Anal. Chem. 1993, 65, 1257-1266
1257
Wavelength-Specific Resonance-Enhanced Multiphoton Ionization for Isomer Discrimination via Fragmentation and Metastable Analysis Scott T. Fountain and David M. Lubman' Department of Chemistry, The University of Michigan, Ann Arbor, Michigan 48109
In this work, it is shown that resonance-enhanced multiphoton ionization (REMPI) at two wavelengths, 266 and 193 nm, can be used to produce fragmentation from which isomers can be distinguished. In particular, metastable ion energy analysis using a reflectron time-of-flight device is used to analyze the metastable ion products arising from REMPI-induced fragmentation during the flight time to the detector. It is demonstrated that metastable ion production is particularly sensitive to variations in wavelength and that unique metastable ions are produced at the correct choice of wavelength that allows easy discrimination of isomer pairs. Such discrimination is demonstrated for the isomers Leu-Tyr and Ile-Tyr at 193 nm, while it is shown that the isomers cannot be distinguished at 266 nm. Likewise, it is shown that the isomer pair Trp-Ala and Ala-Trp can be discriminatedon the basis of their metastable fragmentations at 266 nm but not at 193 nm. Further, a comparison of fragmentation by REMPI is made to electron impact (EI). It is demonstratedthat metastableions cannot beeasily detected by E1 for these compounds. It appears that metastable ion detection is strongly enhanced in REMPI due to the narrow-band excitation of the ion manifoldand the relative sensitivity which may be up to 6 orders of magnitude greater in REMPI than E1 under comparable conditions. INTRODUCTION Resonance-enhancedmultiphoton ionization (REMPI) has been shown to be a technique with unique properties for chemical analysis. In particular, REMPI has served as an ionization source for mass spectrometry capable of solving difficult problems in detection and structural analysis, including isomer,'-j isobar? and isotope discrimination.7-11 The unique capabilities of REMPI as an ionization source (1)Klimcak, C.; Weasel, J. Anal. Chem. 1980,52, 1233-1239. (2)Lubman, D. M.;Naaman, R.; Zare, R. N. J . Chem. Phys. 1980,72, 3034-3040. (3)Hudgens, J. W.; Seaver, M.; DeCorpo, J. J. J. Phys. Chem. 1981, 85,761-762. (4)(a) Rhodes, G.;Opsal, R. B.; Meek, J. T.; Reilly, J. P. Anal. Chem. 1983,55,280-286.(b) Tembreull, R.; Lubman, D. M. Anal. Chem. 1985, 56,1962-1967. (5)Tembreull, R.; Sin, C. H.; Li, P.; Pang, H. M.; Lubman, D. M. Anal. Chem. 1985.57.1186-1192. (6)Miller, C. M.; Nogar, N. W.; Goncarz, A. J.; Shields, W. R. Anal. Chem. 1982,54,2377-2378. (7)Leutwyler, S.; Even, U. Chem. Phys. Lett. 1981,81,578-581. (8)Bwsl, U.;Neusser, H. J.; Schlag, E. W. Sprrnger Ser. Opt. Scr. 1979 - - . - , 164-174 - - . - . .. (9)Miller, J. C. Anal. Chem. 1986,58, 1702-1705. (10)Lubman, D. M.;Zare, R. N. Anal. Chem. 1982,54, 2117-2120. 0003-2700/93/0365-1257$04.00/0
are due to the spectroscopic dependence of the ionization process and the use of narrow-band tunable laser radiation. The process generally used in previous work has been an initial resonant two-photon ionization (RZPI),where the first photon is in resonance with a real electronic state and the second photon produces ionization, if the total energy exceeds the ionization potential of the molecule. Although ions are the final product detected, the intensity of the signal observed reflects the absorption/excitation spectrum of the So S1 transition. The result is that, in principle, molecules can be selectively ionized in a mass spectrometer on the basis of their electronic absorption spectra. In practice this technique must usually be coupled with supersonic jet expansions in order to cool the molecules to obtain sharp spectral features required to realize the potential selectivity of the technique.12 The selectivity obtained by this methodology has been used for both identification and discriminationof geometricisomers of cresol4 and dich1orotoluene.j Indeed, the selectivity of this methodology has even been used to distinguish between rotational isomers of molecules such as hydroquinone and catechoP3and tryptophan.14 Although potentially a powerful tool for analysis, this methodology does require a tunable laser and a detailed knowledge of the spectroscopy of the target molecule. In addition, it may not be broadly applicable to molecules which do not have sharp spectral features even under jet conditions either because their spectra are intrinsically diffuse and unres01vable'~J~ or because the molecules become increasinglydifficult to cool as the size of the molecule increases.17 Alternatively, REMPI can induce fragmentation which can be used for structural analysis. In this process, the molecular ion absorbs additional photons where the total energy exceeds the appearance potential for formation of fragment ions.lG21 These ions can in turn absorb additional photons and fragment further. This mechanism is known as the "ladder switching" model.lg The important consequence of this model is that REMPI/D leads initially to a sharp energy distribution in
-
(11)Lubman, D. M.; Tembreull, R.; Sin, C. H. Anal. Chem. 1985,57, 1084. (12)Levy, D. H.;Wharton, L.; Smalley, R. E. In Chemical and Biochemical Applications of Lasers; Academic: New York, 1977;Vol. 2, P 1. (13)Dunn, T. M.;Tembreull, R.; Lubman, D. M. Chem. Phys. Lett. 1985,121,453-457. (14)Rizzo. T. R.; Park, Y. D.; Peteanu, L.; Levy, D. H. J. Chem. Phys. 1986,84,2534-2541. (15)Tembreull, R.; Sin, C. H.; Pang, H. M.; Lubman, D. M. Anal. Chem. 1985.57,2911-2917. (16)Brady, B. B.; Peteanu, L. A.; Levy, D. H. Chem. Phys. Lett. 1988, 147,538-543. (17)Li, L.; Lubman, D. M. Appl. Spectrosc. 1989,43,543-548. (18)Bernstein,R.B.J.Phys. Chem. 1982,86,1178-1184,andreferences cited therein. (19)Boesl, U.;Neusser, H. J.; Schlag, E. W. J. Chem. Phys. 1980,72, 4327. (20)Gobeli, D. A.; Yang, J. J.; El-Sayed, M. A. Chem. Reu. 1985,85, 529-554. (21)Baer, T . In Advances in Chemical Physrcs; Prigogine, I., Rice, S., Eds.; John Wiley & Sons, Inc.: New York, 1986;Vol. 64,pp 111-202. 0 1993 American Chemical Society
1266 0 ANALYTICAL CHEMISTRY, VOL. 65, NO. 9, MAY 1, 1993
the parent ion, especially if the molecules are initially cooled in a jet expansion. The result is that the ions are excited to a narrow range of intermediate stateswhich should yield very specific products. This is in contrast to electron impact (EI), for example,where a range of energiesis placed intoa molecule. The fragmentation produced via REMPI can thus sometimes provide structural information that can allow for isomer di~crirnination.~-*~-~~ However, in most cases, isomerization occurs before fragmentation so that, as shown by Li,Z5such discrimination cannot be achieved at a given wavelength. In this work, we will demonstrate the use of multiwavelength REMPI for discrimination of isomers and as a means more generally for providing enhanced information for structural analysis. The key to this work is the idea that narrow-band laser radiation can be used to induce a very specific excitation of ions created in the REMPI process. The result should be that, by changing the energy placed in the ion, the fragmentation and the rate at which these ions form can be selectively varied. This could be expected to occur since, a t any given energy, the different fragments appear accordingto the rate of formation and whether the appearance potential for their formation has been exceeded.2l By changing the energy placed into the ion, the position on the k(E)curve for each fragment ion will also change, thus enhancing the formation of some ions a t the expense of others. In addition, as the energy is increased, new fragments appear as their appearance potential is reached. Thus, by varying the laser energy (at constant power), one can vary the fragmentation induced and the rate at which each of these fragments is formed. This phenomenon has been observed even over a very narrow energy range for benzene and 1218*26 and azuleneand naphthalene,2where small changes in energy, i.e., 1-2 A, produced a variation in fragmentation ratios. In more recent work, significant changes in wavelength resulted in considerable changes in the fragments produced for p-nitroaniline and pnitroanisole27 and p-chloroaniline and diphenyl ether,2s where specific wavelengths were found to enhance the formation of specific fragments at the expense of others. More critically though, small changes in excitation energy placed into an ion have been shown to result in relatively rapid changes in the metastable ion fragments formed. In the case of the REMPI/D of aniline it has been shown that the energy distribution for the metastable ion decay processes occurring can be tuned by varying the wavelength of the absorbed photons. The result was that the fragment ions C5H5+and CzH3N+,for example, were too small for metastable ion mass determination at 264 nm, but were strongly enhanced by excitation at 277 nmSz9 Herein, we demonstrate the use of two wavelengths of laser radiation at 266 and 193 nm for specific excitation and fragmentation of several isomeric dipeptide pairs. It is demonstrated that this specific excitation can be used to discriminate among isomers both by variations in fragmentation and more importantly by differences in metastable ion (22) Parker, D. H.; Bernstein, R. B. J. Phys. Chem. 1982,86, 60. (23) Kuhlewind, H.; Neusser, H. J.; Schlag,E. W. J.Phys. Chem. 1985, 89, 5600. (24) Williams, E. R.; McLafferty, F. W. J . Am. SOC.Mass Spectrom. 1990, I , 361-365. (25) Li, L.; Zhang, J.-Y.; Nagra, D. S.;Wang, A. P. L. Int. J. Mass Spectrom. Ion Processes 1991, 110, 103-122. (26) Zandee, L.; Bernstein, R. B.; Lichtin, D. A. J. Chem. Phys. 1978, 69., 3427. . (27) Zhu, J.; Lustig, D.; Sofer, I.; Lubman, D. M. Anal. Chem. 1990, 62,2225-2232. (28) Kinsel, G. R.; Segar, K. R.; Johnston, M. V. Org. Mass Spectrom. 1987, 22, 627-632. (29) Kuhlewind, H.; Neusser, H. J.; Schlag, E. W. J.Chem. Phys. 1985, 82, 5452. ~~
formation.m Using a reflectron e n e r g y a n a l y ~ e rit,is ~ shown ~~~ that the difference in metastable ion fragment formation is particularly sensitive to variations of the laser excitation energy. Indeed, it is demonstrated that this methodology can provide important structural information for identification and can be used to distinguish between isomers of leucine and isoleucine, for example. In previouswork, REMPI had been shown capable of distinguishing these isomers on the basis of minor differences in fragmentation a t low mass produced at 266 nm by a high-intensity laser source.34,35 In related work, differences in the REMPI/D fragmentation at 266 nm and in the ion peak shape due to metastable ion formation in a linear time of flight (TOF) were used to distinguish between isomers of CBZLeu-Ala-OCH3and C B Z Ile-Ala-OCH3.m Differences in metastable decay rates based upon the ion peak shape observed in a linear TOF have also been used to discriminate isomers of o-, m-,and p-chloroaniline,37 and isomers of o-alkylated thymidines have been discriminated on the basis of the decompositionrate of several metastable ions in their secondary ion time-of-flight mass spectra.32 In this work, large changes in metastable ion formation are demonstrated for easy discrimination of Leu and Ile and other isomers as a function of wavelength using REMPI/D in combination with reflectron energy analysis. EXPERIMENTAL SECTION The experimental apparatus was similar to that described previ~usly.~~ It consisted of a reflectron time-of-flight mass spectrometer (Re-TOFMS) equipped with a pulsed supersonic valve source (R. M. Jordan, Co., Grass Valley, CAI. The pulsed valve (PSV) was used both to cool the analyte molecules and to transport the sample to the ionizationregion of the spectrometer. An external reservoir and mechanical pump were connected in series with the PSV so that the nozzle back pressure could be varied from 50 to 6800 Torr. Samples were introduced using pulsed IR (10.6 pm) laser desorption from a Macor probe with entrainment into the jet expansion. The pulsed desorption methodology has been described previously.3a1 Differential pumping between the PSV and the ion source was provided through a stainless steel 1.4-mm-diameter skimmer. A 6-in. diffusionpump was used toevacuate the region above the skimmer while both the ion source and flight tube were evacuated through a 4411.diffusion pump to a working pressure in the flight tube of -3 X lo4 Torr. Several different laser systems were used for ionization. A Questek Model 2120 ArF excimer laser was used to provide 193nm laser radiation. A Quanta-Ray DCR-3frequency-quadrupled Nd:YAG laser provided 266-nm laser radiation whereas wavelengths 280.6 and 286.4 nm were generated by a Nd:YAG pumped frequency-doubled dye laser (Quanta-Ray,PDL-1). Two triggering schemes were used. The ArF excimer laser was triggered externally since in the self-triggering mode there was no "pretrigger" pulse. The PSV trigger monitor was used to trigger (30) Kuhlewind, H.;Neusser,H. J.;Schlag,E. W. Int.J.MassSpectrom. Ion Phys. 1983,51, 255-265. (31) Boesl, U.; Neusser, H. J.; Weinkauf, R.; Schlag, E. W . J. Phys. Chem. 1982,86, 4857-4863. (32) LaFortune, F.; Ens, W.; Hruska, F. E.; Sadana, K. L.; Standing, K. G.; Westmore, J. B. Int. J. Mass Spectrom. Ion Processes 1987, 78, 179-194. (33) Brunelle, A.; Della-Negra, S.; Depauw, J.; Joret, H.; LeBeyec, Y. Rapid Commun. Mass Spectrom. 1991,5, 40-43. (34) Engelke, F.; Hahn, J. H.; Henke, W.; Zare, R. N. Anal. Chem. 1987,59, 909-912. (35) Li, L.; Lubman, D. M. Appl. Spectrosc. 1988,42, 411-417. (36) Segar, K. R.; Johnston, M. V. Org. Mass Spectrom. 1989,24,176182. (37) Kinsel, G. R.; Johnston, M. V. Anal. Chem. 1988,60,2084-2089. (38) Lustig, D. A.; Lubman, D. M. Int. J.Mass Spectrom. Ion Processes 1991, 107, 264-280. (39) Tembreull, R.; Lubman, D. M. Anal. Chem. 1986,59,1082-1088.
(40)Grotemeyer, J.; Walter, K.; Boesl, U.; Schlag, E. W. Org. Mass Spectrom. 1986, 21, 645. (41) Boesl, U . ;Grotemeyer, J.;Walter, K.; Schlag,E. W. Anal. Instrum. 1987, 16, 151.
ANALYTICAL CHEMISTRY, VOL. 65, NO. 9, MAY 1, 1993
I
RE-TOF
I - I
POWER SUPPLY
I
I
I
I
Repeller
--
BIAS IN
REMOTE
PULSER
Flguro 1. (a) Pulsing scheme for the electron gun source; (b) timedependent voltage applled to extractlon plate.
the excimer laser at 10 Hz through a delay generator (California Avionics Laboratories, Inc., Palo Alto, CA) allowing ionization coincident with the molecular beam. When the Nd:YAG laser was used, triggering of the experiment was provided via the internal flash lamp oscillator operating at 10 Hz. A variable internal delay in the PSV, triggered by the oscillator, provided the necessary time adjustmentto allow laser ionization coincident with the arrival of the pulsed jet. The PSV delay could be finely adjusted to optimize sample ionization. In all the UV ionization experiments, the laser beam was focused into the ionization source through a converging and diverging lens telescope providing a beam diameter of 2-3 mm. REMPI spectra typically gave a resolution of at least 1400 using the reflectron. Electron impact (EI) spectra were obtained with an electron gun assembly (R. M. Jordan, Co.). The electron gun (EGUN) consisted of a removable filament assembly directly connected to a 2.75-in. Conflat flange. This configuration allowed easy conversion of the experiment from REMPI to E1 ionization. The filament assembly was a self-contained unit consisting of a collector, focusing element, and filament, making replacement straightforward. Ceramic feedthroughs on the flange doubled as spacers, allowing the filament assembly to sit flush against the ion extraction region. Variable electron beam energies of up to 100eV were obtained with the EGUN with a maximum emission current of 1.4 mA. Since the EGUN produced a continuous electron beam, it was necessary to extract ions by pulsing the ion source. This was accomplished via a remote pulser placed on the extraction plate (See Figure la). Voltage was applied to both the repeller plate and the extraction plate (the latter through the remote pulser) from the same power supply through an SHV 'T". This guaranteed an identical voltage application to both the repeller and extraction plates independent of any supply fluctuations. During ionization an identical voltage was applied to both source plates; i.e., no pulse was applied to the extraction plate (See Figure lb). Thisvoltagewas typically +1500V. After a variable duration (-4 gs) a voltage pulse of -300 V was applied to the remote pulser, dropping the extraction plate to +1200 V, thereby extracting the ions into the flight tube. The rise time of the extraction pulse was 10-15 ns. The ion source remained in this extraction configuration until the remote pulser was again triggered. A resolution typically of 700-800 was obtained using E1 ionization with pulsed ion extraction. Energy analysis of both stable and metastable ions was accomplished using the reflectron as an "energy filter".31-:1,1 Initially a complete reflectron time-of-flight (Re-TOF) mass spectrum was obtained using the reflectron in the 'standard" mode; Le., the reflectron voltages for both grid regions were chosen so that all ions were reflected to the detector with improved resolution. Ions produced in either the source (stable ions) or flight tube (metastable ions) were observed under standard reflectron conditions. Typical reflectron voltages were +800 and +1350 V for the first and second grid regions (VRI and VRd, respectively, when an extraction voltage of +1300 V was used. N
1259
After a complete ion spectrum (Le., showing both stable and metastable ions) was obtained,thevoltage on the second reflectron grid was lowered so that only metastable ions produced by decay in the field-free drift region of the mass spectrometer could be detected. Ions formed in the ion source, having the full extraction energy, were not reflected back to the detector. This procedure provided two complimentary spectrawhich qualitativelydefined the origin of each ion in the spectrum as being from either the ion source or metastable ion decay in the drift tube. Note that the voltages applied to the first and second grid regions of the reflectron were empirically determined so that a minimal lowering of VR2would produce the most rapid signal falloff for the stable ions while still providing reasonable resolution. Once a metastable ion spectrum was obtained, the reflectron was converted to a 'hard" reflection ion mirror. This configuration allowed energy analysis of both metastable and stable ions. For hard reflection, voltage was applied to only one grid region of the reflectron while the other was grounded. A reflection voltage slightly greater than the extraction voltage was used. By lowering the voltage of this single grid and observing the signal falloff for an ion, it was possible to determine the energy necessary to reflect the ion back to the detector. This provided a measure of the ion's kinetic energy. Measured energies were typically reproducible within less than 0.5%. By measuring the energy of both stable and metastable ions, it was possible to determine the fraction of energy a metastable ion acquired from a parent/daughter dissociation beyond the ion source. The kinetic energy of a daughter ion (m*) produced by a metastable ion decay of a parent ion (M+) in the flight tube is given by (m*/M+)[&+], where EM+is the kinetic energy of the parent ion prior to d i s s o c i a t i ~ n . ~By ~~~l using the reflectron as an energy analyzer, it was possible to determine the kinetic energy of both the metastable ion and parent ion, assuming that the parent ion had a stable component; i.e., EM+was determined by the source voltage used for ion extraction. Given additional information on either the parent or metastable ion mass (e.g., metastable ion broadening or ion flight time), it was possible to determine the fragmentation mechanism(s) of the molecule involving metastable ion decay. A dual 40-mm microchannel plate detector was used to detect the ions. The signal was displayed and averaged on a LeCroy 9400A 175-MHzdigital oscilloscope (DOSC). A 20-shot average was typically used. A 50- or 100-gs time window was averaged on the DOSC. Time zero for the ion flight times were given by the laser 'sync out" for the REMPI experiments and the remote pulser trigger for the E1 experiments. The averaged signal was then stored on an IBM-compatible PC using the LeCroy mass storage program. After the spectrum was stored, a precise measurement of the flight time of each peak was determined by expanding the spectrum directly on the DOSC. Using a background signal of known mass, the spectrum was then calibratedto allow conversion of the ion flight times to mass units. The E1 spectra typically contained calibration peaks at m/z 32 (02+),28 (N2+),14 (N+), and 4 (He+) when He was used as the PSV carrier gas. The REMPI experiments produced much less background signal. However, it was common to observe a low-intensity signal from the PSV carrier gas resulting from photoinduced electron impact ionization in the source. In addition to this weak photoinduced E1 background, the REMPI spectra often contained artifacts, possibly due to diffusion pump oil back-streaming. These background peaks were then used for mass calibration. Since the flight time of an ion is directly proportional to the square of its mass, it is possible to calculatethe linear regression expressing the ion flight times as a function of (ion using the known background signal. The regression typically provided a correlation coefficient greater than 0.999 999. The ion flight times were then converted to atomic mass units. The internal nature of this calibration scheme (i.e., individualcalibration using peaks within each spectrum) is believed to provide accurate ion flight time to rn/z conversion. It should be noted that the delay between ion extraction and DOSC triggering is not negligible and can range from 0.05 to 4.5 gs, depending on the triggering scheme. This delay is accounted for in the y-intercept constant of the regression. For accurate mass calibration, this offset must be considered in the total flight time of the ions.
ANALYTICAL CHEMISTRY, VOL. 65, NO. 9, MAY 1, 1993
1260
206 nm MPI of b u - l y r
? H,N-CH.
-
M+
?
C-NH-CH-COOH
yr-cn- C - w - c n - c m
P" H&"'CH,
($ ffi 107 276
f
I.) All lorn
I.' ~
M+
1 ,
1 .
J,..
ti.) Metarnabla
M+-->86 A -
20
30
40
A
266 nm MPI 01 I b l y r
193 nm MPI of L.u-Tp
A
J
30
40
40
50 BO nrnM(-mIgM ( u w )
.
50 80 TlmM1-fllgM ( W )
30
20
70
I
70
I
4
M+->ffi
20
80
70
80
50
Tlmeof-fllght (uaec)
70
80
30
20
40
286 nm MPI of bu-Tyr (domnp)
50 d0 Tlme-of-FllgM (usec) 193 nm MPI of l k l y
170 30 I
Ill) Melastables
20
30
I
170-->113,
40
M
+--> 86
\
50 80 nmwf-flight (urec)
/276->170
70
I 80
11
,
fi
20
30
I.
L .
11
I.
248
M+
1
40
50 80 Tlmo.of-FIIgM(w.0)
70
I
Flguro 2. Resonanceenhanced multiphoton lonlratlonlfragmentationmass spectra and accompanylng reflectron metastable Ion analysis for the following: (a) Leu-Tyr at 266 nm; (b) Ile-Tyr at 266 nm; (c) Leu-Tyr at 266 nm accompanied by thermal decomposition products; (d) Ile-Tyr at 266 nm accompanied by thermal decomposition products; (e) Leu-Tyr at 193 nm; (f) Ile-Tyr at 193 nm.
The samples in this study were either prepared in a thick glycerol matrix or made intoa thin film on the probe. The sample preparation method used depended on which process gave the best desorption. In the former, a drop of glycerol was placed on the Macor probe tip and the probe dipped into the analyte. A stainless steel spatula was used to mix the analyte and glycerol directly on the probe tip. Approximatelyequal volumes of analyte and glycerol were used to form a thick paste. The sample thickness at the tip of the probe was -0.5 mm. After analysis the probe tip was rinsed with methanol, the Macor filed down slightly, and the probe rinsed a second time. This eliminated contamination from one sample to another. In the thin-film process, a dilute solution of the dipeptide in methanol (1-10 mg/mL) was made and a drop placed on the probe tip. The methanol was allowed to air-dry, leaving a thin dipeptide film behind.
Disposable gloves were used during sample handling, and all sample preparation was performed under a fume hood. With the exception of Ile-Ty-r, which was purchased from Schweizerhd, Inc. (South Plainfield, NJ), and Tyr-Ile, which was purchased from ICN Biochemicals (Cleveland, OH), all compounds were purchased from Sigma Chemical Co. (St. Louis, MO).
RESULTS AND DISCUSSION In panels a-f of Figure 2 are shown REMPI/D mass spectra for the isomers Leu-Tyr and Ile-Tyr. At a reduced laser power density, soft ionization can be observed with production of only M*+at 266 nm for these compounds, although the M + is difficult to produce with any significant intensity at 193 nm. However, in this study the laser power densities have been raised to (1.3-5.4) X 1 0 7 W/cm2 in order to produce
ANALYTICAL CHEMISTRY, VOL. 65, NO. 9, MAY 1, 1993
fragmentation for isomer identification. In Figure 2a,b such fragmentation has been produced a t X = 266 nm (4.8 X lo7 W/cm*) for the isomers Leu-Tyr and Ile-Tyr desorbed from a thin film and expanded in a COZsupersonic jet expansion. The fragmentation patterns that result and the accompanying metastable ion spectra are essentially identical for each isomer. In Figure 2c,d we have repeated the same experiment, except that the desorption laser power has been increased to produce thermal decomposition peaks which are then ionized by REMPI. The resulting spectra are a combination of REMPI mass spectra that contain both thermal decomposition peaks and the fragmentation of the original dipeptide The resulting fragmentation spectra for the two isomers are very similar with only minor variations in the relative intensities of some of the peaks. The latter appears to be due to variations in both the desorption and ionization processes, which are highly dependent on the laser intensity, rather than any intrinsic differences in the laser ionization/fragmentation mass spectra of the two isomers. In addition, the metastable fragmentations observed are essentially identical for the two isomers. In contrast are Figure 2e,f where the same experiment was performed at X = 193 nm (3.2 x lo7 W/cm2). Although some variation in relative intensities between the two isomers is observed, the fragmentation patterns are very sensitive to the laser power density. This is due to the fact that these fragments are formed from the absorption of many photons and the power dependence for formation should thus be an I n dependence. Even small changes in power density may result in large changes in measured intensity for the different fragments, depending on how many photons are needed for formation. Thus, the use of small differences in relative intensity in the REMPI fragmentation patterns is not a reliable method for distinguishing these isomers. However, the accompanyingmetastable ion spectra are clearly distinct and can be used to easily differentiate these two isomers at X = 193 nm. Thus the metastable ion spectra are highly dependent on the wavelength chosen and discrimination of isomers can be achieved based upon the correct choice of laser wavelength. The metastable ions observed in panels a-f of Figure 2 were generally of sufficient intensity to allow energy analysis which can provide detailed information on the mechanism of formation of these fragments as shown in Scheme IA,B and Table I. For example, the 266-nm laser-induced REMPI metastable ion spectra of Leu-Tyr and Ile-Tyr produced five quantifiable metastable transitions, some of which are due to fragmentation of the thermal decomposition products (compare a with c and b with d in Figure 2). Two primary decay channels produced the metastable ions observed at 266 nm: (1)the formation of an immonium ion ([H,N=CH(R)]+)from the molecular ion and its subsequent metastable decay to m/z 30 and (2) loss of the tyrosine R functional group from the cyclized dipeptide, produced as the thermal decomposition product during laser desorption, followed by cleavage of the Leu/Ile R group to give further loss of mlz 57 (Scheme IA). Since neither of these decay channels directly involves internal rearrangement or cleavage of the Leu or Ile functional group, the resulting 266-nm laser-induced REMPI metastable ion spectra were indistinguishable. In contrast to the lower energy REMPI mass spectra, distinct metastable ion spectra were observed for Leu-Tyr and Ile-Tyr when X = 193 nm was used. In this case, the metastable fragments (42) Li, L.; Lubman, D. M. Rapid Commun. Mass Spectrom. 1989,3, 12. (43) Rose, M. E.;Johnstone, R. A. W. Mass Spectrometryfor Chemists and Biochemists; Cambridge University Press: Cambridge,England, 1982. (44) Aplin, R. T.; Jones, J. H.; Liberek, B. Chem. Commun. 1966,21, 794-796. (45) McLafferty, F. W.; Levsen, K.; Wipf, H.-K. Org. Mass Spectrom. 1974,8, 117-128.
1261
Scheme 18
A
/
0
I
I
NHz-CH-C-NH-CH-COzH
I
Cab
B
CHz-IC,H,I-OH
al Both Isomers
bl Ile-Tvr Only ~~
NH2-CH--R2 H-C-CH, I
-
86
0
0
Y-E:C I H31
?HZ
FHz
CH3
CH3
m.
/!CH3 CH3
1 69
(A) Metastable ion fragmentation mechanisms of Leu-Tyr and Ile-Tyr observed using 266-nm REMPI. Metastable transitions are indicated by m*. (B)Metastable ion fragmentation observed using 193-nmREMPI for (a)both Leu-Tyr and Ile-Tyr and (b) Ile-Tyr only. (I
observed resulted from rearrangement and fragmentation of the Leu and Ile center (see Scheme IB). Both isomers showed a metastable ion decay of the N-terminal immonium ion, NH2=CH(C4Hg)+, to give NH2=CH2+. However, Ile-Tyr contained an additional metastable ion due to the formation of CH=C(CH~)CHZCH~+ from a unique stable ion at mlz 86. Thus, by increasing the REMPI energy from 4.66 (266 nm) to 6.42 eV (193 nm), it was possible to induce distinct metastable ion fragmentation within Leu-Tyr and Ile-Tyr, allowing isomeric discrimination based on their metastable ion spectra. In comparison, in Figure 3a-d are shown in the REMPI/D mass spectra of the isomers Tyr-Leu and Tyr-Ile at X = 266 and 193 nm. Note that the X = 193 nm spectra show no Me+ signal, but rather a peak at (M+- 1)at m/z293. In this case the REMPI/D mass spectra are very similar for the two isomers at both wavelengths. No information is provided in the metastable ion spectra of either isomer at X = 193 nm that allows isomeric discrimination. The minor differences in the metastable ion spectra in Figure 3c,d are due to variations in sensitivity and are not reproducible. Quantifiable metastable ions were observed under more intense desorption conditions, but thermal decomposition became the dominant mechanism. Under no conditions were unique metastable fragments observed for Tyr-Leu and Tyr-Ile using 193-nm REMPI. The difficulty in producing a spectrum for Tyr-Leu or Tyr-Ile that contained a quantifiable molecular ion may contribute to this effect. However, the metastable ion spectra of Tyr-Leu and Tyr-Ile at 266 nm provide information allowing isomeric discrimination. Both isomers
1262
ANALYTICAL CHEMISTRY, VOL. 65, NO. 9, MAY 1, 1993
Table I. Major Metastable Transitions (MPI)" compound
MW
Tyr-Leub
294
266 nm M+ M+
Tyr-Ileb
294
M+
193 nm see text
220 (277 - R
-
+
L ~
-
see text
M+ --* Leu-Tyr
294
M+
Ile-Tyr
294
M+ --*
-
86 (Ai) 86 (Ad 86
30
-
--*
30 69
(AI) Ala-Trp
M+ --*
275
130 --*
103
(RT~~) Trp-Ala
M+
275
M+
+
-
130 Metastables from thermal decomposition products not included.
193 nm MPI of Tyr-Leu 2i
36
IL
86 107-108
M+-->277
20
M+-->l88
30
40
I
I1
70
266 nm MPI ol Tyr-llr
193 nm MPI of Tyr-llr I$N-CH-C
CH,
La
138
-cI'IAi''om t
$I -
'
I
M+-->lBB
30
40
50
eo
\
M+->277
70
I
70
I
36
? -NH-CH-CWII I
CH-CH,
a4
-t
187-188
ii.) Melaslableo
40
6 k: I
-
a4
86
30
2T
I
7%
4
20
CH-CH,
. '
L
Tlmsof-Flight ( U w )
-NH-CH-CWH
CH
20
80
ti.) Melastables
50 60 Tlmeof-Fllghl (usec)
" H$-CH-C
1I
\
60
50
77
193-nm spectrum shows ( M + - 1) as highest mass peak.
266 nm MPI of Tyr-Leu
11.) Melastables
77
258 (weak signal) (M+ - "3) 103
(RT~~) a
-
86 107-108
-
188
i.)Alllorn It
.Id
.
1
1
ik) Melpltabb
70
Tlmsof-FllgM (usec)
I
20
30
40
50
60
Tlmsof-Fllghl (usec)
Flguro 3. Resonancwnhanced multiphoton lonlratlonlfragmentatlonmass spectra and accompanying reflectron metastable Ion analysls for (a) Tyr-Leu at 266 nm, (b) Tyr-Ile at 266 nm, (c) Tyr-Leu at 193 nm, and (d) Tyr-Iie at 193 nrn.
show a strong metastable fragment signal due to loss of NH3 from the N-terminal end of the dipeptide in addition to a metastable ion at m/z 188due to cleavage of the tyrosine side chain (Scheme IIA). An additional metastable ion decay is observed in Tyr-Leu, however. This decay may be due to further fragmentation of the (M+ - NH3) metastable ion to give a species a t mlz 220 via cleavageof the leucine side chain (Scheme IIB). However, no stable m/z 220 fragment was observed in the total ion spectrum of Tyr-Leu. Since simple cleavage of the C-terminal side chain in Tyr-Leu would
produce a leaving group with a tertiary carbon, this may account for its observance in Tyr-Leu but not in Tyr-Ile. By again using a variable REMPI/D wavelength, unique metastable species were produced in two isomeric dipeptides. It should also be noted that the isomer pairs Leu-Tyr/Tyr-Leu and Ile-Tyr/Tyr-Ile, where tyrosine is on either the C- or N-terminal end of the dipeptide, are easily distinguished by clearly different fragmentation patterns at 266 nm. Wavelength-dependent metastable ions were also observed for dipeptides differing in their primary structure, i.e., their
ANALYTICAL CHEMISTRY, VOL. 85, NO. 9, MAY 1, lSS3
Scheme 11’
A
.
1 HIN-CH,-C
CH-C
II
188
1
B -NH-CH-COOH
-NH - CH-COOH
CH-c
I1
B- - No~ = C H - C O O H
CH
I
OH
I
0 I
OH
I
(A) Metastable ion fragmentation mechanisms of Tyr-Leu and Tyr-Ile observed using 266-nm REMPI. (B)Unique metastable ion decay observed in the 266-nm REMPI spectra of TyrLeu. 0
amino acid sequence. For example, the dipeptide pair AlaTrp/Trp-Ala was distinguishable using 266-nm REMPI/D. The REMPI/D mass spectra for these isomers are shown in Figure 4a-d. In Figure 4c,d, ionization and fragmentation was performed for the isomers Ala-Trp and Trp-Ala a t A = 193 nm. Both the REMPI-induced fragmentation patterns and metastable ion peaks obtained are very similar. The relative differences in fragmentation are highly dependent on laser power, and even small changes in the light intensity make this method an unreliablemeans of discriminatingthese compounds. However, a t a lower photon energy of A = 266 nm (Figure 4a,b) these isomers are easily discriminated. The fragmentation patterns obtained for these isomers are different, and this is reflected by the presence of an intense metastable ion peak a t m/z 187 for Ala-Trp, which is not present for Trp-Ala. Likewise,an intense m/z 258metastable ion is observed for Trp-Ala, which is not observed for AlaTrp. Thus, the correct choice of wavelength, which may be very specificto each particular molecular system, can provide structural information that allows unique discrimination among its isomers. Energy analysis of these ionic species reveals the mechanism for the appearance of the metastable ion fragments as shown in Scheme 111. In both cases at 193 nm, stable formation of the quinolinium ion a t m/z 130 was observed due to cleavage and rearrangement of the tryptophan side chain (Scheme IIIA). This ion was observed as a highly abundant fragment for both dipeptides using 193-nm REMPI. Metastable ion decay of the quinolinium ion was observed via loss of HCN to give C8H7+ at m/z 103. Further metastable ion decay of C,3H7+ to give m/z 77 was also observed. Since these metastable ions were daughter and granddaughter ions of the quinolinium ion, they provided no information on the structure of the dipeptide aside from indicating the presence of tryptophan; i.e., both Trp-Ala and Ala-Trp produced mlz 103 and 77 metastable ions of similar intensity. Under 193nm MPI conditions in which Ma+was observed, Trp-Ala also
126s
showed a very weak (- 1% of the base peak) metastable ion decay a t m/z 258 corresponding to the loss of NH3 from the N-terminal end of the dipeptide. However, the appearance of this metastable fragment was dependent upon the stability of the molecular ion. In cases where the molecular ion was not observed, the (M+ - 17) fragment was absent from the Trp-Ala spectrum. Under such conditions, the 193-nm metastable ion spectra of Trp-Ala and Ala-Trp were identical. By changing the MPI wavelength from 193 to 266 nm it was possible to induce unique metastable ion fragmentation in both Trp-Ala and Ala-Trp. Trp-Ala showed a very intense metastable ion a t m/z 258 (M+- 17) not observed in Ala-Trp (see Scheme IIIBa) due to a loss of NH3 from the N-terminal end of the dipeptide. Likewise, Ala-Trp showed a strong metastable ion signal a t m/z 187 correspondingto a fragment analogous to the highly conjugated (M+- 17) metastable ion observed in Trp-Ala (Scheme IIIBb). Unique metastable ion decay channels were thus achieved in these dipeptides by varying the REMPI wavelength. In this case, the REMPI photon energy was lowered to induce distinct metastable ion decay, thereby allowing discrimination of two dipeptides differing in their primary structure. Comparison to Electron Impact. In comparison to the results in the previous section obtained by REMPI are the fragmentation patterns and metastable ions produced by EI. The E1 mass spectra of Leu-Tyr, Ile-Try, Try-Leu and TryIle, are shown in Figure 5 a 4 , respectively. These spectra were obtained by pulsed laser desorption from a glycerol matrix, followed by entrainment into a supersonic jet expansion and subsequent ionization by the electron gun source (70 eV). The background due to E1 ionization of the carrier gas was significantly reduced when He rather than COn was used. The mass of He is too low to interfere with any analyte fragment for the molecules studied herein, and the high ionization potential of He (IP = 24.6 eV) results in a He background signal of relatively low intensity compared to COZ(IP= 13.8eV). Inaddition,anX-Yvoltagepulsebetween the ion source and detector can be used to selectively pulse out most of the He background without eliminating key fragment peaks. For these reasons, He was the carrier gas of choice when E1 ionization was performed. In the E1 mass spectra of Figure 5 a 4 , all the ionic fragments that were observed are shown; Le., no detectable M*+was observed a t 70 eV. Even at a reduced energy of 25 eV, no signal for Ma+was observed. The E1 mass spectra contain ion peaks similar to those observed by REMPI, which might be expected since both ionization processes involve fragmentation of M*+. However, there are also significant differences in the spectra obtained in the two processes since the selection rules for ionization by 70-eV electrons are very different from those for 266-nm photons.43 The differences between the isomers Leu-Tyr/Ile-Tyr and Tyr-Leu/Tyr-Ile are minimal and are due to the variations in the desorption process rather than any real intrinsic differences between the two isomers. More significant,though, is that metastable transitions could rarely be detected within the sensitivity range used to detect metastable ions via REMPI. The narrowband laser radiation used in REMPI/D excites the ion to a narrow range of energy states, which results in enhancement of specific metastable ion forrnation.28~~9 A change in laser wavelength results in a change in this energy which enhances a different set of metastable fragments, as demonstrated in Figures 2-4. However, in electron impact the broad energy distribution placed in the ion results in a broad range of excitation. The result is that a range of metastable ion fragmentations is produced over this broad excitation. However, the actual production of any one of these metastable ions is sufficiently low so that it is not observed within the
ANALYTICAL CHEMISTRY, VOL. 65, NO. 9, MAY 1, 1993
1264
193 nm MPI of Ala-Trp
266 nm MPI of Ala-Trp
0
H,N-CICC-NH.CH~COOH
30
40
50
/I)
A
- w '
20
U#4-CU-C-NH.CU.Cm
M + ..D l 87
11,) Meteatablea .'
7
7
M+
60
70
80
20
Metasl8ble~
30
h
40
13&> 103
I
50
60
70
nrnsof-RIgM (urn)
Tirnsof-FllgM (us-)
193 nm MPI of Trp-~la
266 nm MPI of TlpAla
M+--~258 ~
103-->77
-
ll.)MetestaMeO
*
ti.)
Melastables
fO3->77
i30..>103
Flguro 4. Resonance-enhanced multiphoton ionization/fragmentatlonmass spectra and accompanying reflectron metastable ion analysis for (a) Ala-Trp at 266 nm, (b) Trp-Ala at 266 nm, (c) Ala-Trp at 193 nm, and (d) Trp-Ala at 193 nm.
Scheme 1118 130
103
77
1
1
1
QljlHOLlNlUM iOH
HN
bl Ala
Only
107
a (A) Metastable ion decay of the quinolinium ion formed via 193-nm REMPI of Ala-Trp and Trp-Ala. (B)266-nm REMPI metastable ion decay channels observed for (a) Trp-Ala and (b) Ala-Trp.
sensitivity of the experiment. The result is that REMPI can be used to enhance metastable ion formation and changes in metastable ion production as a function of wavelength can be used to identify differences in structure for isomer discrimination. It should be noted, though, that dipeptides differing in their amino acid sequence could occasionallybe distinguished
using LD-EI. Such discrimination using LD-E1 required formation of a unique ion of high relative abundance for at least one of the dipeptides. Immonium ion formation from small peptides has been reported previously using E1 and appears to be a general fragmentation mechanism for the dipeptides studied herein.44-45 Using electron impact ionization, it has been shown that immonium ion formation is enhanced when the Xle (Leu or Ile) amino acid is at the N-terminal position of an oligopeptide;this effect can be used to discriminate dipeptides differing in their amino acid sequence.45 For example, the LD-E1 spectra of both LeuTyr and Tyr-Leu contain a peak at mlz 107, due to cleavage of the tyrosine side chain in both species (see Figure 5a,c). However, the analyte base peak of the Leu-Tyr spectrum at miz 86 corresponds to formation of an immonium ion at the N-terminal end of the dipeptide. Analogous N-terminal immonium ion formation is observed for Tyr-Leu, giving a fragment at mlz 136. Since each of these dipeptides produces a unique ion of high relative abundance (indicating the N-terminal amino acid), these dipeptides could be discriminated. A similar case was observed in the LD-E1 spectra of Leu-Phe and Phe-Leu (see Figure 6a,b). Both spectra contained a fragment at mlz 91, indicating cleavage of the phenylalanine side chain. However, the base peaks of LeuPhe and Phe-Leu were mlz 86 and 120, respectively, both due to formation of the N-terminal immonium ion. These fragments unambiguously differentiate the two dipeptides. In cases where unique fragmentation was not observed between dipeptides differing in their primary structure, discrimination was not possible. For example, the 70-eV E1 spectra of Ala-Trp and Trp-Ala both show a strong fragment at mlz 130 due to cleavage of the tryptophan side chain. However, no additional fragments were observed for either
ANALYTICAL CHEMISTRY, VOL. 65, NO. Q, MAY 1, 1993
1265
70 OV El Of T p L . U
70 oV El of Lw-lyr
?
U2N-CH' C- NU- CU-COD(
H,N-CH-C
I
g-
B --NH-CH-COOI( I
OH
o
L
I.)All
L
.d
.
.
I,,
I
1
IN.) MetsRabk
5
10
15
20
25 30 35 Tlms-of-FlIgM (uwc)
107
I.)All
107
.
40
45
50
55
.LY
.
.
138
170
1
1
I
11.) Melastables
4
10
15
20
25
30 35 40 45 Tlmeof-FllgM (uwc)
55
50
80
I
65
70 OV El Of Tyr-lh
70 OV El of IlaTyr
I Hi
?C-tW-CH-CCUl
W-CU.
H@-CH-C
e
-NH-CU-COOH
I
cu-cn, I
7"cn,
CY OH
9 107
I.)All
511,) Metastables 10 15
5
10
1'5
20
25 30 35 n m w - m g h t (uwc)
40
ds
50
55
20
25
30
35
40
45
50
55
60
I
5
Tlme-ot-FllgM (UMC)
Flgure 5. Electron Impact mass spectra at 70 eV for (a) Leu-Tyr, (b) IlaTyr, (c)Tyr-Leu, and (d) Tyr-Ile.
molecule so that discrimination was not possible based on LD-E1 fragmentation. Also, metastable ions were generally not observed in these molecules under study, although lowintensity metastable ion signals were sometimes observed for Leu-Phe and Leu-Tyr. Sensitivity Measurement of REMPI vs EI. The use of metastable ion analysis appears to be critical in distinguishing various isomeric pairs via REMPI, since metastable ion formation is sensitive to the specific energy placed in the ion and the structure of the ion itself. Although attempts were made to study metastable ion formation at different E1 energies, we were unable to clearly detect such ions within the sensitivity provided by EI-TOFMS. As mentioned previously, the specific energy placed in an ion by REMPI enhances formation of specific metastable ions as opposed to EI, where a broad range of energy is provided. In addition, the ability to observe metastable ions by REMPI is enhanced by the high sensitivity and resulting S/Nratio per spectrum as opposed to El. A comparative sensitivity study was thus performed between REMPI and E1 to demonstrate this point. In order to measure the sensitivity of LD-E1 in this experiment, tryptamine was used as the target sample since the Mo+can be produced and thus directly compared to the REMPI results. The M*+is not observed in the dipeptides studied in this work by EI. Several solutions of tryptamine in methanol were made using p-nitroaniline as an internal standard. A microsyringe was used to transfer 1-rL aliquots of each sample solution to the Macor probe tip for analysis. The methanol solvent was allowed to dry and LD-E1 performed in He or COScarrier with a 70-eV E1 source with 1.4-mA emission current. Three averaged spectra of 20 shots each were obtained at each concentration. Two different methods of sensitivity calculation were used. In one case,
the average S/N ratio for each tryptamine concentration was calculated and a plot of SIN vs tryptamine concentration was made using linear regression. The sensitivitywas determined by calculating the amount of tryptamine detectable with a S/N = 3. In this method, an internal standard was not used and a detection limit of 7.2 pg was determined. In the second case, the averaged tryptamine signal was normalized with the p-nitroaniline internal standard signal. A plot of (tryptamine signal)/@-nitroaniline signal) vs tryptamine concentration was made and linear regression calculated. The average noise signal and p-nitroaniline signal were calculated and assumed to be constant. These values were substituted in a rearranged linear regressionformula so that the minimum concentration of tryptamine could be calculated at a S/N = 3. This method takes into consideration the internal standard but assumes the background and p-nitroaniline signal is constant. Under these conditions, a detection limit of 6.8 pg was calculated. In comparison, detection via REMPI waa studied for tryptamine. In this case five tryptamine solutions over three decades' concentration in methanol were made with indole as an internal standard. As in the previous measurement, 1-rL samples were applied to the probe tip and the MeOH was air dried. LD-REMPI was performed, and three 20-shot averages were taken for each sample. A plot of In (signal tryptamine/signal indole) vs In (concentration tryptamine) was shown to be linear over three decades with an r = 0.93. This measurement was initially performed off the resonant absorption of tryptamine at 280.63 nm, and a detection limit of 180pg was obtained with a S/N = 3. This is an improvement of nearly 4 X lo4 times that obtained by E1 under similar conditions. These results are similar to those reported in previous work using a quadrupole mass spectrometer.*
1288
ANALYTICAL CHEMISTRY, VOL. 65, NO. 9, MAY 1, lQ93 To ov El ol W P h .
r
?- N H - W - C W
U+-CH-C
I
I:
88 91
-i.)Nim I