21
Anal. Chem. 1981, 53, 21-25
rameten already described. Memben of the morphine group are now distinguishable also by the many changes in the ‘La band, by the separation of the Xo values, and by evolution of the positive band at 300 nm for morphine, nalorphine, and 6MAM. There is also a fairly general decrease in molar ellipticity values on raising pH which is greatest for heroin and difficult at this time to interpret in terms of structural changes. 3MAM is a singular exception to the trend. The evolution of the positive band at 300 nm is interesting from a structural point of view. It is certainly associated with the formation of the phenolate ion as the pH is raised. There is good reason to believe that the entire spectrum is red shifted with increasing pH, and the maximum of the ‘Lbband occurs at 300 nm. The spectrum a t pH 8.6 is then a superposition of the spectra for the -OH form and the -0- form. The change in sign of the ‘Lb band may be accounted for by the following explanation. From X-ray diffraction it is known that the aromatic ring is forced out-of-plane by the steric restrictions imposed on the molecules by the atoms at the annular bridge head (11,12). On the other hand, delocalization of the oxygen negative charge might assume steric priority causing the aromatic ring to become flat and therefore to dominate in the structure determination. Changes in sign would properly be interpreted as movement of substituents from one sector into another in a viable quadrant-rule model. The critical pivot points are at C(5) and C(13). These points are probably critical too in transmitting the effect of modifying the C-ring over to the transitions in the aromatic ring. The effect is not observed in UV absorption spectra where one can see that the spectrum of thebaine in ethanol is very similar to those of the morphine group compounds (5). Distinction by CD is as much a consequence of the inherent rigidity of the molecule so that small structural changes produce enlarged conformational reorientations. In summary CD spectropolarimetry has been successfully applied to the distinction of 10 opiates. Compared to the alternative methods described from this laboratory, where the
compounds are placed either in KBr pellets (3) or in a cholesteric liquid crystalline solvent (4), the present method is straightforward and easily has the advantage of becoming quantitative. It is necessarily restricted to the study of chiral compounds, and in this regard the anisotropic cholesteric solvent has the advantage, being applicable to the study of both achiral molecules and racemic mixtures. CD using KBr pellets is of particular value where the compound is insoluble and where it is important to preserve the integrity of the sample. Although distinction is greatest in pH 8.6 buffer solutions, greater versatility is obtained by looking at results from both acidic and basic media. This may become necessary as the data base is extended to include other opiates and related compounds. Using different media might well be essential also in the analysea of mixtures, where work has already begun. LITERATURE CITED Mason, S. F. 0 . Rev. Chem. Soc. 1961, 57, 287.
Fulton, C. C. “Modem Mlaocrystal Tests for Drugs”; Interscience: New York, 1969. Bowen, J. M.; Rndle, N. Anel. Chem. 1980, 52, 573. Bowen, J. M.; Crone, T. A.; Hennann, A. 0.;Purdle, N. Anel. Ctmm., in press. S!&, T. J. J. F W W l c &I. 1974, 75, 193. Weiss, U.; Rull, T. 8 u U . Soc. Chlm. Fr. 1965, 3707. DeAngelIs, (3. G.; WHdmen, W. C. Tetrahedron 1969, 25, 5099. Sunshine, I. “Handbook of Analytical Toxicology”; Chemical Rubber Co.: Cleveland, OH, 1969. Slek, T. J.; Osiewicz, R. J. J. Forenslc Scl. 1975, 20, 18. Slek, T. J.; Oslewlcz, R. J.; Bath, R. J. J. Foremlc Scl. 1976, 21, 525. Gylbert, L. Acta Ctystalbgf., Sect. B 1979, 29, 1830. Bye, E. Acta Chem. Sand., Ser. B 1976, 830, 549.
RECEIVED for review July 25,1980. Accepted October 8,1980. We wish to thank the National Science Foundation for the support of this work under Grant No. NSF CHE-7909388.We are also indebted to Mallinckrodt, Inc., and to Research Triangle Institute and the National Institute for Drug Abuse for their assistance in obtaining samples.
Excitation Frequency Dependence of Resonance-Enhanced Inverse Raman Band Shapes Jeanne P. Haushaiter and Michael D. Morris’ Department of Chemhtry, Universm of Michigan. Ann Arbor, Michigan 48 109
The theoretical equation governing the shape of a resonance-enhanced inverse Raman band Is derived. The band shape is shown to depend on the dtfference between the pump laser frequency and the frequency of the resonant electronic transhion. The theory is shown to describe acrk#ne orange and @-carotenespectra obtained with the argon ion laser lines between 488 nm and 514.5 nm as pump lasers.
During the last 3 years inverse Raman spectroscopy and Raman gain spectroscopyhave emerged as useful methods for obtaining the b a n spectra of highly luminescent materials. The utilization of various modulation/demodulation schemes (1-3) provides much better detection limits than direct measurements of absorption or gain. We have recently shown 0003-2700/81/0353-0021%01.00/0
that one modulation/demodulation method, ac-coupled inverse Raman spectrometry, (1) gives detection limits comparable to those obtained by spontaneousRaman spectroscopy (4).
Recently, we reported the first resonance-enhancedRaman spectra obtained by a modulation scheme (5). In that paper we demonstrated that resonance-enhanced inverse Raman spectra of a highly luminescent molecule, acridine orange, could easily be obtained by ac-coupled inverse Raman spectrometry. However, the spectra were inverted. That is, they appeared in apparent gain rather than in absorption. Only tentative explanationswere offered for the unexpected results. In this paper we report the dependence of inverse Raman band shapes on pump laser frequency and derive the governing equation which describes the band shapes. We show that, in general, inverse Raman bands change from positive Lorent0 1980 American Chemlcal Society
22
ANALYTICAL CHEMISTRY, VOL. 53, NO. 1, JANUARY 1981
zians, through dispersive shapes, to negative Lorentzians as electronic resonance is approached. This phenomenon is shown to be a result of the dependence of the third-order susceptibility on pump laser frequency and not on the lifetime of the singlet excited state as was suggested previously (5). THEORY The general third-order susceptibility equation has been given by Lotem, Lynch, and Bloembergen (6)in a very useful form. This or an equivalent expression has been used successfully to describe the shapes of CARS bands under resonance-enhancement conditions (7-10).For the case of inverse Raman spectrometry, only one term of the general expression, 2 and 3 of ref 6, is resonant in the region of an electronic transition. This term is given as eq 1. Here x ( ~is) the x(3)
6NL
= h3[WR
- (fd1 - w2) - i r R ]
c
h (whg - 0 1
X pghphtpthphg
- irhg)(whg - (J1 - irhg)
(1)
third-order susceptibility, N is the number of molecules per unit volume, and L is a local field correction. The frequency of the Raman transition is WR and its damping constant is r R . The pump laser is at frequency w1 and the probe laser is at a lower frequency,w2. The sum is taken over all combinations of vibronic states, h, of the molecule. The ground state is g and the Raman transition terminates in state t. The frequency of an electronic transition is ob and its damping constant is rb PghlPht, Pn, and Phgare the transition dipole moments. Polarization subscripts are not included here. Identical expressions, differing only by numerical factors, can be written for each of the two independent polarization components of the susceptibility. Equation 1 can be greatly simplified by considering only a single electronic excited state. In that case the summation of eq 1 is replaced by a single term, in which the electronic transition frequency is denoted by we and the electronic damping constant is re. This approach is often quite adequate and has been used in most descriptions of resonance-enhanced CARS (10). Further simplification results by noting that the terms in eq 1 are products of electronic enhancement factors, which are the transition moments, and frequency factors, which are the terms in the denominator. Only the frequency factors need be considered to describe the shape and relative enhancement of a Raman band as electronic resonance is approached. The electronic enhancement factors determine only the absolute magnitude of the inverse Raman effect. With these simplifications, we can write eq 2 to describe the shape of inverse Raman bands near an electronic resonance.
Here transition moments, local field correction, and constants have been lumped into a single constant term, K . Inverse Raman spectroscopy involves a net change in molecular energy levels. Thus, the observed signal will be directly . this proportional to the imaginary component of x ( ~ ) In respect there are several key differences between inverse Raman spectrometry and CARS. First, no nonresonant background term is expected in inverse Raman spectroscopy. Second, inverse Raman signals will remain proportional to concentration, as are spontaneous Raman signals until absorption of either laser beam becomes significant. Third, no interference between closely spaced bands is expected. Unresolved bands which are the algebraic sums of their components may occur, of course.
By use of the rules for manipulation of complex numbers eq 2 can be separated into its real and imaginary components. The imaginary component, xi3),is given in eq 3.
xb3) = NK
X
- ail2- rRre2+ 2r&we- OR - ( 0 1 - wz)l [(we - w1)' + r,2I2[(WR - (w1 - w2))' + rR']
rR(we
(3) Several special cases of eq 3 are of interest. For we >> ol, the band shape should be a positive Lorentzian. This limiting case, or more correctly, the analogous equation including the summation over electronic states, describes the nonresonance Raman regime. Here the correspondence in band shapes, relative intensities, and polarization properties between spontaneous Raman spectra and inverse Raman spectra assumed by us and others ( I , 2) is valid. At exact resonance, where w1 = we, the band shape should be a negative Lorentzian. Negative Lorentzian bands have been reported previously by us (5) for this case. When w1 = we - re,a purely dispersive band shape is expected. When w1 = we + rea dispersive band shape of the opposite sense is expected. Neither band shape has previously been reported for inverse Raman spectra. Equation 3 also predicts that resonance enhancement and, therefore, band shapes should depend only upon the difference between w1 and we, and should be independent of w2. For a given molecule, all of the bands should have the same shape if the spectrum is obtained by scanning o2with o1constant. It should be noted that there is some disagreement about what value should be used for the electronic damping constant in descriptions of resonance Raman intensities. In accord with workers in resonance CARS, we use the half-bandwidth of the electronic transition as measured by absorption spectroscopy. The choice of redetermines the range of pump frequencies over which the change from Lorentzian through dispersive to negative Lorentzian is predicted. Bloembergen's treatment of coherent Raman spectroscopic theory (6) is incomplete in that it neglects stimulated transitions between electronic states. Druet, Taran, and Bord6 have recently extended coherent b a n theory to include such processes ( I I , 1 2 ) . However, their discussion is of Dopplerbroadened and Doppler-free systems only and is not strictly valid for the liquid phase. It is reasonable to assume, however, that a similar treatment would apply to the liquid phase. EXPERIMENTAL SECTION The instrumentation and experimental procedures for accoupled inverse Raman spectrometry have been previously described (4,5).Absorption spectra were obtained on a Varian/Cary 219 spectrophotometer. Acridine orange (AldrichChemical Corp.) was used in methanol and distilled water solutions. @-Carotene(Aldrich) was used in ethanol solution. All solutions were filtered through 0.22-pm membrane filtersjust before use to minimize light scattering from particulate impurities. As in our previous work, a decrease in argon ion intensity is plotted as positive-going, while an increase is plotted as negative-going. RESULTS AND DISCUSSION Absorption spectra for &carotene in ethanol and for acridine orange in water and methanol are presented in Figure 1. All solutions used were at concentrations below lo4 M. Therefore, no dimerization of acridine orange in aqueous solution is observed. Figure 2 shows the change in the shape of the strong acridine orange Raman bands in the 1300-1400-~m-~ region as the wavelength of the argon ion laser is changed from 488 to 514.5 nm. The progression from a negative Lorentzian band
ANALYTICAL CHEMISTRY, VOL. 53, NO. 1, JANUARY 1981
23
a
v
b
Absorption spectra: (a) acridine orange, 1 X lo-' M in water; (b) acridine orange, 2 X lo-' M in methanol; (c) @-carotene, 2 X lo-' M In ethanol. Flgure 1.
-50.
-25.
-0.
25.
50.
A w , cm-'
Simulations of an isolated resonance Inverse Raman band (rR= 5 cm-', rE= 684 cm-'): (a) w , - w3 = -84 cm-'; (b) w we = 267 cm-'; (c) w1 - we = 475 cm-'; (d) w1 - w, = 972 ,I?'-< Figure 3.
Table I. Dependence of Band Intensity Ratio I137lCm-1/ on Concentration and Pump Laser Frequency w I , cm-' concn 20492 20141 19932 19436
+ m
I132km- I
.-? c 3
1
rn L
c L
2 X lo-' 1 x 10-5 4X 2X
.I
n
L
m
111
m c rn L U
a
+-> .4
u7
C
tOJ
C
H
1300.
1325.
1350.
1375.
1400.
Am, cm-l Figure 2. Acridine orange inverse Raman band shapes: (a) w1 = 20492 cm-'; (b) wl = 20 141 cm-'; (c) w, = 19932 cm-'; (d) w 1 = 19 436 cm-'. Curve d was recorded In aqueous solution; all others were recorded in methanol. through a dispersive shape to a band with a large positive Lorentzian component is very clear. In all of the inverse Raman spectra shown here, the measured signal is attenuated argon ion laser intensity. A negative signal represents a decrease in the laser attenuation, not a gain in intensity above the initial value. Such a net increase in argon ion laser intensity would appear to violate the law of conservation of energy. The experimental curves also show the decreasing effect of ground-state depletion as the pulsed dye laser is moved farther from the electronic absorption. Ground-state depletion, a major effect at 488 nm, contributes relatively little to the observed signal at 514.5 nm. By contrast, ground-state depletion is the dominant effect at 476.5nm. If this laser line
0.65 0.95 0.93 0.93
0.97
1.2=
0.97
Abnormally high ratio is an artifact of system optics.
is used, the dye laser must be placed sufficiently far into the absorption band that depletion is the only signal observed. No Raman signals are visible above the depletion background. Figure 3 shows simulationsof a single Raman band by using eq 3. For these plots the following parameters are used: we = 20408 cm-' and re= 684 cm-' as measured from acridine orange spectra. l?R is assumed to be 5 cm-'. Comparison of the experimental curves and theoretical band shapes shows that the agreement is very good. Thus, our theoretical description of this system is an adequate approximation. We have previously suggested (5)that experimental peak height ratios should be corrected for depletion effects, since the effect of ground-state depletion is to increase the transmission of the argon ion beam. Data in Table I show the effects of ground-state depletion very clearly. We have measured the peak height ratio of the 1371-and 1328-cm-' bands of acridine orange as a function of pump laser frequency and concentration. From this table it is clear that as absorption of the dye laser decreases and, therefore, as ground-state depletion decreases, the peak height ratio approaches 0.95. This is within experimental error of the ratio, 0.9,predicted from multiplex CARS data (13), which are taken under conditions where band intensity distortions are not present. Figure 4 shows the 1158 cm-' band for @-carotenein ethanol. Data were taken at the argon ion 448 and 514.5 nm lines. These spectra demonstrate that the behavior observed for acridine orange is also observed for other molecules, including those which are only weakly luminescent, as is 8-carotene.
24
ANALYTICAL CHEMISTRY, VOL. 53, NO. 1, JANUARY 1981 19,438cm.'
ffl
! d
.-.
c 3
r L
m
tL .4
n L
m
L
W D
rn c L U
l
a
tr .4 ffl
c
t111
c
Inverse Raman spectra of 2 X 10" M &caroteneh ethanol, 1158-~rn-~ band: (a) w , = 20492 cm-'; (b) w1 = 191 136 Cm-'.
H
Ftgure 4.
-50.
-25.
-0.
25.
Aw,cm-I Complete Inverse Raman spectrum of 2 X orange in water, w1 = 19436 cm-l. Flgwe 6.
51 2 , L
c-'
rn
n W
m
c
m
L U
r
F 4 ffl
c W
2 1 1
H
1100.
I
1250.
1400.
1550.
1700,
Aw, cm-' Flgure 5. Simulations of &carotene band: (b) we - ol= 1615 cm-'.
(a) wI - w1 = 561 cm-';
Figure 5 shows simulations of these bands using the parameters we = 21 052 cm-', re= 720 cm-', and r R = 5 cm-'. Again, the data agree well with the predictions of eq 3. Equation 3 predicts that the shape of a resonance-enhanced inverse Raman band depends only on the frequency difference between the pump laser and the electronic transition giving rise to enhancement, Le., on we - w l . That prediction is readily tested by examination of the resonance inverse Raman spectrum of a given molecule over an extended frequency range. Figure 6 shows the inverse Raman spectrum of acridine orange over the range 1100-1600 cm-', using a pump laser frequency of 19 436 cm-'. This spectrum was taken in water to avoid obscuring some of the weaker acridine orange bands by larger methanol bands. Inspection of Figure 5 demonstrates that all bands in the spectrum have the same shape, except insofar as they are so close together as to overlap. CONCLUSIONS We have demonstrated that resonance-enhanced inverse Raman spectra can be explained by use of the same thirdorder susceptibility theory which has been succes~fuuyapplied to resonance CARS. We have shown that inverse Raman spectra are not strictly analogous to spontaneous Raman spectra. Under resonance-enhancement conditions, inverse Raman bands depart from the Lorentzian shape characteristic of spontaneous Raman spectra. However, the band shapes can be described by straightfoward equations. It is possible to extract the usual Raman parameters, CORand rR, by fitting to these equations. Negative and dispersive bands, while predicted by theory, are difficult to reconcile with the usual energy-level diagrams employed by chemical spectroscopists. The same problem applies to descriptions of other coherent Raman processes. The reason for the difficulty appears to be that energy-level
50.
M aaidine
diagrams are appropriate for spontaneous processes, but they do not represent coherent phenomena well, for which the collective behavior of the entire sample is important. Dispersive and negative band shapes can be interpreted as caused by changes in the phase relation between the driving (laser) electric fields and the response of the molecular system. This behavior is related to the familiar phenomenon of anomalous dispersion, the rapid variation of the refractive index of a system in the vicinity of an electronic transition. Although non-lorentzian band shapes are obtained, inverse Raman bands retain their proportionality to concentration and to spontaneous Raman band intensities if suitable corrections for ground-state depletion and/or absorption are applied. The theoretical treatment and experimentaldata presented here explain why positive Gaussian band shapes have been obtained for o-nitroaniline, as described in our earlier work (5). In that work, argon ion 488-nm excitation was used as the pump beam. The excitation frequency, 20 492 cm-', is well removed from the band maximum at 24 220 cm-' (14). It is risky to identify the maximum of that broad band with the origin of the electronic transition. However, any reasonable assignment would put the origin sufficiently far from 20 492 cm-' that a pure or nearly pure positive Lorentzian band shape would be expected. Although we have presented data for only two systems in this paper, we believe that similar behavior should be generally observable in the resonance inverse Raman spectroscopy of molecules in the liquid phase. We have already obtained non-Lorentzian band shapes as described by eq 3 for bile pigments (15)and flavins (16). These results will be described in separate communications. Our present theoretical treatment can be modified to describe resonanceenhanced Raman gain spectra. In the sense that resonance-enhanced inverse Raman theory parallels resonance-enhanced CARS theory, resonance-enhanced Raman gain theory should resemble that for the coherent Stokes Raman spectrometry (CSRS) case (17). Work on this extension and on a more rigorous theory of liquid-phase resonance inverse Raman and Raman gain spectroscopies is in progress. ACKNOWLEDGMENT We thank Kathy J. Dien and Clifford E. Buffett for help with the design of computer programs for this work. LITERATURE CITED (1) . . W s . M. D.: Wallan. D. J.: R k . 0.P.: Haushalter, J. P. Anel. Chem. 1978, '50, 1796-1799. (2) Owyoung, A. I€€€ J. Quantum Electron. 1878, OE- 14, 192-202. (3) Levlne, G. F.; Bethea, C. G. Appl. Phys. Lett. 1980, 36, 245-247.
25
Atla/. Chem. 1881, 53, 25-29
(4) Hawhalter, J. P.; Rltz, 0. P.; Wallan, D. J.; Dien, K.; M I S , M. D. Appl. Spectrosc. 1080, 3 4 , 144-146. (5) Haushalter, J. P.; Buffett, C.
E.; Morris, M. D. Anal. chem. 1080, 52,
1284-1287.
(13)Tretzel, J.; SchneMer, F. W. Chem. Phys. Lett. 1078, 59,514-518. (14) Kumar, K.; Carey, P. R. J. Chem. Phys. 1075, 63, 3697-3707. (15) Dien, K. F.; M a , M. D., unpublished observatlons. (16) Haushalter, J. P.; Schopfer, L. M.; M s , M. D., unpublished observa-
(8) Lotem, H.; Lynch, R. T., Jr.; Bloembergen, N. phys. Rev. A 1076, 74,
1746-1755. (7) Hudson, B.; Hetherington, W.; Cramer, S.; chebay, I.; Klauminzer, 0. K. Pfm.Net/. Acad. Scl. U . S . A . 1076, 73, 3798-3802.
(8) Lynch, R. T., Jr.; Lotem, H.; Bloembergen, N. J. Chem. phys. 1977, 66, 4250-4251. (9) Carrelra. L. A,; Maguire, T. C.; Malloy, T. B. J . Chem. phys. 1077,
66,2621-2628. (IO) Dutta, P. K.; Spko, T. D. J. Chem. Phys. 1078. 69,3119-3123. (11) Druet, S. A. J.; Taran, J.-P. E.; Bord6. C. J. J. phys. (Orsay, Fr.) 1070, 40, 819-840. (12) Druet, S. A. J.; Taran, J.P. E.; Bord6, C. J. J . phys. (CrsayFr.) 1980, 41, 183-184.
tlons.
(17) Carrelra, L. A.; Qoss, L. P.; Malloy, T. B., Jr. J. Chem. phys. 1078, 69,855-882.
REC-
for review July 23,1980.Accepted October 16,1980. This work was supported in part by the National Science Foundation through Grant CHE 79-15185. The computer used was purchased with support from the National Science Foundation through Grant CHE 78-20065.
Acid-Enhanced Field Desorption Mass Spectrometry of Zwitterions T. Keough” and A. J. DeStefano The Procter & Gamble Company, Miami Valley Laboratorles, P.O. Box 39175, Cincinnati, Ohio 45247
In the field desorption mass spectrometrlc analysis of a wlde varlety of zwltterlons, sensltMty for (M H)’ adduct kms can be slgnlficantly Increased by protonating the zwitterkn to form a quaternary “onlum” salt prior to analysis. Protonation can be effected in one step by simply adding p-toluenesulfonic acld to a solutlon of the zwltterlon. Fleld desorption mass spectra obtained by thls procedure typically exhlbh only a single ion Corresponding to the protonated zwfflerion. The reduced emmer current required for these analyses elhinates the fragment kns or adduct kns (fonned by thermaiy Wllated bimolecular processes) that domlnate the spectra of underlvatlzed zwltterlons. Appllcatlons to zwkterionlc surfactants, amlno aclds, and phospholipids are dlscussed.
+
Field desorption (FD) mass spectrometry is an effective technique for the analysis of a wide variety of nonvolatile and thermally unstable compounds (I), often providing intact molecular or adduct (e.g., M + H+, M + Na+) ions that cannot be observed with other readily available ionization techniques (2). Zwitterionic compounds are biologically and commercially important and are often best characterized by FD. For example, the molecular weights of some thermally unstable amino acids (3) and phospholipids (4) have been directly determined by FD, and we have characterized ammoniocarboxylates (5)and hydroxyammoniocarboxylatea (6)by this technique. Unfortunately, many zwitterionic compounds are thermally unstable a t the emitter temperatures typically required in FD analyses (150-250 “C). The extent of sample pyrolysis on the emitter is strongly dependent upon emitter temperature (5, 6) as are the resulting FD spectra. Hightemperature FD spectra often exhibit little, if any, of the (M + H)+ adduct ion and are dominated by fragment ions and ions formed by intermolecular processes (such as alkyl group transfer). Thus, poor sensitivity and thermal instability of the sample (or the (M + H)+ adduct ion) can severely limit the information content of the FD maas spectra of zwitterionic compounds. We have developed a simple one-step procedure that significantly increases FD sensitivity for the (M + H)+ adduct ions of a wide range of zwitterionic compounds and minimizes 0003-2700/81/0353-0025$01.00/0
the thermal problems just mentioned. This method involves converting the zwitterion to an “onium” salt by adding p toluenesulfonic acid (p-TSA) prior to analysis. The reaction is illustrated in eq 1 for an ammoniocarboxylate. The method
(CH3)3N+(CHJbCOO- + CH3C6H4SO3H
-
[(CH3)3N(CH2)5COOHI+[CH3CGH,S031(1) is based on the previous observation (7) that the FD mass spectra of quaternary ammonium salts yield an intense ion corresponding to the intact cationic portion of the salt. Our approach has yielded excellent FD mass spectra for zwitterionic carboxylates, phosphates, sulfonates, and sulfates, compounds that differ greatly in the basicity of the anionic portion of the molecule. EXPERIMENTAL SECTION Instrumentation. All FD mass spectra were obtained by use of a modified Varian/Atlas SM-1B mass spectrometer that has been described previously (5). The original electron impact source was replaced with an FI/FD source similar to that discussed by Beckey and Schulten (8). Source focusing was accomplished manually, with the aid of micromanipulators,by maximizing (at the collector) the molecular ion current generated by field ionization of acetone. The mass scale was manually calibrated with the aid of a Hall-effect probe after analysis of suitable FI/FD reference compounds (9). All spectra were output to a Honeywell Model 1508 Visicorder. Reagents. The amino acids and distearoylphosphatidylcholine (DSL) were obtained from commercialsources and used without further purification. The ammoniocarboxylates were synthesized and purified by methods developed by Laughlin and McGrady (10). The purity of these samples was verified by NMR. The remaining compounds were also synthesized at the Miami Valley Laboratories, and their purity was verified by NMR and TLC. There was no evidence for significant levels of impurities in the FD mass spectra of any of the compounds studied. Procedure. Concentrated solutions of the zwitterion were prepared in suitable solvents such as water, methanol, or a chloroform/methanol mixture. We did not accurately measure the concentration of the zwitterion in these solutions. However, we typically prepare solutions with concentrations of at least 1 pg/pL to ensure adequate sensitivity for analysis on our instrument. Sampleswere loaded onto high-temperature-activatedFD emitters ( I I , 1 2 ) by dipping or by direct syringe loading (13).The FD emitter was then inserted into the ion source and positioned @ 1980 A r w i c a n Chemlcai Society
