J. Pkys. Ckem. 1995, 99, 15428-15437
15428
Charge Separation Processes of Multiply-Charged Fullerene Ions Cs0-2~~+, with 0 and3 I z I7
I m I 7
P. Scheier, B. Dunser, and T. D. Mark* Institut f i r Ionenphysik, Leopold Franzens Universitat, Technikerstrasse 25, A-6020 Innsbruck, Austria Received: February 28, 1995; In Final Form: June 13, 1995'
-
Using isotope-resolved, two sector field mass spectrometric techniques we have identified and measured quantitatively the energetics and kinetics of the superasymmetric spontaneous decay reactions C60-2~~+ C60-2m-2('-')+ 4-C2+ with m ranging from 0 to 7 and z from 3 to 7 . From the kinetic energy release data determined, the apparent intercharge distance derived and the metastable fractions measured in the two experimental time windows available after the electron induced production of the various multiply-charged ions, we have been able to identify the mechanism of these fragmentation reactions. This novel three-stage reaction sequence, termed here auto charge transfer (ACT) reaction, is initiated by the statistically driven neutral CZevaporation (predissociation) followed by an electron (charge) transfer process from the receding neutral fragment to the remaining highly-charged fullerene ion cage thereby leading finally to the observed Coulomb repulsion between the two charged reaction products.
1. Introduction One of the important tools of experimental physics and chemistry in elucidating the properties of matter has been the study of collision-induced or spontaneous fragmentation of particles as diverse as nuclei, atoms and molecules. Recently, fragmentation of ionized clusters has added a wealth of information for this new state of matter. A truly remarkable feature in this context has been the exceptional stability against fragmentation of a class of particles halfway between molecule and cluster, that is, the fullerenes. This significant property of c.50and other carbon cage molecules has been the subject of intensive investigations since the beginning of fullerene research. It appears to be well accepted now that upon energization the main dissociation process of c60and c.50' is the sequential loss of Cz units',2which are bound by 7.6 and 7.1 eV, respectively.3 Electron impact ionization studies of C a have shown,3 however, that in addition to the ionization energy of 7.6 eV4 and the binding energy of 7.1 eV at least about 31 eV (kinetic shift) are necessary to promote the unimolecular dissociation reaction Ce0
+
e
-
Cs0+*
+
2e
(1)
c2
(2)
U C58*
+
Similar reactions were studied recently for doubly-, triply- and quadruply-charged C60 and C70 ion^.^.^ Moreover, it became also possible to produce and to identify fullerene ions with up to seven charges.' In the course of these mass spectrometric studies spontaneous (charge separation) decay reactions
were observed and identified thereby confirming the existence of these highly-charged molecular species. A first account of these results8 conceming the decay of highly-charged parent fullerene ions via reaction 3 including a new type of decay mechanism (Le., auto charge transfer, ACT; see also ref 9) has been presented recently at the Highly Charged Ions Conference.'O @
Abstract published in Aduunce ACS Ahstrucrs, October 1. 1995.
The present paper is a systematic extension of these earlier studies including not only the quantitative investigation of the superasymmetric decay reaction 3 but also of similar reactions involving multiply-charged fragment fullerene ions C60-2~~' (with m from 1 to 7). These studies include (i) the isotoperesolved identification of the various decaying precursor ions and the respective decay channels, (ii) the measurement of the kinetic energy release as a function of z in two different time windows after the production of the precursor ions, and (iii) the determination of apparent intercharge distances between the two charged fragments at the instance of separation. Moreover, in order to confirm the proposed ACT reaction mechanism the kinetics (Le., metastable fractions) of decay reactions 3 is studied in the two field-free regions of the two-sector field mass spectrometer system used. These results allow us to conclude that the charge separation reactions observed for parent and fragment multiply-charged fullerene ions proceed via a multistep reaction sequence. Initiated by unimolecular C2 evaporation (C2 predissociation) the next step in this sequence is a charge transfer reaction between the separating fragments (ACT reaction) followed by a Coulomb repulsion between the nascent charged fragment ions. At the outset of the present study a complete isotopically resolved mass spectrum produced by electron impact ionization of C a will be presented and discussed thereby revealing several interesting features of the fragmentation of ionized C60.
2. Experimental Section The present measurements were carried out with a doublefocusing, sector-field mass spectrometer of reversed geometry with a maximum mass resolution of 25 000 and a mass range of 10 000 at a nominal acceleration voltage of 3000 V." Figure 1 shows schematically the experimental setup. The Cm powder (purity of 99.99%) was evaporated in a temperature-controlled oven (set at 890 K in the present study) and introduced as an effusive beam via a small orifice into the modified Nier-type ion source of the mass spectrometer.'' Because highly-charged fullerene ions are produced in sufficient abundances only via multiple-electron collision processes,' it was necessary to use besides high-energy electrons (around 200 eV) high electron currents (up to 1 mA). In order to increase the trapping time
0022-365419512099-15428$09.00/0 0 1995 American Chemical Society
J. Phys. Chem., Vol. 99, No. 42, 1995 15429
Multiply-Charged Fullerene Ions C a - h z +
Finally, the correlated study of metastable transitions in both field-free regions allows the study of the time dependence of the corresponding decay rates.I3J4
’f10,= 0
detector
3. Results and Discussion 3.1. Identification and Quantitative Analysis of Ions Produced by (Multi)electronImpact Ionization of Pure C ~ O .
Figure 1. Schematic view of the experimental setup.
of the ions in the negative space charge region of the electron beam all lenses in the ion source had to be optimized to reduce the extracting field. The ion source chamber is evacuated with a 500 Us turbo molecular pump, is sealed with gold and copper rings, and can be heated up to 600 K. Therefore, the residual gas pressure is below Pa and consists mainly of nitrogen and oxygen. Even with the effusive fullerene beam (at an oven temperature of 890 K) the pressure in the ion source is less than 2 x Pa. It is very important to have a very good vacuum, because multiplycharged ions with charge states higher than 5 have extremely high cross sections for charge transfer to neutral components in the residual gas. Moreover, the whole analyzing part of the mass spectrometer is metal sealed and pumped with an oil-free turbo pump equipped with magnetic bearings. Therefore, the pressure from the ion source to the detector is always better than Pa. Essential to the present study is the possibility to study quantitatively spontaneous or collision-induced dissociations in the two field-free regions of the mass spectrometer. In order to record the corresponding (metastable) decay peak in the first field-free region the HV-scan technique is used.12 Thereby the change of the kinetic energy is compensated by proper tuning of the acceleration voltage U. The value U* of the high voltage of the center of the metastable peak is correlated with the mass per charge ratio of the precursor and daughter ion, mllzl and mdz2, respectively, via
(4) with U the nominal acceleration voltage. Decay reactions resulting only in little daughter signal or which are very fast can be measured in the present setup better in the first fieldfree region because on the one hand the length of the first compared to the second field-free region is more than twice as large and on the second hand the first field-free region starts directly after the acceleration lenses and is therefore in time very close to the ionization process. Moreover, the HV-scan technique12 gives the opportunity to investigate metastable transitions even during the acceleration out of the ion source. For the investigation of the decay reactions in the second fieldfree region the MIKE scan techniquei2is used, which involves the proper tuning of the electric sector field E. A simple equation as in the HV scan case relates the value E* at the center of the metastable peak to the sector field voltage of the precursor ion E and the mass per charge ratios of the precursor and daughter ion, i.e.,
In the present study, multiply-charged carbon cluster ions, Cm-hz+ (up to septuply-charged Ca), are produced by electron impact of pure Ca. Figure 2 shows as an example the logarithm of the ion current as a function of the mass per charge ratio obtained with an electron energy of 200 eV and an electron current of 200 PA. An almost exponential decrease of the even numbered singly- and doubly-charged carbon cluster ions is clearly visible and has been previously interpreted as being due to sequential “monomer” (C2) evaporations in the ion Whereas the contribution of C ~ Oto + Ca2+ at the mass per charge ratio of 360 Da is negligible, peaks of more highly-charged ions may be seriously contaminated by fragment ions having a lower charge but the same mass per charge ratio. The only way to identify the contributions of differently charged ions which overlap each other is a careful analysis of the isotopic pattern of the peaks.I8*I9 Carbon consists of the two stable isotopes and I3C, and therefore every cluster size has its typical peak pattern; Le., multiply-charged carbon clusters have always signal contributions at noninteger masses. In general, the peak containing one 13C (second isotopomer) can be used to identify and to determine the total abundance of the corresponding ion by summing over all the other possible isotopomers. Such an isotope-corrected mass spectrum can be seen in Figure 3, where the total signal intensity of each ion-the sum over all isotopic peaks-is plotted vs the cluster size for different charge states. Several interesting features can be deduced from the isotopically corrected results shown in Figure 3. 1. Fragment size distributions for singly- and doubly-charged CnZ+exhibit a bimodal shape with a minimum at around size 30. At least for singly-charged fragment ions above size 33 only even-numbered fragment ions are observed (in accordance with earlier observations), whereas below this size both evenand odd-numbered fragments are present. According to Bowers and co-workers20carbon cluster ions with even numbers and above size 30 can be related to three-dimensionalcage structures (fullerenes), whereas smaller carbon clusters exist as chains or ring structures. The quasi-exponential decrease of the evennumbered singly-and doubly-charged fragment ions between n = 60 and n 30 has been ascribed to sequential evaporations of C2 units (see above). A similar dependence on n can be seen to exist for even numbered triply-, quadruply-, and quintuply-charged fragment ions, and it may be conjectured that a similar evaporation mechanism is responsible for this finding. 2. Whereas below the minimum at around n = 30 evenand odd-numbered triply-, quadruply-, and quintuply-charged fragment ion abundances still decrease with decreasing n each of these series ending at a larger minimum size of 20,27, and 36, respectively, the abundances of the doubly- and singlycharged fragment ions increase quasi-exponentially with decreasing n. The reason for the decrease of the abundance of the highly-charged ions and the occuffence of a minimum size may be due to an increasing instability of these multiply-charged ions with decreasing size due to the increasing action of the Coulomb repulsion.
-
Scheier et al.
15430 J. Phys. Chem., Vol. 99, No. 42, 1995
-
I
1
I '
.
I
I
,
I!
.
!
,
/
'
I-I
1
10
20
I
7 30
40
,
, !
I
II j
I
1
I
I
--
1
50
I 1
60
Cluster size (n)
Figure 3. Isotope-corrected (see text) mass spectrum of Cm showing even-numbered (open signals) and odd-numbered (filled symbols) singly-and doubly-charged (upper part) and triply-, quadruply-, and quintuply-charged (lower part) ion series.
3. The conspicuous increase of ion abundances with decreasing ~t(below the minimum at around n = 30) of the singlyand doubly-charged ions is surprising. Similar bimodal distributions (Le., the U-shaped fragment mass spectrum) have already been observed by Hertel, Campbell, and co-workers in the case of photofragmentation and collisional fragmentation of C ~ O by , ~ LeBrun ' et a1.22bombarding Cm with highly-charged Xe ions in the MeV range and by Gaillard and c o - ~ o r k e r s * ~ for the fragmentation of mass-selected 60 keV/amu H,+ ions induced by single collision with He atoms. Moreover, nuclear multifragmentation displays very similar features (e.g., for nuclear fragments emerging from heavy nuclei bombarded by GeV proton^^^.^^). The falloff (Le., the decrease of ion abundances with increasing n ) at small masses has been described in some of these earlier studies by a power law. In analogy to these cases, the present experiment may be also interpreted by a power law for the abundance of the singlyand doubly-charged carbon cluster ions Cnz+(with n I 30) proportional to n P , where x 2.6 (as given by the dashed line
-
-
in Figure 3). This behavior is similar to the case of nuclear f r a g m e n t a t i ~ nwhere , ~ ~ x 2.6 in inclusive (impact parameter integrated) reactions, to the hydrogen cluster ion case,13where x 2.63, and to the c 6 0 multifragmentation experiment of LeBrun et al.,22 where x 1.3. In the case of nuclear multifragmentation reactions the power law arises as a consequence of the finiteness of the system and of the integration over various excitation energies.26 Further experiments are needed to clarify the underlying reaction mechanism of the present electron impact induced pattem at the low-mass side of the distribution. 4. It is interesting to point out that at least for odd-numbered doubly-, triply-, and quadruply-charged fragment ions the above discussed falloff appears to be continued above the minimum point in the U-shaped distribution (as indicated by the dashed line in Figure 3) and that therefore odd-numbered carbon cluster ions exist above the previously observed size limit of the singlycharged ion at size 33 (the largest odd-numbered carbon cluster ion in the present study being C514+). This again would indicate that these odd-numbered ions may exist in a different structure than their even-numbered counterparts at around the same size. In further experiments we hope to identify the structure and production mechanism of these large odd-numbered highlycharged carbon cluster ions. 5. Finally, it is quite noteworthy that the relative abundance of multiply-charged parent Cb0 ions is remarkably high (as compared to the situation in an ordinary molecule) and that the abundance of doubly- and triply-charged fragment ions is even higher than that of the corresponding singly-charged fragments. This feature is in agreement with recent accurate partial ionization cross section measurements of these ions in our laboratory,*' although the probability for multiple electron collisions was rather high in the present study due to the high electron current used. 3.2. Spontaneous Cz+ Emission from Multiply-Charged Parent Ions C& (3 5 z 5 7): Identification of the Reaction and KER. Unfortunately, the signal intensity for multiplycharged fullerenes even at highest transmission of the mass spectrometer (low mass resolution) is rather low. Therefore, it was virtually impossible to check systematically all possible decay channels by scanning the whole momentum and energy space. Nevertheless, we tried to cover in the present search the most likely decay reactions. As it tumed out besides the well-known C2 evaporation
-
- c58;-+ c,
C60;+
the second strongest spontaneous fragmentation process occurring for multiply-charged Cm ions with a charge state of 3-7 is the superasymmetric charge separation reaction (3) involving the loss of a charged C*+ fragment ion. The only other spontaneous decay reaction identified in the present study, albeit with much lower reaction probability, is the loss of a charged C4+ fragment ion (for details of this reaction see ref 28). Figure 4 shows the measured ion kinetic energy spectra for the five metastable decay reactions (involving the loss of Cz-) observed for C603+ through Cso7+, respectively. Whereas the peaks on the left-hand side are obtained by HV scans of the metastable transitions in the first field-free region, the plots on the right-hand side are MIKE scans in the second field-free region and the corresponding ordinary mass peaks of the parent ion. The HV scans exhibit the broad dished peak shape which is typical1*of metastable transitions associated with rather large and sharp kinetic energy release and relatively long flight path before collection. Moreover, as can be seen the minimum of the dished peaks is close to the exact position of the metastable
Multiply-Charged Fullerene Ions
J. Phys. Chem., Vol. 99, No. 42, 1995 15431
C60-2~~~
8
0.8
300
0.15
n
8
0.10
Y
0.6 200
0
Y
0.0s 0
8
0.4
5c 100
0.2
0
g 0.00
19
U
8
0.o
0 5
Acceleration wl
8
$
; W
Sectotfield voltage (E)
80 60
3 40
9
20
E
A 0 2300
2320
2340
2360
505
505
n
8 - 6
'
.
'
' I
'
660
510
610
615
620
Sectorfield voltage (E)
Acceleration voltage (V) '
655
Sectotfield voltage (E)
Acceleration wltage (V)
101'
510 650
O
8
E
E 4 0
sc
2
0 2540 2560 2580 2600 2620 2640
505 -510
Acceleration wltage (V)
2620
2640
2660
2680
590
-?85
595
600
Sectorfield voltage (E)
2700
Acceleration wltage (V) Figure 4. Metastable peaks associated with the spontaneous decay reactions C&
-
+
C5&z-1)+ C2+ for triply-, quadruply-, quintuply-, sextuply-, and septuply-charged fullerene parent ions in the first field-free region (left-hand side) and second field-free region (right-hand side) as recorded by HV and MIKE scan techniques (see text), respectively. Also shown for comparison in the case of the MIKE spectra are the respective precursor ion peaks.The solid lines drawn through the data points are data fits involving a smoothing procedure.
transition (given by eq 4),which is designated with a dashed line in Figure 4. Whereas the width of the peaks is due to the
kinetic energy release (KER) during the decay in the forward and backward direction, respectively, the minimum in the peaks
15432 J. Phys. Chem., Vol. 99, No. 42, 1995
Scheier et al.
I
3
4
5
6
7
Charge state z (e) Figure 5. Kinetic energy release as a function of precursor ion charge state z: squares, first field-free region; circles, second field-free region. is caused by discrimination for those ions which are accelerated by this energy perpendicular to the flight direction. In the case of the MIKE scans discrimination is less severe due to the much shorter final flight path, and therefore no minimum is observed in the respective peaks shown in Figure 4 (again the exact position of the metastable transition as given by eq 5 is indicated by a dashed line in Figure 4). It is furthermore important to mention that the shapes of these peaks is, according to arguments presented in detail in ref 12, a clear indication that these decay reactions are purely spontaneous without interference from collision-induced dissociations. This has been confirmed in the present study by additional evidence from the dependence of these peaks on the gas pressure in the field-free regions (for details of this technique see ref 29. As the precursor ion peak already has a certain width in the HV or MIKE scan (see the parent ion peak in the MIKE scans in Figure 4), the width of the daughter ion peak is only in part a result of the KER. In a first approximation the KER is proportional to the fwhm of the daughter ion peak diminished by the fwhm of the parent ion where the latter has to be corrected by the factor U*IU or E*IE (U* and E* are the acceleration or sector field voltage at the peak center of the daughter ion, U and E of the precursor ion, r e s p e ~ t i v e l y ) . According ~~ to Beynon and co-authors,12 for the simple charge separation reaction (3) the translational energy release can be calculated from the corrected width of the metastable peaks AVc and AE, using the relations
(7)
where z1 and ml are the charge state and mass of the precursor ion, z2 and m2 are the charge state and mass of the detected fragment ion, m3 is the mass of the undetected fragment ion CzL, V is the correct acceleration voltage, and E is the correct sector field voltage for the detection of the precursor ion. Figure 5 shows the derived KER for all of the studied decay reactions 3 of c60’+as a function of the charge state z. Within the experimental error bars no dependence of the KER on the time since production of the precursor ion can be seen; i.e., the energy released is the same for the two experimental time windows (first and second field-free region, respectively) available. Conversely, the energy released depends strongly on the charge state. Above the threshold charge state of z = 3 there is an almost linear increase of the KER with z, but at higher charge states the KER values appear to deviate from this linear relationship. As the KER is imparted to the two
separating fragment ions by the Coulomb repulsion (see below), this deviation from linearity can be interpreted by screening3’ of the charges delocalized and distributed over the fullerene cage (calculated charge distributions are shown in Table 1) with respect to the receding C2+ ion; Le., the polarizability of the electron cloud covering the fullerene ion diminishes the Coulombic force of the charges located on the opposite side of the receding C2+. The higher the charge state the more charges will be shielded by this effect. 3.3. Spontaneous CZ+Emission from Multiply-Charged Fullerene Fragment Ions C,j0-2,~+ (with z = 4-6 and 1 5 m 5 7): Identification of the Reaction and KER. In the course of the present study involving the stability of ions produced by electron impact ionization of c 6 0 we were able to monitor in the two field-free regions of our mass spectrometer further charge separation reactions. In this case the decaying ion is not a multiply-charged parent c 6 0 ion, but one of the fragment ions C60-2mZf produced already in the ion source (see section 3.1). As in the case of the parent ion the dominant charge separation reaction involves the loss of a C2+ fragment ion via
Because of the much weaker parent and fragment ion signals, it was not possible to investigate all conceivable decay reactions. Nevertheless we were able for the most abundant class of decay reactions, which always turns out (see also above) to be the one starting from the quadruply-charged ion, to record a complete set of MIKE scans (decay in the second field-free region) of the fragment ion series from C5x4+ down to C464f (see Figure 6). Moreover, Figure 7 shows as an example HV scans for the decay of C& ions in the first field-free region with z = 4,5, and 6 (also shown in this figure are neighboring metastable peaks belonging to charge separation reactions involving the emission of C4+; see ref 28). As in the case of the parent c 6 0 ion the corresponding peak shapes for the decay reactions of these fullerene fragment ions in the two field-free regions exhibit the typical shapes known for a metastable transition. Moreover, the data shown in Figure 6 have been analyzed with help of eq 8 and the obtained KER values are given in Figure 8 as a function of the precursor ion size. Although within error bars rather similar to the value of the corresponding Cm4+,the KER values appear to show a slight dependence on the size of the precursor ion; Le., there exists a small drop from C604+ to C5x4+and then a continuous increase from a value of 4.37 to a value of 5.32 for C464+. Analysis of the three peaks shown in Figure 7 with the help of eq 7 gave within the experimental error bars KER values equal to those obtained for the corresponding c 6 0 ions (see Figure 5). 3.4. Decay Mechanism (ACT). Assuming that the kinetic energy released in the metastable transition is a consequence of the Coulomb repulsion between the two fragment ions, the charge separation of the transition state can be calculated with help of the Coulomb law. For the simple case of two singlycharged separating fragment ions (and this is the only case treated previously) and under the assumption that these fragment ions can be treated as point charges, previous authors have calculated the intercharge distance from the relation”
In the present case, however, one of the separating fragment ions carries more than one charge and these charges are located
J. Phys. Chem., Vol. 99, No. 42, 1995 15433
Multiply-Charged Fullerene Ions Cm-zmzf
TABLE 1: Cumulative Coulombic Repulsion Vtot and Schematic Charge Configuration Existing between the Charges on the Polycations Cmz+ (z = 2-12) z=2, linear Vtot = 2.06 eV
2=3, triangle Vtot = 7.13 eV pairs of charges: edges:
pairs of charges: edges:
1 1
faces:
0
3 3 faces: 1 z=5, trigonal bipyramid V, = 26.64 eV
pairs of charges:
6
pairs of charges:
edges: faces:
6
edges:
4
faces:
15 12 8
pairs of charges: edges: faces:
z=4, tetrahedron Vtot = 15.1 eV
z=6, octahedron Vtot = 41.08 eV pairs of charges: edges: faces:
6 2=7, pentagonal bipyramid V, = 59.42 eV
z=8, twisted cube
2 9
Vtot = 80.89 eV
Vtot = 105.91 eV
pairs of charges: edges:
28 16
,airs of charges: :dges:
faces:
10
%aces:
~ 1 0like , z=8 with 2 poles V, = 134.51 eV
10 9
21
15 10
36 21 14
Fll , ,V = 166.91 eV
Dairs of charges:
45
:dges:
24
lairs of charges: :dges:
55 27
Faces:
16
aces:
18
~ 1 2icosahedron ,
Vtot
= 202.14 eV
lairs of charges: :dges: 'aces:
66 30 20
at different positions on a large molecule and therefore the two separating fragment ions cannot be approximated by point charges. In order to account for this more complicated situation we assume that according to a model of Bohme and co-workers3* the charges are delocalized and freely movable on a spherical surface created by the sea of n-electrons covering the fullerene cage. As it is not clear whether the charges will be located on the outermost radius of the electron cloud (= hard sphere radius) or not, we used the radius of the carbon cage (3.5 A)33 as proposed by Bohme and c o - ~ o r k e r s .When ~ ~ calculating the total potential energy of the system it is necessary to move first the charges to positions where this Coulomb energy is a minimum. These minimum potential configurations have been
determined for charge states 2-12 (see Table 1) and at least for charges up to 8 correspond to simple geometrical figures. For example, the most stable Cm7+ charge distribution consists of a pentagonal bipyramid upon its surface. The total Coulombic repulsion V,, thus determined is also given for CmZ+ ( z = 2-12) in Table 1. It is clear if one charge is removed from its surface position (Le., a C2+ ion is emitted) to a certain distance away from the surface the remaining charges will relax to different positions on the surface accordingly. Only in the limit of great distances the fullerene fragment ion will change from the charge configuration of the precursor ion to the ground state distribution of a fullerene ion having one charge less (e.g., by such a charge separation reaction the pentagonal bipyramidal
15434 J. Phys. Chem., Vol. 99, No. 42, 1995
Scheier et al.
400
8
25 20
6 s
15
c,
300
8200
3u A 100 0 505
10
2I2
8
140
35
120
30
510
650
660
655
L
2o
4
15
o
10
c ,o
0 665
CD
2
58
20
0
0 505
510
650
660
655
665
Sectorfield voltage (E)
Sectorfield voltage (E)
250
30
400
z 5
g
:l
$ 4 0
5E
25
200
.
,
AE-118V
100
.
d
Sectorfield voltage (E)
Sectorfield voltage (E)
25
20
F 100 50
0
Sectorfield voltage (E)
Sectorfield voltage (E) I
120
6
100
5
6 1
Q m
4,
8"
3 2
0 E4
8
2 a
0
Qg 8 c
2o 0 505
0
510
645
650
655
660
Sectorfield voltage (E)
30 20
0.2 0.1
10 0.0
"0
1.
505
I
.
.
..
510 640
-
I
.
645
.
..
I
.
650
.
..
I
.
655
.
.
Fi8 3
.I-o,l -0.1
Sectorfield voltage (E)
+
Figure 6. Metastable peaks associated with the spontaneous decay reaction Cm-2m4+ Cm-2m.23+ C2+ for 0 5 m 5 7 in the second field-free region as recorded by MIKE scan technique. Also shown for comparison the respective precursor ion peaks. The solid lines drawn through the data points are data fits involving a smoothing procedure.
G o 7 + will eventual end up in an octahedral
C5s6+ of slightly smaller size). Taking into account the screening effects discussed above and the relaxation of the charges on the fullerene cage as the reaction products separate, we have calculated from the experimental KER data the corresponding intercharge distances of the transition states. The results thus obtained are shown in
-
Figures 8 and 9, respectively. The data points shown in Figure 9 for the decay reaction C& C58("')+ C2+ decrease slightly with increasing charge from a value of approximately 8 to 7 A. Similarly, there is also a slight dependence of this intercharge distance on the size of the precursor ion C W - ~ , ~ ~ + (in reaction 9) as exhibited by the data in Figure 8. Quite surprisingly, these apparent intercharge distances derived from
+
J. Phys. Chem., Vol. 99, No. 42, 1995 15435
Multiply-Charged Fullerene Ions Cm-%z+
1- cs86+1
n 0.3
s
-
1
-
-
I
1
-
-
1
csJ++ c,'
1 n
8-
3
0.2
-
3
6:
5
-
E
.a
4:
0.1
W
0
c
3
0.0
+d
v)
'
2-
3 2550
2600
2650
2700
2750
Acceleration voltage (V)
'5
I
1
c,,S+-c,,*+
I
+ c,'
--, 4
i
i
i
Charge z (e) Figure 9. Apparent intercharge distance as a function of precursor ion charge state z: squares, first field-free region; circles, second fieldfree region.
I
''\>~\
2450
2500
2550
charee transfer
2600
Acceleration voltage (V)
c,?
c,l+-
Distance r(A)
+ c,+
I I I I AVr68V
. 0
A:
it is interesting to note that this is also true for highly excited fullerenes (up to 6ooo IC) as shown in recent molecular dynamics
I 2411 V
I
I
I
calculation^.^^ , 2350
On the basis of this rather characteristic experimental fingerprint (i.e., the significant discrepancy between the radius of Cm and the apparent intercharge distance of the separating Acceleration voltage (V) fragment ions) and further considerations and experimental Figure 7. Metastable peaks associated with the spontaneous decay details (see below), we propose the following three stage decay reactions C++ C&-I)+ + CZ+for z = 4,5, and 6 in the first fieldmechanism to be responsible for the observed superasymmetric free region as recorded by HV scan technique. Also shown are fission reactions 3 and 9 of the highly-charged Cmi+ and neighboring decay peaks corresponding to charge separation reactions Cm-%z+ fullerene ions: involving the loss of C4+.2*The solid lines drawn through the data points are data fits involving a smoothing procedure. Reaction 3 and in analogy also reaction 9 proceed via three different stages (visualized schematically in Figure 10 for the case of Cm4+). This reaction sequence is initiated by the statistically evaporation of a neutral C2 unit (reaction I 6). It is followed in the second stage by an electron transfer -6 process between the receding C2 fragment and the remaining o o l o o highly-charged fullerene cage. This charge transfer occurs at -o -4 the above-determined intercharge distance of about 7-8 A thus -2 finally leading in a third stage to the Coulombic repulsion between the two nascent charged fragments thereby imparting to the fragments the kinetic energy responsible for the width of 46 48 50 52 54 56 58 60 the metastable peaks. Size of precursor ion C60-2m In the following we would like to discuss reasons and further experimental facts supporting this novel reaction sequence. Figure 8. Kinetic energy release (left-hand side, circles) and apparent Clearly the initial clue came from the observed large apparent interchargedistance (right-hand scale, squares) as a function of precursor intercharge distance, which cannot be reconciled with the ion size. properties expected in the case of a single-step fissioning reaction (where the eventually charged fission products may the experimental data are approximately a factor of 2 larger be distinguishable already at a rather early stage of the fission than the radius of the parent neutral, I(&) = 3.5 A33and the process) known from nuclear physics and described by the radius of any of the higher charged C a polycations.M Moreover, 2300
2400
2450
-.
Ez6104-o-ool
In
15436 J. Phys. Chem., Vol. 99, No. 42, 1995
celebrated liquid drop model (LDM).36 Despite its successful application in the case of doubly-charged alkali cluster ions3' and for ionized van der Waals cluster^,^^^^^ the LDM would predict much smaller intercharge distances than the ones derived here. On the other hand, the present intercharge distance of the transition state fits perfe.ctly well the values expected in the case of the suggested ACT reaction; Le., the typical reaction window for charge transfer reactions lies between 2 and 6 A40.41 thus yielding together with the c 6 0 radius of 3.5 A a predicted range of 5.5-9.5 A for the transition state distance. Charge transfer reactions between an incoming doubly charged ion and a neutral reactant are described in terms of curve-crossing models, where the charge transfer occurs at the (avoided) crossing of the attractive r-4 ion induced dipole interaction curve (representing the incoming channel) and the repulsive r- Coulomb potential curve (representing the outgoing reaction channel). The present reaction sequence can be interpreted by this model reversing the order of the incoming and outgoing channel (see Figure IO). The average kinetic energy released in the Cz evaporation reaction has been determined to lie around 0.4 eV',2 and appears to decrease with increasing charge state.42 Thus as pictured in Figure 10 the charge transfer which can take only place at the crossing of the respective curves will occur within a rather small reaction window (defined by the possible different states in both channels) and therefore the energy released in the subsequent Coulomb repulsion will be quite well-defined. This is in accordance with the experimental results, Le., the shape of the metastable peaks (see above). Moreover, the slight decrease of the transition state distance in Figure 9 with increasing charge state can be interpreted by a corresponding shift of the reaction window due to a change in the shape of the respective potential energy curves. On the other hand, the slight decrease of the transition state distance in Figure 8 when going from C S Sto~ ~ C464+ may be rationalized by the shrinking of the cage radius with decreasing precursor ion size. Another important confirmation for the concept of the ACT reaction sequence can be obtained from the time dependence of the decay reactions. For all charge states higher than 2 and lower than 7 both the neutral and charged C2 evaporation could be observed. The neutral C2 evaporation is a typical statistically driven decay which can be described by RRKM theory or a similar statistical model. Therefore, the decay rate is determined by the internal energy, the number of internal degrees of freedom, and the binding energy.'-3 Moreover, the time dependence of the decay of an ion ensemble formed by electron impact ionization exhibits a nonexponential behavior due to the presence of ions with a range of internal energies (see also the considerations by I U O ~and S ~experimental ~ resultsI4). This time dependence can be determined in the present experiment by measuring the metastable fraction, mf (that is, the daughter ion signal divided by the respective precursor ion signal) in the first and second field-free region, respectively. In order to compare the time dependence of the C2 evaporation reaction 6 and the charge separation reaction 3 we have plotted in Figure 12 the ratio between the metastable fraction in the first and second field-free region, mfJmf2, vs the charge state for both reactions considered. In accordance with the proposed ACT reaction sequence both sets of data are independent of the charge state and match each other within the experimental error bars. This is very strong experimental evidence in favor of the three-stage mechanism, because it points out that both reactions follow the same reaction kinetics (i.e., it is highly unlikely that the charge separation reaction 3 if proceeding via a single-step fissioning reaction would not change its kinetic properties when the
'
Scheier et al.
23'0 .5
0.5 0.0
i 3
~
4
5
6
Charge z (e) Figure 11. Ratio of the metastable fractions, mf, in the first and second field-free region, mfl/mfi, as a function of charge state for the neutral Cz evaporation reaction 6 (squares) and for the charge separation reaction 3 (circles).
I
3
4
5
6
7
Charge z (e) Figure 12. Metastable fraction in the first field-free region vs charge state for the neutral Cr evaporation reaction 6 (squares) and for the charge separation reaction 3 (circles).
Coulomb energy responsible for the shape of the decay barrier changes for instance from 15 eV in the case of the quadruplycharged ion to 109 eV for the septuply-charged ion). Moreover, as can be seen in Figure 12 where the metastable fraction measured in the first field-free region is plotted vs charge state the relative reaction probability for the neutral Cz evaporation and for the charge separation reaction is strongly depending on the charge state. For low charge state the neutral evaporation is the dominant reaction path, at charge state 4 the two reactions have about equal probability, and at higher charge states destruction of the highly-charged ions proceeds preferentially via the charge separation reaction. This again is in line with the present three-stage model, where it is expected that the probability of the electron transfer (Le., the second reaction step) increases drastically with the number of charge states on the precursor ion thereby shifting the branching ratio between the two reactions in favour of the charge separation reaction with increasing charge state. Finally, additional support and a striking argument in favour of the present interpretation is the complete absence of the charge separation reaction 3 for doubly-charged fullerene ions. In the case of a single-step fissioning reaction there is no reason for this ominous absence, whereas if reaction 3 proceeds via the three-stage sequence outlined above the electron charge transfer (second stage) is endothermic for precursor ions with less than three charges and thus cannot proceed (for more details on the energetics of these reactions see ref 44). 4. Conclusion
In concluding, we were able to investigate for the first time quantitatively the mechanism, the energetics, and the kinetics of spontaneous charge separation reactions of multiply-charged parent and fragment fullerene ions. The discovery of a
Multiply-Charged Fullerene Ions Cm-2mZS superasymmetric charge separation reaction for highly-charged fullerenes opened a new door to the study of the ion chemistry of polycations. The determination of the KER from the shape of MIKE! and HV scans of the corresponding metastable transitions and the apparent intercharge distance of the transition state derived concomitantly prompted the development of a three-stage reaction sequence which we have termed auto charge transfer (ACT) reaction. This sequence is initiated by the evaporation of a neutral C2 via statistical predissociation followed by a charge transfer between the receding CZunit and the remaining highly-charged fullerene fragment cage thus leading finally to the observed Coulomb repulsion of the two charged reaction products. The present multistage mechanism is similar to (i) a semiquantitive avoided-crossingmodel recently introduced by Gill and R a d ~ m (see ~ ~also refs 46 and 47)and to (ii) a Coulomb explosion model of Eland et al.9 invoking the free charge exchange between separating fragments from small decaying dications.
Acknowledgment. Work was supported in part by the Osterreichischer Fond zur Forderung der Wissenschaftlichen Forschung, the Bundesministerium fur Wissenschaft und Forschung, and the Osterreichische Nationalbank, .Vienna. References and Notes (1) Radi, P. P.; Hsu, M. T.; Rincon, M. E.; Kemper, P. R.; Bowers, M. T. Chem. Phys. Lett. 1990, 174, 223. (21 Lifshitz. C.: Iraai. M.: Peres. T.: Fischer, J. E. Int. J. Mass Spectrom. Ion Proc. 1991, 107, 54. (31 Foltin, M.; Lezius, M.; Scheier, P.; Miirk, T. D. J. Chem. Phys. 1993, 98, 9624. (4) Lichtenberger, D. L.; Jatcko, M. E.; Nebesny, K. W.; Ray, C. D.; Huffman, D. H.; Lamb, L. D. Chem. Phys. Lett. 1992, 198, 454. (5) Scheier, P.; Dunser, B.; Worgotter, R.; Lezius, M.; Robl, R.; Miirk, T. D. Int. J. Mass Spectrom. Ion Processes 1994, 138, 77. (6) Worgotter, R.; Diinser, B.; Scheier, P.; Miirk, T. D. J. Chem. Phys. 1994, 101, 8674. (7) Scheier, P.; Miirk, T. D. Phys. Rev. Lett. 1994, 73, 54. (8) Scheier, P.; Diinser, B.; Miirk, T. D. Phys. Rev. Lett. 1995, 74, 3368. (9) Eland, J. H. D.; Sheahan, J. R. Chem. Phys. Lett. 1994,223,531. (10) Miirk, T. D.; Scheier, P. Nucl. Instrum. Methods, B 1995, 98,469. (1 1) Lezius, M.; Scheier, P.; Foltin, M.; Kolibiar, M.; Cleveland, C. L.; Landman, U.;Miirk, T. D. Int. J. Mass Spectrom. Ion Processes 1993,129, 67. (12) Cooks, R.; Beynon, J. H.; Caprioli, R. M.; Lester, G. R. Metastable Ions; Elsevier: Amsterdam, 1973. (13) Scheier, P.; Miirk, T. D. Phys. Rev. Lett. 1987, 59, 1813. (14) Ji, Y.; Foltin, M.; Liao, C. H.; Mtirk, T. D. J. Chem. Phys. 1992, 96, 3624. (15) Radi, P. P.; Bunn, T. L.; Kemper, P. R.; Molchan, M. E.; Bowers, M. T. J. Chem. Phys. 1988.88, 2809. (16) Radi, P. P.; Hsu, M.-T.; Brodbelt-Lustig, J.; Rincon, M.; Bowers, M. T. J. Chem. Phys. 1990, 92, 4817. (17) Yu, D. H.; Anderson, L. H.; Brink, C.; Hvelplund, P. Z. Phys. 0 1994, 29, 53.
J. Phys. Chem., Vol. 99,No. 42, 1995 15437 (18) Scheier, P.; Robl, R.; Schiestl, B.; Miirk, T. D. Chem. Phys. Lett. 1994, 220, 141. (19) Scheier, P.; Miirk, T. D. Int. J. Mass Spectrom. Ion Proc. 1994, 133, L5. (20) van Helden, G.; Hsu, M.-T.; Gotts, N.; Bowers, M. T. J. Phys. Chem 1993,97, 8182. (21) Campbell, E. E. B.; Hohmann, H.; Furrer, S.; Callegari, C.; Kittelmann, 0.;Ringling, J. Contributions SASP 94; Mtirk, T. D., Schrittwieser, R., Smith, D., Eds.; Maria Alm, 1994, p 37. Gaber, H.; Hiss, R.; Busmann, H. G.; Hertel, I. V. Z. Phys. 1992, 024, 302. (22) LeBrun, T.; Beny, H. G.; Cheng, S.; Dunford, R. W.; Esbensen, H.; Gemmell, D. S.; Kanter, E. P. Phys. Rev. Lett. 1994, 72, 3965. (23) Ouaskit, S.; Farizon, B.; Farizon, M.; Gaillard, M. J.; Chevarier, A.; Chevarier, N.; Gerlic, E.; Stem, M. Int. J. Mass Spectrom. Ion Processes 1994, 139, 141. Farizon, B.; Farizon,M.; Gaillard, M. J.; Gerlic, E.; Ouaskit, S. Int. J. Mass Spectrom. Ion Processes 1995, 144, 79. (24) Hirsch, A., et al. Phys. Rev. 1984, C 29, 508. (25) Campi, X. Nucl. Phys., A 1989, 495, 259c. (26) Bauer, W. Phys. Rev. 1988, C38, 1927. (27) Diinser, B.; Lezius, M.; Scheier, P.; Miirk, T. D. Phys. Rev. Lett. 1995, 74, 3364. (28) Diinser, B.; Scheier, P.; Miirk, T. D. Chem. Phys. Lett. 1995,236, 271. (29) Miirk, T. D.; Scheier, P.; Leiter, K.; Ritter,, W.; Stephan, K.; Stamatovic, A. Int. J. Mass Spectrom. Ion Processes 1986, 74, 281. (30) Baldwin, M. A,; Demck, P. J.; Morgan, R. P. Org. Mass Spectrom. 1976, 11, 440. (31) Walter, C. W.; Bae, Y. K.; Lorents, D. C.; Peterson, J. R. Chem. Phys. Lett. 1992, 195, 543. (32) Petrie, S.; Wang, J.; Bohme, D. K. Chem. Phys. Lett. 1993, 204, 473. (33) Hawkins, J. M.; Meyer, A,; Lewis, T.; Loren, S.; Hollander, F. J. Science 1991, 252, 312. (34) Hrusak, J.; Schwarz, H. Chem. Phys. Lett. 1993, 205, 187. Yamaguchi, K.; Hayashi, S.; Okumura, M.; Nakano, M.; Mori, W. Chem. Phys. Lett. 1994, 226, 372. (35) Xu, C.; Scuseria, G. E. Phys. Rev. Lett. 1994, 72, 669. Jing, X.; Chelikowsky, J. R. Phys. Rev. B 1992, 46, 15503. (36) Yannouleas, C.; Bamett, R. N.; Landman, U. Comments At. Mol. Phys., in press. (37) Brechignac, C.; Cahuzac, P.; Calier, F.; deFrutos, M. Phys. Rev. Lett. 1994, 72. 1636. (38) Kreisle, D.; Echt, 0.;Knapp, M.; Recknagel, E.; Leiter, K.; Miirk, T. D. Phys. Rev. Lett. 1986, 54, 1551. (39) Echt, 0.; Miirk, T. D. In Clusters of Atoms and Molecules; Haberland, H., Ed., Springer: Heidelberg, 1994; Vol. 11, p 183. (40) Smith, D.; A d m i N. G.; Alge, E.; Villinger, H.; Lindinger, W. J. Phys. 1980, B 13, 2787. (41) Hansel, A,; Richter, R.; Lindinger, W.; Herman, Z. Int. J. Mass Spectrom. Ion Processes 1992, 117, 213. (42) Diinser, B.; Matt, S.; Muigg, D.; Senn, G.; Scheier, P.; Miirk, T. D., unpublished results.. (43) Klots, C. E. J. Phys. Chem. 1988, 92, 5864. (44) Scheier, P.; Miirk, T. D. Int. J. Mass Spectrom. Ion Processes, in press. (45) Gill, P. M. W.; Radom, L. J. Am. Chem. SOC. 1988, 11, 5311. (46) Senekowitsch, J.; Neil, S. 0.;Meyer, W. Theor. Chim. Acta 1992, 84, 85. (47) Weiske, T.; Koch, W.; Schwarz, H. J. Am. Chem. Soc. 1993, 115, 6312. JF'950588+
