J. Phys. Chem. 1982, 86, 606-612
808
N-0.H nonbonded pairs, was examined, and the uncertainties of this origin were included in the error estimates. The limits of uncertainty in the structural parameters listed in Table VI were based on our estimates of random and systematic errors.31
Comparison of Structural Parameters The structural parameters obtained in the present study are compared in Tables VI1 and VI11 with those reported previously for hydrazine and related molecule^.^^-^^ It turns out that the N-N and N-H bond lengths reported by Morino et al.’ are remarkably accurate. The dihedral angle, 9l0, is also compatible with the values determined by microwave and infrared studies.2-8 The presence of a significant difference between the inner and outer N-N-H bond angles has now been demonstrated experimentally. The same trend has also been predicted by a recent ab initio ~alculation.’~ The outer N-N-H angle is similar to the H-N-H angle of ammonia,34whereas the inner angle is about 5’ larger, The structure of hydrazine determined in the present study is still far from complete. Even a combined analysis of electron diffraction and high-resolution spectroscopy has been able to solve the problems described in Introduction only partly. One of the future steps toward our goal seems to be to combine ab initio calculations, particularly in making even a crude set of estimates of the isotopic differences in the vibrationally averaged nuclear positions on the basis of the cubic force field, so that one can take all the available isotopic rotational constants into the analysis. Acknowledgment. The authors are grateful to Professor M. Tsuboi and Dr. Y. Hamada for providing data on the force field and the structural parameters derived from their (31)K. Kuchitau, ‘Molecular Structures and Vibrations”, S. 3. Cyvin, Ed., Elsevier, Amsterdam, 1972,Chapter 10. (32)S.Teunekawa, J.Phys. SOC. Jpn., 33, 167 (1972). (33)K. Takagi and T. Kojima, J. Phys. SOC.Jpn., 30, 1145 (1971). (34)W.S.Benedict and E. K. Plyler, Can. J.Phys., 35,1235(1957).
calculations prior to publication.
Appendix. Microphotometry The electron diffraction intensity recorded on a photographic plate is measured by a microphotometer (Rigaku Denki MP3).19 The plate is spun by a synchronous motor during the photometry at precisely 150 rpm in order to smooth out fluctuations due to unevenness of the emulsion, and it is driven at a constant speed of 2 mm/min along the diameter of the halos. The sampling is carried out at 1-8 intervals for 400 ms. A pair of 6-V batteries (102 A h for each) connected parallel is used for the light source (a dc lamp). Random fluctuations of the voltage from linearity are less than 0.00190 and the gradient, about 2 pV/s, is compensated for by subtraction of a linear function. The electric current from a photomultiplier, about 0.1 PA, is grounded through a high-precision resistor of 2.5 MQ. The output voltage is integrated to eliminate random noises in the photocurrent for 400 ms, during which time the plate makes exactly one full turn. The voltage measured to six decimals is read from a digital voltmeter as parallel binary signals. They are serialized by an interface and fed into a microcomputer (NEC, TK85). In our previous measurements of photographic densities the sampling interval was set to As = 7/10 A-l. About 100 data were obtained for each photographic plate taken with a camera distance of 107 mm. In the present scheme more than 3000 data can be measured with intervals of about 0.01 A-l, and about 600 data obtained by averaging are used for subsequent analyses. Several test measurements on the same photographic plates have shown that the random standard error in the molecular intensity has been reduced to about one-half of that in our previous measurements. Supplementary Material Available: Experimental data of the tobl intensity for hydrazine obtained by gas electron diffraction (2 pages). Ordering information is given on any current masthead page.
Tlme-Dependent Mass Spectra and Breakdown Graphs. 2. The Klnetlc Shift in Pyridine Chava Llfshltz L?Spartment of Physical Chemkhy, The Hebrew Universtty of Jerusalem, Jerusalem 91904, Israel (Received June 12, 1981; I n Nnal Form: October 3, 1981)
-
Appearance potentials, breakdown curves, and metastable peak shapes were determined as a function of time, by trapped ion mas8 spectrometry (TIMS), for the reaction C6H5N+ C4H4++ HCN, in pyridine. The experimental data were compared with QET calculations of time-resolved appearance potentials, breakdown curves, and crossover shifts. Good agreement was obtained between the present electron impact data and previous photoionizationand photoion-photoelectron coincidence data. The long-time (milliseconds) appearance potential value limit is AP(C4H4+)= 11.95 f 0.2 eV at 423 K.
Introduction Two factors causing appreciable errors in obtaining thermochemical data from appearance potential measurements have existed since the early days of mass spectrometry:l (1)the swalled “kinetic shift”,the excess (1)W. A. Chupka, J. Chem. Phys., 30,191 (1969);L. Friedman, F. A. Long, and M. Wolfsberg, ibid., 26,714 (1957). 0022-3654/82/2086-0606$0 1.25/0
energy required to produce detectable dissociation in S, and (2) the temperature of the gas in the ionbiW region. The introduction2Of SUperSOniC beinto photoionization m a SpedrometW has bled the Cooling of gases to VerY (2)G.R. Parr and J. W. Taylor, Rev. Sci. Instrum., 44,1578(1973); C.Y.Ng, B. H. Mahan, and Y. T. Lee, J.Chem. Phys., 65,1956(1976), W. M.Trott, N. C. Blais, and E. A. Walters, ibid., 69,3150 (1978).
0 1982 American Chemical Society
The Kinetic Shift in Pyridine
low temperatures without condensation and thus surmounting the second problem. Several methods have been employed in order to determine "kinetic shifts". Appearance potential measurements for metastable ions3by the semilog plot technique4 and ion cyclotron resonance (ICR) spectrometry5 involve extension of the time scale to allow the observation of slow dissociations at threshold energies. The methods have a disadvantage in that a limited time range is available to them. In the ICR experiment, loss of signal intensity was observed as the residence time was de~reased.~ Erroneous data may result when metastable ion appearance potentials are determined relative to those of normal fragments, due to the different detection s e n s i t i ~ i t i e s . ~ ~ ~ We have employeda trapped ion mass spectrometry (TIMS) to measure the "kinetic shift" in benzene. Ions were trapped in an electron space charge from -1 p s to 1 ms and appearance potentials for normal fragments were determined under conditions of constant sensitivity by the semilog plot technique. The method has been extended to benzonitrile and toluene by Gordon and Reidqg Rosenstock et al.1° found it disturbing that the appearance potential measured by TIMSg showed no sign of leveling off at long residence times. The many discrepancies and difficulties noted, when trying to evaluate kinetic shifta from appearance potential measurements, led us to propose1' that high-quality normalized second-derivative electron impact ionization efficiency curves, measured at different times following the ionization process, could furnish the required information on kinetic shifts. The first results on time-resolved breakdown graphs, based on second-derivative electron impact ionization efficiency curves, were reported by ud2 for chlorobenzene. The analysis of the data required calculating time-resolved breakdown graphs, from energy-dependent specific rate coefficients i z Q for the unimolecular dissociation reaction, on the basis of the quasiequilibrium theory (QET)12 Additional studies involved 1 , 5 - h e ~ a d i y n e ~and ~ J ~demonstrated the long-time (milliseconds) isomerization of the 1,bhexadiyne ion to benzene, followed by dissociation to a phenyl ion and hydrogen atom pair. The idea of employing time-resolved breakdown graphs for kinetic shift measurements was adopted by Stockbauer and Rosenstock,'"16 who have used the technique of
The Journal of Physical Chemistry, Vol. 86, No. 5, 1982 607
r -
A
V (5,usec) =3.0+ 0.1
voits
-
(3)1. Hertel and Ch. Ottinger, 2.Naturforsch. A, 22,40(1967);R. G. Cooks, I. Howe, and D. H. Williams, Org. Mass Spectrom., 2,137 (1969); J. H. Beynon, J. A. Hopkinson, and G. R. Lester, Znt. J. Mass Spectrom. Zon Phya., 2,291 (1969);P. Brown, Org. Mass Spectrom., 3,639 (1970); R. D. Hickling and K. R. Jennings, ibid., 3,1499(1970);M. A. Baldwin, ibid., 14,601 (1979). (4)F. P. Lossing, A. W. Tickner, and W. A. Bryce, J. Chem. Phys., 19, 1254 (1951). (5) M.L. Gross, Org. Mass Spectrom., 6 , 827 (1972). (6)T.W. Bentley, R. A. W. Johnstone, and B. N. McMaater, J. Chem. Soc., Chem. Commun., 510 (1973). (7)J. H. Beynon, R. G. Cooks, K. R. Jennings, and A. J. FerrerCorreia, Int. J. Mass Spectrom. Zon Phys., 18, 87 (1975). (8)C. Lifshitz, A. MacKenzie Peers, M. Weiss, and M. J. Weiss, Adu. Mass Spectrom., 6 , 871 (1974). (9)S. M. Gordon and N. W. Reid, Znt. J. Mass Spectrom. Zon Phys., 18,379 (1975). (10)H.M. Rosenstock, K. E. McCulloh, and F. P. Lossing, Znt. J . Mass Spectrom. Zon Phys., 25,327 (1977). (11)C. Lifshitz, Adu. Mass Spectrom., 7A,3 (1978). (12)C.Lifshitz, M. Weiss, and S. Landau-Gefen, Paper presented at 25th Annual Conference on Mass Spectrometry, Washington, DC, June 1977. (13)C. Lifshitz and S. Gefen, Znt. J. Mass Spectrom. Zon Phys., 35, 31 (1980)(Part 1). (14)R.Stockbauer and H. M. Rosenstock, Znt. J. Mass Spectrom. Zon Phys., 27, 185 (1978). (15)H.M.Rosenstock, R.Stockbauer,and A. C. Parr, J . Chem. Phys., 71,3708 (1979).
10-4'10
50
6'0
70
i o
I
IONIZING PULSE HEIGHT , V O L T S
Flgue 1. Semilogarithmic plots of ion currents vs. ionizing pulse height; open circles are results for a 5-ps delay time and filled circles are for 1000 ps; 79, C,H,N+; 52, C4H,+.
photoion-photoelectron coincidence (PIPECO). This enabled measurements at a high-energy resolution over a limited (several microseconds) time range. A systematic analysis of the breakdown curves and residence time effects has been carried out by using quasiequilibrium Since trapped ion (electron impact) mass spectrometry covers a wide time scale but has a poor energy definition while variable-time photoion-photoelectron coincidence mass spectrometry has excellent energy definition but a limited time range, it makes sense to compare data obtained by the two methods for the same molecule. Pyridine has been studied recently by photoionization mass spectrometry," by PIPEC0,17 and by variable-time PIPECO.la In the following, we shall present recent data of our own on the dissociation
-
C5H5N+ mlz 79
+
C4H4+ HCN
(1)
mlr 52
in pyridine as obtained by TIMS.
Experimental Section The ion-trapping technique used to vary the reaction time is similar to the one which has been described by Herod and Harri~0n.l~A continuous electron beam of about 5 eV energy and 1-4-~Atrap current, provided by thermionic emission from a rhenium filament, is used to trap the ions produced when a pulse of up to 45 V (negative with respect to the ionization chamber) and 1-2-ps duration is applied to the filament. At a known and variable time after the ionizing pulse, a positive voltage pulse is applied to the repeller electrode to remove ions for mass analysis. Ions have been trpped in our system for up to 3 ms. The ionizing electron energy is varied by varying the filament pulse height. The electron energy distribution is obtained from second-derivative ionization efficiency (16)H. M.Rosenstock, R.Stockbauer,and A. C. Parr, J. Chem. Phys., 73,773 (1980). (17)J. H.D.Eland, J. Berkowitz, H. Schdte, and R. Frey, Znt. J. Mass Spectrom. Zon Phys., 28,297 (1978). (18)H.M.Rosenstock, R. Stockbauer, and A. C. Parr, Int. J. Mass Spectrom. Zon Phys., 38,323 (1981). (19)A. A. Herod and A. G. Harrison, Znt. J. Mass Spectrom. Zon Phys., 4,415 (1970).
Lifshitz
P
. .
I
c, H~ N+ -+ c,H',+
HCN
( Pyridme)
.
0
I/130=lo-2
I
/
I
~
~
I
~ 1
~ I
I
200 300 400 Delay T i m e , microseconds
I00
1
500
Flgure 3. Dependence of AP-IP (the difference between the ap,H,' and the ionization potential of pyridine) pearance potential of C on the delay time between ionization and repelling pulses.
' C h Z l N G PULSE HEIGtiT
VCLT5
L
1
10
,
20
1
30
Delay Time, milliseconds
i I0-'t
F w e 4. Figure caption as in Figure 3 open circles are data for I / I 3o = filled squares are data for I / 1 3 0 =
curves for CO+ and Ar'. The width at half-height of the distribution is -2.2 eV. It can be narrowed somewhat by using a longer pulse length of 2-ps duration. The ion source temperature is 150 "C. Ion source concentrations were -3.5 x 10" molecules/cm3.
-
c I 10-41
20
40
6C
83
IONlZING ?!JLSE t'EIGHT,
IO3
dO-TS
i '
IONIZING
4'0
60
80
100'
PULSE 4EIGHT. V O L T S
Flgure 2. Semilogarithmic plots of ionization efficiency curves for several delay times between the ionizing pulse and the repeilar pulse: (a) 5 11s; (b) 55 ps; and (c) 1 ms. The curve labeled 34.3 is for the 52' which takes place in the metastable transition 79' (ffl') field-free region of the single-focusing CH4 mass spectrometer.
-
Results and Discussion ( a ) Time-Resolved Appearance Energies and the Kinetic Shift. ( i ) Experimental Measurements. Semilogarithmic plots of ion currents for the parent ( m / z 79) and fragment ( m / z 52) ions from pyridine, from a typical experiment at short and long trapping times, are shown in Figure 1. A kinetic shift of about 0.8 eV transpires for this time range. One of the major problems with this type of experiment was that the parent and fragment curves were not parallel to each other at long delay times, even though they were quite parallel at the short delay times employed. This is shown in Figure 2. A fairly large number of experiments of this type were run. The resultant differences between the apperance potential (AP) of the fragment and the ionization potential (IP) of the parent are plotted as a function of delay time in Figures 3 and 4. These were obtained from the semilogarithmic and respectively, where I is the plots at 1 / 4 0 = ion current and IN is the current at an ionizing pulse height of 30 V. The appearance potentials are observed to level off at long delay times; the values obtained from I/130 = 10" do so more rapidly than the ones based on l/IN= lo-*. The long-time limit is AP(C,H4+) - IP(C5HSN) = 2.3 f 0.1 V (I) The appearance potential of the fragment C4H4+ was determined relative to the ionization potential of the parent C5H5Nin the above experiments. It is known, however,
The Journal of Physical Chemistry, Vol. 86,No. 5, 1982 609
The Kinetic Shift in Pyridine ,
I
,
,
,
,
+
% %118 l0-$.O
’
510
7’0
I
’
9.0
IONIZING PULSE HEIGHT, VOLTS
0
200
400
600
800
IO00
T I M E , MICRCSECCh3S
Figure 6. Calculated dependence of appearance potential on ion source residence time for the pyridine reaction (see text).
monochromator value.1° Eland et al.17 have set limits of 11.8-12.0 eV on the true threshold for C4H4+formation, on the basis of their photoionization and PIPECO study of pyridine. Rosenstock et a1.18 prefer, on the basis of variable-time PIPECO, a fragmentation threshold of 12.15 f 0.02 eV at 0 K. Our result, obtained at 423 K (eq 111), is in agreement with both sets of data. However, the experimental error limits are too large to distinguish between the two. (ii) Calculations. Gordon and Reidg have indicated a way for calculating the dependence of appearance energies on delay time. The equation one should employ in the case of pyridine is
I/
[C4H4+Imin = [C5H5N+Io
ma-IP
P ( E ) ( l-
dE (IV)
IONIZING PULSE HEIGHT,VOLTS
Figure 5. Time-resotved ionization potential measurements. Semilogarithmic plots of ion currents vs. ionizing pulse height: (a) pyridine and aniline: (b) pyridine and ethane.
that the ionization potential of pyridine measured by the semilogarithmic plot method is several tenths of an electronvolt higherms2lthan the actual ionization potential of the molecule (IP(pyridine) = 9.25eV).17 We have verified this by measuring the ionization potential of pyridine relative to ethane (IP = 11.52 eV22),aniline (IP = 8.32 eV23),and methyl iodide (IP = 9.55 eV22). Some semilogarithmic plots are shown in Figure 5. The resultant ionization potential for pyridine, irrespective of trapping time, is IP(pyridine) = 9.65 f 0.10 eV (11) By combining (I) and (11) one obtains the long-time limit for the appearance potential (or “appearance energy”) of C4H4+from pyridine AP(C4H4+)= 11.95 f 0.2 eV
(111) This value is considerably lower than (short-time) electron impact values.20 It is even 0.4 eV lower than the electron (20)J. L.Franklin, J. G. Dillard, H. M. Rosenstock, J. T. Herron, K. Draxl, and F. H. Field, Natl. Stand. Ref. Data Ser., Natl. Bur. Stand., No.26 (1969). (21)M.R. Basila and D. J. Clancy, J.Phys. Chem., 67,1551 (1963). (22)H. M.Rosenstock, K. Draxl, B. W. Steiner, and J. T. Herron, J. Phys. Chem. Ref. Data,Suppl. 1, 6 (1977). (23)G.F. Crable and G. L. Kearns, J. Phys. Chem., 66, 436 (1962).
Here [C4H4+Imin is the minimum C4H4+concentration which gives rise to a detectable ion signal above background noise; [C5H5N+Iois the initial parent ion concentration; P(E) dE is the fraction of parent ions having energies between E and E dE;and k(E)is the energy-dependent specific rate coefficient for reaction 1in pyridine. The value of [C4H4+Imin is evaluated by inserting into eq IV the experimentally determined value of AP-IP at a fixed (short) delay time. AP for 1/130= was chosen since at I/& = the signal is slightly above background noise in pyridine. The value for [C4H4+]--the detection limit-should be the same irrespective of the delay time and can thus be employed to calculate the dependence of AP upon the delay time. In the original calculationsg for benzonitrile, P(E) was essentially constant above the onset of fragmentation. A close inspection of the photoelectron spectrum shows that this is certainly not the case in pyridine.17 The fragmentation threshold is somewhere within a Franck-Condon gap close to a sharply rising part of the PES curve. We have calculated the dependence of AP on time using eq IV, adopting the photoelectron spectrum of pyridine17for P(E) and the experimental lt(E)17 with extrapolated values down to an ionization energy of 11.5 eV. The resultant calculated curve is shown in Figure 6. The experimental AP’s for I / I N = drop more rapidly with time (Figures 3 and 4)than the calculated values do.24 This may be an
+
(24)A point to notice is that the zero of the time scale for the experimental figures (figures 3 and 4) and for the calculated one (Figure 6)are not the same. At zero delay time the ions are estimated to spend 0-5 ps in the ion source.
The Journal of Physical Chemistry, Vol. 86, No. 5, 1982
610
Lifshitz PY RlDlNE
1000
2 - 800 m
600 40 0 200 00
' = I
f3 01 a
6
7
8
9
IO
II
12
13
14
Nominal E l e c t r o n Energy ( e V )
Flgure 7. Experimental breakdown graph for pyridine at 6-ps delay time. The breakdown curves for ions 79+, 52+, and 34.2' give the probabiltty (percent abundance) as functlon of nominal electron energy. The curves were obtained from smoothed ionization efficiency curves by using a polynomial fit. taking the second derivatiie and normalizing the sum of the second derivatives to 100% The relative abundance of the metastable ion (m.s.) at m l z 34.2 has been multiplied by a factor of 200.
.
indication that the Franck-Condon gap in pyridine is "filled-in" in our electron impact experiment, (1) by the thermal energy distribution of neutral pyridine at 423 K and (2) by resonant autoionization. (b) Time-Resolved Breakdown Graphs. ( i ) Experimental Measurements. The breakdown graph is obtained experimentallyfrom normalized second-derivative electron impact ionization efficiency ~ u r v e s . ~ Results ~S of a typical experiment at 6-pus delay time are shown in Figure 7. The curve labeled m.s. is for the ion at the noninteger mass m* = 34.3, due to reaction 1 taking place in the field-free region of the MAT CH4 mass spectrometer. The crossing region between parent and fragment ion curves is much broader than in the photon impact experiments,"J8 mainly due to the very broad electron energy distribution employed in our experiment. The results of an experiment at short (5 p s ) and long (1000 ps) delay times are 8hown in Figure 8. The shift of the crossing point between the parent and fragment breakdown curves (the so-called "crossover shift"18) toward lower energy at the longer trapping time is clearly visible. The shift from 5 to 1000 p s is 0.6 eV. (ii) Calculations. Breakdown graphs were calculated on the basis of two models. Model A is the "tight" complex model of Eland et al." which fits their experimental K(E), with a threshold of Eo = 11.8 eV. Model B is the one adopted by Rosenstock et al.18s26 with Eo = 12.15 eV. The calculated breakdown curves at 10 and 1000 ps, based on model B, are shown in Figure 9. The calculated breakdown graph (Figure 9) would have been obtained experimentally (provided the model is correct) for an infinitely narrow electron energy distribution and for vibrationally and rotationally cold pyridine molecules. In order to compare the experimental data to the calculated ones, the latter have to be convoluted with the calculated thermal energy distribution for pyridine at 423 K and with experimental electron energy distribution. The assumption made involves the additivity of the thermal energy of the neutral molecule and the energy transferred, i.e., the amount of energy transferred in electron impact ionization is assumed to be independent of the initial amount of thermal energy. Evidence has been presented18 that the vibrational thermal energy distribution of pyridine may in fact be distorted upon vertical ionization. The resultant curves calculated as a function of nominal electron energy, for models A and B respectively, are included in Figure 8 for comparison with the experimental breakdown curves. The short delay time of 5 ps corresponds to an actual reaction time of -10 ps. The overall behavior of the (25) W. A. Chupka and M. Kaminsky, J. Chern. Phys., 35,1991 (1961). (26) We thank Dr. H. M. Rosenstock for sending us the set of calculated rate coefficients, k(E),for reaction 1.
Nominal Electron E n e r g y , e V
Flgure 8. Top: Experimental breakdown curves for parent and fragment ions (79 and 52) at two delay times: 5 and 1000 ps. The c w e s were obtained as explained in the caption for figure 7. The crossover shift (the shift in the crossing point between parent and fragment curves) is 0.6 eV (see text). Mlddle: Calculated breakdown curves at two reactlon times: 10 and 1000 p s , according to model A following Eland et ai. The model assumes a 0 K fragmentationthreshold of 11.8 eV, and an activation energy of 2.55 eV. The tight complex r a t a energy dependence was graphically extrapolated down to threshold energy. The model corresponds to an equivalent activation entropy of --2.5 cal mol-' K-'. The calculated curves for 0 K and a delta function electron energy distribution were convoluted with the 423 K thermal energy distribution of pyridine and with the experimental electron energy distribution. The crossover shift is 0.5 eV. Bottom: Calculated breakdown curves (model B) for 423 K and for the experimental electron energy distribution. The crossover shii is 0.36 eV.
12.0
12 5 I3 0 Internal Energy,eV
Flgure 9. Calculated breakdown curves at two reaction times: 10 and 1000 ps. The model is model 6, due to Rosenstock et al.," and assumes a 0 K fragmentation threshold of 12.15 eV; the activation energy is .Eae= 12.15 - 9.25 = 2.9 eV and the equivalent activation entropy (calculated at 1000 K) is 4.35 cal mol-' K-'. The crossover shift (from point c to point a) is 0.36 eV.
experimental breakdown curves and the calculated ones is similar although there are discrepancies in the actual behavior of the curves particularly at high nominal energies. An important point to note is that the calculated crossover shift for a certain model does not change upon convolution with the thermal and electron energy distri-
The Kinetic Shift in Pyridine
The Journal of Physical Chemistty, Voi. 86,No. 5, 1982 611
TABLE I : Crossover Shifts (eV) time range,a w s exptl shift calcd shift (model A) calcd shift (model B )
IO?
5-1000 0.6 t 0.1
10-1000 100-1000 0.4 k 0.1 0.11 5 0.05 0.51 0.225 0.36 0.155
0.605 0.43
a The
experimental time range corresponds to delay times; the calculated shifts are for actual reaction times.
1
Pyridine
Pyridine C5H5N'-C4Hi
+HCN
( 0 )zero delay time
IO
20 30 40 50 60 70 00 M i c r o seconds
Figure 11. Decay curve of the metastable peak intensity (normalized to the precursor intensity) as function of delay time. Pyridine
Thermal E n e r g y Distribution 423'K
34 5
340
MOSSUnits
( b ) l00psec delay time
k
Pyridine C5H5N'-
C4Hi t HCN
t
\
I
c 0
Mass Units
Figure 10. The metastable peak shape at rn = 34.2 as function of delay time.
butions under the above restrictive assumption. If this value can be obtained with any degree of accuracy experimentally, it should be compared with theory, disregarding the exact behavior of the curves at the wings of the breakdown graph. Table I shows a comparison of some calculated and experimental crossover shifts. The agreement is quite good, although the experimental error limits are too large to distinguish between the two models. While the delay time range of 100-1000 ps corresponds more closely to the actual residence time range, the crossover shift is smaller than for the other two ranges and can be less accurately determined. The Metastable Peak at m* = 34.2. The metastable peak shape (and kinetic energy release) can be determined as a function of delay time. The results of a preliminary experiment on reaction 1 are shown in Figure 10. Due to the decay in relative intensity of the metastable (see Figure ll),the signal-to-noiseratio is much poorer at 100 ps than at 0 delay time. Computer accumulation of the data is clearly necessary in order to improve the signal-to-noise ratio at the longer delay times and is now planned. A rough estimate of the ratio of maximum kinetic energy releases at 0- and 100-ps delay times (from the widths at the base of the peaks above the apparatus noise level) yields 3.5 (VI
i?jl=/v&M,
The maximum kinetic energy releases were found equal,
\
01
02
03 04 eV
05
06
07
Figure 12. Thermal energy distribution of pyridine at 423 K. All the vibrational degrees of freedom have been included in the calculation; no rotations have been taken into account.
within experimental errors, to the kinetic shifts, being -110 meV and rBN I 370 meV. The appearance potential of the metastable peak as obtained from Figure 2a (- 11.3 eV) is clearly in error due to using normal ions as reference standards for the semilog plot method. We have also used the metastable at m* = 26.13 due to the reaction
-
C2H6+ C2H4+ + H2 m / r 30
(2)
mlz 28
as a reference standard (AP(26.13) = 12.08 eVn) and obtained AP(34.2) = 11.8 f 0.1 eV. This is equal, within experimental error, to the long-time limiting value of the AP for the normal fragment ion, which is somewhat surprising, but is in keeping with recent findings on other systems.28
Conclusion Reaction 1 is characterized by a large kinetic shift. This kinetic shift can be observed experimentally in photoionization experiments"J8 as well as in electron impact experiments (present data). The actual kinetic shift for a certain time range may be obtained experimentally from electron impact semilog plots. Alternatively, the crossover shift can be determined from breakdown graphs, obtained from second-derivative electron impact ionization efficiency curves. (27)W.A.Chupka and J. Berkowitz, J.Chem. Phys., 47,2921(1967). (28)J. L.Holmes, personal communication.
612
J. Phys. Chem. 1982, 86, 612-617
The long-time limiting value for AP(C,H,+) = 11.95 f 0.2 eV is in agreement with the photoionization and PIPECO v a l ~ e s . ' ~ JIf~ we take into account the thermal energy distribution of pyridine at 423 K (Figure 12), we can estimate the threshold at 0 K to be -0.2 eV higher. It seems therefore that our apearance potential is in better agreement with the more recent value of 12.15 f 0.02 eV for the 0 K threshold, obtained from variable-time PIPECO.18 It is interesting to notice that reactions such as (l), which are characterized by large kinetic shifts and whose thermochemical onsets occur in Franck-Condon gaps, cannot be easily studied by ordinary photoionization methods. Even the very careful two s t u d i e ~ ' ~give J ~ an uncertainity in the threshold energy of 0.35 eV. What seems to be needed is trapped photoion mass spectrometry (TPIMS).
This could combine the excellent energy resolution of photoionization with the wide time range available to TIMS. We are currently constructing a TPIMS device which includes a cylindrical ion trap (CIT),29*30 a pulsed vacuum-UV light source and monochromator, and a quadrupole mass spectrometer. This instrument will be able to trap photoions for up to milliseconds and will enable us to measure kinetic shifts at constant detection sensitivity with excellent energy resolution. One of the first molecules to be studied will be pyridine. (29)R. F. Bonner, J. E. Fulford, R. E. March, and G . F. Hamilton, Int. J. Mass Spectrom. Ion Phys., 24,255(1977);J . E.Fulford, R. E. March, R. E. Mather, J. F. J. Todd, and R. M. Waldren, Can. J . Spectrosc., 25, 85 (1980). (30)We thank Dr. R. E. March for very helpful advice in the design of the cylindrical ion trap.
Interaction of Massive Water Cluster Ions with Neutral Gases H. Udseth, H. Zmora,+ R. J. Beuhier, and L. Friedman' Department of Chemistry, Brookhaven National Laboratory, Upton, New York 11973 (Received: June 26, 1981; In Final Form: August 24, 1981)
The attenuation and disintegration of water cluster ion beams, interacting with neutral gases, were studied for a range of cluster energies and sizes. Scatteringcross sections and rates of water molecule loss were measured. For these experiments, a time-of-flightmass spectrometer capable of analyzing dc ion sources has been developed and is described. The observed cross sections for attenuation of water cluster ions are consistent with the assumption of structures having densities comparable with that of liquid water.
Introduction Prior to the development of high molecular weight mass spectrometric techniques, interest in water cluster ions has generally been limited to solvated protons or hydroxyl ions containing less than 10 or 15 water molecules. Such clusters can be prepared in mass spectrometer ion sources operating at a few torr or less and have been the subject of a variety of thermodynamic and kinetic studies.'-5 Larger water cluster ions have been generated in our laboratory for studies of energy transfer in ion impact processes and as model compounds in the development of ion source techniques devoted to the synthesis of cluster ions of high molecular eight.^,^ A particularly interesting application of high molecular weight cluster ions is in the exploitation of their low charge-to-mass ratio for injection of fuel atoms into thermonuclear devices? An important question that arises in connection with this application deals with the interaction of relatively large cluster ions with residual gases in a vacuum system and is concerned with mechanisms of cluster ion beam attenuation as the result of cluster ionneutral molecule collisions. Large cluster ions produced by condensation of molecules at very low temperatures may have very low densities and consequently larger cross sections for a variety of gas-phase collisional interactions. Croas sections for the degradation of the respective clusters into lower molecular weight ionic species or of endothermic dissociativeprocesses were expected to provide information on cluster ion densities as well as the applied problem of 'On leave from Soreq Nuclear Research Center, Yavneh, Israel.
beam transport in vacuum systems. In the present work we have studied the interaction of water clusters with neutral gases at various energies. In order to be able to analyze high mass clusters (>loo00 amu), a specially built time-of-flight mass spectrometer has been employed. Since the water source in its present configuration could not be pulsed, a novel, three-grid beam-chopping system was incorporated into the TOF spectrometer.
The TOF Mass Spectrometer ConstructionDetails. A 2 m length time-of-flight (TOF) mass spectrometer capable of analyzing in dc ion signals was built and incorporated in interaction studies of water clusters with neutral gases. The spectrometer consists of a dc ion source coupled to a three-grid beam-chopping assembly followed by an acceleration region, a drift tube, and a secondary electron detector. (1)M. DePaz, J. J. Leventhal, and L. Friedman, J . Chem. Phys., 51, 3748 (1969). (2)P. Kebarle, "Higher Order Rea$ion-Ion Clusters and Ion Solvations", in "Ion-Molecule Reactions , J. L. Franklin, Ed., Plenum Press, 1972. (3)J. J. Solomon, N. Meot-Ner, and F. H. Field, J . Am. Chem. SOC., 96,3727 (1974). (4)I. N. Tang and A. W. Castleman, J. Chem. Phys., 60,3981(1974). ( 5 ) J. Q. Searcy and J. B. Fenn, J. Chem. Phys., 61,5282 (1974). (6)R.J. Beuhler and L. Friedman, Nucl. Inst. Meth., 170,309(1980). (7)R. J. Beuhler and L. Friedman, to be published. (8) W. Henkes, V. Hoffman, and F. Mikosch, Reo. Sei. Instrum., 48, 675 (1977).
0 1982 American Chemical Society
