1
Microsolvation of the indole cation (In+) in a nonpolar environment: IR spectra of In+–Ln complexes (L = Ar and N2, n ≤ 8)

2
Infrared photodissociation (IRPD) spectra of complexes composed of the indole cation (In+ = C8H7N+) and several neutral ligands (L = Ar and N2) were recorded in the vicinity of the N–H stretch vibration (ν1) of bare In+ in its 2A″ electronic ground state.

3
The analysis of systematic size-dependent ν1 band shifts and photofragmentation branching ratios in the spectra of In+–Arn (n ≤ 5) and In+–(N2)n (n ≤ 8) provides information about the stepwise microsolvation of In+ in a nonpolar hydrophobic environment, including the existence of structural isomers and the determination of ligand binding energies.

4
The IR spectra of the In+–L dimers reveal two transitions, which are attributed to ν1 fundamentals of the H-bound and π-bound isomers on the basis of their complexation-induced ν1 frequency shifts, Δν1.

5
In both cases, the H-bound isomer is found to be more stable than possible π-bound isomers.

6
The Δν1 shifts are used to derive the first experimental estimate for the proton affinity of the indolyl radical (∼920 ± 30 kJ mol−1).

7
The IR spectra of In+–Arn (n ≤ 5) suggest that the preferred microsolvation path for this cluster system begins with the formation of the H-bound In+–Ar dimer core, which is further solvated by (n − 1) π-bound ligands.

8
In contrast, the spectra of In+–(N2)n with n ≤ 8 suggest that this cluster grows by the formation of an In+–(N2)2 trimer core with two H-bound N2 ligands, to which (n − 2) π-bound N2 molecules are attached.

9
The In+–Ln complexes were generated in an electron impact (EI) ion source, which predominantly produces the most stable isomer of each cluster ion.

10
For several In+–Ln complexes, the geometry of the most stable isomer produced in this ion source differs drastically from the structures previously observed by resonant photoionization of the corresponding neutral precursors, demonstrating the severe restriction of photoionization techniques (given by the Franck–Condon principle) for the spectroscopic characterization of cluster ions.

11
Most of the In+–Ln complexes investigated exhibit a distinct ionization-induced change in the preferred substrate–ligand recognition pattern.

Introduction

12
Many biophysical and biochemical phenomena, including molecular recognition, protein folding, and biological activity, strongly depend on the environment.1,2

13
Solvation effects are particularly important for charged species, because of the larger strength and longer range of ion–solvent interactions compared to corresponding forces between neutral molecules.3–7

14
Biological molecules are often (locally) charged due to either (de-)protonation or charge separation.

15
Isolated complexes of the type X±q–Ln are simple and attractive model systems to investigate the effects of stepwise solvation of an ion with charge ±q (X±q) by a well-defined number (n) of ligands (L) with both experimental and theoretical techniques.

16
The present work reports infrared (IR) spectra and quantum chemical calculations of complexes of the indole cation (In+ = C8H7N+) solvated by several inert ligands (L = Ar and N2) to characterize the interaction of In+ with a hydrophobic environment.

17
Indole is the aromatic chromophore of the amino acid trypthophan, which plays an important role in solvation-dependent photophysical processes of proteins.

18
Hence, size-selected In+–Ln clusters serve as model systems to investigate the interaction between tryptophan-containing biomolecules and surrounding solvent molecules under controlled solvation conditions.

19
In general, aromatic molecules (X = A) with acidic functional groups (YHk) offer two principal recognition sites for a solvent molecule.

20
The ligand L can either interact with the aromatic π-electron system (π-bond) or form a hydrogen bond to one of the protons of the YHk group (H-bond).

21
Other binding sites are usually less stable.

22
The preferred recognition site depends strongly on the electronic excitation and charge of A, the acidity of the YHk group, and the type of ligand (e.g., nonpolar or polar).

23
For example, complexes of neutral aromatic molecules with the spherical rare gas atoms (e.g., A–Ar) generally feature π-bound equilibrium structures in the singlet electronic ground state (S0).

24
Illustrative examples inlude Ar complexes of phenol (Ph–Ar, YHk = OH)8,9 and aniline (An–Ar, YHk = NH2).10

25
The dominant contribution to the attraction in neutral A–Ar complexes is the dispersion interaction between Ar and the aromatic π-electron system of A, which is most favorable for π-bonding.

26
So far, no H-bound isomer has been detected for any aromatic A–Ar complex in the S0 state.

27
The situation changes for complexes with molecular nitrogen, A–N2.

28
The quadrupole moment of N2 leads to additional electrostatic interactions with the polar YHk group of A. Consequently, the H-bound A–N2 isomer often competes energetically with the π-bound one.

29
For example, the intermolecular potential of Ph–N2 has a H-bound global minimum,8 whereas the less acidic An molecule prefers π-bonding with N2.11

30
The interaction in A+–Ln cation clusters is usually rather different from that in the corresponding neutral A–Ln complexes, because of the substantial additional electrostatic and inductive attraction arising from the positive charge distribution in A+.3,4,6

31
Hence, A+–Ln and A–Ln complexes often have rather different equilibrium structures and binding energies.

32
For example, the Ph+–Ar12–15 and An+–Ar16 cation dimers have H-bound global minima in the ground electronic state (D0), whereas the π-bound structures, which are global minima in S0, are only local minima on the intermolecular potential in D0.

33
A similar situation is observed for Ph+–N28,12,13,17 and An+–N2 dimers.18

34
In general, the interaction strength in A+–N2 is larger than that in the corresponding A+–Ar dimer, mainly because of the additional charge–quadrupole interaction in the former complexes.

35
The substantial difference in the topolopgy of the interaction potentials in neutral A–Ln and ionic A+–Ln clusters, with respect to both their equilibrium structure and binding energy, leads to important consequences for spectroscopic studies of the cation clusters prepared by photoionization of the corresponding neutral precursor.

36
Frequently, A+–Ln complexes are generated by the formation of neutral A–Ln complexes in a supersonic molecular beam expansion, followed by resonance enhanced multiphoton ionization (REMPI).

37
For the spectroscopic characterization of A+–Ln, the REMPI process is usually combined with photoionization efficiency measurements (PIE), mass analysed threshold ionization (MATI), zero kinetic energy photoelectron (ZEKE) spectroscopy, or photodissociation (PD) spectroscopy.

38
However, all these ionization techniques suffer from the severe restrictions imposed by the Franck–Condon principle, which prevents the efficient population of the most stable isomer of A+–Ln in cases where the global minimum structures of the neutral and the cation complex are rather different.

39
This scenario is quite common for complexes of (substituted) aromatic molecules with polar and nonpolar molecules.19

40
To overcome the limitations of the REMPI technique for cluster ion generation (“REMPI ion source”), the present work employs electron impact (EI) ionization within a supersonic expansion for the production of cold A+–Ln complexes.7

41
This EI cluster ion source was shown to produce predominantly the most stable isomer of a given A+–Ln cluster ion, because the reaction sequence begins with EI ionization of A (to yield A+), which is followed by three-body cluster aggregation reactions.12–16,18,20,21

42
Hence, the EI source forms predominantly the most stable A+–Ln isomer, independent of the most stable geometry of the corresponding neutral A–Ln species.

43
This is in contrast to the REMPI source, which often generates only local minima of A+–Ln.

44
For example, EI-IR spectra of Ph+–Arn12–15 and An+–Arn16 were interpreted with global minimum structures of these complexes, which escaped detection in the corresponding REMPI-IR spectra as well as other photoionization spectra (PIE, ZEKE, MATI).

45
Similar to most aromatic molecules with an acidic YHk group, the planar indole molecule offers at least two principal binding sites for neutral ligands, namely H-bonding to the N–H group or π-bonding to the aromatic π-electron system.

46
This recognition pattern yields two major isomers for dimers, namely the H-bound and π-bound In(+)–L structures shown in Fig. 1(a) and (b), respectively.

47
(Although there are possibly more than one stable π-bound isomer on the potential of In(+)–L, they are not distinguished in the present work.) Accordingly, three combinations are possible for the In(+)–L2 trimer, one with two H-bonds (Fig. 1(e), 2H isomer), one with two π-bonds (Fig. 1(d), 2π), and one with one H-bond and one π-bond (Fig. 1(c), H/π).

48
In the present work, the notation xH/yπ is used to describe an In(+)–Ln isomer with x H-bound and y π-bound ligands (x + y = n).

49
Spectroscopic information about the structure of neutral In–Arn clusters (with n ≤ 2) in the two lowest singlet electronic states, S0 and S1, is available from laser-induced fluorescence excitation (LIF),22,23 dispersed LIF,24 and REMPI spectroscopy.24,25

50
All available spectral information is consistent with the observation of a π-bound In–Ar dimer (Fig. 1(b)) and a 2π isomer of In–Ar2 (Fig. 1(d)) in both electronic states.

51
Such structures were also predicted by simple Lennard-Jones pairwise additive atom–atom potentials.24

52
No evidence for an H-bound In–Ar isomer has been reported.

53
The analysis of the rotationally resolved spectrum of the S1 ← S0 transition of π-bound In–Ar yields a structure, in which the Ar ligand is approximately located R ≈ 3.4 Å above or below the indole plane.23

54
The attraction between Ar and the lone pair of the N atom results in some preference for Ar toward the five-membered ring of In.

55
The interaction in the S1 state (R = 3.40 Å) is slightly stronger than in S0 (R = 3.43 Å),23 with experimental dissociation energies of D0 = 477 ± 10 and 451 ± 10 cm−1, respectively.26

56
The larger interaction in the excited state arises from an increase in the polarisability of the aromatic π-electron system upon electronic excitation, which enhances the dispersion attraction.

57
The S1 ← S0 transitions of In–Arn with n = 0–2 display nearly additive red shifts upon Ar complexation, ΔS1 = −26 cm−1 (n = 1) and −53 cm−1 (n = 2),22 indicating that both Ar ligands in the 2π isomer of In–Ar2 form almost equivalent π-bonds on opposite sides of the molecular plane, with little interaction between both Ar ligands.

58
Spectroscopic information about the interaction of In+–Arn (with n ≤ 2) in the cation ground state (D0) is provided by PIE,27 MATI,26 and ZEKE25 spectroscopic studies.

59
Similar to the corresponding neutral complexes, all spectral features in the cation spectra could be rationalized by a π-bound In+–Ar isomer and a 2π isomer of In+–Ar2.

60
No evidence for the formation of an intermolecular H-bond in In+–Arn clusters has been reported.

61
This observation is not surprising because all cation spectra recorded so far were obtained by REMPI of the neutral clusters featuring only intermolecular π-bonds.

62
In general, the interaction in the cation ground state of π-bound In+–Ar (D0 = 537 ± 10 cm−1)26 is significantly stronger than in the neutral electronic states because of the additional charge–induced dipole attraction.

63
In addition, the additive red shifts in the ionization potentials of In–Arn with n = 0–2 indicate that both π-bonds in the 2π isomer of In+–Ar2 are nearly equivalent.27

64
Little activity in the intermolecular bending (νb) and stretching vibrations (νs) observed in the dispersed LIF, LIF excitation, ZEKE, and MATI spectra of π-bound In–Arn (n = 1, 2) suggests only small changes in the geometry of the intermolecular π-bond(s) along both the bending and stretching coordinates upon electronic excitation and subsequent ionization.

65
In contrast to In–Arn, neither spectroscopic nor theoretical information is available for the neutral and ionic states of In(+)–(N2)n complexes.

66
The major goal of the present work is the characterization of the intermolecular interaction in In+–(N2)n and In+–Arn complexes in their cation ground states by EI-IR spectroscopy in the vicinity of the N–H stretch fundamental of bare In+ (ν1 = 3454 cm−1).28

67
Of particular interest is the competition between H-bonding and π-bonding in the In+–L dimers and its consequences on the microsolvation process (e.g., the geometry of the most stable isomer of larger In+–Ln clusters).

68
Comparison of the EI-IR spectra obtained for In+–Ar1,2 with previous photoionization data demonstrates the deficiencies of REMPI techniques for the production of the most stable isomers of this type of cluster ions.

69
In addition to the EI-IR spectra, the intermolecular interaction in the H-bound isomers of In+–(N2)n (n ≤ 2) and In+–Ar is characterized by density functional calculations, as no previous theoretical studies of the cation PES of these clusters are available.

70
Comparison of In+–(N2)n and In+–Arn with corresponding complexes of Ph+ and An+ reveals the effect of the acidic YHk group on the recognition pattern of aromatic ions with hydrophobic solvent molecules, including the preferred sequence for cluster growth and the determination of ligand binding energies.

Experimental

71
IRPD spectra of mass-selected In+–Ln complexes were recorded in a tandem quadrupole mass spectrometer coupled to an electron impact cluster ion source and an octopole ion trap.7

72
Briefly, In+–Ln cluster ions were produced in an ion source which combines a pulsed supersonic expansion with electron impact ionization close to the nozzle orifice.

73
The expanding gas mixture was generated by seeding indole vapor (heated up to T ≈ 340 K) in Ar and/or N2 carrier gas at backing pressures ranging from 5 to 10 bar.

74
As outlined in detail in ref. 13, the dominant production mechanism of In+–Ln begins with electron ionization of In, which is followed by three-body association reactions: In + e → In+ + 2eIn+–Ln−1 + L + M → In+–Ln + M (M = L, In) This reaction sequence produces predominantly the most stable isomers of In+–Ln clusters and to smaller extent less stable isomers.13

75
A typical mass spectrum of the ion source is shown in Fig. 2 for the generation of In+–N2.

76
The spectrum is dominated by Nn+ and In+.

77
Only little fragmentation of In+ is observed.

78
Major cluster series include [X − (N2)n]+ with X = N2, N3, In, Ar, and the impurity H2O. The intensity ratio of In+ and In+–N2 is of the order of 200∶1, indicating the formation of a weakly-bound noncovalent complex.

79
The central part of the supersonic plasma expansion was extracted into an initial quadrupole mass spectrometer, which was tuned to the mass of In+–Ln.

80
The mass selected In+–Ln beam was then deflected by 90° and injected into an octopole ion guide, where it interacted with a tunable IR laser pulse.

81
Resonant vibrational excitation of In+–Ln induced the evaporation of weakly bound ligands: In+–Ln + hν → (In+–Ln)* → In+–Lm + (n − m)L Only the rupture of the weak intermolecular bonds was observed upon IR excitation.

82
The In+–Lm fragment ions were mass selected with a second quadrupole mass filter and detected with a Daly ion detector.29

83
The IRPD spectrum of In+–Ln was obtained by monitoring the In+–Lm fragment current as a function of the laser frequency.

84
For In+–Ln clusters with n > 1, several fragment channels m were possible, and in these cases IRPD spectra were recorded simultaneously in the two dominant channels.

85
As an example, Fig. 3 shows the mass spectrum obtained for resonant excitation of In+–(N2)6 at ν1 = 3413 cm−1.

86
For this cluster, the major fragment channels were m = 1 (30%) and m = 2 (65%).

87
In general, the IRPD spectra recorded in different fragment channels were rather similar.

88
Consequently, only the spectra obtained in the dominant channel are shown in the figures.

89
Tunable IR laser radiation with a bandwidth of 0.02 cm−1 was generated by a single-mode optical parametric oscillator laser system pumped by a seeded Nd:YAG laser.

90
The IRPD spectra were calibrated to an accuracy of better than 1 cm−1 by simultaneously recording optoacoustic spectra of ammonia and analysing water absorptions along the IR laser path.30,31

91
All IRPD spectra were normalized for laser intensity variations recorded with an InSb detector.

Quantum chemical calculations

92
Density functional calculations were carried out for In+, In+–Ar, In+–N2, and In+–(N2)2 at the unrestricted B3LYP/6-31G* level of theory32 to investigate the intermolecular interaction between In+ and H-bound ligands as well as the effects of the intermolecular H-bond(s) on the properties of In+.

93
Previous calculations for the related complexes involving An+ and Ph+ revealed that this theoretical level is sufficient to describe the properties of the H-bonds of these ions to Ar and N2 with reasonable accuracy (in particular the frequency shifts of the proton donor stretch vibrations) and to facilitate vibrational assignments.12,15,16,18

94
In general, all coordinates were relaxed during the search for stationary points.

95
Intermolecular interaction energies were counterpoise corrected for basis set superposition error.33

96
Harmonic frequencies were scaled by a factor of 0.9589 to bring the calculated frequency of the N–H stretch vibration of In+ into agreement with the corresponding experimental value, ν1 = 3454 cm−1.28

97
Table 1 summarizes the results of the calculations of In+–Ln relevant for the present work.

98
The table includes structural and energetic properties of the intermolecular H–L bond, such as length (RH–L), angle (θN–H–L), and dissociation energy (De), as well as fundamental attributes of the intramolecular N–H bond, namely length (rN–H) and stretching frequency (ν1).

99
The proton-bound In+–L minima have nearly linear H-bonds of L to the acidic N–H proton donor group of In+, leading to planar equilibrium geometries with Cs symmetry (Fig. 1(a)).

100
The calculated intermolecular bond angles and separations are 170° and 2.64 Å for Ar and 177° and 2.16 Å for N2, respectively.

101
As expected, the interaction in In+–Ar is significantly weaker than in In+–N2 (De = 291 and 1201 cm−1).

102
The effects of H-bonding on the intramolecular N–H bond of In+ are an elongation (ΔrN–H), a reduction in the stretching frequency (Δν1), and an enhancement in the IR oscillator strength (ΔI1).

103
As expected, the magnitude of these effects is correlated with the strength of the H-bond (ΔrN–H = 0.0011 and 0.0045 Å, Δν1 = −19 and −68 cm−1, and ΔI1 = 100 and 290% for Ar and N2).

104
The experimental IR spectra discussed in section IV suggest that In+–(N2)2 has a structure, in which both N2 ligands are H-bonded to the N–H group of In+.

105
All efforts in searching for such a 2H isomer of In+–(N2)2 resulted in a planar equilibrium geometry with two nonequivalent N2 ligands (Fig. 1(e)).

106
The H-bond to the first N2 ligand, which is closer to the six-membered ring of In+, is characterized by RH–N = 2.29 Å, θN–H–N = 155°, and De = 805 cm−1.

107
The interaction with the second ligand is considerably weaker (RH–N = 2.70 Å, θN–H–N = 122°, De = 549 cm−1).

108
The effects of both ligands on the N–H bond of In+rN–H = 0.0033 Å, Δν1 = − 39 cm−1, ΔI1 = 180%) are smaller than those of the single ligand in H-bound In+–N2.

109
Thus, because of noncooperative interactions, the addition of the second H-bound ligand destabilizes the intermolecular bond to the first ligand, which in turn leads to a stabilization of the intramolecular N–H bond.

110
Complexes of In+ with π-bound ligands were not investigated theoretically, because density functional calulations do not properly describe interactions, which are dominated by dispersion forces.

111
However, as π-bound ligands in In+–Ln have only little effects on the properties of the N–H bond, the changes of the ν1 frequency and IR intensity of In+ upon formation of π-bonds are assumed to be small12,15,18.

Experimental results and discussion

112
The IRPD spectra of In+–Arn (n ≤ 5) and In+–(N2)n (n ≤ 8) are shown and analysed in Figs. 4–8, and the band maxima and widths of the transitions observed are listed in Table 2, along with their assignments.

In+–L dimers

113
Fig. 4 compares the IRPD spectra of In+–L (L = Ar, N2) in the vicinity of the ν1 vibration of bare In+ (ν1 = 3454 cm−1).28

114
Two transitions are observed in each spectrum in the frequency range investigated (3250–3600 cm−1), and they are attributed to the ν1 fundamentals of (at least) two different isomers, denoted H and π.

115
The more intense bands at 3425 (Ar) and 3379 (N2) cm−1 are assigned to the H-bound isomers (Fig. 1(a)).

116
They feature significant red shifts of Δν1 = −29 and −75 cm−1 upon complexation, which compare favorably with the calculated values of −19 and −68 cm−1, respectively.

117
The shift is larger for H-bound In+–N2 than for In+–Ar, consistent with the stronger interaction in the former dimer.

118
The weaker bands at 3451 (Ar) and 3450 (N2) cm−1 are close to the In+ transition (3454 cm−1)28 and display only modest red shifts of Δν1 = −3 and −4 cm−1.

119
This observation suggests an assignment of these bands to the ν1 fundamentals of the π-bound isomers of In+–L (Fig. 1(b)), because in these structures the ligands have only little influence on the N–H bond of In+ with respect to both frequency and IR intensity.

120
There are actually other binding sites which also have little impact on the N–H bond, such as H-bonding to the aromatic C–H groups.

121
Although at present these alternative binding sites cannot be completely ruled out both from the experimental and theoretical point of view, π-bound isomers are presently the favored assignment for the transitions near 3450 cm−1 and this interpretation is used as working hypothesis for the rest of this paper (see also section V).

122
In addition to the band shifts, the observed ν1 band profiles support the isomer assignment in Fig. 4 and Table 2.

123
The ν1 bands of the H-bound dimers are significantly shaded to the blue, with sharp heads in the P branch and slowly decreasing signal to higher frequency of the band maximum.

124
Such band profiles are typical for the excitation of proton donor stretch vibrations in H-bound dimers,4,12,13,34–37 because the intermolecular bond becomes stronger and shorter in the vibrationally excited state leading to larger rotational constants.

125
In the limit of adiabatic separation of inter- and intramolecular degrees of freedom, the red shifts Δν1 correspond directly to the increase of the interaction energy upon ν1 excitation.

126
In contrast to the H-bound dimers, the π-bound isomers feature rather symmetric ν1 band profiles, because ν1 excitation has nearly no influence on the strength of the intermolecular π-bond, consistent with the small complexation shifts Δν1.

127
The ratios of the integrated ν1 band intensities observed in the IRPD spectra of the In+–L dimers can be used to estimate the relative abundance of the H-bound and π-bound isomers in the expansion, assuming the calculated relative ν1 IR oscillator strengths for the H-bound In+–L dimers listed in Table 1.

128
The π-bound In+–L isomers are assumed to have the same ν1 IR oscillator strength as In+.12,16,18

129
This procedure results for In+–Ar (In+–N2) in a relative abundance of nHnπ ≈ 2∶1 (7∶1), on the basis of the experimental intensity ratio of 4∶1 (28∶1) and the theoretical ratio of the oscillator strength of 2∶1 (4∶1).

130
The larger abundance of the H-bound In+–L dimers suggests that they are more stable than the corresponding π-bound isomers, because the electron impact ion source produces predominantly the most stable isomer of a given cluster.

131
Moreover, the energy difference between both isomers appears to be larger for In+–N2 than for In+–Ar leading to a more efficient production of the π-bound isomer in the latter case.

132
Similar to the previous studies of Ph+–Ar,12,14 the relative intensity ratio of the ν1 bands of H-bound and π-bound In+–Ar depended considerably upon the expansion conditions, confirming that both transitions arise from different isomers.

133
The spectrum in Fig. 4 was obtained when the conditions were optimized for the production of π-bound In+–Ar.

134
Moreover, the relative concentration of the π-bound isomer decreased for lower effective temperatures of the ion source.

135
This observation is taken as clear evidence that the H-bound isomer of In+–Ar corresponds to the global minimum of the potential, whereas the π-bound structure is a less stable local minimium.

136
Similar conclusions apply also to the In+–N2 dimer.

137
The dissociation energy of π-bound In+–Ar in the ground vibrational state was determined as D0 = 537 ± 10 cm−1.26

138
Hence, the shift of Δν1 = −3 cm−1 implies a binding energy of D0 = 540 ± 10 cm−1 (+0.6%) in the ν1 excited state.

139
As the interaction in H-bound In+–Ar should be somewhat stronger than in π-bound In+–Ar (because of the larger abundance in the EI ion source), the dissociation energy calculated at the B3LYP/6-31G* level, De = 291 cm−1, substantially underestimates the true interaction energy.

140
Comparison with the related Ph+–Ar dimer yields an estimated binding energy of D0 ≈ 670 ± 140 cm−1 in the ground state (see section IV.B).

141
On the basis of this value, the interaction increases by +4.5% upon ν1 excitation (Δν1 = −29 cm−1), resulting in D0 ≈ 700 ± 140 cm−1.

142
No experimental information is available for the binding energy of H-bound and π-bound In+–N2.

143
For H-bound Ph+–N2, the dissociation energy calculated at the B3LYP/6-31G* level, De = 1787 cm−1,12 is close to the experimental binding energy, D0 = 1640 ± 10 cm−1.17

144
Assuming similar accuracy for the corresponding H-bound In+–N2 dimer, the theoretical value, De = 1201 cm−1 (Table 1), leads to an estimated dissociation energy of D0 ≈ 1100 cm−1.

145
On the basis of this value, the observed red shift of Δν1 = −75 cm−1 corresponds to an increase of the interaction strength of ≈7% upon ν1 excitation (D0 ≈ 1175 cm−1).

146
The dissociation energy of π-bound In+–N2 may be estimated as D0 ≈ 700 cm−1 from those of the corresponding π-bound isomers of Ph+–N2 (D0 = 750 ± 150 cm−1)13 and An+–N2 (D0 = 700 ± 200 cm−1).18,38

147
This value is also in line with the analysis of the photofragmentation data of larger complexes outlined in section IV.B.

Larger In+–Ln clusters

148
Fig. 5 shows the IRPD spectra of In+–Arn (n ≤ 5) recorded between 3380 and 3480 cm−1.

149
All spectra were obtained in the In+ fragment channel, indicated as n → 0.

150
The spectrum of In+–Ar displays two bands at 3425 and 3451 cm−1, which are assigned to the ν1 fundamentals of the H-bound and π-bound isomers (section IV.A).

151
Similar to In+–Ar, the spectrum of the In+–Ar2 trimer reveals also two transitions (at 3424 and 3450 cm−1), which are barely shifted upon addition of the second Ar ligand.

152
Consequently, they are assigned to ν1 bands of In+–Ar2 isomers, which are obtained by adding a π-bound ligand to the H-bound and π-bound In+–Ar dimer isomers, respectively.

153
The intense band at 3424 cm−1 corresponds to the H/π isomer (Fig. 1(c)), whereas the weak band at 3450 cm−1 arises from the 2π isomer (Fig. 1(d)).

154
The relative intensity ratio of the ν1 bands of the 2π and H/π isomers is of the order of 1∶20 and is significantly lower than the corresponding ratio of the π-bound and H-bound dimers (1∶4).

155
This observation confirms that the intermolecular H-bond between Ar and In+ is more stable than the π-bond.

156
The spectra of In+–Arn with n = 3–5 are dominated by a single transition, which is assigned to ν1 of an isomer with one H-bound and (n − 1) π-bound ligands, denoted H/(n − 1)π.

157
No evidence for an isomer with only π-bound ligands is observed for n ≥ 3, again demonstrating that the H-bond is more stable than the π-bond.

158
Hence, the preferred solvation sequence in In+–Arn complexes begins with the formation of a H-bound In+–Ar dimer core, which is further solvated by (n − 1) π-bound ligands.

159
In general, the widths of the ν1 bands of In+–Arn decreases with increasing cluster size from 14 (n = 1) to 5 cm−1 (n = 5).

160
This effect is attributed to (i) lower internal energy, (ii) smaller rotational constants, and (iii) smaller effects of ν1 excitation on the intermolecular modes of the larger clusters (i.e. sequence bands of ν1 involving intermolecular modes better overlap with the ν1 fundamental).

161
Fig. 6 shows the dependence of the ν1 frequencies of In+–Arn as a function of the cluster size n.

162
The plot mirrors the growth of the solvation (sub)shells of the different cluster isomers, nπ and H/(n − 1)π.

163
In the case of the nπ isomers of In+–Arn (observed up to n = 2), the incremental red shifts are of the order of −2 ± 1 cm−1.

164
In the more stable H/(n − 1)π cluster series (observed up to n = 5), the large ν1 red shift of the H-bound ligand (−29 cm−1, n = 1) is followed by a small red shift for the first π-bound ligand (−1 cm−1, n = 2) and a small blue shift for the second π-bound ligand (+2 cm−1, n = 3).

165
No change in the ν1 frequency is observed for clusters in the size range n = 3–5 within experimental resolution.

166
In general, the incremental complexation-induced shifts of π-bound ligands (|Δν1| ≤ 3 cm−1) are significantly smaller than the widths of the observed transitions (fwhm = 5–14 cm−1), making it difficult to extract more definitive information on the structure of the complexes.

167
Fig. 7 compares the IRPD spectra of In+–(N2)n with n ≤ 8 recorded in the dominant In+–(N2)m fragment channel, indicated as n → m.

168
Similar to In+–Ar, the spectrum of In+–N2 displays two bands at 3379 and 3450 cm−1, which are assigned to the ν1 fundamentals of the H-bound and π-bound isomers, respectively (section IV.A).

169
In contrast to In+–Ar2 and In+–N2, however, the spectrum of the In+–(N2)2 trimer reveals only a single transition at 3404 cm−1.

170
As this transition is significantly red shifted from ν1 of π-bound In+–N2ν1 = −46 cm−1), it is not attributed to the 2π isomer of In+–(N2)2 (Fig. 1(d)).

171
On the other hand, the band also displays a large blue shift from ν1 of H-bound In+–N2ν1 = 25 cm−1), indicating that the second N2 ligand has a significant stabilizing effect on the N–H bond.

172
Consequently, the second ligand can not be π-bonded to H-bound In+–N2, excluding an assignment of the 3404 cm−1 transition to the H/π isomer of In+–(N2)2 (Fig. 1(c)).

173
This observation initiated our theoretical search for In+–(N2)2 trimer structures, in which both N2 ligands interact with the acidic proton of In+.

174
All efforts in locating such minimum structures, using the B3LYP calculations described in section III, converged toward the planar In+–(N2)2 geometry with two nonequivalent H-bonds shown in Fig. 1(e) (2H isomer, Table 1).

175
Interestingly, the ν1 frequency of this In+–(N2)2 isomer is predicted as 3414 cm−1, in nice agreement with the observed value of 3404 cm−1.

176
In addition, the calculated shifts from ν1 of H-bound In+–N2 and bare In+, Δν1 = +28 and −40 cm−1, are compatible with the corresponding experimental values, Δν1 = +25 and −50 cm−1.

177
Hence, the 3404 cm−1 band is attributed to the 2H isomer of In+–(N2)2.

178
The absence of any pronounced absorptions of In+–(N2)2 near the ν1 bands of H-bound and π-bound In+–N2 suggests that the 2π and H/π isomers of In+–(N2)2 are significantly less stable than the 2H isomer.

179
The low concentration of the 2π isomer in the expansion confirms that H-bound In+–N2 is more stable than π-bound In+–N2.

180
The ν1 blue shift of In+–(N2)2 with respect to H-bound In+–N2 indicates that ν1 excitation reduces the binding energy of the second ligand by 25 cm−1.

181
Moreover, addition of the second ligand strengthens the N–H bond and weakens the first H-bond in H-bound In+–(N2)2, leading to the observed increase in the ν1 frequency.

182
This trend is in line with the results of the B3LYP calculations (Table 1).

183
The fact that the ν1 band of the 2H isomer is shaded to the red indicates that the rotational constants in the ν1 state of In+–(N2)2 are smaller than in the ground state.

184
The spectra of In+–(N2)n with n = 3–8 are dominated by a single band, which is close to the ν1 band of the 2H isomer of In+–(N2)2.

185
Consequently, this transition is assigned to the ν1 fundamental of In+–(N2)n structures, in which the 2H isomer of In+–(N2)2 is further solvated by (n − 2) π-bound ligands, denoted 2H/(n − 2)π.

186
Similar to In+–Arn, the widths of the ν1 transitions tend to decrease as the size of the cluster increases.

187
Fig. 8 shows the dependence of the ν1 frequencies of In+–(N2)n as a function of the cluster size n.

188
Similar to In+–Arn, this plot clearly reflects the structure of the subshells of the first solvation shell and the sequence of In+ solvation.

189
The less stable nπ isomer is only observed up to n = 1, with a red shift (−4 cm−1) that is larger than the one for In+–Arn.

190
The preferred solvation sequence in the more stable 2H/(n − 2)π cluster series of In+–(N2)n begins with the formation of a strongly H-bound In+–(N2)2 trimer core, in which the first H-bond (−75 cm−1, n = 1) is much stronger than the second H-bond (+25 cm−1, n = 2).

191
The 2H In+–(N2)2 trimer core is then further solvated by π-bound ligands.

192
These induce small incremental blue shifts in the ν1 frequency, which decrease almost monotonically from 3 (n = 3) to ≈1 cm−1 (n = 8).

193
Thus, the N–H bond of the H-bound In+–(N2)2 core is slightly stabilized by further sequential addition of π-bound ligands.

194
This observation is in line with a destabilization of the two H-bonds of the In+–(N2)2 core by filling the π-bound solvation (sub)shell.

195
Table 3 summarizes the photofragmentation branching ratios measured for resonant ν1 excitation of the most stable isomers of In+–Arn and In+–(N2)n.

196
In agreement with previous studies on related systems,13,18,37–42 the range of fragment channels (m) for a given parent cluster size (n) is rather narrow.

197
This information may be used to estimate ligand binding energies within the framework of a simple model.

198
This model assumes that the absorbed photon energy (ν1) is available for subsequent ligand evaporation.

199
Moreover, the ligands are classified into H-bound and π-bound ligands, with dissociation energies of D0(H) and D0(π), respectively.

200
Equivalent ligands are assumed to have the same binding energy.

201
In+–Arn clusters up to n ≤ 5 evaporate all n ligands upon ν1 excitation, implying that the sum of the dissociation energies of one H-bound and four π-bound ligands is below 3425 cm−1.

202
The dissociation energy of π-bound In+–Ar has been determined as D0(π) = 537 ± 10 cm−1,26 yielding an upper limit for the dissociation energy of the H-bound ligand, D0(H) < 1320 cm−1.

203
This limit can be further reduced because several observations suggest that the dissociation energy of H-bound In+–Ar should be significantly lower than the one of H-bound Ph+–Ar, D0(H) = 670 ± 140 cm−1.13

204
First, the relative ν1 red shift of H-bound In+–Ar is much smaller than the one of H-bound Ph+–Ar (0.8 vs.

205
2.0%).14

206
Second, the abundance ratio of H-bound and π-bound In+–Ar observed in the EI-IR spectra is much lower than the one observed for Ph+–Ar under comparable experimental conditions.

207
For example, when the conditions are optimized for the production of the π-bound isomers, this ratio is nHnπ ≈ 2∶1 for In+–Ar and ≈7∶1 for Ph+–Ar.

208
This observation indicates that the difference in binding energies between the H-bound and π-bound dimers, ΔD0(H–π) = D0(H) − D0(π), is larger for Ph+–Ar than for In+–Ar.

209
Moreover, the 2π isomer of In+–Ar2 could clearly be detected (in addition to the more stable H/π isomer).

210
In contrast, for Ph+–Ar2 only the H/π isomer was observed, whereas the 2π isomer was below the detection limit.

211
This observation indicates again that ΔD0(H–π) of In+–Ar is smaller than for Ph+–Ar.

212
As D0(π) of Ph+–Ar (535 ± 3 cm−1)9 and In+–Ar (537 ± 10 cm−1)26 are comparable, D0(H) of Ph+–Ar (670 ± 140 cm−1)13 provides an upper limit to D0(H) of In+–Ar.

213
In conclusion, D0(H) of In+–Ar can be bracketed by D0(π) of In+–Ar (537 ± 10 cm−1) and D0(H) of Ph+–Ar (670 ± 140 cm−1), resulting in D0(H) ≈ 670 ± 140 cm−1.

214
For the analysis of the photofragmentation branching ratios of In+–(N2)n, the ligands are classified into three categories with three distinct binding energies, D0(H1) > D0(H2) > D0(π).

215
For this scenario, the photofragmentation data yield D0(π) = 600 ± 100 cm−1.

216
Information about the values of D0(H1) and D0(H2) are less accurate, with 600 cm−1 ≤ D0(H2) ≤ 1450 cm−1 and 650 cm−1 ≤ D0(H1) ≤ 2300 cm−1, and the additional restriction 1300 cm−1 ≤ D0(H1) + D0(H2) ≤ 2900 cm−1.

217
These values are compatible with the theoretical predictions of De(H1) = 805 cm−1 and De(H2) = 549 cm−1 for In+–(N2)2 (Table 1).

Further discussion

218
Because of the additional charge, the intermolecular PES of the In+–Ar cation differs drastically from that of the corresponding neutral dimer, leading to very different equilibrium structures.

219
Electronic spectroscopy at the level of rotational resolution suggests that the π-bound structure corresponds to the most stable isomer of neutral In–Ar in the S0 state.23

220
No evidence for a less stable H-bound In–Ar isomer has been reported so far.

221
The π-bound structure is favored by the dispersion forces, which dominate the attractive part of the potential in S0.

222
In contrast, the EI-IR spectrum of In+–Ar demonstrates that the intermolecular H-bond is the preferred recognition site in the cation ground state (D0), whereas the π-bond is less favorable: D0(H) ≈ 670 ± 140 cm−1 and D0(π) = 537 ± 10 cm−1.26

223
Comparison with the related Ph+–Ar and An+–Ar dimers suggests that the isomerization barrier from π-bound to H-bound In+–Ar is of the order of Vb ≈ 100–200 cm−1.14,16

224
The preference for H-bonding in the In+–Ar cation arises from the additional induction forces between the charge distribution of In+ and the polarisability of Ar.

225
The AIM (atoms-in-molecules) population analysis at the UMP2/6-311G(2pd,2df) level yields a large positive partial charge of qH = 0.50 e on the acidic N–H proton of In+, which leads to substantial additional stabilization of the H-bound isomer on ionization.

226
In contrast to the acidic N–H proton, the aromatic C–H protons carry only little positive charges (qH = 0.11–0.15 e), resulting in the formation of relatively weak C–H–Ar hydrogen bonds.

227
Similar to C6H7+–Ar,38 this result supports the interpretation that the C–H–Ar bonds in In+–Ar are less favorable than π-bonding.

228
The drastic ionization-induced change in the interaction potential and the equilibrium structure of In–Ar explains why the most stable H-bound In+–Ar geometry has completely escaped detection in all previous spectroscopic studies of this cation dimer.

229
These include PIE,27 MATI,26 and ZEKE25 measurements, which are all based on resonant photoionization of the neutral π-bound In–Ar dimer via the S1 state.

230
For this REMPI process, the Franck–Condon factors for populating the H-bound global minimum of In+–Ar are very small.

231
As a consequence, the ionization potentials determined by these photoionization methods should not be considered as the adiabatic ionization potential, as claimed by the authors in .refs. 25–27

232
The ionization-induced change in the preferred recognition site in A(+)–Ar and also A(+)–CH4 dimers from π-bonding to H-bonding has now been observed for a variety of aromatic molecules A(+) featuring acidic functional YHk groups (Y = O, N).

233
It appears to be a quite general phenomenon and has so far been demonstrated for phenol,12–14 halogenated phenols,15 aniline,16 indole, and recently also for naphthol.43

234
H-bonds to Ar have also been inferred as most stable binding motif for protonated aromatic molecules with acidic functional groups, such as protonated phenol,34 protonated halogenated phenols,44 and protonated naphthol.43

235
In contrast, (protonated) aromatic cations without functional groups, such as (protonated) benzene, apparently prefer π-bonding with Ar and CH4 in the cation ground state, because the C–H bonds are much less acidic and carry only small positive partial charges, so that dispersion (favoring π-bonds) overrides the induction forces (favoring H-bonds).38,45,46

236
As the optimal In+–Ar interaction is for any orientation stronger than the Ar–Ar attraction (≈100 cm−1), the cluster growth in In+–Arn is mainly governed by the In+–Ar dimer potential.

237
Hence, the preferred solvation sequence in In+–Arn complexes begins with the formation of a H-bound In+–Ar dimer core, which is further solvated by (n − 1) π-bound ligands.

238
This process is very similar to the cluster growth observed in Ph+–Arn and An+–Arn, which also starts with the solvation of the acidic protons of the OH and NH2 groups by H-bound ligands, before further π-bound ligands are attached to the aromatic ring.13,16

239
Interestingly, the less stable 2π isomer of In+–Ar2 could be weakly observed, whereas the concentrations of the corresponding isomers of Ph+–Ar2 and An+–Ar2 were below the detection limit.13,16

240
This observation may indicate that the difference in the binding energies of the H-bound and π-bound dimers, ΔD0(H–π) = D0(H) − D0(π), decreases along the series Ph+–Ar > An+–Ar > In+–Ar.

241
Similar to Ph+–N2 and An+–N2, the H-bound isomer corresponds to the global minimum on the In+–N2 dimer potential.

242
The anisotropy of the long-range charge–quadrupole and charge–induced dipole interactions aligns the N2 ligand in such a way that the molecular axis points toward the positive charge, resulting in a nearly linear N–H⋯N–N configuration in H-bound In+–N2.4,12,18,35,38,42,47

243
The EI-IR spectra clearly demonstrate that the π-bound isomer of In+–N2 is substantially less stable than the H-bound one.

244
Interestingly, the π-bound isomers of An+–N2 and Ph+–N2 were not detected in the corresponding EI-IR spectra, suggesting that π-bonding becomes more competitive in In+–N2.

245
Actually, N2 complexes of (protonated) aromatic ions without acidic functional groups, such as C6H6+ or C6H7+, appear to prefer π-bonding with N2 in the cation ground state.38,45,46

246
According to the current interpretation of the EI-IR spectra, the preferred structure of the first solvation subshell of In+–(N2)n differs qualitatively from those of In+–Arn, Ph+–(N2)n13 and An+–(N2)n.18

247
In the latter three cluster systems, each acidic proton of the YHk group is solvated by a single ligand and subsequently π-bound ligands are attached.

248
In contrast, in In+–(N2)n the acidic N–H proton appears to be solvated by two H-bound ligands before π-bonding sets in.

249
The reason for this different behaviour is unclear at the present stage.

250
Reliable high-level quantum chemical calculations and/or higher-resolution spectroscopic data are required to definitively determine the structures of the first solvation shell of these aromatic ion–ligand complexes.

251
Previous spectroscopic studies revealed that the magnitude of the complexation-induced red shift in the proton donor stretch vibration, ΔνX−H, of H-bound X–H+–L dimers is correlated with the difference in the proton affinities (PA) of the involved bases X and L.4,13,35,48–50

252
In general, the smaller PA(X) − PA(L), the stronger the intermolecular H–L bond and the larger ΔνX−H.

253
Table 4 and Fig. 9 summarize the relative red shifts |ΔνX−H|/νX−H for a series of H-bound XH+–N2 and XH+–Ar dimers, including XH+ = SiOH+, Ph+, In+ and An+.

254
In general, the shifts are larger for N2 compared to Ar, because PA(N2) > PA(Ar) (494 > 369 kJ mol−1).51

255
Similarly, for a fixed ligand L the shifts decrease in the order XH+ = SiOH+ > Ph+ > In+ > An+ because of increasing PA(X).

256
To the best of our knowledge, no experimental value has been reported for the PA of the indolyl radical.

257
Linear extrapolation from the known PA values of the anilino and phenoxy radicals (PA = 950 and 873 kJ mol−1, respectively) and the measured frequency shifts yields an estimated value of PA ≈ 920 ± 30 kJ mol−1 for indolyl.

258
This example demonstrates that cluster ion IR spectroscopy may be used to probe thermochemical properties of transient radicals.

259
In general, the attraction in weakly H-bound X–H+–L dimers is based on electrostatic and inductive forces.4,38,52

260
For a series of X–H+–L dimers with L = Ar and N2, the strength of the H-bond was shown to correlate with the partial charge located on the intermediate proton, qH,38 and the In+–L dimers confirm this trend.

261
Thus, the dissociation energies of H-bound In+–L dimers are systematically lower than those of the corresponding Ph+–L dimers, because qH in In+ (0.50 e) is smaller than in Ph+ (0.66 e).

262
For example, D0(H) = 6520 ± 50 and 4790 ± 10 cm−1 for Ph+–H2O and In+–H2O, respectively.26

263
This trend is also compatible with PA(indolyl) > PA(phenoxy).

Concluding remarks

264
IR photodissociation spectra of In+–Arn (n ≤ 5) and In+–(N2)n (n ≤ 8) have been analysed in the vicinity of the N–H stretch vibration (ν1) to characterize the preferred recognition sites of In+ interacting with hydrophobic ligands.

265
Analysis of the size-dependent complexation-induced frequency shifts (Δν1) and photofragmentation branching ratios provide valuable information about the microsolvation of In+ in a nonpolar environment.

266
The IR spectra of all In+–Ln clusters investigated are dominated by the ν1 fundamental of the most stable isomer of each cluster ion.

267
The In+–Ar and In+–N2 dimers have H-bound equilibrium structures (global minima), whereas the π-bound isomers are identified as less stable local minima.

268
In general, the interaction strength in In+–N2 is larger than that in the corresponding In+–Ar dimers, mainly because of the additional charge–quadrupole interaction in the former complexes.

269
The Δν1 shifts yield a first experimental estimate for the proton affinity of the indolyl radical as ∼920 ± 30 kJ mol−1.

270
This example demonstrates that cluster ion IR spectroscopy may be used to probe thermochemical properties of bio-active transient radicals.

271
The IR spectra of In+–Arn (n ≤ 5) indicate that the preferred microsolvation path for this cluster system begins with the formation of the H-bound In+–Ar dimer core, which is further solvated by (n − 1) π-bound ligands.

272
In contrast, the spectra of In+–(N2)n with n ≤ 8 are interpreted with cluster structures, in which a In+–(N2)2 trimer core with two H-bound N2 ligands is generated and subsequently solvated by (n − 2) π-bound N2 molecules.

273
Unfortunately, neither Ar nor N2 matrix isolation studies are available for In+,53 preventing a detailed comparison of the In+–Ln cluster band shifts with the bulk limit (n → ∞).

274
The In+–Ln clusters have been generated in an EI cluster ion source, which predominantly produces the most stable isomer of a given cluster ion.

275
In the case of In+–Arn, the structures found differ qualitatively from those deduced from (REMPI) photoionization techniques, demonstrating that the EI ion source is more generally applicable than REMPI for the spectroscopic characterization of global minima of cation clusters.

276
In general, the structures of the most stable In+–Arn clusters differ qualitatively from those of the corresponding neutral species, emphasizing the large impact of ionization on the intermolecular potential and the preferred recognition pattern of the interaction between aromatic molecules and neutral ligands.

277
The ionization-induced change in the preferred recognition site in A(+)–Ar dimers from π-bonding to H-bonding has now been demonstrated for a variety of aromatic molecules A(+) featuring acidic functional YHk groups (Y = O, N) and appears to be a quite general phenomenon.

278
This ionization-induced variation in the recognition pattern is of importance for biological aromatic molecules with acidic functional groups surrounded by a hydrophobic environment.