Ion formation pathways of crown ether – fullerene conjugates in the gas phase †

The ion formation of crown ether–[60]fullerene conjugates of the type (crown H)–C60–H with crown = 12cr4, 15cr5 and 18cr6 has been studied with matrix-assisted laser desorption/ionisation (MALDI) and electrospray ionisation mass spectrometry (ESI MS). In total five different ways of ion formation are presented, including metalation (MALDI, ESI), protonation and oxidation (both in MALDI) in the positive-ion mode and deprotonation (MALDI, ESI) and reduction (MALDI) in the negative-ion mode. In line with thermochemistry, the deprotonation and electron transfer processes involve the C60 moiety as the charge-carrying entity, while protonation and metalation occur at the crown ether. Particular emphasis has been placed on the study of metal cation attachment in MALDI varying the crown ether size in the conjugate and using different alkali metal chlorides in the target preparation. Dissociation reactions of the metalated conjugates are influenced by the interaction strength of the metal cation to the crown ether fullerene conjugate. The data confirm an increase in bond strength with smaller metal cations, supporting the notion of charge densitydriven interactions.

Please note that technical editing may introduce minor changes to the text and/or graphics, which may alter content. The journal's standard Terms & Conditions and the Ethical guidelines still apply. In no event shall the Royal Society of Chemistry be held responsible for any errors or omissions in this Accepted Manuscript or any consequences arising from the use of any information it contains. fullerenes would not necessarily be easily ionised and in fact are ionised only by difficult-tocontrol oxidation and reduction processes. 2, 3 While the Wilson-conjugate was readily metalated, 1 the need to perform a synthesis prior to the ESI analysis was not attractive.
However, as a proof of concept study the paper was well accepted. In the Wilson-conjugate the crown ether was linked to C 60 via a methylene (spiro) bridge, occupying a former double bond of the fullerene. The Cretan partner in this collaboration has recently reported a new type of crown ether-fullerene conjugate in which the crown iswithout the use of a linkerdirectly attached via a C-C single bond to the fullerene, which also attains a hydrogen atom as the second substituent to the former fullerene double bond. 4-6 The present work is motivated by the desire to study different ion formation mechanisms for these new compounds and to evaluate the gas phase behaviour of the resulting ions. The two most widely-used ionisation methods to date, electrospray ionisation 7 and matrix-assisted laser desorption/ionisation (MALDI) 8 have been employed. ESI cannot only be applied to these in the well-established way of metalation in the positive ion mode, but the acidic nature of the fullerene hydrogen [9][10][11][12][13] allows through deprotonation also the use of the negative ion mode by the formation of quasi-

ESI Experiments
Most electrospray measurements were conducted with a quadrupole ion trap (esquire6000, Bruker Daltonics). For spectra with higher resolution a quadrupole time-of-flight (QTOF) mass spectrometer with MS/MS capability (micrOTOF-Q II, Bruker Daltonics) was used. As the crown ether-fullerene conjugates are prone to oxidation while in solution, 32 only very small amounts (0.05-0.2 mg) of the compounds were dissolved in dichloromethane (DCM) for each experiment. As the weight measurements on the available laboratory balance were not reliable below 0.1 mg, at least a twofold molar excess of alkali metal chloride was added as methanolic solution (1 g/l). To achieve stable spraying conditions, the sample was then further diluted with acetonitrile (ACN) in a 1:1 volume-to-volume ratio. The concentration of the crown ether-fullerene conjugate in the final solution was about 2·10 −5 mol/l.
The target was prepared as follows. The crown ether-fullerene conjugates were dissolved in DCM and mixed with appropriate volumes of alkali metal chloride salt solutions (1 g/l in MeOH). These mixtures were diluted with DCTB (5 g/l in DCM) to give molar sample-tosalt-to-matrix ratios of 1:2:100. The solutions were deposited on a stainless steel target and dried in air.
The instrument parameter laser power relates to the attenuation of the laser light and is used here in arbitrary units (%), as the actual laser fluence is unknown. For all MALDI spectra a laser power of 0-2 % was used, resulting in sufficiently abundant ions without inducing fragmentation.
The selection of a particular precursor ion for the post-source decay (PSD) spectra 33 is achieved by "ion gating", i.e. after delayed extraction and a short drift region, the ions are Page 4 of 30 Physical Chemistry Chemical Physics Physical Chemistry Chemical Physics Accepted Manuscript deflected by an electric field. The applied high voltage is only switched off for a short period of time to allow ions of a selected mass range to pass unhindered. The fragments travel basically with the same velocity as the precursor ions, but due to their different masses they possess different kinetic energies. As the lengths of the product ions' flight path in the reflectron depends on their kinetic energy, they can be separated and mass analysed using the reflectron. Focusing the product ions on the detector, however, only works for a certain range of kinetic energies. Therefore, the reflectron voltage has to be decreased incrementally and for each step a spectrum for the resulting mass range is recorded. These segment spectra are finally stitched together to form a complete post-source decay spectrum. 33 For the fragmentation study, the PSD spectra of the three crown ether C 60 adducts with Na + , K + , Rb + and Cs + were each measured at least three times for the laser powers 5, 10, 15 and 20 %. For each segment spectrum 200 single shot spectra were averaged. As the calibration of the measurement depends on a sufficiently abundant and resolved precursor ion signal, the laser power in the first segment had to be adjusted for some experiments. Most notably for the spectra at high laser fluence, the parent ion signal of the more unstable complexes was nearly extinguished and the laser power had to be lowered to allow calibration. The laser power in all other segments was kept constant at the values mentioned above.
The MALDI spectra in negative-ion mode were acquired using a TOF/TOF instrument (ultraflex II, Bruker Daltonics), which is capable of measuring PSD in the negative ion mode.
It also incorporates an Argon-filled collision cell to enhance fragmentation (placed between the ion source and the precursor ion selector) and a "lift" device to allow the acquisition of a full PSD spectrum in a single scan without incrementally decreasing the reflectron voltage. 34 This is referred to hereafter as the LIFT method.

Discussion
In Figure 1 the crown ether-[60]fullerene adducts are depicted together with the five different molecular and quasi-molecular ions that could be generated in this study. The adducts are composed of a 1,2-addition of a hydrogen atom and a covalently linked crown ether, the synthesis of which was recently published. 4 The crown ether moiety was varied from 12-crown-4 (12cr4) to 15-crown-5 (15cr5) and 18-crown-6 (18cr6).
As a brief summary of the discussion that will follow we emphasise that both the cation and anion radical are readily produced by electron transfer MALDI in both ion modes. ESI was used as the other ionisation method. In the negative ion mode the deprotonated molecule [M − H] − is abundantly formed. Cationisation can easily be achieved by both approaches, while the protonated molecule was only occasionally observed as a minor by-product of the MALDI process. Efforts to protonate the crown in ESI remained unsuccessful, cationisation was the sole reaction.

Negative-Ion Mode
The negative-ion formation is displayed in Figure  The dissociation behaviour of the radical anion was probed by LIFT which is a collisioninduced/laser activation experiment on a TOF/TOF instrument. 34 The spectrum reveals that the electron resides on the C 60 moiety while the crown ether acts as "ligand" and is released in the dissociation (Fig. 2b) (Fig. 2c). The formation of these species is perfectly in line with the fact that C 60 HR compounds of this type are acidic. 9-13 Evidently deprotonation is more attractive than electrochemical reduction as alternative ion formation mechanism which is occasionally operative for other, non-acidic derivatised fullerenes. 3 Upon activation, the crown ether radical is released which identifies the fullerene as expected as the charge carrier (Fig. 2d). In the following, the formation of positive ions applying MALDI will be discussed.

Positive-Ion Mode
The positive-ion mode MALDI mass spectra of C 60 H(15cr5 − H) are shown as indicates that the C 60 moiety does not play a dominant role as a charge-carrying species under the applied conditions. A noticeable exception is the radical cation, i.e. the molecular ion, of the crown ether adduct, which is seen as a relatively low abundant ion in the high mass region. Sodiation, however, although no sodium was added in this experiment is much more pronounced than the formation of the radical cation. Sodium is a well-known, omnipresent  37 , to allow electron transfer from (crown − H) • to DCTB +• , but also lower than that of C 60 H • , so that it may not be ionised in competition to (crown − H) • .
The ionisation energies (IEs) of the free crown ether molecules have been established by photoelectron spectroscopy and amount to 9.3 eV for 12cr4, 9.6 eV for 15cr5 and 9.7 eV for 18cr6. 46, 47 These values are considerably higher than the IE of DCTB and what can be assumed for the IE of C 60 H • . The latter is not known, but can be expected to be lower than the IE of C 60 (IE = 7.6 eV). Unfortunately, the ionisation energies of the crown ether radicals are unknown and the proposed mechanism has to remain speculation. However, considering the C-H bond, the IE of the [M − H] • radical is often considerably lower than the IE of the corresponding organic molecule M. 48 Therefore, it can be safely assumed that the IE of the crown ether radical is clearly lower than the IE of the DCTB, so that DCTB +• ions are in fact capable of ionising the crown ether radical.
As a result of aging, the crown ether adducts show a certain degree of oxygen addition to C 60 .
The ease by which fullerene derivatives can oxidise at ambient conditions has been studied recently using C 60 H 2 as example. 32 Following the signal of MNa + at m/z 963 in Fig. 3a, the signals at m/z 979, m/z 995 and m/z 1011 are most likely caused by this oxidation. Therefore, the signal at m/z 979 represents predominantly (if not exclusively) the sodium adduct of M with oxygen addition (MONa + ) rather than K + addition to the crown ether adduct (MK + ). The mass difference between sodium (MNa + ) and a potassium (MK + ) adduct amounts to 16 mass units, which also accounts for the addition of an oxygen to the sodium adduct (MONa + ). The same situation exists for the mass difference of a lithium (MLi + ) to a sodium (MNa + ) adduct.
Unfortunately the instrumental resolution is not sufficiently high to distinguish these isobaric ions. As a consequence the oxidation prevents the accurate quantitation of Na + adducts following Li + adducts and of K + adducts following Na + adducts. However, if the addition of the heavier alkali cation is of sizable proportion, it will be noticed.

10
The dominating ion formation process (Fig. 3a) is, however, the Na + addition to the neutral molecule, even though no sodium was added. If the alkali metal is deliberately added, the attachment of the respective metal cation is the most prominent process in each case with C 60 H(15cr5 − H). In the low mass region the bare metal cation signal gains more importance with increasing size and overtakes the ionised radical of the crown ether as the most intense signal (in that region) from K + onwards.
[ Figure 4, near here please] In the present experiments, metalated ions can in principle be formed via two major pathways, including gas phase metal ion attachment following and/or accompanying the ablation process or these ions may be pre-formed and deposited on the target from solution. Both pathways cannot be clearly distinguished in the present experiments and may also change in their relative importance depending on the crown ether adduct and the metal ion. The ion distribution of the MS 1 spectrum is thus influenced by a multitude of competing processes and, therefore, not suited to evaluate the bond strength between metal cation and crown ether C 60 adduct. However, the careful comparison of the ion distributions obtained for the three crown ether adducts without and with addition of the five metal cations allows several observations to be made that relate to the interplay of metal cation and the crown ether adduct.
Of the three crown-adducts C 60 H(12cr4 − H) shows the highest abundance of the real molecular ion, the radical cation. Electron transfer of the analyte with matrix ions effectively competes with metal cation attachment as ion formation process. This is particularly evident when no salt is added, here the radical cation exceeds the abundance of the sodiated ions ( Fig. 4a). For the adducts with the 15cr5 and 18cr6 ligands, the sodiated adduct exceeds the radical cation ( Fig. 4b and c, respectively). Evidently for the bigger-sized crowns, sodiation even with only minute amount of the omnipresent Na + is much more attractive than the electron transfer reaction. The addition of Li + and Na + to C 60 H(12cr4 − H) reduces the radical cation. The addition of K + and larger cations results in sizable amounts of the radical cation, indicating that electron transfer ionisation efficiently competes with cation attachment even indicating that electron transfer is more efficient than metal cation uptake. For C 60 H(18cr6 − H), however, the metalated crown ether radical, starting with Na + addition, is more prominently produced while the crown cation becomes less important. It seems more likely to assume that the crown ether radical experiences metalation after the dissociation of the neutral adduct rather than representing the dissociation of the metalated analyte (loss of • C 60 H). The dramatic increase of the cationised crown fragment occurs only for C 60 H(18cr6 − H) which would mean that the fullerene/crown bond becomes considerably weaker than in the case of C 60 H(12cr4 − H) and C 60 H(15cr5 − H). There is, however, no structural reason that would lead to a considerable weakening of that bond. It seems reasonable to assume that the rate for cation addition exceeds the rate of electron transfer for 18cr6, which is reversed for the other crown ethers.

Post-source decay
The post-source decay (PSD) experiment, details of which are given in the supporting information (Fig. S3) and in the Experimental section, can be referred to as MS 2 experiment or as tandem mass spectrometry, as it represents the recording of the product ions of a chosen precursor following its dissociations. Examples are shown in Figure 5. Three examples were chosen as they represent a particular fragmentation behaviour which is typical for the ions under investigation. Figure 5a depicts  ) a partial and complete loss of the crown ether unit. This clearly reveals, as mentioned earlier, that the charge is located on the C 60 unit. Only a small signal for the charged crown ether fragment is observed. This charge distribution is essentially in accordance with the ionisation energies of C 60 (7.6 eV) 44,45 vs. crown ether (9.3-9.7 eV) 46,47 . In striking contrast is the decay behaviour of the sodiated crown ether adduct C 60 H(15cr5 − H)Na + , displayed in We note that any of these additional fragment ions is still retaining the Na + within the decaying crown. Finally, the bare Na + ion is detected as a product ion that may result from the sodiated precursor ion and any of the other daughter ions. As a final type of dissociation the Cs + cationised crown ether adduct is displayed (Fig. 5c). Here, the same general decay pattern as observed for the Na + adduct is taking place, however, there is only one successive decay product after the fullerene loss and the bare metal ion is clearly more pronounced. Both these latter features indicate that the Cs + ion is much more weakly chelated by the crown in comparison with Na + .
[ Figure 6, near here please] The dissociation of the precursor ion is caused by the internal energy content of the ion and by collisional activation that may occur during the flight. 50 The dissociation pattern shows a profound dependency on the laser power variation. That is the laser activation influences the internal energy content of the ion 50 and thus influences the decay behaviour of the ion. PSD spectra were recorded at varying laser power settings to account for the deviation in product ion abundances. The resulting fragmentation dynamics (Fig. 6) is used to evaluate the relative affinity of the metal ion to the crown ether adduct. The abundance of the precursor ion could not be taken into account, as its isotope pattern was distorted at higher laser power and for some measurements the laser power in the first segment, i.e. the high mass end, had to be lowered in order to gain a sufficient signal for calibration. Therefore, to evaluate the cation/crown ether bond strength the variation of the ratio of three different types of ions relative to all PSD fragment ions were considered. Firstly, the abundance ratio of the intact,     ). An enlargement of the low mass region of each spectrum is shown on the right side, containing the crown ether ion, its fragments and the bare alkali metal ion. The peak annotations shown in grey correspond to product ions generated by fragmentation of C 60 H(15cr5 − H)Na + in a) and C 60 H(15cr5 − H)K + in b) which were inadvertently included in the isolation due to the ineffectiveness of the ion gate. Figure 6. Results of the PSD experiments for the three crown ether-C 60 adducts with Na + , K + , Rb + and Cs + . The intensity of each fragment ion species, namely (cr − H)M +• , its fragments and the bare alkali metal ion M + , divided by the sum of intensities of all fragment ions is plotted against the applied laser power. As each experiment was repeated at least three times, the arithmetic average is used with the error bars representing the standard deviation. The polynomial fit is of no mathematical relevance and is only used to guide the eye.