Aggregation-Induced Dissociation of HCl(H2O)4 Below 1 K: The Smallest Droplet of Acid

See allHide authors and affiliations

Science  19 Jun 2009:
Vol. 324, Issue 5934, pp. 1545-1548
DOI: 10.1126/science.1171753


Acid dissociation and the subsequent solvation of the charged fragments at ultracold temperatures in nanoenvironments, as distinct from ambient bulk water, are relevant to atmospheric and interstellar chemistry but remain poorly understood. Here we report the experimental observation of a nanoscopic aqueous droplet of acid formed within a superfluid helium cluster at 0.37 kelvin. High-resolution mass-selective infrared laser spectroscopy reveals that successive aggregation of the acid HCl with water molecules, HCl(H2O)n, readily results in the formation of hydronium at n = 4. Accompanying ab initio simulations show that undissociated clusters assemble by stepwise water molecule addition in electrostatic steering arrangements up to n = 3. Adding a fourth water molecule to the ringlike undissociated HCl(H2O)3 then spontaneously yields the compact dissociated H3O+(H2O)3Cl ion pair. This aggregation mechanism bypasses deep local energy minima on the n = 4 potential energy surface and offers a general paradigm for reactivity at ultracold temperatures.

Brønsted acid/base dissociation, leading to proton transfer and solvation (1, 2) of the charged fragments in nanoenvironments (313), is one of the most fundamental chemical processes, at the root of myriad subsequent reactions. The process plays a central role not only in bulk aqueous chemistry but also in heterogeneous reactions in polar stratospheric clouds (14, 15), the spontaneous synthesis of molecules in interstellar space at ultracold temperatures (16), and dissociative adsorption and ionization at surfaces (1722). In particular, the dissociation mechanism of the strong acid HCl has been the focus of recent studies in ambient bulk water (1, 2, 23, 24), as well as at low temperatures in confined geometries; for example, on amorphous ice surfaces, on ice nanocrystals, and in microsolvation environments most relevant to atmospheric or interstellar environments (15, 18, 19, 25). In view of Arrhenius' activation law, however, a persistent puzzle has been the means whereby reactants at ultracold temperatures surmount energetic barriers that far exceed the available thermal energy. Traditionally it is assumed that photochemical excitations or tunneling are the major triggers enabling such reactions.

Despite impressive progress on the fundamental steps of HCl dissociation on ice (18, 25), infrared (IR) spectra of microsolvated HCl remain underanalyzed. Suhm and co-workers (26, 27) observed several IR bands in a HCl/H2O/He expansion and were able to assign two IR bands between 3000 and 2000 cm−1 to undissociated HCl(H2O)n, with n = 1 and 2. A prominent and unusually broad absorption band centered at ≈2570 cm−1 was attributed to dissociated structures of unknown cluster sizes. Multiple theoretical studies (2832) lend support to the conjecture that four H2O molecules might provide the smallest possible H-bonded water network, in which the most stable structure incorporating HCl is the charge-separated solvent-shared ion pair (SIP) H3O+(H2O)3Cl, as depicted in Fig. 1 (this species is sometimes termed a solvent-separated ion pair in the cluster literature). Still, experimental spectroscopic evidence is so far lacking and, in addition, the pathway to dissociation at low temperatures remains elusive.

Fig. 1

Free-energy surface (37) of HCl(H2O)4 spanned by two collective reaction coordinates: the Cl to O coordination number CNCl−O and the dihedral angle ϕ given by the four O atoms; the contour lines are 3 kJ/mol apart. The relevant free-energy minima were found to be associated with the following fully optimized structures as marked by arrows: UD, CIP, PA, and SIPC1 and SIPC3, with C1 and C3 symmetry, respectively, as determined by the orientation of the water molecule in the lower right corner. The proton derived from HCl is marked in black in all structures. The free-energy barriers are as follows: from UD to CIP, ≈930 K; from CIP to SIP, ≈650 K; and from PA to SIP, ≈100 K; the reaction coordinates used do not distinguish between the conformations (i.e., C3 versus C1) that characterize the SIPC3 and SIPC1 species, the latter being only ≈2.7 kJ/mol higher in potential energy.

Here we report the full dissociation of a single HCl molecule as a result of molecular aggregation with exactly four water molecules in superfluid He nanodroplets at 0.37 K. These droplets provide a gentle matrix for step-by-step aggregation of molecular clusters, via a controllable pickup process governed by Poisson's statistics, and they allow for concurrent high-resolution spectroscopy (3335). IR and mass-spectroscopic evidence, in conjunction with theory, shows the formation of a SIP within a minimal H-bonded network in which three water molecules solvate both H3O+ and Cl as shown in Fig. 1, thus producing what we believe is the smallest droplet of dissociated acid. We find that stepwise aggregation can induce reactions in an energetically downhill fashion, preventing the system from being trapped in deep local minima out of which it would need substantial activation to escape. At the same time, sufficiently small energy barriers can be overcome by using kinetic energy gained upon lowering the potential energy during aggregation. Transcending the specific system here, such aggregation-induced reactions are expected to play a pivotal role in explaining chemical reactivity under cryogenic conditions.

Ab initio molecular dynamics simulations (36), as well as frequency calculations, were performed in the framework of density functional theory using the BLYP functional (37). In addition, Helmholtz free-energy calculations were carried out with ab initio metadynamics (37) in the extended Lagrangian formulation, using two collective variables (CVs) to drive the isomerization reaction of the HCl(H2O)4 system from the undissociated (UD) initial state to the most stable global minimum; that is, the dissociated SIP species. One CV is defined to be the coordination number CNCl−O of Cl with respect to all O’s, and the other CV, φ, is the dihedral angle formed by the four O atoms and thus measures the two-dimensional (2D) versus 3D character of the aggregate. Aggregation simulations (37) probing the effects of electrostatic steering were carried out with both fragments dipole-aligned in their initial configuration, while keeping the nearest heavy atoms separated in equilibrating canonical simulations at 0.5 K. Subsequent release of the constraint and microcanonical ab initio molecular dynamics propagation yield the product complexes within a few picoseconds at most, which is much faster than the expected He cooling rate.

Comprehensive mapping of the free-energy surface for the n = 4 species HCl(H2O)4 clearly demonstrates that the UD species reaches the SIP global minimum via a partially dissociated contact ion pair (CIP) as a locally stable intermediate (Fig. 1). The SIP structure H3O+(H2O)3Cl is the most compact network wherein a hydronium core, H3O+, forms an Eigen complex (1, 2); i.e., H3O+(H2O)3, by donating three H bonds to the three remaining water molecules, which in turn can nicely microsolvate Cl by creating an anionic solvation shell. The network structures of pure protonated water clusters H3O+(H2O)n have been thoroughly characterized by IR spectroscopy (913) as a function of n including the bare Eigen cation n = 3. Frequency analyses of HCl(H2O)n species predict a prominent IR active mode around 2708 cm−1 corresponding to a symmetric O-H stretching motion in the H3O+ core that specifically characterizes the lowest-energy SIP minimum. Both UD and CIP structures do not show any calculated signal in this spectral regime for n = 4. For the UD structure, the closest vibrations are the water O-H stretches (>2900 cm−1) and the H-Cl stretch (<1700 cm−1), whereas the nearest frequencies in the CIP structure are predicted to be more than 200 cm−1 and ≈150 cm−1 higher and lower in frequency, respectively.

We have studied microsolvation of HCl in He nanodroplets via high-resolution IR spectroscopy, using the Bochum He-nanodroplet apparatus in combination with a home-built continuous wave optical parametric oscillator with full frequency coverage in the range from 2600 to 3400 cm−1, high output power (up to 2.7 W), and high resolution (0.0001 cm−1). Briefly (37), He droplets with an average size of 8500 He were formed in a supersonic expansion of precooled (16 K) gaseous He through a 5-μm-diameter nozzle. In a partitioned second chamber, the He droplets were doped with HCl or DCl as the first dopant and H2O as the second dopant. They were detected in a quadrupole mass spectrometer after passing through a 0.9-m spectroscopic interaction region. Absorption of an IR photon and subsequent cooling to the ground state result in the evaporation of several hundred He atoms, yielding a depletion of the mass spectrometer signal due to a reduced droplet ionization cross section. We used two different scanning modes: Frequency scans yield spectroscopic information on all complexes in the investigated frequency range, using the high-pass mode of the mass spectrometer [mass-to-charge ratio (m/z) > 8], whereas optically selective mass spectroscopy (OSMS) (38) returns information on a single species and its fragments. In He nanodroplets, soft ionization often allows detection of the parent as well as large fragments. Electron bombardment of a He cluster results in the ionization of a He atom, subsequent charge migration, and excitation of either a He cluster state (Hen)+* or the dopant molecule.

We experimentally searched the frequency range from 2645 to 2915 cm−1 for IR absorption bands of HCl/water and DCl/water complexes aggregated in He droplets that might indicate dissociation. Bands of HCl monomer, HCl homodimer, and the undissociated HCl-H2O heterodimer in He droplets have been observed and assigned previously by us (39). In addition to those, we found two strong absorption bands centered at 2669.854(1) and 2667.804(2) cm−1, subject to an intensity ratio of about 3:1 (Fig. 2A) upon sequential pickup of HCl and H2O. These signals disappeared when pure HCl or H2O was added alone to the He droplets.

Fig. 2

IR depletion spectra (arbitrary units) of the symmetric hydronium stretch of H3O+(H2O)3Cl in He nanodroplets. (A) H2O and HCl in a natural abundance isotopic mixture. (B) H2O and isotopically pure H35Cl. (C) H2O and DCl in a 35Cl/37Cl natural abundance mixture. Measured spectra are shown in black and overall fits in red; dashed blue lines show Lorentzian line profiles fitted to asymmetric top stick spectra. The H2O/HCl spectrum in (A) shows two absorption bands, which are assigned to two different structures of H3O+(H2O3)Cl: SIPC3 and SIPC1, as shown. Four absorption bands contribute to the DCl(H2O)4 absorption in (C). The stick spectra in (C) show the calculated frequencies (37) for the symmetric O-H stretching motion in the H3O+ core of H3O+(H2O)2(HDO)Cl, with the length of the bars being proportional to the IR intensities. The predicted frequencies of the symmetric O-H stretching mode in the H3O+ core of the SIPC3 conformer are marked, respectively, by red and orange vertical bars when D (stemming from DCl) is substituted for H in the H bond between an intact water molecule and Cl (HB OH), or substituted for H in the free OH group of such a water molecule (free OH); blue (HB OH) and purple (free OH) bars correspond to the same scenarios for SIPC1. The brown dashed line corresponds to residual signals caused by H/D exchange in the gas feeding system.

In order to assign the observed bands to a specific cluster, we measured the dependence of the depletion signal on the partial pressure of both dopants and used OSMS (38). The variation with pressure supported our assignment to a cluster aggregated from one HCl and four H2O molecules (37). A comparison of mass spectra (Fig. 3) acquired at the HCl/water resonances (at 2667.8 and 2669.8 cm−1) to the off-resonance mass spectrum (at 2673.8 cm−1) shows a clear increase in depletion at the masses corresponding to pure hydronium and to hydronium/water clusters H3O+(H2O)n, with n = 0, 1, 2, 3 (m/z = 19, 37, 55, and 73). No increase is observed at the undissociated HCl mass (m/z = 36), the undissociated parent cluster mass [m/z = 108 for HCl(H2O)4], and larger dissociated cluster masses with n > 3 (m/z = 91, 109, ...). In the case of a DCl/H2O mixture (Fig. 3), comparing an on-resonance spectrum at 2671.3 cm−1 to an off-resonance spectrum at 2666.4 cm−1 reveals an increase for masses corresponding to H3O+, H3O+H2O, H3O+HDO, and H3O+HDO(H2O)n, with n = 1, 2 (m/z = 19, 37, 38, 56, and 74).

Fig. 3

The top and bottom panels show a comparison of on-resonance (red) and off-resonance (black) optically selective mass spectra for HCl and DCl water clusters, respectively. The corresponding frequencies are shown in Fig. 2 as black and gray arrows for the on- and off-resonance measurements, respectively. The masses correspond to the parents H3O+(H2O)3 and H3O+(H2O)2(HDO) and their fragments. Tick marks represent integer m/z values.

For a dissociated charge-separated cluster, we expect to detect the positively charged ion and its ionized fragments; negative ions such as Cl are deflected or neutralized after electron impact ionization and charge transfer, and thus are not detected by the mass spectrometer. The highest mass affected by the change from on- to off-resonance is attributed to the parent ion, which has m/z = 73 in the case of HCl/water and m/z = 74 for DCl/water. All observations are fully consistent with an assignment to the solvent-shared ion pair H3O+(H2O)3Cl. An undissociated cluster would, in contrast, have been expected to give rise to OSMS signals at the HCl and HCl(H2O)4 parent cluster masses, as observed for undissociated HCl/(H2O)1. No signal at the mass (m/z = 20) of deuterated hydronium, H2DO+, was detected, indicating that D replaces one of the H atoms in the solvent layer (in agreement with our simulations) and not one of the H atoms in the hydronium core.

The two peaks separated by ≈2 cm−1 may then be explained in two ways: either as an isotopic splitting due to the 35Cl:37Cl natural abundance ratio (close to 3:1) within one SIP structure, or as a symmetry splitting due to the presence of different SIP conformers for n = 4. The calculations predict a negligible Cl isotopic shift (less than 0.0002 cm−1). This prediction was confirmed experimentally: Substitution of the natural isotopic mixture with isotopically pure H35Cl (Fig. 2B) yielded a nearly identical spectrum. Thus, the three water molecules in the Eigen-type SIP network are arranged in either C3 or C1 conformation with respect to their free OH bonds (see SIPC3 and SIPC1 in Fig. 1). The C1 conformer is statistically sixfold degenerate, whereas there are only two SIPC3 structures, a relationship in agreement with the observed 3:1 intensity ratio by combinatorics. In accord with experimental results, the calculated symmetric hydronium O-H stretching frequency is slightly blue-shifted in SIPC1 as compared to SIPC3. We therefore assign the two observed peaks to the SIPC1 and SIPC3 species. Given the 2.7 kJ/mol (≈320 K) energy gap separating the SIPC1 structure from the SIPC3 global minimum, Boltzmann weighting at 0.37 K would fully suppress the blue peak due to SIPC1, thus suggesting the observation of a nonequilibrium distribution.

Substitution of HCl by DCl leads to a substantial spectral shift and further splitting (Fig. 2C). Strong absorption peaks are now found at frequencies of 2669.152(6), 2671.300(1), and 2673.198(2) cm−1; a further transition is just evident at the detection limit at 2675.70(3) cm−1. The observed H/D shift is consistent with a predicted blue shift of up to at most 4 cm−1 for the various possible D positions in the H2O molecules of the SIP structures. In contrast, substitution in the H3O+ core would result in major red shifts in the range of 70 to 120 cm−1, in disagreement with the experiment (37). The predicted transitions (also in Fig. 2C) compare favorably to the observed spectrum.

Given this experimental evidence that the fully dissociated SIPC1 and SIPC3 complexes form in He droplets, the puzzle remains how the huge barriers of ≈930 and ≈650 K along the two-step dissociation mechanism following the UD → CIP → SIP pathway on the n = 4 free-energy surface (Fig. 1) can be surmounted at only 0.37 K. Such high barriers are indeed expected in view of the required fundamental rearrangement of the H-bonded network from a 2D to a 3D topology: The original UD complex is a flat five-membered ring, whereas the SIP dissociation product is compact, with three water molecules sandwiched between the hydronium core and Cl anion.

To account for our observations, we first note that within the He droplets, the dopants are cooled down to 0.37 K at typical rates of 1010 K/s (35). The associated cooling times are shorter than the time scales (35) for migration of the dopants within the droplet (≈10 μs), as well as the time scale between single pickup processes (≈0.1 ms). Thus, precooled monomers aggregate sequentially, a process that may yield nonequilibrium structures not found at higher temperatures (40, 41). We then consider that the n = 4 free-energy surface possesses an extremely shallow minimum, corresponding to the partially aggregated (PA) species in Fig. 1: an undissociated cyclic HCl(H2O)3 structure wherein the added fourth H2O molecule accepts a H bond from the free OH of the central H2O in the ring. Thus, the solution to the puzzle is provided by a process of aggregation-induced dissociation. A sequence of aggregation simulations (37) using electrostatic steering initial conditions (Fig. 4) yields cluster growth around undissociated HCl up to the ringlike n = 3 structure; adding the fourth H2O then leads to the open PA structure, which readily dissociates into either SIPC1 or SIPC3 according to combinatorial statistics without ever transiting through the much-discussed UD or CIP trap states. According to this HCl(H2O)3 + H2O → PA → SIP mechanism, the acid's proton is transferred to one of the three solvent H2O molecules, in full agreement with the OSMS HCl/DCl data. This mechanism would translate into an indirect, solvent-mediated proton transfer from HCl to H3O+ in bulk solvation environments (1, 2).

Fig. 4

Simulation of a stepwise aggregation process leading to aggregation-induced dissociation of HCl(H2O)n from n = 3 to n = 4 according to the sequence HCl + 4 H2O → HCl(H2O) + 3 H2O → HCl(H2O)2 + 2 H2O → HCl(H2O)3 + H2O → H3O+(H2O)3Cl. The optimized structures for the n = 4 PA and SIP are explicitly labeled; only the global minimum (the SIPC3 conformation) is shown here but using a different perspective from that in Fig. 1. The proton derived from HCl is marked in black in all structures. The solid red lines show the evolution of the potential energy on the scale given by the binding energy {that is, E[HCl(H2O)n] – E[HCl] – nE[H2O]} obtained from representative aggregation simulations (37) as a function of the ab initio molecular dynamics step. Vertical dashed red lines symbolize cooling of the aggregation product (obtained via full structure optimization) inside the He droplet (schematically represented at left) before the next aggregation simulation commences. Horizontal dashed green and dotted blue lines mark the binding energies of SIPC3 and SIPC1, respectively.

Supporting Online Material

Materials and Methods

Figs. S1 to S3

Table S1


  • * Present address: Department of Chemistry, Koc University, Rumelifeneri Yolu Sariyer, 34450 Istanbul, Turkey.

  • Present address: Dipartimento di Chimica, Università degli Studi di Sassari, CNR-INFM SLACS and INSTM, Via Vienna 2, 07100 Sassari, Italy.

  • Present address: Faculty of Chemistry, Nicolaus Copernicus University, Gagarina 7, 87-100 Toruń, Poland.

References and Notes

  1. Materials and methods are detailed and references are given in the supporting material available on Science Online.
  2. The authors are grateful to W. Sander for helpful discussions and K. Merz for support in the synthesis of isotopically pure HCl. We thank D. Habig, S. Henkel, M. Ortlieb, and T. Poerschke for help with the experiment and data evaluation. Support by the Deutsche Forschungsgemeinschaft (grants HA 2394/13 and MA 1547/8 within FOR 618 "Molecular Aggregation") and Fonds der Chemischen Industrie is gratefully acknowledged, as well as computer resources from Bovilab@RUB and Rechnerverbund-Nordrhein-Westfalen (LiDO).
View Abstract

Stay Connected to Science

Navigate This Article