Ionic Lewis superacids in the gas phase. 'Part 1. Ionic

International

Journal of Mass Specirometry

and Ion Processes,

21

124 (1993) 2 l-36

Elsevier Science Publishers B.V., Amsterdam

Ionic Lewis superacids in the gas phase. ‘Part 1. Ionic intermediates from the attack of gaseous SiF,f on n-bases Felice Grandinetti”, Giorgio Occhiucci”, Ornella Ursini”, Giulia de Petrisb and Maurizio Speranzab ‘Istituto di Chimica Nucleare e Servizio FTIUS de1 CNR, Area della Ricerca de1 CNR, 00016 Monterotondo Stazione, Rome (Italy) bDipartimento di Studi di Chimica e Tecnologia delle Sostanze Biologicamente Attive Universik di Roma “La Sapienza”, 00185 Rome (Italy)

(First received 12 June 1992; in final form 20 September 1992) Abstract The reactivity of the SiF: cation towards oxygen bases (H,O, CH,OH, and CH,CH,OH) and nitrogen bases (NH, and CH, NH,) has been studied using Fourier-transform ion cyclotron resonance mass spectrometry. The SiF: ion exclusively attacks the n-centre of the selected bases, yielding excited onium intermediates that undergo fragmentation by elimination of either an alkyl cation (CH,OH, CH,CH,OH, and CH,NH,) or an HF molecule (HsO, NH,, and CH,NH,). In H,O, various protonated fluorosilicic and silicic acids are formed which can be readily converted into their esters and amides by reaction with alcohols and ammonia, respectively. The structure and the reactivity of several such species have been investigated by mass-analysed ion kinetic energy-collision induced dissociation spectroscopy and ab initio calculations. The extreme affinity of SiF: toward n-type electrons ranks it as a powerful gaseous “Lewis superacid”, suitable for generating long-lived, highly reactive ions, e.g. CH: , in “non nucleophilic” gaseous media, such as, for instance, SiF,. Keywordr: gas phase ion chemistry; FT-ICR; trifluorosilyl cation; gaseous Lewis superacids; gaseous protonated

silicic

acids.

Introduction

In the last few decades, a great research effort has been addressed to the understanding of the nature and reactivity of carbenium ions in the gas phase and in solution with the aim of establishing a direct link between theoretical and experimental approaches in ion chemistry, and of determining the effects of perturbing solvation and ion-pairing Correspondence to: M. Speranxa, Dipartimento di Studi di Chimica e Tecnologia delle Sostanze Biologicamente Attive, Universita di Roma “La Sapienza”, 00185 Rome, Italy.

0168-I 176/93/$06.00

factors on the intrinsic properties of ions in the condensed phase. The synergic development of the experimental procedures in gas and condensedphase ion chemistry has been decisive in the success of this effort. In solution, a determinant impetus was given by the discovery that stable, long-lived electrondeficient ions, including carbenium ions, can be obtained in low-nucleophilicity solvents by the action of superacids on suitable molecules [l]. In these systems, the lifetime of cationic species which are elusive transients in solvolytic media, is greatly

0 1993 Elsevier Science Publishers B.V. All rights reserved

22

F. Grandinetti et al./Int. J. Mass Spectrom. Ion Processes 124 (1993) 21-36

enhanced by the low nucleophilicity of the medium so as to allow their spectroscopic characterization [2]. The same concept finds its extreme expression in the gas phase, where the lack of solvent and of their counterion (a far-removed electron) allows generation and investigation of virtually all carbenium ions, including those escaping spectroscopic detection even in superacid media, i.e. the majority of primary alkyl cations, vinyl and alkynyl ions, etc. In this area, the use and the gradual development of sophisticated mass spectrometric techniques have been essential. Nevertheless, the evaluation of the structure and the reactivity of a gaseous ion by these techniques as compared to those in solution often meets with considerable difficulties due to the inherently limited structural and time resolution of mass spectrometry, as well as to its typical operating conditions. Indeed, ionic species are normally generated in the gas phase with substantial excitation energy that, at the low pressures typical of mass spectrometry (< 1 Torr), cannot be readily removed [3]. As a result, frequent is the observation of ions in mass spectrometry following high-energy reaction pathways which have no counterpart in solution [4]. An obvious means to circumvent this problem implies the use of experimental methodologies, such as a stationary radiolysis [5’J,operating at much higher pressures (up to several atmospheres) where most ionic species can be thermalized by unreactive collisions with an inert moderator prior to reaction [5,6]. Sometimes, however, it may happen that radiogenie ions are reactive enough to attack some bulk components of the gaseous mixture before reacting with the substrate of interest, normally present in the mixture at low concentrations in order to minimize its direct radiolysis [7]. Thus, setting up an approach to highly reactive, structurally defined carbenium ions, thermally equilibiated in a “non-nucleophilic” gaseous medium, would be extremely desirable. Taking the generation of carbenium ions in superacid solution as a model procedure, a viable approach could involve the use of a “gaseous ionic superacid” catalyst to generate carbenium ions in a non-

100 -

-100 200

I

300

D”(R f-F -1 (kcallmol)

Fig. 1.Dissociation energies of gaseous RF (P’(R+ . F-)) and enthalpies of the gas-phase F- transfer from RF to SiFT [8].

nucleophilic gaseous matrix. In Fig. 1, an ordering of decreasing strengths of potential cationic acids, R+ ,mainly Lewis-type acids, is reported, expressed by their order of decreasing fluoride-ion affinities and by the directly related (D”(R+ 71. F-)) enthalpies of fluoride-ion transfer to a reference Lewis acid, i.e. SiF: (AW). According to the gasphase acidity scale of Fig. 1, SiF: is recognized as a potential “Lewis superacid” in the gas phase, isoelectronic with the well-known AlF3- Lewis superacid in solution, which is able to generate virtually all carbenium ions, except the unsubstituted ethynyl cation HCE C+, by fluoride-ion abstraction from the corresponding alkyl fluorides. This potentiality, coupled with the exceedingly low nucleophilicity of its conjugated base SiF, [9], expressed by its very low proton affinity ((PA) = 116 kcalmol-‘) [lo], makes the SiF: /SiF4 pair an ideal candidate for acting as a Lewis superacid to catalyze the formation of reactive carbenium ions in a gaseous, non-nucleophilic medium. As a first preliminary insight into the applicability of SiF: as an ionic Lewis superacid in the gas phase, a mass-spectrometric investigation of its reactivity towards gaseous n-type bases, such as H,O, CH,OH, C,H,OH, NH, and CH,NH,, and

F. Grandinetti et al./lnt. J. Mass Spectrom. Ion Processes 124 (1993) 21-36

their mixtures, has been undertaken in view of the fact that these bases are normally present in radiolytic samples, either as additives or as radiolyti~lly formed impurities. Experimental Materials SiF, was a research grade gas from UCAR Specialty Gases N.V., with a minimum purity of 99.99mol.%. (SiFX)*O and HF were the major impurities present in commercial SiF4 which were removed by passage of the gas through an ihydrous KF : SiO, packed column kept at 0°C. The n-type bases used were commercially available and all were degassed byseveral freeze-pump-thaw cycles at 77 K. In the case of gases, purification by preliminary distillations in a vacuum line was utilized. Procedure The Fourier-transform mass spectrometry experiments were carried out with a Nicolet FTMS-1000 instrument equipped with a 2.54cm cubic trapping cell situated between the poles of a superconducting magnet operated at 2.0T. Typical experimental conditions were as follows: nominal electron beam energy 30eV; electron beam pulse 5 ms; emission current 400nA; total pressure, 3 x IO-’ to 2 x 10e6Torr; resolution 1000 full width half height at mass 100. Sample pressures (uncorrected) were measured with a Granville-Phillips 280 Bayard-Alpert ion gauge. The inlet system and trapping cell were kept at room temperature. The ion sequences taking place in the reaction cell were examined by multiple resonance experiments by choosing suitable reaction time sequences to maximize the abundance of a given precursor and appropriate frequency windows to remove all the undesired ions from the cell. Detection of the daughter ion pattern after a suital e delay time gives direct information about the reaction sequence concerning the precursor under investigation and the rate of formation of its daughter species.

23

The chemical ionization mass spectrometric (CIMS) measurements were carried out with a ZAB2F mass spectrometer (VG Micromass, Ltd.), whose CI source was fitted with a MKS Baratron Model 221 A capacitance manometer. Typical operating conditions were as follows: emission current, 1 mA; repeller voltage, 0 V; source temperature 18O’C; accelerating voltage, 8 kV; total pressure in the ion source, 0.3 Torr. The ZAB-2F instrument was used as well to record mass-analyzed ion kinetic energy spectra (MIKES), at a typical energy resolution of 4000 full width at half maximum. The collision-induced dissociation (CID) spectra of the ions were taken by admitting He into the collision cell of the ZAB-2F sp~trometer and increasing the pressure until a 70% reduction of the main beam was obtained. computational details Ab initio quantum-mechanical calculations were performed by using an IBM-902 1/VF version of the GAUSSIAN 88 suite of programs [l 11. The standard internal 6-31G* [12(a)] and 6-311G** [12(b)] basis sets were thoroughly employed. Geometry optimizations were performed, at the SCF 6-3 1G* level of theory, by gradient based techniques [13], and the corresponding analytical frequencies were computed in order to ascertain the real nature (minimum or higher-order saddle point) of each critical point on the surface. The SCF 6-31G* zeropoint energies were scaled by 0.89 in order to account for known inadequacies at this computational level 1141.Single-point calculations, at the MP2 [15] level of theory were performed with the 6-31 lG** basis set, in order to include the correlation energy effects on the relative stabilities of the investigated species. Results and discussion SF: ion attack on oxygen bases SiF: is by far the predominant ion observed in the Fourier-transfo~ mass spectrometer (FTMS)


24

g

80

f

Time

(ms)

Fig. 2. Time dependence of ion abundances following ionization of a 1: 1 mixture of SF,: H,O (P,,, = 3 x IO-‘Torr) and isolation of SiF: : (a) 1; (b) 2; (c) 3; (d) 4; (e) 5.

cell containing pure SiF, at 3 x 10P7Torr. It is known to be inert toward the parent molecule [9], but highly reactive toward H20, contained in the FTMS cell at its background level (about lo-’ Torr), yielding SiF?OH+ [lo]. When, under the same conditions, equimolar amounts of SiF, and H,O are admitted through independent inlets into the FTMS cell, sequence (1) is observed with the stationary concentrations of all the ionic intermediates l-5 involved depending upon the reaction time as described in Fig. 2. Species 5 is formed appreciably only at relatively long reaction times (> 500 ms) from addition of Si(OH): to an H,O molecule. Best fit of the experimental curves of Fig. 2 provides an estimate of the rate constants of the individual steps of sequence (1): SiF+ s

SiF20H+ 3 2

1 s

SiF(OH)t

Si(OH): 3 4

3 Si(OH), OH: 5

whose values are reported in Table 1. From this Table, it emerges that the first three steps of sequence (1) are highly efficient, whereas the last one is comparatively inefficient. A plausible reason for the high efficiency of the first three steps of

sequence (1) is founded on the exceedingly good leaving-group ability of HF coupled with the persistence of a large fraction of the positive charge on the Si atom after each substitution. When, in fact, SiCl: is used instead of SiF:, the SiClzOHf daughter species is formed which appears reluctant to further react with water [16]. Sequence (1) provides a convenient access to species that formally correspond to protonated meta-fluorosilicic acids, i.e. 2 and 3, as well as protonated me&-silicic, i.e. 4, and o&o-silicic acid, i.e. 5, whose occurrence in the free state was never proved to date. Their partial characterization has been carried out by MIKES-CID analysis. When, in fact, an SiF,: H,O gaseous mixture is introduced into the CIMS source at about 0.3 Torr, all the members of the ion family of sequence (1) were observed, together with substantial stationary concentrations of the corresponding adducts 68 with H,O. The MIKES-CID spectral data are listed in Table 2. Accordingly, while protonated meta-silicic acids 2-4 do not exhibit any appreciable unimolecular decomposition, their ortho-silicic SiF,OH: 6

SiF, OHOH: 7

SiF(OH)* OH: 8

analogues 5-8 display intense peaks corresponding to a loss of an HX (X = OH or F) moiety. Even considering the different internal energy possessed by ions 2-8, this difference reflects a comparatively high stability of the Si+ ***XH interaction in the sampled meta ions 2-4, relative to the sampled ortho ions 5-S. In order to gather a better insight into the nature of protonated meta-silicic acids, e.g. 4, ab initio quantum-mechanical calculations have been performed on the structural and energetic features of [H,, Si, O,]’ ions as well as on their interconversion processes. Two distinct [H3, Si, 03]+ stable isomers, I and II, which are connected by the transition state III, have been identified on the SCF 6-31G* potential energy surface. Structures I-III and their corresponding optimized geometries are shown in Fig. 3. The relative stabilities of these species were evalu-

25


TABLE 1 Rate constants of ion/molecule reations k”

Reaction

Species

Efficiencyb

AH”

(k/k,,

(kcalmol-‘)

1

H,O

SiF: + H,O + SiF,OH+ + HF SiFrOH+ + Hz0 -+ SiF(OH)t + HF SiF(OH): + H,O + Si(OH): + HF Si(OH): + H,O + Si(OH),OH:

1.7 1.5 1.4 0.03

1.0 0.88 0.82 0.017

CH, OH

SiF: + CH,OH + CH: + SiF,OH + CH>OH+ + SiF,H CH; + CH,OH + CHrOH+ + CH, CHrOH+ + CH,OH + CH,OH$ + CH,O

0.6 1.1 1.8 2.5

0.35 0.65 0.69 1.19

-41 -63 - 10

SiF: + CH,CH,OH

0.3 1.0 2.4

0.20 0.60 0.98

-26

Si(OH): +HZ~. Si(OH),OH: +c,,ou~ Si(OH),OMe+ + H,O

0.06 0.40

0.034 0.24

,I(OH),OH$ + CH,OH + H,O + Si(OH)3MeOH+ Si(OH), MeOH+ + HZ0 -t CH, OH + Si(OH),OH:

0.25 0.09

0.15 0.05

CH,CH,OH

CH,CH: CH,OH/H,O

+ SiF,OC,H,+ + H, --) CH,CHl + SiF,OH + CH,CH,OH + CH,CH,OH; + C,H,

“Phenomenological rate constants (x 109cm3 molecule-’ s-l). bk,no values calculated according to T. Su and M.T. Bowers, Int. J. Mass Spectrom. Ion. Phys., 12 (1973) 347; 17 (1975) 211.

TABLE 2 MIKE-CID spectra of an SiF., : H,O mixture ml2

Intensities” HF$iO+ (2) (m/z 83)

85 83 82 81 80 79 78 66 64 63 62 61 47 45 28

H, FSiO: (3) (mlz81)

H, SiO: (4) (m/z 79)

H,SiO: (5) (m/z 97)

H, F, SiO’ (6) (m/z 103)

H,F,SiOz (7) (m/z 101)

H,, FSiO: (8) (mlz 99)

70.3 b

b

8.6 b

8.8 b

28.3

b

28.7 b

30.6 44.3 16.2 15.0

22.2 2.3

13.2 38.2 13.1 4.9 14.3 19.6 8.2 1.7

7.5 31.4 36.5 29.8 2.3

16.0 29.6 15.0

6.7 1.7 0.6

12.3 14.2 15.7 6.8 6.4 11.6 3.9 0.8

“All intensities are normalized respect to the sum. bPeaks showing unimolecular contribution are excluded. All unimolecular fragmentations display gaussian-shaped peaks.

16.4 13.4 15.6 10.0 7.4


26

0

According to Fig. 3, significant differences are associated with structures I and II. The considerable %-OH, bond distance (1.76 A) in II characterizes it as a tightly bound ion/dipole complex between Si02Hf and H20. Consistent with this, the corre.. sponding interaction energy is computed as amounting to 66 kcal mol-‘, at the MP2/6-3 1lG** //6-31G* level of theory (Table 3). Irrespective of the theoretical level employed, ion I is considerably more stable than ion II. At the post-SCF level of theory, the energy difference is computed as 42.4 kcalmol-‘, thus reducing by about 10 kcal mol-’ the corresponding SCF energy gap. From the total energy of meta-silicic acid, reported in Table 3, the PA of this molecule can be evaluated as 210.7 kcalmoll’, if the formation of structure I is assumed. Kinetic aspects of the I --) II isomerization process have been investigated by characterizing the relevant transition structure III. The SCF 63 1G* geometry of this species (Fig. 3) is consistent with a 1,3-H migration. The negative eigenvalue of the corresponding Hessian matrix, - 2003 cm-‘, is associated with the in-plane motion of the migrating hydrogen, which is almost equally distant from 0( 1) and O(3) (1.18 vs. 1.14 A, respectively). From the data in Table 3, the I + II isomerization is a high-energy process, whose MP2/631 lG** activation barrier is computed to be as large as 63 kcal mol-‘, relative to the most stable ion I. Inspection of the energy profile of Fig. 4

/H

I

1.336

H\olSi-yo 129.4

-I

I

H

H

11,c,

I* C3h

1.512

Si++

I& C,

P H

Fig. 3. SCF 6-3lG* optimized geometries of [H,, Si, O,]+ species.

ated by performing MP2/6-3 11G** calculations at the SCF geometries and the corresponding energy values are given in Table 3. Both structures I and II correspond to minima on the surface, as derived from the calculation of the corresponding vibrational frequencies. The C,, symmetric ion I can be formally viewed as the Si = 0 oxygen-protonated form of the meta-silicic acid, whereas protonation at one of its Si-OH oxygen atom yields the C,- symmetry structure II. TABLE 3 Absolute energies (atomic units), relative stabilities (kcalmol-‘) the H,O molecule Species

and ZPEs’ (kcalmol-‘)

for the [H,, Si, OJ+, the OSiOH+ ions, and

Basis set

ZPE (SCF 6-31 G*)b

6-31 G*

6-311 G**

MP2/6-311 G*

I

-515.10845 (0.0)

- 5’15.21147(0.0)

- 515.88588 (0.0)

25.8

;:I (TS) Si02H+ H,G H,SiO,

--515.02395 514.97763 (+53) - 438.89578 - 76.01075 - 514.75922

-515.08017 (+53.7) -515.12590 - 438.96511 - 76.04692 - 514.85335

-515.78019 (+63) -515.81888 (+42.4)

26.3 23.0 9.4 12.8 19.8

“Zero point energy. ?xaled by 0.89.

-y;;:;;;;} - 515.54023

(+ 108.0)

27

F. Grandinetti et a/./M. .I. hhass Spectrom. Ion Processes 124 (1993) 21-36 TABLE 4

t

E (kcal mot’)

MIKE spectra of adducts between 4 and reference bases (B)

1oo-

so-

Adducts

Fragmentsaqb

m/z

4

Others”

607060-

B

BH+ 5

BH+ -Hz0

so-

159

40-

3.5 96.5

3020lo-

NO

O-

reaction

186

11.5 09

137 CH,COCH,

13.8

87.6 (m/z 156)

coordinate

Fig. 4. MP2/6-31 lG*//6-31G* potential energy profile of [H,, Si, O,]+ species.

indicates that the transition state III is 45 kcal mol- ’ below the Si02Hf + H,O dissociation limit, thus supporting the view of two very stable isomeric structures for protonated meta-sihcic acid separated by a substantial activation barrier. A check of the performance of the post-SCF level of theory in these systems is provided by a comparison between the theoretical estimate of the PA of meta-silicic acid (210.7 kcal mol-‘) and that measured by Cooks’ kinetic method [17]. Thus, adducts between 4 and many reference bases have been isolated in the CIMS source (P x 0.3 Torr) of our reverse-geometry mass spectrometer and their MIKE fragmentation spectra recorded. Within the reasonable assumptions that competitive unimolecular dissociations of the adducts involve similar frequency factors and vanishingly small reverse activation energy (as suggested by the lack of appreciable kinetic energy release of the ionic fragments (see footnote b of Table 2), an estimate of the PA of meta-silicic acid can be drawn from the data of Table 4. In fact, the dramatic change in the fragmentation pattern of the adducts with reference bases B more basic than pyrimidine (PA = 210.5 kcal mol- ’ ), wherein predominant formation of BH+ is observed, and with those less basic than nitrosobenzene (PA = 204.8 kcal mol-‘), wherein 4 is mainly formed, points to a PA of 208 &3 kcal mol-’ for meta-silicic acid which well fits the

233

2.5

44.6

41.6 (m/z 119)

32.5

60.0 (m/z 2 15) 5.0 (m/z 121)

cl

“The values are intensities which are normalized with respect to the sum. bAll peaks display a gaussian-type shape, without appreciable kinetic energy release.

theoretical prediction. A visualization of the different fragmentation patterns is given in Fig. 5. Concerning the ortho-silicic acid, direct application of Cooks’ procedure failed under the CIMS conditions adopted, owing to the scarce tendency of 5 to form adducts with most reference bases used. Nevertheless, a rough estimate of the PA of ortho-silicic acid was obtained by taking advantage of the pronounced tendency of adducts between 4 and ketones RCOCH,, such as acetone, acetophenones, etc., to unimolecularly fragment via the competing pathways shown in reaction (2): Si(OH), + CH, = C-R

+ c

CH,

(HO),%+ * O=C R

[

Si(OH),OH:

+ HCr

CR

1 (2)

The completely different fragmentation patterns

F. Grandinettiet al./Int. J. Mass Spectrom. Ion Processes 124 (1993) 21-36 H’

?a

BH+- H20

Fig. 5. MIKES of [4 * B]+ adducts: B = la) pyrimidine; lb) nitrosobenzene; 2a) acetone; 2b) mera-chloroacetophenone.

observed for acetone and meta-Cl-acetophenone (Fig. 5) are consistent with a PA for or&o-silicic acid which is intermediate between that of meta-Clphenylacetylene (196.1 kcal mol-‘) and that of propyne (182.0 kcalmoll’) [18]. Having established that a convenient route for production of meta- and ortho-silicic acid in the gas phase involves the attack of SiF: on H,O (sequence (1)) it is of great interest to verify whether esters of the same acids can be generated by attack of SiF: on alcohols in the FTMS cell.

The gas-phase reaction pattern of SiF: with CH30H in an 8 : 1 mixture of SiF,: CH,OH (P,, = 4.5 x 10-‘Torr) is rather intriguing, being basically different from that of water. Indeed, the reaction network depicted in Scheme 1 is observed with the stationary concentrations of all the ionic involved depending upon the intermediates reaction time as described in Fig. 6. Best fit of the experimental curves gives the rate constants of the individual steps as reported in Table 1, which indicate that SiF: efficiently attacks CH,OH

29

F. Grandinetti et aLlInt. J. Mass Spectrom. Ion Processes 124 (1993) 21-36

(4

FsSiOC2H4+

SiF,+ + CH,CH20H

SiF3+ + CH,OH

- SiFsOH

SiF3H (4

+ CH,OH - CH,o

CH,CH,+ (c)

I CH30H2+

+ CH3CH20H - C2H4

CH,CH20H2+

Scheme1.

Scheme2.

yielding both CH: and CH20H+ , the latter arising from attack of CH$ on CH,OH as well [ 191. The CHzOH+ eventually loses a proton to CH30H. Formation of CH: from path (a) of Scheme 1 implies the highly exothermic attack of SiF: on the oxygen of CHjOH [20], yielding the excited oxonium intermediate 9, wherein simple O-R bond cleavage (channel (3a)) predominates over any conceivable concerted HF elimination (channel (3b)), which instead is the exclusive fragmentation

channel observed in 6. Since both thermochemitally allowed, predominance of channel (3a) over channel (3b) in 9 is accounted by the more favorable entropy factors. In 6, simple O-R (R=H) bond fission (channel (3a))‘is much more energetically demanding [21] than the same process in 9 (R = CH,) and, thus, only the concerted HF elimination (channel (3b)) takes place. In principle, several mechanisms could be involved in the formation of CH,OH+ from path

H I

SiF,+ + ROH /“\

1

F,SiOH + R+

(3a)

c F2&OR + I-IF SiF, I R = H (6); CH, (9); CH,CH, (10)

R

(1)

(3b)

100

Q

e $ C

80

0 ‘0

60

5

2

40

s '3 a 5 K 2o

Time (ms) 0

200

400

600

600

1

Time (ms) Rg. 6. Time dependence of ion abundances following ionization ofan 8: 1 mixture of SiF,: CH,OH (P,O,= 4.5 x IO-‘Torr) and isolation of SiF,+ : (a) 1; (b) CH,OH*

; (c) CHZ ; (d) CH,OH$

Fig. 7. Time dependence. of ion abundances following ionization of a 5: 1 mixture of SiF,: CH,CH,OH (P,, = 6 x IO-‘Torr) and isolation of SiF: : (a) 1; (b) F,SiOC,H: ; (c) C,H: ; (d) CH,CH,OH: .

30

F. Grandinetti et aLlInt. J. hfass Spectrom. ion Processes 124 (1993) 21-36

(b) of Scheme 1, including a direct hyd~de-ion transfer from CH,OH to SiF: or, alternatively, the unimolecular fragmentation of excited 9 with loss of SiF,H (reaction (4)). Evidence will be given below in favor of the ?atter hypothesis.

fra~entation complex is prevented since it is about 3 kcal mol-’ endothermic [lo]. Occurrence of F,SiOC,H,+ in Scheme 2 (path(a)) implies its origin from 10via elimination of an Hz molecule. The same H, loss is not operative in the

-

SiFs+.t CH,OH -+

CH$H+

+

SiF,H

(9)

(1)

Formation of SiF,H from attack of SiF: on Hz0 is prevented, since it is thermochemically forbidden (AH“ = + l~kcaimol-I).

excited 9 from methanol, thus suggesting that at least one of the atoms of HZ originates from the p carbon of 10 (channel 5b,c). 1.2

siF3+

(1)

(4)

+

CH3CH@H

+

-

(10)

I2’3

CH, =

CH -

6 -

SF3

(5c)

I H

The gas-phase attack of SiF: on CH,CH,OH in a 5 : 1 mixture of SiF4 : CH,CH*OH (P,,, = 6 x lo-‘Torr) induces the reaction network shown in Scheme 2, with the stationary concentrations of all the ionic intermediates involved depending upon the reaction time as described in Fig. 7. Best fit of the experimental curves gives the rate constants of the individual steps as reported in Table 1. In analogy with the H,O and CH,OH systems, SiF: is found to e~ciently attack CH3CH,0H yielding the F1SiOC2H4+ and CH,CH$ fragments in about 1: 3 proportion. While F,SiOC2H$ is inert toward its parent molecules, CH,CH: rapidly loses a proton to CH,CH,OH yielding CH,CH,OH: . As for CH30H, formation of CH,CH: (path(b) of Scheme 2) is att~buted to the O-R (R = CH, CH2) bond fission in the intermediate 10, excited by the exothermicity of its formation process. Proton transfer between CH,CH: and SiF,OH in the ensuing eI~tros~tically-bonded

The lack of any detectable formation of CzH50+ from the attack of SiF: on CH3CH,0H indicates that direct hydride-ion abstraction from CH, CH2 OH to SiF: , although thermochemically allowed (AH* = - 68 kcal mol-‘), is kinetically superseded by the reaction channels of Scheme 2, proceeding through formation of the oxonium intermediate 10. By extension, direct hydride-ion transfer from CH30H to SiF: (AHa = - 47 kcal mol-t) is thought to be a minor path in the methanol systems, thus supporting 9 in reaction (4) as the major ionic precursor of CHzOHf . In conclusion, completely different reaction patterns are observed in the attack of SiF: on H,O and ROH (R=CH3, CH,CH,), as a result of different fragmentation pathways followed by the corresponding oxonium intermediates 6,9 and 10. Consecutive SiF: addition/HF elimination processes predominate in H,O systems, with formation of a variety of protonated &h&c and fluorosilicic


31

fication of the two competing processes (reaction (6)), whose individual rate constants are reported in Table 1:

Si(OH)a+ (6b)

- . 0

100

200

Time

300

400

500

(ms)

zjig.8.Time dependence of ion abundances following ionization of a 5 : 3: 1 mixture of SiF,: H,O: CH,OH (P,,,, = 2 x 10m6 ‘Torr) and isolation of Si(OH)t after 1OOmsdelay time: (a) 4; (b) IS; (c) 11, (d) 12 + 13 (see text); (e) 16; (f) 14.

Ion 5 (m/z 97), in turn, undergoes a reversible CH30H/H20 displacement process to Si(OH)J CH30Hf (12) (m/z 111) yielding the equilibrium 5112 = 1.32 concentration ratio shown in Fig. 9, ~~esponding to an eq~lib~~ constant for reaction (7) of 2.75, at 300K: H I

acids. The search for their methyla~ and ethylated esters in the CH,OH and CH,CH20H samples, respectively, is frustrated by the rapid occurrence of unimolecular fragmentation in the corresponding oxonium ions 9 and 10, yielding inter alia very unstable carbenium ions, i.e. CH: and CH,CH$ , which supports the role of SiF: as a Lewis super,acid in the gas phase. Better issue of this search was obtained by adding to the SiF4 : ROH (R = CH,) mixture an appropriate amount of H,O. In fact, when a 5 : 3 : 1 mixture of SiF;, : H,O : CH,OH (P,,, = 2 x low6 Torr) is admitted through independent inlets into the FTMS cell, a complex reaction pattern takes place, dominated essentially by the extremely efficient first three steps of sequence (l), accompanied by the first competing steps of Scheme 1. After a reaction time of looms, by far the most abundant ionic species is 4 which is the precursor of tonic species at m/z 93, 97, 111, 121, and 125, observed at longer reaction times. When, in fact, 4 is isolated by multiple resonance procedures and allowed to react with the present neutrals, ions at m/z 93, 97, 111, 121 and 125 are actually observed whose stationary concentrations as a function of the reaction time are given in Fig. 8. Analysis of the curves of Fig. 8 allows identi-

Si(O~)~OH~ + CH,OH =GH,O + Si(OH)~OC~~ (12)

(5)

(7) The Keq = 2.75 value for the equilibrium in reaction (7) corres~nds to a CH,OH vs. H,O binding energy difference for 4 of AD” [22]. This (Si(OH): . - - ROH) w 0.6 kcal mol-’ relatively small binding energy difference reflects

loo-

or..,..,..,..,..

600 800 1 >o Time (ms) Fig. 9. Time dependence of ion abundances following ionization of a 5 : 3 : 1 mixture of SiF, : H,O : CH,OH (Pm, = 2 x 10v6Torr) and isolation of Si(OH), OH: after 500 ms delay time: m/z = 97 (5); m/z = 111 (12). 0

200

400

32


only a very limited positive inductive effect of the methyl group in stabilizing 12 relative to 5. In covalently bonded 0-protonated methyl ethers, in fact, the fully developed positive inductive effect of the methyl group stabilizes them by 8-10 kcal mall’ relative to the corresponding 0-protonated alcohols [8]. This effect is reflected in a CH30H vs. H,O binding energy gap for alkyl cations of the order of 11-l 6 kcal mol-’ . A plausible reason for such a different methyl group effect can be found in the structure of 12 (and 5) resembling an ROH molecule loosely bound to the Si(OH): moiety. Accordingly, the positive charge in 12 (and 5) is expected to be essentially located on the Si atom and, consequently, the better electron-releasing properties of the methyl group have little effect in stabilizing 12 relative to 5. Ion 11 rapidly adds H,O, yielding an ion at m/z 111, and slowly CH,OH to give a daughter species at m/z 125 (Fig. 8). The observation that the 5/12 equilibrium distribution of Fig. 9 is not reproduced in the m/z 97/m/z 111 yield ratio of Fig. 8, measured at the same reaction time (500ms), is indicative of different connectivities between the m/z 111 ion arising from the equilibrium in reaction (7), i.e. 12, and that generated by addition of Hz0 to 11, i.e. 13. Besides, the lack of significant interconversion between 12 and 13 indicates involvement of a substantial activation barrier. By extension, the m/z 125 species is suggested to have the same connectivity as in 14.

_.

si+G (12)

_. OMe

HO-

Si+ cc .OMe

OH

MeO’

_. OMe Sl+ - _ -0Me

-

In conclusion, the picture arises of SiF: as a gaseous Lewis superacid, displaying an exceedingly high affinity for oxygenated bases, such as ROH (R =H, CH,, or CH, CH,OH). In the alcohols, SiF: addition/R+ elimination sequences represent a major reaction channel which predominates over the conceivable SiF: addition/HF elimination path. In H,O, where the first process is thermochemically unfavored, the only observed sequence is the latter one, eventually yielding protonated silicic acids. Their protonated esters, which cannot be produced by direct attack of SiF: on the corresponding alcohols, are accessible in the gas phase by multiple ROH/H,O displacement reactions in species 2-5. SiFt ion attack on nitrogen bases

The behavior of SiF: toward NH, is essentially the same observed in the H,O samples. A sequence (8), similar to sequence (l), takes place in pure NH,, which is accompanied by comparatively slow proton loss of the intermediates species 17 and 18 to NH,:

i

ii+---’ OH -OH

ii+--*’ OH -OH

I&O’

(14)

(13)

As previously shown for MeOH/H, 0 ion microclusters [23], uptake of an MeOH molecule by 11, is expected to induce a reduction of the cluster size by preferential loss of water. Accordingly, formation of Si(OMe): (16) (m/z 121) from 11 is regarded as a sequential addition of MeOH with loss of water through the Si(OH)(OMe): inter-

in the

(16)

I

OH

m

US)

I

I I

HO’

mediate (15), which is barely detectable FTMS cell.

SiF: $$

SiF,NH$

(1)

s

(17)

Si(NH,):+NH” (19)

SiF(NH,):

s

(18)

Si(NH,),NH: (20)

(8)

F. Grandinetti et aLlInt. J. Mass Spectrom. Ion Processes 124 (1993) 21-36

1000

Time (ms)

0

1000

2000

Time (ms) Fig. 10. Time dependence of ion abundances following ionization of a 5 : 3 : 1 mixture of SiF, : NH, : H, 0 = (P,,, = 1 x 10m6Torr) and isolation of SiF: (See Scheme 3): (a) 85; (b) 82; (c) 79; (d) 80; (e) 77; (f) NH:; (g) 83; 0 81; C.i)78; (1) 97; (m) 76; (n) 95; (0) 94; (p) 93. Figures in boldface refer to the m/z values of the ionic species.

The reaction pattern is further complicated by the presence of background HzO, which interferes with sequence (8) by inducing the formation of oxygenated species, whose relative abundance depends upon the H,O concentration. When, in fact, a 5 : 3 : 1 mixture of SiF, : NH3 : H,O (P,,, = 1 x 10e6Torr) is admitted through independent inlets into the FTMS cell, a very complex reaction pattern is observed with the stationary concentrations of the ionic intermediates involved depending upon the reaction time as des-

33

cribed in Fig. 10. Analysis of the curves of Fig. 10, coupled with individual multiple resonance experiments, points to the reaction network of Scheme 3 as responsible for the observed trends. In it, the first horizontal and vertical steps take place at rates approaching the collision limits (1.6-l .9 x 10m9cm3 molecule-’ s-l), whereas the efficiency of further addition/elimination processes in the ensuing intermediates progressively decreases. The reaction network of Scheme 3 provides a convenient access to entire families of novel ionic species whose features deserve further investigation. Among others, special interest is attached to the isobaric adducts (shown in squares), whose occurrence as distinct isomeric species is expected to be easily demonstrable by mass spectrometric procedures, provided that their interconversion is slow relative to detection time as for 12 e 13. The gas-phase attack of SiF: on CH3NH2 in a 6: 1 mixture of SiF,: C&NH2 (P,,, = 1 x 10e6 Torr) induces a complex reaction pattern resembling in part that of the CH, OH analog and in part that of the NH, homolog (Scheme 4). In fact, as for CH,OH, SiF: efficiently attacks CH,NH, yielding CH:, which rapidly abstracts a hydride ion from the parent CH3NH2 yielding CH3NH: which eventually loses a proton to CH,NH,. However, at variance with the oxonium intermediate 9, the excited adduct arising from the exothermic attack of SiF: on CH,NH, preferentially eliminates HF instead of SiF,H, thus yielding SiF,NHCH: and eventually SiF(NHCH,): . in this latter species, further CH,NH* addition/HF elimination to produce a conceivable Si(NHCH,): ion is not observed, since it is precluded by the faster proton transfer to CH,NH,. The same reason accounts for the failure to observe tetracoordinated adducts, analogous to those depicted in the square of Scheme 3. Conclusions The gaseous Lewis superacid SiF: displays a similar behavior toward oxygen and nitrogen bases which ranks it as a voracious electrophile, able to readily produce free methyl and ethyl cations from


34

H

H ‘0’

F.. +. &1-P F (8s)

P.... +. -OH fS

+H20 -HF

(13)

+H20 _HF

HE*>& -

OH

+%O -HF

Eb2ii-OH

-

I

I

+NHs -HF

+NH,

+NH,

NH,

pQF’

I

-HF

07)

qN.3

I

-HP

HaN.... +. H,N &I -

NH2

+H,o

(70

+

H2N

NH,

I

--ySit,

HaN 03)

Scheme 3.

SiF,+

+

+ CH,NH, - CHzNH

CH3NH2 -

Scheme 4.

I I

HO-~~sikoH HO

(79)

(91)

\ +NH, -HP

+ H,O

CH3NH3+


CH, OH, CH3 NH,, and CH, CH2 OH, respectively. When these channels are inhibited by the specific nature of the n-type base (Hz0 or NH,), an alternative pathway predominates involving elimination of HF. In this case, a variety of protonated forms of gaseous fluorosilicic and silicic acids, which can undergo further substitution by alcohols or amines yielding the corresponding esters and amides, are formed whose features could be investigated for the first time by classical mass spectrometric procedures. The mass spectrometric evidence is consistent with protonated forms of o&o-silicic acids or esters as loosely bound adducts between the corresponding protonated meta-silicic analogues and a water or alcohol molecule, wherein O/O proton transfer involve substantial activation energies. However, mass spectrometric and theorletical evidence excludes similar loosely bound istructures for protonated meta-silicic acids.

8

9 10 11

12

Acknowledgements

The authors express their gratitude to F. Cacace turd S. Fomarini for their interest in the present work. Acknowledgment is also due to the Ministero dell’Universit8 e della Ricerca Scientifica e Tecnologica (MURST) and the Consiglio Naziopale delle Ricerche (CNR: Progetto Finalizzato “Chimica Fine II”) of Italy for financial support.

13 14

15 16 17 18

References G.A. Olah, G.K.S. Prakash and J. Sommer, Superacids, Wiley, New York, 1985. For comprehensive reviews, see: G.A. Olah and P.v.R. Schleyer, (Eds.), Carbonium Ions, Wiley, New York, Vol. 1 (1968), Vol. 2 (1970), Vol. 3 (1972), Vol. 4 (1973), Vol. 5 (1976). J.M. Farrar, in J.M. Farrar and W.H. Saunders, Jr., (Eds), Techniques for the Study of Ion/Molecule Reactions, Wiley, New York, 1988, Chapter VII. See for example: J.D. Morrison, K. Stanney and J.M. Tedder, J. Chem. Sot., Perkin Trans. 2, (1981) 838. See also: F. Cacace, J. Chem. Sot., Perkin Trans. 2 (1982) 1129.

19 20

35

(a) F. Cacace, Act. Chem. Res., 21 (1988) 215. (b) M. Speranza, Mass Spectrom. Rev., 11 (1992) 73. M. Speranxa, Int. J. Mass Spectrom. Ion Processes, 118/ 119 (1992) 395. For instance, phenylium ion can be readily generated in the gas phase by attack of a radiogenic CH: (from CH,) on fluorobenzene. Its lifetime is, however, severely limited under these conditions by the fact that it may efficiently add CH, to yield eventually protonated toluene (cfr. M. Speranxa and F. Cacace, J. Am. Chem. Sot., 99 (1977) 3051; M. Speranza, M.D. Sefcik, J.M.S. Henis and P.P. Gaspar, J. Am. Chem. Sot., 99 (1977) 5583). S.G. Lias, J.E. Bartmess, J.F. Liebman, J.L. Holmes, R.D. Levin and W.G. Mallard, J. Phys. Chem. Ref. Data, 17 (suppl. 1) (1988). S.N. Senzer and F.W. Lampe, J. Appl. Phys., 54 (1983) 3524. W.D. Reents, Jr. and A.M. Mujsce, Int. J. Mass Spectrom. Ion Processess, 59 (1984) 65. M.J. Frisch, M. Head-Gordon, H.B. Schlegel, K. Raghavachari, J.S. Binkley, C. Gonzalez, D.J. DeFrees, D.J. Fox, R.A. Whiteside, R. Seeger, C.F. Melius, J. Baker, R. Martin, L.R. Kahn, J.J.P. Stewart, E.M. Flueder, S. Topiol and J.A. Pople, GAUSSIAN 88, Gaussian Inc., Pittsburgh, PA, 1988. (a) P.C. Hariharan and J.A. Pople, Chem. Phys. Lett., 66 (1972) 217. (b) R. Krishnan, M.J. Frisch and J.A. Pople, J. Chem. Phys., 72 (1980) 4244. H.B. Schlegel, J. Comput. Chem., 7 (1986) 359. J.A. Pople, H.B. Schlegel, R. Krishnan, D.J. DeFrees, J.S. Pinkley, M.J. Frisch, R.A. Whiteside, R.F. Hout and W.J. Hehre, Int. J. Quantum Chem. Symp., 15 (1981) 269. C. MolIer and M.S. Plesset, Phys. Rev., 46 (1934) 618. S. Murthy and J.L. Beauchamp, J. Phys. Chem., 96 (1992) 1247. R.G. Cooks and T.L. Kruger, J. Am. Chem. Sot., 99 (1977) 3980. As a matter of fact, the 182.0-196.1 kcal mol-’ uncertainty range of the proton affinity of ortho-silicic acid can be further reduced to 186.3-196.1 kcalmol-‘, if one can demonstrate that the neutral fragment from unimolecular dissociation of the adduct between 4 and acetone is allene (PA = 186.3 kcalmol-‘) and not propyne (182.0kcal mall’). Work is in progress to ascertain the nature of the C,H, neutral fragment from this dissociation. N.G. Adams and D. Smith, Chem. Phys. Lett., 54 (1978) 530. Boundaries for the heat of formation of SiF,OH can be established by the occurrence. of reaction (3a) when R = CH,, and by its non-occurrence when R = H. Accordingly, the heat of formation of SiF,OH is between - 333 and - 448 kcal mol-’ . By taking the PA of SiF, OH equal to 160 kcalmol-’ [lo], boundaries for the heat of formation of 6 can be set as - 127 and - 242 kcal mol-’ , respectively. Thus, the reaction exothermicity for the

36

21

F. Grandinetti et al./&. J. Mass Spectrom. ion Processes 124 (1993) 21-36 formation of 6 from attack of SiF;’ on H,O is calculated to range from 45-160 kcal mol-’ . Formation of 9 and 10 from attack of SiF$ on the more nucleop~lic ROH(R= CH, or CH,CH,) is expected to be even more exothermic. Simple heterolytic O-R (R=H) bond fission (reaction (3a)) in 6 is estimated to be about lOOkcalmol-’ more endothermic than the ~mol~ular loss of CH$ from 9. The homolytic O-R (R=H) bond cleavage in 6 appears unfavored as well relative to reaction (3b), as demonstrated by the absence of the SiF,OH+ molecular ion in the corre-

22

23

sponding Fourier transfo~-ion cyclotron resonance spectra. Actually, - 0.6 kcal mol-’ represents the free-energy change for the equilib~um reaction (7). This value should be close to the enthalpy change for the same reaction and, thus, the - AD’(Si(OH~~+ . . . ROH) = DO(Si(OH~)+. * . H,O) - D*(Si(OH,)+ . . MeOH) binding energy difference, since only a small change of entropy is expected for reaction (7). P. Kebarle, R.N. Haynes and J.G. Collins, J. Am. Chem. Sot., 89 (1967) 5753.