrometry; John Wiiey & Sons: New York, 1987; pp 88 1-899. (17) Briggs, D.; Wooiton, A. E. S I A . Surf. Interface Anal. 1982. 4 . 109-1 15. ~~~~

~

Anal. Chem. 1989, 6 1 , 1093-1099 (18) Benninghoven, A,; Rudenauer, F. G.; Werner, H. W. I n Secondary I o n Mass Spectrometry; John Wiley & Sons: New York, 1987; Chapter 5.1.3. (19) Ziegler, J. F.; Biersack, J. P.; Cuomo, G., TRIM-The Transport of Ions in Matter, based on The Stopping and Range of Ions in Matter; Pergamon Press: New York, 1985. (20) Benninghoven, A,; Riidenauer, F. G.: Werner, H. W. I n Secondary I o n Mass Spectrometry; John Wiley & Sons: New York, 1987; Chapter 5.1.6.2. (21) Sunner, J.; Ikonomou, M. G.;Kebarle, P. I n t . J. Mass Spectrom. Ion Processes 1988,82. 221-237. (22) Beuhler. R. J.: Friedman. L. Int. J . Mass Snectrom. Ion Processes 1987,78,1-15. (23) Benninghoven, A.; Riidenauer, F. G.; Werner, H. W. I n Secondary Ion Mass spectrometry: John Wiley 8 Sons: New York, 1987: Chapter 5.1.6.7.

1093

(24) Pachuta, S. J.; Cooks, R. G. I n Desorption Mass Spectrometry; ACS Symposium Series; American Chemical Society: Washington, DC, 1985; pp 1-42. (25) Benninghoven, A.; Rudenauer, F. G.; Werner, H. W. I n Secondary I o n Mass Spectrometry: John Wiley 8 Sons: New York, 1987; Chapter 5.1.6.5. (26) Van Ooij, W.; Brinkhuis. R. H. G. Proceedings of the VIth International Conference on Secondary Ion Mass Spectrometry, Versailles, France, 1987.

RECEIVED for review September 21,1988. Accepted February 13, 1989. Work supported by the Office of Health and Environmental Research, Office of Energy Research, DOE under Contract 4AA901.

Ion Mobility Spectrometry of Halothane, Enflurane, and Isoflurane Anesthetics in Air and Respired Gases G . A. Eiceman,*J D. B. Shoff, C. S. Harden, and A. P. Snyder U S . A r m y Chemical Research, Development, and Engineering Center SMCCR-RSL, Aberdeen Proving Ground, Maryland 21020 P. M. Martinez and M. E. Fleischer Department of Chemistry, New Mexico State University, Las Cruces, New Mexico 88003 M. L. Watkins Memorial General Hospital, Las Cruces, New Mexico 88005

Three common gaseous anesthetics, halothane, enflurane, and isoflurane, were Characterized by using ion mobility spectrometry (IMS)/mass spectrometry, and the dependence of product Ion distributions on temperature and concentration was evaluated. At 40 'C and 500 ppb, negative ion mobility spectra In air largely consisted of monomer or dimer adducts with Br- or CI- formed through dissociative electron capture of molecular neutrals. With increased temperature or decreased vapor concentrations, declustering and dissociation of product ions became pronounced. Ion-molecule reactions in the drift region of the IMS were evident as distortions in peak shape In the mass-resolved mobility spectra and in variable reduced mobllltles for the same Ions. A portable hand-held I M S was used for convenient, real-time detection of enflurane In respired gases following a controlled inhalation episode.

INTRODUCTION Volatile halogenated anesthetics (VHAs),based principally on fluorinated and chlorinated two- or three-carbon alkane or ether structures, have become medical and occupational health concerns due to chronic or acute inhalation exposures (1). These anesthetics can be found in blood (2-4), respired air (5),and operating-theater atmospheres (6-8) under normal clinical conditions, and subsequent investigations provided strong evidence that VHAs caused episodes of organ toxicity Permanent address: Department of Chemistry, Box 30001-Dept. 3C, New Mexico State University, Las Cruces, NM 88003.

(9) and mortality (10) in surgical patients. Furthermore, possible liver damage to hospital surgical personnel from chronic, subclinical environmental exposure to VHAs has been suggested (11). Poisonings from solvent abuse (12) represent related but somewhat unpredictable medical obstacles. Gas chromatography (GC) has been used for VHA determinations in blood (13)and in operating-theater atmospheres (14) since the early 1970s and exhibited good selectivity and high sensitivity (2, 3,5-8,12-14), which were largely due to the characteristics of electron capture detectors (ECDs). While GC was well-suited for analysis of discrete samples, continuous monitoring of VHA concentrations in air or blood was found to be technically cumbersome. Other technologies (15-21) were proposed for VHA or solvent sensing; however, they too proved to be insensitive, unselective, or too expensive for practical clinical utility. Thus, analytical techniques are needed (8,12)for the determination of VHAs in ambient air and respired gases within a short time period as well as on a continuous basis with sub-part-per-million detection limits. Relatively inexpensive, continuous, organic vapor detectors based on ion mobility spectrometry (IMS) instrumentation have been developed for portable air monitoring (22,23). In IMS, analyte vapors are drawn into a reaction region where analyte molecules are collisionally ionized through proton or electron exchange from reactant ions created in the reaction region. The resultant product ions are then characterized by using gaseous ionic mobilities in a drift region that contains a weak electric field. While ion integrity during transport through the IMS drift region cannot always be assumed (24, 25), a more common disabling factor is unfavorable chemistry of product ion formation at atmospheric pressures in the IMS reaction region. This can be manifested as poor electron (or

0003-2700/89/0361-1093$01.50/0 C 1989 American Chemical Society