Research Article

Spectroscopic Determination of the OH Solvation Shell in the OH·(H2O)n Clusters

See allHide authors and affiliations

Science  28 Feb 2003:
Vol. 299, Issue 5611, pp. 1367-1372
DOI: 10.1126/science.1080695


There has been long-standing uncertainty about the number of water molecules in the primary coordination environment of the OH and F ions in aqueous chemistry. We report the vibrational spectra of the OH·(H2O)n and F·(H2O)n clusters and interpret the pattern of OH stretching fundamentals with ab initio calculations. The spectra of the cold complexes are obtained by first attaching weakly bound argon atoms to the clusters and then monitoring the photoinduced evaporation of these atoms when an infrared laser is tuned to a vibrational resonance. The small clusters (n ≤ 3) display an isolated, sharp feature near the free OH stretching vibration, the signature of open solvation morphologies where each water molecule binds independently to the ion. Pronounced changes in the spectra are observed at n = 4 in the hydroxide ion and at n = 5 in the fluoride ion. In both cases, new features appear in the region typically associated with interwater hydrogen bonding. This behavior establishes that the primary hydration shells occur at n = 3 and 4 in hydroxide and fluoride, respectively.

The hydroxide ion (OH) is one of the two essential ionic species in aqueous chemistry (1), but its microscopic, molecular-level character, when associated with water, is still actively discussed (2–5). The major difficulty introduced by OH arises from its very high proton affinity, which causes the ionic H-bond to explore the symmetrical [HO⋯H⋯OH] Zundel-type structures. Indeed, such shared proton motifs lie at the heart of the anomalous mobility displayed by OH in water (6).

Most simulations indicate that three or four water molecules are in direct contact with OH in aqueous solution, and Parrinello (3, 7), for example, has concluded that population of the higher energy tetrahedral (trihydrate) form is the key to proton transport. We present vibrational spectra of the cold OH·(H2O)1-5 cluster ions that display sharp, well-defined bands in the OH stretching region. These spectra are then analyzed with the aid of ab initio calculations to identify the structural motifs at play in the size regime of the putative shell closing. To clarify the distinctive aspects of OH hydration, we contrast its behavior with similar spectra obtained for cold F·(H2O)n clusters in the same size range. This system was chosen because the Fhydrates are predicted (8, 9) to adopt morphologies similar to those anticipated to occur in the OH hydrates.

Previous studies and experimental design.

In previous work on these systems, thermochemical measurements (10) of the sequential water binding energies for the OH·(H2O)n clusters indicate a break at n = 4, although similar measurements on the fluoride hydrates do not suggest a shell closing (10, 11). The break in the OHhydrates was interpreted to reflect a shell closing atn = 3, but this hypothesis is somewhat tempered by the lack of discontinuities in the binding energies in OD·(D2O)n analogs (12). Theoretical work (13,14) indicates that the OH·(H2O)3 minimum-energy configuration corresponds to a C 3 or pyramidal structure with three symmetrically displaced water molecules, each tethered to the OH ion with one hydrogen. There is, however, another isomer quite close in energy where the third water molecule attaches to two water molecules rather than to the ion to form the beginning of a second shell. The geometry of the n= 4 cluster is much less clear, and some authors suggest that this form (15, 16), rather than the trihydrate, corresponds to the closed-shell species. The proposed closed-shell structure for the tetrahydrate is similar to that predicted forn = 3, but with four water molecules in direct contact with the ion. Other structural candidates for the tetrahydrate involve H-bonding the fourth water molecule to the first-shell water molecules of the n = 3 cluster, an arrangement that gives rise to many energetically close isomeric forms (17, 18).

The various isomeric candidates for the X·(H2O)n structures should be readily distinguishable based on the pattern of OH stretching bands in the mid-infrared (IR). Calculations indicate that open structures display a sharp free OH band near 3700 cm−1, whereas the networked structures yield more intense OH bands in the 3400 to 3600 cm−1 range. Several experimental difficulties unique to the anionic systems have hampered the application of vibrational spectroscopy to strongly bound anion-water systems such as the OH and F hydrates. The basic problem is that mass-selective action spectroscopy, which intrinsically requires photon absorption to fragment the cluster, is difficult to apply in a spectral range where more than one photon is required to induce fragmentation (i.e., hν ≪ D o, where h is the Planck constant, ν is the excitation frequency, and D o is the cluster dissociation energy). This is especially true when there are strong spectral shifts after absorption of the first photon. Thus, in the previous mid-IR study of OH·(H2O)n, Lee and co-workers (17) were only able to study the larger, more weakly bound n = 4 and 5 clusters. Similarly, for the bare F·(H2O)n clusters, action spectroscopy was limited to n ≥ 3 (19). In each case, the action spectra (obtained by recording the photoinduced water evaporation efficiency while varying the excitation wavelength) were extremely broad, as is typical when predissociation spectroscopy is applied to the bare hydrated anions (19–21) [X·(H2O)n +hν → X·(H2O)n-1 + H2O]. This broadening is caused by two effects. First, warm clusters are selectively detected in strongly bound systems; and second, the H-bonded networks are intrinsically weakened when attached to anions (22, 23). The net result is that the previous spectra (17) obtained for the bare OH·(H2O)4 and OH·(H2O)5 species were broadened so as to resemble that of liquid water. As such, many isomeric forms were invoked (17) to contribute to the nearly continuous absorption envelope.

The advantages of argon atom “messenger” spectroscopy.

One way to obtain the spectra of cold ion-molecule complexes is to first attach a weakly bound, inert species such as an argon atom and then monitor the production of fragment ions created when photon absorption causes the argon atom to dissociate (24). A cluster that survives extraction from the ion source with an argon atom attached necessarily has an upper limit on its internal energy content (25). As a result, the size-selected ion ensemble can be quenched into only a few (in most cases one) isomeric forms that can then be mass spec trometrically separated from the warmer clusters before analysis. There is a second important advantage in that the weakly bound argon atoms are readily evaporated even though absorption of a single mid-IR photon does not provide sufficient energy to dissociate the OH·(H2O)n core ion. The argon predissociation spectra are recorded by monitoring fragment ion production with respect to excitation wavelength:Embedded Imagewhich has the advantage that absorption events can be detected with great sensitivity using particle-counting techniques against zero background. We have recently shown this argon “nanomatrix isolation” spectroscopy to be an effective tool in the identification of the minimum-energy structures for a variety of hydrated anionic systems, (20,26–30), and we have already reported its successful application to the binary OH·H2O and F·H2O complexes (31,32).

Experimental details.

The measurements were carried out with a tandem time-of-flight photofragmentation spectrometer (33). Mass-selected cluster ions were excited with the output from a pulsed laser (Nd:YAG-pumped optical parametric oscillator/amplifier; YAG, yttrium-aluminum-garnet). Each reported spectrum is the accumulation of 30 to 40 individual scans and has been normalized for the wavelength variation in laser pulse energy. The signal was obtained by monitoring the fragment ion corresponding to the loss of an argon atom from the mass-selected parent ion, isolated with a second (reflectron) mass spectrometer. All spectra presented here result from predissociation of complexes containing one Ar atom; however, similar patterns were also recorded by monitoring N2O predissociation from the size-selected OH·(H2O)n·N2O clusters.

The OH·(H2O)nspecies were synthesized by using two different schemes to ensure that the results were not compromised by accidental mass degeneracies in the primary mass spectrometer. In the first method, we used our supersonic flowing afterglow ion source (34), where the species of interest were created by injection of neutral precursor reagents on the low-pressure side of a supersonic expansion containing trace water vapor seeded in argon. The F·(H2O)n clusters were generated with NF3 as the F source; for OH, CH4 and N2O were introduced through independently controlled pulsed valves. These reagents form the OH ion by the well-known ion-molecule reactions:Embedded Image Embedded ImageThis approach primarily yielded clusters of the type O·(H2O)n·Arm, with about a 10% contribution from the desired OH·(H2O)n·Armspecies. To further establish the integrity of our species identification, we synthesized the OH·(H2O)n·Armparents by ionizing the water/Ar expansion using a counterpropagating 1-keV electron beam under conditions that yield OH by electron-induced dissociation of water molecules in the high-density region of the expansion. This approach created a much less cluttered distribution of masses, but with much lower yields of the desired OH·(H2O)n·Armclusters. The predissociation spectra of the OH·(H2O)n·Armspecies were observed to be independent of synthetic method. This definitively establishes the OH hydrates as the carriers of the spectra and rules out the possibility that different isomers are prepared under these very different source conditions.

Analysis of the spectra.

The argon predissociation spectra of the OH·(H2O)1-5 clusters in the OH stretching region are displayed in Fig. 1with band locations collected in Table 1, and can be compared to the F·(H2O)1-5spectra presented in Fig. 2. These spectra of cold OH·(H2O)2-5 and F·(H2O)3-5 are markedly simpler than those previously reported for bare OH·(H2O)4,5 and F·(H2O)3-5, especially in the region above 3200 cm−1. Of particular importance is the low background absorption in the argon predissociation spectra; the spectra of the bare complexes (17, 19) displayed strong, continuous absorption from 3000 to 3700 cm−1. The strongest transitions in the spectra are generally caused by excitation of the ionic H-bonded OH groups (denoted IHB), with sharper structure appearing in the 3400 to 3600 cm−1 region characteristic of interwater (IW) H-bonding. The structural interpretation of the IW bands has been discussed extensively in the analysis of the halide and superoxide hydrates (26, 29, 35,36). The vibrational fundamental of isolated OH occurs at 3555 cm−1(37) and is calculated (13) to blue-shift toward the free OH band in the small hydrates, as discussed further below.

Figure 1

Argon predissociation spectra of the OH· (H2O)n·Ar complexes. (A) OH·H2O. IHB2←0 denotes overtones of the ionic H-bonded OH stretching fundamental, which is strongly coupled with bending modes to give two dominant transitions (13,31). (B) OH·(H2O)2, (C) OH·(H2O)3, (D) OH·(H2O)4, and (E) OH·(H2O)5. F and IHB indicate free and ionic H-bonded OH stretches, respectively; IWDDand IWDDA denote the interwater H-bonded OH stretches associated with water molecules in double-donor and donor-donor-acceptor environments.

Figure 2

Argon predissociation spectra of the F·(H2O)n·Ar complexes. (A) F·H2O. IHB2←0 denotes the overtone of the ionic H-bonded OH stretching fundamental. (B) F·(H2O)2, (C) F·(H2O)3, (D) F·(H2O)4, and (E) F·(H2O)5. Abbreviations are as in Fig. 1.

Table 1

Vibrational transition energies for the OH·(H2O)n·Ar complexes (±5 cm−1).

View this table:

The most striking aspect of both the OH·(H2O)n and F·(H2O)n spectra is the sudden appearance of bands in the 3400 to 3600 cm−1 IW region with increasing size. These abrupt changes occur atn = 4 for OH and n = 5 for F. Initially, at small size, only a single feature appears on the high-energy side arising from excitation of water molecules with a dangling or free (F in Figs. 1 and 2) OH group. [A second peak in the OH·(H2O)2spectrum is due to the fundamental of the hydroxide ion, which lies close to the free OH stretch in this size range (13).] Our observation of a single free OH stretch feature indicates open structures with negligible IW H-bonding, as is expected (8, 9) for the small water clusters of each ion. That is, both ions break up the n = 2 andn = 3 water networks such that they are, in effect, “internally” solvated. This is in contrast to the situation in the larger halide ions (29, 35), where their spectra display clear signatures of cyclic water networks and are better described as precursors for surface solvated ions (38,39).

In the OH·(H2O)ncase, the IW bands introduced at n = 4 persist in then = 5 spectrum, which displays several new features in addition to the spectral signature established in the n= 4 complex (Fig. 1, D and E). Thus, the structure adopted by then = 4 cluster is preserved in the n = 5 case. The sudden onset of the IW bands strongly suggests that the fourth and fifth water molecules H-bond to the first solvation shell water molecules rather than directly to OH. Similar considerations in the F hydrates indicate that the second shell occurs upon addition of the fifth water molecule (Fig. 2E).

To identify the nature of the structures at play, we turn to predictions of the various stable arrangements of water molecules around these ions recovered in previous ab initio calculations. The energetic ordering of the low-lying isomers varies with the method and level of theory used (8,13, 17–19, 40). Nonetheless, the structures give rise to markedly different patterns of OH stretching fundamentals, which can be useful in the identification of the form prepared in the laboratory. Unfortunately, the reported patterns are often either incomplete or are not presented with vibrational assignments. Therefore, we have recalculated the OH·(H2O)3,4 clusters at the same level (B3LYP/6-31+G*) used by Lee and co-workers to treat the n = 4 complex (17), whose results are similar to those obtained by Novoa et al.(40). The resulting band locations and structures are consistent with these earlier reports.

The observed spectrum of OH·(H2O)3 in the critical, IW networking region is compared in Fig. 3A with the calculated OH stretching fundamentals for the openC 3 isomer inferred in the experimental discussion above [as opposed to the ring isomer (13), which places one water molecule in the second shell]. TheC 3 cluster geometry is displayed in Fig. 4A. The absence of notable absorption in the 3200 to 3700 cm−1 region (below the free OH band) confirms that this arrangement dominates the OH·(H2O)3·Ar ensemble generated in the ion source, because the ring isomer should display strong bands in the 3400 to 3500 cm−1 range (13).

Figure 3

Evolution of the interwater H-bonding bands of OH·(H2O)nclusters at the shell closing. (A) Comparison of OH·(H2O)3 spectrum with the frequencies calculated for the internally solvated (C 3) OH·(H2O)3 structure (Fig. 4A). (B) Comparison of the OH·(H2O)4 spectrum with the frequencies calculated for the cage-type OH·(H2O)4 structure (Fig. 4B). Abbreviations are as in Fig. 1. Calculated free OH transitions in the lower trace are magnified four times.

Figure 4

Structures of the OH·(H2O)3,4 clusters showing (A) OH·(H2O)3, the completed first solvation shell, and (B) OH·(H2O)4, the onset of the second solvation shell. These were calculated at the B3LYP/6-31+G* level as described (17). Abbreviations are as in Fig. 1.

The IW region of the OH·(H2O)4 spectrum displays a sharp band at 3615 cm−1 and a group of bands near 3450 cm−1 (Fig. 3B). On the basis of the reported band patterns, it appears that the topology necessary to recover both features involves addition of a fourth water molecule to theC 3 trihydrate. One such form creates a symmetrical (C s), compact structure in such a way that the fourth water molecule forms three H-bonds to the first-shell water molecules in a donor-donor-acceptor (DDA) arrangement as depicted in Fig. 4B (16, 17). Figure 3B compares the observed OH·(H2O)4 spectrum with frequencies calculated for the structure shown in Fig. 4B [isomer OHW4V reported by Lee and co-workers (17), which differs from isomer MS4c reported by Pliego and Riveros (16) and an unreported third isomer by the orientation of the two free OH groups]. Because none of the other isomer classes so far reported (17) display activity in both the 3600 cm−1 and 3400 to 3500 cm−1regions, these spectroscopic features were emphasized by Riveros (16) as a key structural diagnostic. The IW bands in the argon predissociation spectra are in excellent agreement with the predicted location and intensity of the expected features for the isomer depicted in Fig. 4B, and we therefore assign this configuration to the cold n = 4 complex. The assignments of the various calculated bands to vibrational modes of thisC s structure (Table 1) indicate that the symmetric and asymmetric OH stretching bands of the DDA molecule are responsible for the strong absorption in the 3450 cm−1range. The double-donor (DD) water molecule in the primary shell contributes the strong 3615 cm−1 band by excitation of the OH group H-bonded to the water molecule in the second shell.

Presumably, the equivalence of three pairs of H atoms in the C s structure accounts in part for the relative simplicity of the observed spectrum. This structure has not been identified as the energy minimum of the n = 4 complex, so that a goal of future work will be to refine calculations to establish whether its preferential formation is driven by thermochemistry or kinetics. This compact, tricyclic isomer is strongly reminiscent of the charge-separated form of the HBr·(H2O)4 cluster (41–44), where calculations suggest that three bridging water molecules separate the Br and H3O+ ion pair in a C 3symmetry species (45).

Turning to the F·(H2O)n series, the open spectral pattern persists to n = 4 before the IW bands appear promptly at n = 5. The spectrum of then = 4 cluster, dominated by the IHB and free OH transitions, strongly suggests an open structure with equivalent water molecules. However, this single “free OH” feature (Fig. 2D) appears clearly below the position established in the smaller hydrates, indicating that there is an observable interaction between the water molecules. This result is consistent with the pyramidal structure recovered by Bryce et al. (46), but they favored a second shell structure at the global minimum. Kim (8), who reported a C 4 pyramidal form, pointed out that a C 4h structure would be dominant after inclusion of zero point effects. TheC 4 topology accommodates some IW H-bonding involving the “free” OH groups and accounts for the observed red-shift by formation of a loosely bound, cyclic water tetramer. This pyramidal C 4 structure was actually calculated to be the minimum energy configuration of F·(H2O)4 by Lisy et al. (19), but they assigned a different four-coordinate “3+1” structure to the warmer n = 4 complex observed in their experiment. We have also acquired the spectrum of bare F·(H2O)4, and the free OH feature of this warmer cluster indeed shifts to the blue of the feature in the argon solvated cluster and is centered about the same value as the free OH bands found in the smaller hydrates. Thus, the pyramidalC 4 structure evident in the cold F·(H2O)4·Ar complex is destroyed under conditions that yield predissociation spectra of the bare species.

The nature of the second shell initiated in the F·(H2O)5 cluster was also considered by Lisy and co-workers (19), who found three low-lying minima. The minimum-energy structure was found to be four coordinate, with the fifth water molecule residing in the second shell, attached to three first-shell water molecules. The calculated spectrum (19) arising from this form appears quite similar to that observed using argon predissociation, with a roughly evenly spaced progression of IW bands. This isomer features a web of four water molecules with a motif similar to that indicated in the structure of OH·(H2O)4.

One interesting aspect of these spectra is the breadth and location of the IHB bands associated with the H atoms in direct contact with the ion. These are much sharper in the case of the Fhydrates, reflecting the lower proton affinity of Fcompared with that of OH (1555 versus 1634 kJ/mol for F and OH, respectively) (47). In the binary complexes, the IHB fundamentals lie below the cut-off of our spectrometer (2400 cm−1), and the observed IHB features in each case (Fig. 1A and 2A) are assigned to 2←0 overtone transitions (31, 32). The IHB vibration is strongly mixed with the HOH bending modes in the case of OH·H2O, which results in an effective doubling of the IHB overtone transition (31). In the larger F·(H2O)n complexes, the IHB bands are reasonably well defined and systematically blue-shift in going from n = 1 to 4, but then red-shift again in going from n = 4 to 5. This result further confirms the conclusion that n = 5 adopts a second solvation shell structure. The water molecule in the second solvation shell should act to strengthen the ionic H-bond of water molecules in the first shell through the cooperative mechanism (48, 49), explaining the large incremental red-shift.

In contrast to the behavior of the F·(H2O)n clusters, the IHB bands in the OH·(H2O)n spectra are extremely broad, with full width at half maximum (FWHM) values ranging from 150 to 250 cm−1, and are very strongly red-shifted. This breadth is obviously not anticipated by the ab initio treatment, and when taken together with the large red-shifts, it appears that these protons remain largely delocalized between the O atoms in the OH and the three water molecules in the first shell. The IW and F bands are sharp, indicating that there is an intrinsic mode-specificity in the bandwidths. It is therefore of interest to establish the mechanism for this broadening, because it dictates the pathways for vibrational energy flow in these strongly H-bonded systems.

A curious feature in the OH·(H2O)n spectra is the location of the OH stretch. In going from isolated OH to the binary OH·H2O complex, for example, only one stretching vibration is observed at 3653 cm−1, whereas two high-frequency transitions split by about 50 cm−1 are calculated to occur as a result of the asymmetric ground state structure (Table 1) (13). The quenching of these two transitions to a single line was attributed to the rapid equilibration of the two “free” OH groups caused by delocalization of the shared proton in the vibrational zero point level of the [HO⋯H⋯OH]double-minimum potential-energy surface (31). The observed feature is much closer to that of a typical free OH stretch (3707 cm−1) than to the vibrational origin of isolated OH (3555 cm−1) (37).

The OH band location is almost unchanged upon addition of the second water molecule (Fig. 1B), although a new band occurs quite close to the free OH, presumably resulting from the free OH stretches of the two open water molecules of the OH·(H2O)2 complex. However, only one band is observed again for the trihydrate, possibly indicating a rapid blue-shift of the OH stretch so that it accidentally falls at the same position as that of the free OH vibrations on the three water molecules. Another possibility is raised by the calculated vibrational pattern, which predicts that the cross section for excitation of the OH stretch is markedly suppressed in the n = 3 to 5 clusters (by a factor of 50 to 100 compared with the free OH bands). It is not obvious what is causing this diminution in transition moment. Finally, a sharp band at 3610 cm−1 has been identified in Raman spectra of concentrated alkali metal hydroxide solutions, MOH (M = K, Na) and assigned to the OH moiety in the bulk (50). This is about 90 cm−1 below the position established in the small hydrates studied here, indicating that this feature is either due to ion pairs or that there remains a substantial size-dependent shift to be found in the larger clusters. Such studies are presently under way in our laboratory.

Further directions.

Argon predissociation spectroscopy of the size-selected OH·(H2O)n and F·(H2O)n clusters has revealed sharp bands in the high-energy region of the OH stretching spectrum and very broad bands in the ionic H-bonding region below 3000 cm−1. The upper bands reflect the networking morphology of the attached water molecules. The sudden onset of IW networking with an additional water molecule added to OH·(H2O)3 and F·(H2O)4 indicates that these sizes correspond to the maximum number of water molecules in the first hydration shells of the ions in the minimum-energy structures. A characteristic cage motif in the X·(H2O)4 moiety appears to be important in the OH·(H2O)4 and F·(H2O)5 cases, and the structure of this species has been assigned with the aid of ab initio calculations. Having established the overall character of hydration in these archetypal systems, it is clear that our understanding of their role in aqueous chemistry would benefit from fresh theoretical treatments, both to refine assignments and to elucidate the implications of the IHB bandwidths on the extent of proton delocalization.


View Abstract

Navigate This Article