rates in Table I with the relative rates of reactivity of many of the same compounds with biacetyl (I), namely: 1,l.W; 2, 0.70; 3, 0.25; 5 , 0.074; 6, 0.031; 7, 0.069. The ordering of compounds is the same with each ion: 1 > 2 > 3 > 5 > 7 > 6. Even more important, the decrease in the rate constant is greater for all compounds with C ~ H & O C O C ~ H Ythan + with CH3COCOCHr+. Most important, the decrease is greater for the larger ketones 5 , 6, and 7 with C3H5COCOC3H5-+ than for the smaller; for example, the relative rate of 2 with C ~ H ~ C O C O C ~ is H 0.61 S - ~ compared to 0.70 for its relative rate with CH3COCOCH3.+, while the relative rate of 5 drops from 0.074 with CH3COCOCHr+ to 0.015 with C3H5COCOC3Hyf. The drop-off in relative rate on going to the bulkier ion is greater for the bulkier ketones. Thus, as predicted, 1,2-dicyclopropylethanedione reacts more selectively than biacetyl: it is more sensitive to the steric environment of the carbonyl group in these molecules. A sensitivity to steric effects, therefore, appears not only as a result of selected changes in the neutral molecule, but also as a result of selected changes in the ion as well. Such considerations should be kept in mind when designing specific reagents for analytical ion cyclotron resonance studies in the future.
ACKNOWLEDGMENT We thank Bill Burnsides for assistance in synthesizing the diketone.
LITERATURE CITED (1) M. M. Bursey, J. L. Kao, J. D. Henion, C. E. Parker, and T. I. S. Hwng, Anal. Chem., 46, 1709 (1974). (2) M. M. Bursey, T. A. Elwood, M. K. Hoffman, T. A. Lehman, and J. M. Tesarek, Anal. Chem., 42, 1370 (1970). (3) E. Stenhaaen. S. Abrahamsson. and F. W. McLaffertv. Ed., "Atlas of Mass SpeEtral Data," John Wiley and Sons, New York. N.Y., 1969. J. Kelder, J. A. J. Geenevasen. and H. Cerfontain, Svnth. Commun.. 2, 125 (1972). R. C. Dunbar. D. A. Chatfield, and M. M. Bursey, lnt. J. Mass Spectrom. /on Phys., 13, 195 (1974). R. E. Ireland and J. A. Marshall, J. Org. Chem.. 27, 1615 (1962). B. Kirshna and K. K. Srivastava. J. Chem. Phys., 27, 835 (1957); J. Chem. Educ., 32, 663 (1960). K. Higashi, Bull. lnst. Phys. Chem. Res. (Tokyo). 22, 805 (1943). P. A. Dobosh. CNINDO, Quantum Chemistry Program Exchange No. 141, Bloomington, Ind. A. L. McClellan, "Tables of Experimental Dipole Moments." W. H. Freeman and Company, San Francisco, Calif.. 1963, p 208. W. Huckel, Justus Liebigs' Ann. Chem., 441, 1 (1925). J. L. Beauchamp, L. R. Anders. and J. D. Baldeschwieler. J. Arner. Chem. SOC., 89, 4569 (1967). J. M. S. Henis. J. Amer. Chem. SOC., 90, 844 (1968). J. M. S. Henis, Anal. Chem., 41, (IO), 22A (1969). M. L. Gross, P.-H. Lin, and S. J. Franklin, Anal. Chem., 44, 974 (1972). T. Su and M. T. Bowers, J. Chem. Phys., 58, 3027 (1973).
RECEIVEDfor review August 8, 1974. Accepted December 2, 1974. This work was supported by the National Institute of General Medical Sciences (GM15994) and the Alfred P. Sloan Foundation. The ICR spectrometer was purchased through funds from Hercules, Inc., the Shell Companies Foundation, the North Carolina Board of Science and Technology (159), and the National Science Foundation (GU 2059).
Proposed Method for Mass Spectrometric Analysis for Ultra-Low Vapor Pressure Compounds Robert T. Mclver, Jr.,' Edward B. Ledford, Jr., and Judith S. Miller Department of Chemistry, University of California, Irvine, Calif. 92664
A trapped ion cyclotron resonance mass spectrometer with greatly improved mass range has been developed. The most novel feature of the instrument is its remarkable ability to trap gaseous ions. At a pressure of 4.7 X lo-' Torr, 80% trapping efflciency is observed after 3 seconds. This feature allows chemical ionization mass spectra to be obtained at very low pressures. A wide variety of positive and negative reagent ions can be generated by electron impact and allowed to react with the vapors of the sample to be analyzed. Calculations indicate that sample vapor pressures as Torr can be detected by this method. low as
One of the major problems presently confronting mass spectrometry is the analysis for low volatility compounds. This problem is especially acute in applications of mass spectrometry to studies of biological importance. In general, biological compounds exhibit high molecular weight, high polarity, and the ability to form strong hydrogen bonds. Such compounds have low vapor pressures and sublime slowly even a t elevated temperatures. The traditional approach to this problem has been to chemically derivatize l
692
Author to whom reprint requests should be addressed. ANALYTICAL CHEMISTRY, VOL. 47, NO. 4, APRIL 1975
samples in order to enhance volatility. However, chemical modification not only increases the molecular weight but also is time consuming, difficult, and uncertain, especially when only small samples are available for analysis. A number of new methods have been introduced recently for the analysis for low volatility compounds. Field desorption mass spectrometry has proved useful for a large number of polar compounds of low volatility ( I , 2). Rapid heating of a sample dispersed on a Teflon surface has been shown to greatly increase the rate of sample evaporation relative to competing surface decomposition reactions (3, 4 ) . Underivatized oligopeptides have been analyzed in a high pressure chemical ionization source by inserting a solid-sample probe directly in the plasma of reagent ions (5). The analytical potential of the ion cyclotron resonance technique has been discussed previously by several authors (6-11). But despite some initial success in developing specific reagent ions, the scope of the investigations was severely restricted by the limited mass range and low mass resolution of the early instruments. In this paper, we describe a trapped ion cyclotron resonance mass spectrometer which not only overcomes these problems but also provides
Path of ions at reynance c 0
0 0 0 0
generator
Trapping voltage SIDE VIEW
Figure 1. Omegatron mass spectrometer developed by Sommer, Thomas, and Hipple
tion efficiency for resonant ions is greatly improved because both the upper and the lower plates of the analyzer cell function as ion collector electrodes; and the homogeneity of the electric fields in the cell is maintained. In the omegatron, the mass resolution may be adversely affected by the ion collector electrode which is placed inside the cell. 2) RF Irradiation. In the omegatron, the resonant ions are ejected by applying an RF voltage to the upper plate of the cell, but, in the trapped ion cell, the R F voltage is applied to two pairs of wires running parallel to the long axis of the cell. High frequency amplifiers maintain a 180’ phase difference in the R F applied to the upper and lower pairs of wires. The wires are placed in the asymptotic region of the hyperbolic electrodes in order to minimize the inhomogeneity of the quadrupolar electrostatic field in the analyzer cell.
Upper Plate, -.5V
Side Plate. ..5v
/
r
#q
ion
t
/
1
Side Plate, + 5V
..--
-
Filament Block and Grid
Path of /Resonant
Plate, -.5V
/Upper
Electron Collector
_*
-\R.F H.
Lower Piate, - 5~
Irradiation
\Lower
Wire
Plate, -.5v Electrometer
End View
Side View
Figure 2. Analyzer cell for trapped ion cyclotron resonance mass spectrometer
a unique capability for studying the gas-phase ion-molecule reactions of ultra-low vapor pressure compounds.
EXPERIMENTAL In 1949, Hipple, Somrner, and Thomas introduced an “omegatron” mass spectrometer based on the phenomenon of cyclotron resonance (12,13).A schematic drawing of the omegatron is shown in Figure 1. Ions are produced in the center of the analyzer cell by electron impact. A strong, homogeneous magnetic field constrains the ions to circular orbits in the plane perpendicular to the magnetic field. The frequency of the circular motion is
yH w, = MI c where w c is the cyclotron frequency, qlm the charge-to-mass ratio, and c the speed of light (14). Radiofrequency voltage is applied to the upper plate of the cell and is divided down through a series of guard rings in order to obtain a linearly polarized R F electric field. If the cyclotron frequency of an ion is equal to the frequency of the alternating electric field, a resonance condition is established and the ion is accelerated to larger orbital radii until it finally impinges on a collector electrode. The current of resonant ions is measured by an electrometer. Nonresonant ions do not reach the collector electrode because the radii of their cyclotron orbits remain bounded. The omegatron has been successfully applied to residual gas analysis in ultrahigh vacuum (15-17) and to airborne sampling of the upper atmosphere (18). However, low mass resolution above about 50 amu has precluded its use as a general purpose mass spectrometer. Figure 2 shows the analyzer cell used in our trapped ion cyclotron resonance mass spectrometer (19, 20). The analyzer cell is enclosed in a high vacuum stainless steel manifold which is placed between the pole caps of a 9-in. electromagnet. At a field strength of 1 2 kG,the homogeneity of the magnetic field is on the order of 2 G over the volume of the analyzer cell. The trapped ion analyzer cell differs from the omegatron in a number of respects: I ) Collection of the Ions. The upper and lower electrodes of the trapped ion analyzer cell function as the collector for resonant ions. These electrodes are shorted together internally and connected directly to an electrometer. Bias voltage is applied to the electrodes by floating the electrometer at the appropriate dc voltage. There are two principal advantages of this design: the collec-
Double resonance irradiation is an important capability of the ion cyclotron resonance technique which enables the sequence of reactions in a complex reaction mechanism to be elucidated (6, 21 ). In our trapped ion cell, the double resonance experiments can be performed using the ion ejection method devised by Beauchamp and Armstrong (22). Ions are ejected from the analyzer cell by applying an R F voltage to the side electrodes a t a frequency which excites the characteristic oscillatory motion of the ions in the direction parallel to the magnetic field. 3) Ion Trapping Efficiency. The residence time for nonresonant ions in the omegatron is only about sec, but Figure 3 shows that residence times of several seconds are achievable in the trapped ion cell. Our initial studies have shown that the ion trapping efficiency of the trapped ion cell is approximately proportional to the square of the magnetic field strength and inversely proportional to the pressure. The trapping efficiency of the analyzer cell was measured by using a pulsed mode of operation. The first pulse in the sequence is a 5-msec duration pulse of the electron beam through the center of the analyzer cell. Ions are trapped by the combination of the magnetic field and an electrostatic potential well established by appropriate dc voltages applied to the six
.1
0
.5
1.0
1.5
Residence
2D
2.5
3.0
Time (secl
Figure 3. Ion trapping efficiency of the trapped ion analyzer cell. Benzene ions, m/e 78+, were trapped at a magnetic field strength of
12 kG ANALYTICAL CHEMISTRY, VOL. 47, NO. 4, APRIL 1975
693
58
59
€0 Ma55
ASC
495
5CC
505
Scale ( a m " )
Figure 4. Demonstration spectra showing the mass resolution and mass range attained with the trapped ion cyclotron resonance mass spectrometer
electrodes of the analyzer cell. Ions are slowly lost from the analyzer cell by diffusion perpendicular to the magnetic field. After a given delay period, an R F pulse of about 10-msec duration ejects remaining ions of a particular charge-to-mass ratio and causes an ion current to register on the electrometer. Finally, all the ions, regardless of mass, are rapidly neutralized on the walls of a side electrode by a quench pulse. By temporarily inverting the polarity of the dc voltage applied to one of the side electrodes, the quench pulse destroys the trapping action of the analyzer cell. This prevents ions from one pulse sequence from overlapping into the next sequence. The whole cycle is then automatically repeated. By slowly varying the delay time between the grid pulse and the ejection pulse, one can follow the abundance of a particular mass ion as a function of time. Experiments such as this are useful for kinetic and equilibrium studies of ion-molecule reactions because the time evolution of the system can be monitored ( 2 3 , 2 4 ) . The remarkable ability of the trapped ion analyzer cell to trap gaseous ions is due to a form of the electrostatic fields within the cell which is fundamentally different from the omegatron. The end plates of the trapped ion analyzer cell are set a t the same dc voltage as the upper and lower plates in order to generate equipotentials which close upon themselves. In the plane perpendicular to the magnetic field, ions are constrained to drift slowly from one end of the analyzer cell to the other. In comparison, the omegatron is designed so that the dc voltage on the end plates is the same as on the side plates. The resulting equipotentials enable ions to drift perpendicular to the magnetic field and to strike the cell plates after a short time. 4) Mass Resolution. The maximum length of time that ions can be trapped imposes an upper limit on the attainable mass resolution (20, 25). For large mass ions, the resolution of the omegatron is limited because the residence time in the cell is only on the order of a few milliseconds ( 1 3 ) .In contrast, the mass resolution of the trapped ion analyzer cell is not limited by this factor because residence times of several seconds are routinely obtained. Figure 4 shows typical mass spectral traces obtained with the trapped ion analyzer cell a t a magnetic field strength of 12 kG. For masses on the order of 50 amu, the full width a t half height gives a mass resolution of about 3000, and a resolution of 700 is attained at 495 amu. As with all mass spectrometers, there is a trade-off between resolution and sensitivity. We have found that factors such as magnetic field homogeneity and frequency stability of the irradiating oscillator appear to be limiting the mass resolution of the present instrument. Considerable improvement in both the mass resolution and mass measurement accuracy appears possible. Most ion cyclotron resonance mass spectrometers presently in existence utilize a marginal oscillator circuit to detect the power absorbed by resonant ions rather than the actual ion current (6, 23). A marginal oscillator is a very sensitive detector but suffers a number of disadvantages and limitations which have prevented the ion cyclotron resonance technique from achieving its full analytical potential. Practical limitations in the design of marginal oscillators have limited the mass range to about 250 amu, and mass resolution is so low that most instruments are unable to resolve unit masses in the 200-amu range. Detection of high mass ions is also hampered by the fact that the intrinsic sensitivity of a marginal oscillator is inversely proportional to the mass of the ion detected (26, 27). 694
ANALYTICAL CHEMISTRY, VOL. 47, NO. 4 , APRIL 1975
There are many advantages of using an electrometer to detect the resonant ions in an ion cyclotron resonance mass spectrometer. The intrinsic sensitivity of the electrometer is independent of the mass of the ion detected, and, as shown above, the mass range of the mass spectrometer is greatly improved. Solid-state electrometers currently available commercially are very reliable and sensitive, and this provides a rugged and inexpensive ion detection system. Mass resolution and ion trapping efficiency increase as the square of the magnetic field strength (28). Thus, in scanning a mass spectrum, both of these parameters can be maximized by setting the electromagnet at its peak field strength and scanning the frequency of the RF voltage. An important additional feature of the frequency sweep mode of operation involves the rate a t which a spectrum can be sampled. Conventional instruments which use a marginal oscillator detector require that the magnetic field strength be scanned rather slowly and smoothly. But frequency scanning using an electrometer detector permits the possibility of rapid, programmed scans over several separate regions of the mass spectrum. One final comparison which can be made concerns problems caused by build-up of sample on the analyzer cell plates. The resolution and sensitivity of previous ICR cells and of magnetic deflection mass spectrometers is adversely affected by adsorption of polar samples on the walls of the system. However, with the trapped ion cyclotron resonance cell, we have not encountered problems of this type even at ambient temperature with highly polar compounds like water. The explanation for this appears to be that both the three-section ICR drift cell and the magnetic deflection mass Spectrometers use electric fields to control the speed and direction of the ions. Build-up of charges on the internal surfaces can modify the effective electric fields and alter the motion of the ions. In contrast, the electrostatic fields in the trapped ion analyzer cell serve only to trap the ions, not to steer them, and the buildup of charge on the electrode surfaces does not degrade the trapping action. The absence of ion optics and any electric field controlled drift processes makes the cell exceptionally rugged and reliable.
RESULTS The trapped ion analyzer cell is particularly well suited for studying gas-phase ion-molecule reactions at ultra-low pressures. F o r t h e bimolecular proton-transfer reaction between reagent ions AH+ and s a m p l e B, AH+
+
B
-
BH'
+
A
(2 )
the e x t e n t of conversion of AH+ t o BH+ depends upon t h e pressure of B and the t i m e allowed for t h e reaction t o occur. The long ion residence t i m e s available with the t r a p p e d ion analyzer cell permit significant e x t e n t s of conversion to be obtained even w h e n the pressure of the sample is very low. A comparison of operational p a r a m e t e r s for high press u r e chemical ionization and trapped ion cyclotron resonance m a s s s p e c t r o m e t r y is shown in T a b l e I (29-34). Since ion residence times in the t r a p p e d ion analyzer cell a r e roughly lo5 t i m e s longer than in a conventional chemical ionization source, s a m p l e pressures lo5 t i m e s lower c a n b e used. Electron multipliers a r e f a r m o r e sensitive t h a n elect r o m e t e r s in detecting ion currents, but t h e high ion collection efficiency of t h e trapped ion cyclotron resonance m a s s s p e c t r o m e t e r provides a significant compensation in d e t e r mining the overall sensitivity of t h e i n s t r u m e n t . All ions generated i n t h e t r a p p e d ion analyzer cell a r e detected because of t h e large ion collector area, but only a b o u t 0.1% of the ions generated i n a conventional chemical ionization source a r e extracted for mass analysis ( 3 4 ) .A t t h e low pressures used for trapped ion cyclotron resonance studies, a wide variety of e i t h e r positive or negative reagent ions c a n b e utilized because of t h e absence of ion clustering reactions. P o l a r reagent gases, A, t e n d to cluster extensively a r o u n d s a m p l e ions, B H + , at pressures of 1 Torr which are used typically in conventional chemical ionization sources.
Table I. Comparison between High Pressure Chemical Ionization and Trapped Ion Cyclotron Resonance Mass Spectrometry Trapped ICR
High pressure C I
I
268
1. Reagent gas pressure 2. Sample pressure 3. Ion residence time 4. Ion collection efficiency 5. Variety of reagent ions
Torr
1 Torr Torr IO-$ sec
lo-* Torr
0.1%
100%
Extensive clustering with polar reagent gases
No
l " " l " " I " " I " " I " " / " " I " " I " " I " " I
200
250
300
I
clustering
242
1 1357 ~; W U L -
-
11
416 h
BH' AHB'
+ A + A = AHB' + A + A + A = A,HB' + A
I " " l " " 1 " " 1 ' " ' I " " I ' ' " I ' ' " I ' ' ' ' I ' " ' I
200
(3)
But these termolecular reactions are not in general obTorr. served a t pressures on the order of T o illustrate these features of trapped ion cyclotron resonance, we have investigated the gas-phase ion-molecule reactions of various reagent ions with morphine, acetyl salicyclic acid, and the diammonium salt of a nucleotide. Figure 5a shows a trapped ion cyclotron resonance mass spectrum obtained for the reaction of (CH3)3NH+ with morphine. This spectrum was obtained using a continuous mode of operation rather than the pulsed mode described previously. In the continuous mode, the electron beam is set a t a constant emission of about 10-8 A, a continuous R F voltage is applied to the analyzer cell, and no quench pulse is used. Figure 5a was obtained by adding trimethyl amine to a pressure of 3 X 10-6 Torr. A t an electron impact energy of 20 eV, the protonated species (CH3)3NH+, mle 60+, forms rapidly. Due to the low pressures utilized in the experiment, (CH?)3NH+does not react further with trimethyl amine to form cluster ions. The morphine sample was Torr via a solid then added to a partial pressure of 3 X sample inlet probe. Even though morphine has a high melting point (254 OC), a temperature of only 125 "C a t the solid sample inlet was needed to vaporize the sample to this pressure. The reagent ion (CH3)3NH+ rapidly reacts with morphine through a proton transfer reaction to give primarily the protonated parent peak,
N-CH, HO
4w
350
10 sec
H
250
300
350
400
450
500
550
602
650
Figure 5. Trapped ion cyclotron resonance mass spectra: (a) morphine ionized by (CH3)3NH+,and (b) diammonium salt of 2'4eoxythymidine-5'-phosphate ionized by (CH3)3NHf
In contrast, isobutane-chemical ionization of morphine yields a base peak a t ( M - 17)+, an (M l)+peak a t a relative abundance of 27%, an ( M - 16)+ peak a t a relative abundance of 19%,and several other minor peaks (18).The site of greatest basicity in morphine is the trialkyl nitrogen position. By comparing the proton affinity of trimethyl amine and isobutene, it is apparent that the reaction between (CH&NH+ and morphine is about 33 kcal/mol less exothermic than the reaction between (CH3)3C+ and morphine. The extensive fragmentation observed with isobutane reagent gas is a t least partially due to the greater exothermicity of the reaction. Negative reagent ion studies under high pressure chemical ionization conditions have thus far been limited primarily to C1- (35).This ion is a weak base and is limited in the scope of sample compounds it will react with. In fact, we observe that C1- does not deprotonate morphine. Using estimated values for the acidity of HCl and morphine, we calculate that this reaction is about 10 kcal/mol endothermic. A more versatile negative ion reagent for proton abstraction reactions is CH30-, a stronger base than C1-. At pressures on the order of Torr, low energy electrons react with methyl nitrite via a dissociative electron capture process to produce CH30- ( 3 6 ) ,
+
CH,ONO + e- = CH,O- + NO (5) If CH3O- ions are reacted with morphine, a rapid proton abstraction reaction occurs yielding only a single peak a t (M- 1)-.
Table 11. Reaction of Various Reagent Ions with Morphine and Acetyl Salicylic Acida pOH
Mol i v t
(cH~)~~uH*
Morphine
285
286+(100) 268'(15)
Acetyl salicylic acid
180
N o reaction
q+ OCII,
CbH6+
CH30'
c1-
268+(100) 286+(75) 162+(13)
285% 00) 162'(2 5) 268+(18)
284-(100)
No reaction
121+(100) 1 63+(70) 138*(23) 139+(18) 181+(18) All entries consist of m l e values, and relative abundances are in parentheses
138'( 100) 1 8('O 52) 120+(24) 163 (22)
137-(100) 179-(44)
No reaction
ANALYTICAL CHEMISTRY, VOL. 47, NO. 4 , APRIL 1975
695
sample pressure which will give a usable signal. Such a calculation can be readily performed by considering a bimolecular ion-molecule reaction such as Equation 2. The current of sample ions, dS/dt, is given by Based upon the measured gas-phase acidity of methanol (37) and as estimation of the acidity of morphine using a substituted phenol as a model, the exothermicity of Reaction 6 can be estimated as 33 kcal/mol. In general, proton abstraction Reaction 7 can be expected to give less fragmentation than proton transfer Reaction 8 of comparable exothermicity.
+ BH = B- + AH* AH+ + B = (BH+)* + A
(7)
A-
(8)
A large fraction of the reaction exothermicity can be expected to remain as vibrational excitation in the species containing the newly formed bond, AH* for Reaction 7 and (BH+)* for Reaction 8. The sample ions (BH+)* generated by proton transfer thus would be expected to have higher internal energy than the sample ions B- generated by proton abstraction. In addition to just simple deprotonation reactions involving one product, CH30- reacts with acetyl salicylic acid to give two negative ion reaction products, O+O
/ \
-
oo-
0,
C
(9)
