Helium nanodroplets doped with copper and water

Copper nanoparticles are promising, low-cost candidates for the catalytic splitting of water and production of hydrogen gas. The present gas-phase study, based on the synthesis of copper-water complexes in ultracold helium nanodroplets followed by electron ionization, attempts to find evidence for dissociative water adsorption and H2 formation. Mass spectra show that H2O–Cu complexes containing dozens of copper and water molecules can be formed in the helium droplets. However, ions that would signal the production and escape of H2, such as (H2O)n−2(OH)2Cum+ or the isobaric (H2O)n−1OCum+, could not be detected. We do observe an interesting anomaly though: While the abundance of stoichiometric (H2O)nCum+ ions generally exceeds that of protonated or dehydrogenated ions, the trend is reversed for (H2O)OHCu2+ and (H2O)2OHCu2+; these ions are more abundant than (H2O)2Cu2+ and (H2O)3Cu2+, respectively. Moreover, (H2O)2OHCu2+ is much more abundant than other ions in the (H2O)n−1OHCu2+ series. A byproduct of our experiment is the observation of enhanced stability of He6Cu+, He12Cu+, He24Cu+, and He2Cu2+.


Introduction
Copper nanoparticles dispersed in water or in the form of coatings have a range of promising uses, including lubrication, ink jet printing, as luminescent probes, exploiting their antimicrobial and antifungal activity, and in fuel cells [1][2][3][4][5][6]. Thermodynamic data predict that water splitting at monocrystalline copper surfaces does not produce significant amounts of molecular hydrogen but the evolution of hydrogen gas from oxygen-free water in contact with copper surfaces has been observed [7].
Theoretical studies of the reaction of water with surfaces of crystalline copper have been reported by Johansson et al. [8] and Lousada et al. [9]. At the lowdensity [110] surface, dissociation of adsorbed H 2 O is spontaneous; it has a lower activation energy than at the higher-density [100] or [111] surface. The activation energy decreases with increasing water coverage. The autocatalytic dissociation of water on the Cu[110] surface has, indeed, been identified by photoelectron spectroscopy at near-ambient conditions [10]. Johansson et al. have suggested that water oxidizes Cu[100] until the surface is saturated with hydroxyl groups, and that H 2 forms by direct combination of hydrogen atoms at the surface [8]. Lousada  monolayer of H 2 O accompanied by the release of hydrogen gas [9]. In industrial applications the surface of a copper catalyst is not crystalline. Some catalytic reactions may be promoted by atoms or small clusters of copper [11,12]. Huseyinova et al. have reported the synthesis of surfactant-free, nearly mono-disperse Cu 5 clusters in water that are stable to UV irradiation, elevated temperature, and a wide range of pH [13]. Their catalytic activity has not yet been measured, but several theoretical studies of water dissociation on Cu 7 have been reported [14][15][16]. Cu 7 represents a prototypical cluster because most DFT-based calculations predict that the ground-state structures of copper clusters Cu m are planar for 3 ≤ m ≤ 6 while Cu 7 forms a pentagonal bipyramid [17].
Based on a DFT approach, Chen et al. conclude that the main driving force for the adsorption of H 2 O at Cu 7 is the overlap between the p-orbital of O occupied by the lone pair and the 3d orbitals of Cu [14]. The reports discussed so far involve neutral complexes. Water splitting in ionic systems is of interest, too, because in the condensed phase metallic clusters may be charged [16]. Furthermore, the influence of the charge will likely decrease as the size of the copper cluster increases.
Several experimental studies of ionic copper-water complexes in the gas phase have been reported; most are restricted to atomic Cu + or Cu 2+ . Holland and Castleman have used high-pressure mass spectrometry to determine thermochemical properties of (H 2 O) n Cu + for n = 3, 4, 5 [18]. Michl and coworkers as well as Armentrout and coworkers measured dissociation energies D n (or sequential binding energies) of (H 2 O) n Cu + for 1 ≤ n ≤ 4 by collision-induced dissociation [19,20]. (H 2 O) 2 Cu + is remarkably stable; it's dissociation energy (1.7 eV) exceeds that of (H 2 O)Cu + by about 0.2 eV [19]. Several theoretical studies have shown that Cu + is, indeed, two-fold coordinated; the oxygen atoms of the first two H 2 O bind to opposite sides of Cu + [21][22][23][24][25]. The interaction is nearly evenly divided among electrostatic, polarization, and charge transfer; the Cu Mulliken charge is only +0.85 e [25]. Properties of (H 2 O) n Cu + have also been explored by vibrational spectroscopy [26][27][28]. (H 2 O) n OHCu + complexes have been characterized by vibrational spectroscopy and collision-induced dissociation [29][30][31][32] as well as theoretical modeling [31].
In contrast to these numerous studies of (H 2 O) n Cu + and (H 2 O) n OHCu + , we are aware of only two gasphase studies that mention complexes containing more than one copper atom. Michl and coworkers observed (H 2 O) n HCu 2 + , (H 2 O) n−1 OHCu 2 + , and (H 2 O) n−1 OCu 2 + but no (H 2 O) n Cu 2 + upon fast-atom bombardment of frosted copper surfaces; unfortunately they did not quantify the yield of these ions nor the range of n values [19] (Here the chemical formulas are written such that the subscript n specifies the number of oxygen atoms; they do not necessarily convey structural information). Stace and coworkers used a pulsed-arc cluster source; they observed predominantly ( As a by-product, we have recorded mass spectra of helium droplets doped with copper but no water, leading to the formation of He n Cu m + . The He n Cu + series features local maxima at n = 6, 12 and 24. Maxima at n = 6 and 12 have previously been observed for Ar n Cu + and Ne n Cu + , respectively, and rationalized by DFT calculations [34]. The significance of those observations for our results will be discussed.

Experiment
Helium nanodroplets were produced by expanding helium (Linde, purity 99.9999%) at a stagnation pressure of 25 bar through a 5 µm nozzle, cooled by a closed-cycle cryostat to between 9 and 10 K, into vacuum. Droplets formed at these conditions contain roughly 10 6 atoms. The exact temperatures and estimated [35] average numbers of helium atoms in the droplets will be specified in the Result section.
The expanding beam was skimmed by a 0.8 mm conical skimmer located 8 mm downstream from the nozzle and traversed an 8 cm long, differentially pumped pick-up cell filled with copper vapor produced in a resistively heated oven. The temperature of the copper oven could not be measured directly; it was adjusted in order to obtain the optimal conditions for formation of either He n Cu m + or (H 2 O) n Cu m + cluster ions. Water vapor was introduced into a second differentially pumped pickup chamber from an external water reservoir.
The beam of doped helium droplets was collimated and crossed by an electron beam with a nominal energy of 70-80 eV. Cations were accelerated into the extraction region of a reflectron time-of-flight mass spectrometer (Tofwerk AG, model HTOF) with an effective mass resolution m/∆m = 3000 (∆m = full-width-at-halfmaximum). The base pressure in the mass spectrometer was 10 −5 Pa. Ions were extracted at 90 • into the fieldfree region of the spectrometer by a pulsed voltage. At the end of the field-free region they entered a two-stage reflectron which reflected them towards a microchannel plate detector operated in single ion counting mode. Further experimental details have been provided elsewhere [36].
Mass spectra were evaluated by means of a customdesigned software [37]. The abundance of ions is derived from the mass spectra by a matrix method. The routine includes automatic fitting of a custom peak shape to the mass peaks and subtraction of background by fitting a spline to the background level of the raw data. Hydrogen and helium are very nearly monoisotopic (the natural abundance of deuterium is 0.0115%; that of 3 He is 0.000137%) but copper has two naturally occurring isotopes, 63 Cu (mass 62.9296 u, natural abundance 69.17%) and 65 Cu (64.9278 u, 30.83%).  at an electron energy of 71 eV. Helium was expanded at 25 bar through a 5 µm nozzle cooled to 9.25 K. At these conditions, the estimated average number of helium atoms in a droplet is 5 × 10 6 helium atoms [35].

Experimental results
The most prominent mass peaks in Figure 1a are due to Cu m + , m ≤ 15. Complexes of Cu m + with a water molecule (due to a water contamination) become apparent for m ≥ 3; for m ≥ 7 complexes with two water molecules appear as well. Another minor contamination is assigned to PH + (mass 31.982 u); it appears when the copper oven is heated. The mass peak of PH + is distinct from that of O 2 + (31.990 u) but PHCu m + and O 2 Cu m + cannot be distinguished for m ≥ 1. Figure 1b zooms into the mass region between Cu + and Cu 2 + . The two naturally occurring isotopes of copper ( 63 Cu and 65 Cu) and the three isotopologues of Cu 2 are marked by open triangles. Solid triangles connected by a solid line mark mass peaks due to He n 63 Cu + ; they are nearly as intense as pure He n + ions (open dots connected by a dotted line). Figure 2 displays the ion abundance of He n Cu + and He n Cu 2 + deduced from the mass spectrum in Figure 1 with a custom-designed software [37]. Uncertainties are reported by the software; a few statistically significant anomalies are marked. Some error bars are very large as a result of mass spectral coincidence with potential contaminants. For example, He 4 63 Cu + and He 4 65 Cu + cannot be distinguished from O 63 Cu + and H 2 O 63 Cu + , respectively (the relative mass differences are ∆m/m ≤ 2 × 10 −4 ).
A mass spectrum of helium droplets doped with copper and water is presented in Figure 3. Data were recorded with a helium stagnation pressure of 25 bar, nozzle temperature 9.65 K, estimated droplet size 5 × 10 5 atoms, electron energy 80 eV. The total pressure (mostly water) in the pickup cell was 1 × 10 −5 mbar. The temperature of the copper oven was about the same as used to record the spectrum in Figure 1. The most prominent mass peaks in Figure 3 are due to He n + and, above about 150 u, (H 2 O) n H + (for the sake of consistency, we write all chemical formulas such that the subscript n specifies the number of oxygen atoms in a complex; they are not meant to convey structural information). Figure 4 zooms into a section of the spectrum. Mass peaks due to (H 2 O) n 63 Cu + are marked by solid triangles,

Copper-water complexes
The main motivation for the present work is the identification of ions that might signal the splitting of water on the surface of copper cluster ions. Several theoretical studies suggest that hydration of Cu 7 may lead to the evolution of molecular H 2 [14][15][16]. Water splitting can be catalyzed by another H 2 O via hydrogen bonding. Water coverage beyond the first monolayer favors the reaction, thermochemically as well as kinetically [16].
These studies pertain to neutral clusters containing seven atoms (Cu 7 is commonly chosen because it is the smallest three-dimensional cluster, but the reaction energetics and kinetics may be similar for other small clusters [14]). In our experiment, neutral water-copper complexes are synthesized in superfluid helium nanodroplet by successive capture of H 2 O and Cu, in random order. This is a statistical process which leads to a broad distribution in the size distribution of (H 2 O) x Cu y , amplified by the considerable size range of the helium nanodroplets. The mass spectra show that x may be as large as ≈50, and y may exceed ≈20. Presumably there is no lower limit to the values of x and y. Water splitting is unlikely to occur during synthesis of (H 2 O) x Cu y in a helium droplet because the droplet temperature is only 0.37 K; the energy released upon aggregation is rapidly removed by evaporation of helium atoms (which costs about 0.6 meV per atom). However, ionization of the dopant is an indirect process which starts with the formation of He + and resonant hole hopping, or of electronically excited, highly mobile He * − [38]. Formation of He + requires about 24.6 eV while the ionization energy of Cu is 7.73 eV; the difference will be released upon charge transfer between He + and Cu. The excess energy would be a few eV smaller if ionization involves He * − . 1 On the other hand, the excess energy would be a few eV larger for large copper clusters whose ionization energies converge towards the copper work function (≈5 eV). Thus an excess energy of 17 eV, give or take a few eV, will be released upon ionization of the dopant. 1 Furthermore, the emitted electrons could possibly carry away a large fraction of the excess energy.
Electron ionization mass spectra of clusters embedded in helium nanodroplets do, in fact, indicate extensive intra-and intermolecular fragmentation. Often the spectra are very similar to mass spectra recorded by direct electron ionization of bare clusters at commonly used (70 eV) energies. For example, electron ionization of water clusters embedded in helium as well as electron ionization of bare water clusters result in predominantly protonated water cluster ions, (H 2 O) n H + , and an abrupt drop in the ion abundance at n = 21 [39,40].
Furthermore, the mass spectrum in Figure 1 provides direct evidence for ionization-induced dissociation. Although the size distribution of neutral clusters is smooth because of the statistical nature of the capture process, the measured abundance distribution of Cu m + closely tracks the size dependence of their dissociation energies, with the electronically closed-shell Cu 3 + and Cu 9 + being much more stable than Cu 4 + and Cu 10 + , respectively [41]. The evaporative model explains why local anomalies in cluster ion abundance distributions reflect anomalies in their relative dissociation energies, provided each cluster ion has lost at least one atom before the distribution is being measured [42,43]. The dissociation energies of Cu 3 + and Cu 9 + are 2.83 and 3.66 eV [41] proving that ample energy becomes available when copper clusters embedded in helium are ionized by electron impact.
Splitting of water adsorbed on copper cluster ions followed by generation and escape of H 2 would be described by the reaction which could possibly be followed by the reaction A mass spectrum cannot distinguish between the reactant and product of reaction (2); the ion would commonly be identified as (H 2 O) n−1 OCu m + . An even better indication of H 2 production would be the reaction Based on a DFT study of hydrated Cu 7 , Stenlid et al. concluded that as many as 4 H 2 might be generated, which would result in (H 2 O) n−8 (OH) 8 Cu 7 or the isobaric (H 2 O) n−4 (O 2 ) 2 Cu 7 [16].
The theoretical studies of water dissociation on Cu 7 focus on neutral copper clusters [14][15][16] but Stenlid et al. [16] briefly mention Cu 7 − and Cu 7 + as well. They point out that Cu 7 + binds water more strongly, accommodates more water in the first solvation shell, and the Cu-Cu distances in the equatorial plane are increased relative to neutral Cu 7 . It is not clear though to what degree the catalytic activities of Cu 7 + and Cu 7 would differ. We have searched for evidence of these types of water splitting reactions in our mass spectra, to no avail. Unfortunately the sensitivity to the detection of (H 2 O) n−x O x Cu m + , x ≥ 1 is hampered by two factors. First, these types of ions cannot be positively identified for even-numbered x in the mass range of interest (≥63 + 32 u) because the mass of ions containing O 2 becomes indistinguishable from the mass of ions containing the contaminant PH. Second, the two naturally occuring isotopes of copper ( 63 Cu and 65 Cu, abundance 69.83% and 30.17%, respectively) produce highly congested spectra. The isotopologues are easily resolved up to, at least, Cu 15 + (see the spectrum in Fig. 1a), but O 65 Cu + and H 2 O 63 Cu + differ by only 0.018 u. The mass resolution of our instrument, m/∆m ≈ 3000, is not quite sufficient to resolve these two ions; resolving them in complexes containing more than one copper atom would be a hopeless task. Experiments with isotopically enriched 63 Cu would alleviate this problem.
Theoretical studies pertain to neutral copper clusters containing 7 atoms while our work pertains to cationic clusters. Furthermore, the mass spectral resolution and the occurrence of isotopologues limits the information that we can deduce for ions containing more than a few copper atoms. Carnegie et al. have performed infrared photodissociation spectroscopy of argon-tagged H 2 OCu + in the O-H stretching region [28]. Iino et al. have used a similar approach to study argon-tagged (H 2 O) n Cu + , n ≤ 4, and untagged (H 2 O) n Cu + , n ≤ 7 [27]. Their work provides no evidence for water splitting.
The catalytic activity of Cu m + may be quite sensitive to its geometric and electronic structure, hence its size. In this context, the flagged feature in Figure 6b is noteworthy. It shows a greatly enhanced yield of dehydrogenated ions for just two species, (H 2 O)OHCu 2 + and (H 2 O) 2 OHCu 2 + . This suggests dissociative adsorption of H 2 O on Cu 2 + which would be a prerequisite for H 2 formation. Theoretical work is needed to elucidate the origin of this feature. Dehydrogenated hydrated copper ions have been characterized by collision-induced dissociation [32] and vibrational spectroscopy [29][30][31] but we are not aware of related work on ions containing two or more copper atoms.
The results presented in Figure 6 are markedly different from those reported earlier [19][20][21][22][23][24][25][26][27][28][29][30][31][32][33]. Michl and coworkers applied secondary-ion mass spectrometry of copper covered with a frozen water film [19]. The copper monomer formed a series of hydrated ions (H 2 O) n Cu + but no dehydrogenated ions (H 2 O) n−1 OHCu + . The copper dimer formed the protonated series, the dehydrogenated series, and the doubly dehydrogenated series (H 2 O) n−1 OCu 2 + , but no (H 2 O) n Cu 2 + . Stace and coworkers formed small copper cluster ions in an arc source; the ions were swept out by a helium carrier gas seeded with water vapor [33]. They detected only one complex containing a single copper atom, HCu + . The hydrated copper dimer appeared in two forms, (H 2 O) n−1 HCu 2 + and (H 2 O) n−1 H 2 Cu 2 + . The hydrated copper trimer appeared in the protonated, dehydrogenated, and doubly dehydrogenated form but no (H 2 O) n Cu 3 + was observed. Thus, while we find that (H 2 O) n Cu m + is (with the exception of m = 2, n = 2 and 3) the most intense ion series, that series was not observed at all in previous work (with the exception of (H 2 O) n Cu + [19]). Our tentative conjecture is that the very large excess energies that are available when ions are sputtered, or formed in a plasma, are responsible for the striking differences.
Finally we address the features marked in Figures 6a and 6c. The dissociation energy of (H 2 O) 2 Cu + is known to be about twice that of (H 2 O) 3 Cu + ; it is even higher than that of (H 2 O)Cu + [18][19][20]. This rather unusual feature has been traced to the peculiar structure of (H 2 O) 2 Cu + whose lowest-energy structure has the H 2 O molecules adsorbed on opposite sides of the Cu + ion with the O atoms facing Cu + . Most theoretical studies agree that Cu + in (H 2 O) 3 Cu + and (H 2 O) 4 Cu + is two-fold coordinated, but the increase to three-or even four-fold coordination in ions containing as many 10 H 2 O is less clear [21][22][23][24][25].
The local maximum in the abundance of (H 2 O) n Cu + at n = 2 (see Fig. 5) signals its high stability. We also see an abrupt drop of the abundance at (H 2 O) 6 Cu + which suggests a sudden decrease in the dissociation energy. The only theoretical study of ions in this size range [25] reported relative energies of structural isomers relative to the ground state structure; dissociation energies of (H 2 O) n Cu + ions in their ground state were not reported. However, it was noted that the coordination of Cu + increased from two or three for n = 5, 6 to three or four for n ≥ 7. It is tempting to conjecture that this change in coordination is accompanied by a weakening in binding.
A pronounced local maximum occurs in the abundance distribution of the (H 2 O) n Cu 3 + series at n = 3. We are not aware of any studies of water adsorption on the copper trimer (nor trimers of other noble metals); theoretical work is needed to elucidate the origin of the apparent enhanced stability of (H 2 O) 3 Cu 3 + . Figure 2 displays the abundance distributions of He n Cu + and He n Cu 2 + . Noteworthy are local maxima at n = 6, 12, and 24 in the He n Cu + series and an abrupt drop at He 2 Cu 2 + ; the features suggest enhanced stability of these ions. The high abundance of He 12 Cu + might indicate icosahedral arrangement of the ligands, a structure observed for some other systems with non-directional bonding such as He n K + [44], He n Ar + [45], He n Ag + [46], and He n Au + [47]. Another factor in the appearance of icosahedral structure is that the size of the ion, i.e. the ion-ligand bond length, has to be about right for the size of the ligands, i.e. the ligand-ligand bond length. For He n Ar + the match is near-perfect, resulting in a highly ordered system with three distinct solvation shells of I h symmetry [48].

Copper-helium complexes
For hard-sphere models, icosahedral packing is preferred if σ * , the ratio of the ion-ligand and ligand-ligand lengths, lies between 0.82 and 0.95 [49]. Ab initio-calculations of the potential energy curves of He-Cu + result in bond lengths (R e ) of 1.93 or 1.95Å [50,51]. The He-He bond length depends very much on the environment because of the large zero point energy. If one uses the helium bulk density for an estimate (which gives results in reasonable agreement with dimer bond lengths for the heavier noble gases) one obtains R e ≈ 2.52Å for He-He, and σ * ≈ 0.77. This value is midway between the range for which icosahedral packing would be favorable, and the range 0.61 ≤ σ * ≤ 0.71 for which octahedral packing would be favorable [49]. Thus the first two maxima in the abundance distribution in Figure 2, at n = 6 and 12, are not inconsistent with expectations based on hard-sphere packing models.
The only realistic theoretical study (at the coupled cluster single double triple level) of the stability of He n Cu + containing more than one helium atom was reported by Li et al. [52]; it extends to n ≤ 3. The authors concluded that He 2 Cu + is linear with D ∞ symmetry, and He 3 Cu + is planar with D 3h symmetry. These features are not consistent with the assumption of non-directional bonding.
Another study worth mentioning is by Froudakis et al. [34]. The authors reported mass spectra of Ne n Cu + and Ar n Cu + . The abundance distribution of Ne n Cu + suggested enhanced stability at n = 4 and 12; DFT calculations confirmed this conjecture. The ground state structure of Ne 12 Cu + turned out to be icosahedral even though for Ne-Cu + the value of σ * is only ≈0.66, within the range where hard-sphere packing models would favor octahedral packing. Ne 6 Cu + had the largest computed dissociation energy in the size range 3 ≤ n ≤ 13; its structure was a strongly distorted octahedron of C 2v symmetry.
To summarize, local abundance maxima at He 6 Cu + and He 12 Cu + have been tentatively assigned to octahedral and icosahedral structures; calculations are needed to confirm this conjecture. Existing theoretical reports of small He n Cu + and larger Ne n Cu + indicate that He n Cu + cannot be modeled by simple pairwise additive potentials. We have made no attempt to provide structural models for He 24 Cu + and He 2 Cu 2 + which also seem to enjoy enhanced stability.

Conclusion
We have attempted to detect evidence for the production and release of H 2 from cationic copper-water complexes in the gas phase. The reaction would have to proceed after (or upon) ionization because of the low temperature of the helium droplets in which the neutral precursors are grown. We failed to detect the telltale of H 2 production, i.e. ions of the composition (H 2 O) n−2 (OH) 2 Cu m + or the isobaric (H 2 O) n−1 OCu m + (or (H 2 O) n−2 O 2 Cu m + which would indicate the production of 2 H 2 ). In future work we plan to use isotopically enriched copper; this would greatly reduce the number of isotopologues for ions containing several copper atoms, and increase the sensitivity for the detection of (H 2 O) n−x O x Cu m + . We did detect ions that suggest highly size-dependent reactions, namely very intense signals for (H 2 O)OHCu 2 + and (H 2 O) 2 OHCu 2 + ; the latter forms a maximum in the (H 2 O) n OHCu 2 + series and exceeds the abundance of (H 2 O) 3 Cu 2 + . Another interesting observation is the large abundance of He 6 Cu + , He 12 Cu + , He 24 Cu + , and He 2 Cu 2 + . It is tempting to speculate that the apparently high stability of the first two of these ions correlates with octahedral and icosahedral arrangements of the solvent atoms. Theoretical work is needed to confirm these conjectures, and to provide a rational for the postulated enhanced stability of He 24 Cu + and He 2 Cu 2 .
This work was given financial support by the Austrian Science Fund (FWF) Wien (Project P26635) and the European Commission (ELEVATE, Horizon 2020 research and innovation program under grant agreement No. 692335). Open access funding provided by Austrian Science Fund (FWF).
Author contribution statement P. Scheier conceived the project; S. Raggl, N. Gitzl, and P. Martini carried out the experimental work and data analysis; O. Echt prepared a draft of the manuscript; all authors discussed the results and commented on the manuscript.
Open Access This is an open access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.