Uptake of Hydrogen Bonding Molecules by Benzene Nanoparticles.

The uptake of molecules on nanometer-size clusters of polyaromatic hydrocarbons (PAHs) is important for the condensation of water on PAH aerosols in the atmosphere and for ice mantle growth on nanoparticles in the interstellar medium. We generate benzene clusters BzN of mean size N̅ ≈ 300 (radius R̅ ≈ 2.2 Å) as a model system for the PAH nanoparticles. Using molecular beams and mass spectrometry detection, we investigate the uptake of water, methanol, and ethanol by these clusters. All picked up molecules are highly mobile on BzN and generate clusters within <3 ms. The relative uptakes for the different investigated molecules can be directly compared and quantified. Water molecules exhibit the lowest relative pickup probability that is ∼30% lower than those for methanol and ethanol, which are approximately the same.

Aromatic carbonaceous compounds such as benzene and polyaromatic hydrocarbon (PAH) molecules are omnipresent in our Earth atmosphere as well as in many regions of interstellar space. They can become part of atmospheric aerosols, which have a huge impact on the Earth’s radiation balance.[1] PAHs and products of their atmospheric degradation contribute to positive radiative forcing by absorption of light, as a part of so-called atmospheric brown carbon.[2] The aerosol particles, in general, play a pivotal role in atmospheric chemistry, because they act as surface catalysts for heterogeneous reactions.[3,4] In addition, the impact of aerosols on air quality and human health, namely the carcinogenicity of PAHs, has been long recognized.[5]

From another point of view, astronomical observations reveal that various aromatic carbonaceous compounds, namely, benzene and PAHs, are present in different galaxies.[6−10] Carbonaceous grains and gaseous PAHs are believed to constitute the majority of carbon available in diffuse clouds.[11,12] Icy mantles and PAHs are involved in photochemical pathways to complex organic molecules in the interstellar medium (ISM).[13] The life cycle of large PAHs in the interstellar medium involves their breakdown by ultraviolet radiation to small PAHs and the generation of fragment clusters.[14,15] The recently launched James Webb Space Telescope should provide further insight into the PAH life cycle,[10] which makes this investigation topical.

Here, we investigate the uptake of molecules by nanometer-size benzene clusters. As species that get picked up by the Bz clusters, we choose the following hydrogen bonding molecules with increasing complexity: water (H2O), methanol (CH3OH), and ethanol (CH3CH2OH). Their uptake can mimic the hydration of PAH soot particles in the atmosphere. Because water and methanol are part of the main molecular species observed in interstellar ice mantles,[16] they can also represent the laboratory analogues to the ice mantles that grow on PAH nanoparticles.

From a fundamental point of view, the mixed benzene/water clusters serve as model systems for studying PAH–water interactions.[17,18] Infrared spectroscopy revealed their structure and the microscopic hydrophobicity of the aromatic ring.[19] Hydration and proton transfer in benzene–water cation clusters were studied with ion mobility experiments and calculations.[20] Recently, clusters of the more complex PAHs, namely naphthalene, with water were investigated.[21,22] In most of the experimental studies so far, the clusters were prepared by co-expansion of benzene (or other PAHs) with water vapor. In contrast to that, here we generate pure nanometer-size benzene clusters and deposit several molecules on them as they pass through a pickup chamber. Such uptake processes can mimic the initial steps of ice mantle growth. In addition, it allows molecular level investigations of the pickup, coagulation, and reactions after electron ionization in the benzene–molecule clusters.

In our experiment, benzene clusters Bz were produced by a continuous supersonic expansion of benzene vapor with argon, resulting in a mean size N̅ of ≈300 and a radius of 2.2 nm as determined below. As they fly through a vacuum chamber where the molecules can be introduced, they can pick up several molecules. Then, the clusters enter a time-of-flight mass spectrometer (TOF), where the mass spectra after electron ionization are recorded. The experimental setup is described elsewhere,[23,24] and more details are given in the Supporting Information. We should mention that due to cluster fragmentation and the limited discrimination-free mass range of our TOF spectrometer, the mean cluster size has been derived from the Poisson distribution of the adsorbed molecular clusters rather than from the mass spectra, as detailed in the Supporting Information.

Figure shows examples of the mass spectra of the benzene clusters after the uptake of the molecules. The mass spectra in the extended mass and intensity range are shown in the Supporting Information. Figure a corresponds to the uptake of water. The blue trace at the bottom shows the mass spectrum of pure benzene clusters without any pickup. This spectrum is shown and discussed in detail in the Supporting Information. When the water molecules are evaporated into the pickup cell, water clusters appear in the mass spectrum, as demonstrated by the top trace (red). The new ion series after the pickup are indicated by the following symbols: black circles for protonated water clusters (H2O)H+ and triangles for water clusters with benzene Bz·(H2O)+. It is interesting to note that the pure water clusters are protonated, while the major series of benzene-containing cluster ions correspond to nonprotonated species, although the dominating peaks of the first three members of the single Bz-containing series are even deprotonated, i.e., Bz·(H2O)OH+ (k = 0–2). Figure b shows a part of the spectrum after the methanol pickup. The spectrum is very similar to the previous one in terms of the generated species: black circles for protonated methanol clusters (CH3OH)H+ and triangles for methanol clusters with benzene (Bz)·(CH3OH)+. Finally, the pickup of ethanol (EtOH) in Figure c exhibits the same features: protonated ethanol clusters (EtOH)H+ and ethanol clusters with benzene (Bz)·(EtOH)+.

Figure 1

Mass spectra of the benzene clusters after the pickup of (a) water (H2O) (the blue bottom spectrum is the pure Bz cluster spectrum without the pickup), (b) methanol (CH3OH), and (c) ethanol (CH3CH2OH). The symbols and dashed lines indicate series of cluster ion fragments of the adsorbed molecules (see the text for details).

Interestingly, the mass peaks corresponding to the protonated, nonprotonated, and dehydrogenated species are present in all series in different abundances for different adsorbed molecules. The details of the cluster ionization and ion–molecule reactions, which can yield the observed cluster ion fragments, will be the subject of future investigations. Here, we concentrate on the overall pickup of different molecules on benzene clusters demonstrated by the ions containing the adsorbed molecules and their fragments independent of their actual composition and structure. Finally, it ought to be mentioned that more spectra (three for EtOH, five for MeOH, and seven for H2O) have been recorded and evaluated under different experimental conditions (see the Supporting Information), and the results presented below represent an average of all of the measurements.

The molecules are picked up one by one, but the spectra contain quite large molecular clusters. Thus, the presence of the molecular cluster ions in the spectra shows that the molecules coagulate efficiently on and/or in benzene nanoparticles during their flight time from the pickup to the ionization region, which is ∼3 ms (1.8 m over the measured cluster velocity of 540 m/s). Note that the observed molecular cluster ions cannot be generated from individual molecules after cluster ionization, leading to cluster fragmentation.

To assess differences in the pickup of the three hydrogen bonding molecules by the benzene nanoparticles, we integrate the intensities of all mass peaks containing molecule M and its fragments. We assume that this integrated intensity IM is proportional to the number of Bz clusters that picked up at least one molecule M in the pickup cell. It should be noted that a benzene nanoparticle that picked up molecules M contributes the same amount to the signal integrated in IM independently of how many molecules M were embedded in the nanoparticle and whether the molecules coagulated. Analogously, we integrate all of the clean Bz+ peaks (including the corresponding peaks displaced by m/z ±1, ±2, etc., as outlined in the Supporting Information). This integral, IC, corresponds to the clusters, which passed through the pickup cell without adsorbing any molecule M. The ratio then reflects the relative probability of the uptake for the given molecules M by the Bz nanoparticles (see the Supporting Information for more details). These ratios are summarized in Table .

Table 1

Relative Uptake Ratios Rup of Different Molecules M by the Bz Clusters

molecule	Rup
water (H2O)	0.21 ± 0.10
methanol (CH3OH)	0.30 ± 0.08
ethanol (CH3CH2OH)	0.29 ± 0.08

We need to discuss several assumptions of the mass spectrometry analysis justifying the comparison of the Rup factor for different molecules as a relative measure of their uptake probability. The mass peak intensity depends on the following factors: (1) the ionization probability of the cluster, (2) the fragmentation after the ionization, and (3) the detection probability of the ion fragment. (1) For large clusters, the ionization cross section can be approximated by the geometrical cluster cross section of Bz. The number of molecules picked up by the clusters is significantly smaller than the cluster size N. Thus, the cross section does not change significantly upon uptake of the molecules, and the ionization cross section of the cluster after the pickup can be assumed to be proportional to the Bz geometrical cross section independent of the adsorbed molecules. (2) According to the mass spectra, the fragmentation pattern is similar for all of the adsorbed molecules. In addition, the fragmentation is not critical as long as all of the ions containing the molecules (H2O, MeOH, and EtOH) and their fragments are accounted for in the signal integration. In this respect, also mass coincidences between fragments containing the molecules do not represent any problem. On the contrary, the mass coincidences between the ions containing the molecules BzM+ and the clean Bz+ ions can introduce errors. There are such coincidences; e.g., three Bz molecules correspond to 13 H2O molecules (see the Supporting Information). Thus, we have included possible uncertainties in the error bars. (3) Finally, it is reasonable to assume that the detection probability of a BzM+ ion does not depend significantly on the kind of molecule M.

In addition, we have to consider the possibility that the molecules M can evaporate from the cluster upon ionization yielding the clean Bz+ peaks and contributing to IC rather than to IM. Nevertheless, even this process would not change the fact that the ratio Rup for the different molecules reflects their relative pickup probabilities, provided that the evaporation of the molecule is similar for all of the species, which is a reasonable assumption in view of their similar binding energies with benzene discussed below.

First, we discuss the uptake of different molecules quantified in Table . Within the relatively large error bars, given especially by the uncertainties in the pickup pressure determination and the mass coincidences, the uptake probabilities for different molecules are similar. Nevertheless, the observed trend in Rup suggests that the uptake probability for water is by 30% smaller than that for MeOH and EtOH, which are almost the same.

The first reason could be that the pickup probability reflects the sticking of the molecule to the cluster in the uptake process, and thus, it would correlate with the energy for the binding of the molecule to benzene. Indeed, the H2O–Bz binding energy is smaller than the MeOH– and EtOH–Bz binding energy (see Table ). However, the differences are smaller than the uncertainty of the theoretical estimates; therefore, we consider other factors below.

Table 2

Binding Strengths (dissociation energies) of Dimers in Kilojoules per Molea

molecule	Ar	Bz		H2O		MeOH		EtOH
H2O	1.7b	13.8e	10.6h	22.8,i 12.2–20.8j	18.5k	19.4–23.6l	21.0k	22.4k	22.4k
MeOH	1.5c	17.0f	12.9h	19.4–23.6l	21.0k	15.0–22.8	21.4k	–	–
EtOH	d	4.8g	13.1h	22.4k	22.4k	–	–	15.1–23.3k,m	23.5k

The first column of each series compiles data from the literature, and the second column lists the theoretical estimates obtained by the same procedure for all systems.

From ref (50).

From ref (51).

Not found.

From ref (52).

From ref (53).

From ref (54).

This work, the same procedure as in footnote k.

Experimental.

From ref (55).

From ref (49).

From ref (48).

From ref (56).

Rup is proportional to the cluster cross section. The actual size of the molecule factors into the evaluation of the cross section via the equation σ = π(RC + rM)2, where RC is the cluster radius and rM is the van der Waals radius of the molecule. In addition, below, we evaluate the mean cluster radius R̅ of ≈2.2 nm from the Poisson distribution of the picked up molecules. The cluster radius is the dominant term; however, the radius of the molecule can make a small difference in the cross section. Considering rM values of 0.19, 0.25, and 0.28 nm for H2O, MeOH, and EtOH, respectively, and RC = R̅, the ratio of the cross sections for the three molecules would be 0.93:0.98:1. Thus, this correction points in the right direction with the cluster collision cross section for water being the smallest, though not by ∼30% compared to those of MeOH and EtOH. Thus, none of the factors considered above could account for the 30% lower value of Rup for water with respect to MeOH and EtOH quantitatively; nevertheless, they follow the observed qualitative trend.

The pickup of molecules enables us to determine the actual size of the large cluster. The uptake is supposed to follow Poisson statistics.[25] Thus, information about the pickup cross sections of the clusters can be evaluated from the mass spectra.[26,27] Let us represent the cluster by a sphere with cross section σ. As it flies through the pickup cell of length L (17 cm) filled with gas at pressure p and temperature T, it collides with the molecules, which can stick to the cluster. According to the Poisson distribution, the probability that the cluster collides with k molecules equalsThe cluster ion intensities of M+ will follow the Poisson distribution P assuming that all k molecules stick to Bz upon collisions and coagulate to a cluster M on it and that the ionization does not result in any fragmentation of M. Thus, we could plot the adsorbed molecular cluster ion intensities as a function of their size k and fit them with eq to derive pickup cross section σ. A good agreement of these fits provides support for the stringent assumptions mentioned above. It should be noted that the approach described above assumes that all of the clusters can be represented by the mean pickup cross section corresponding to mean size N̅. Large clusters have log-normal size distributions, and thus, in principle, a convolution of the log-normal and Poisson distributions should be used in the analysis.[28] Nevertheless, for practical purposes, the Poisson distribution has been mostly considered in the literature.[25−27,29−33] Therefore, we find the use of the Poisson fits in this case justified.

Figure shows the dependencies of the MH+ and Bz·M+ ion intensities on k for the pickup of methanol (a and b) and ethanol (c and d) on benzene nanoparticles. Note that for the protonated pure clusters, i.e., (MeOH)H+ and (EtOH)H+ (panels a and c, respectively), the abscissa represents k + 1 rather than k, because we assume that the protonated fragment originated from the neutral cluster that was larger by one molecule, i.e., (MeOH) → (MeOH)H+. In the right column, we show also the distribution of the Bz-containing fragments Bz·M+. Here we plot the distribution as a function of k, because we assume that the fragmentation after ionization proceeds through benzene evaporation. Similar distributions could also be found for series with more than one Bz molecule (see the Supporting Information).

Figure 2

Cluster ion fragment abundances (symbols) fitted with Poisson distributions (lines). The left panels show the protonated cluster ions of the adsorbed molecules (a) MeOH and (c) EtOH. The right panels show the analogous series with one Bz molecule Bz·M+.

(MeOH)H+ and (EtOH)H+ and the corresponding Bz-containing series follow the Poisson distributions, and we can evaluate the pickup cross sections σ from the fits. Assuming a spherical cluster shape, the corresponding nanoparticle radius R can be evaluated from the equation σ = πR2. In the case of water, the (H2O)H+ ions do not quite follow the Poisson distribution (see the Supporting Information). Clearly, the stringent assumptions are not fulfilled here. Nevertheless, for MeOH and EtOH ions [and even for Bz·(H2O)+ (see the Supporting Information)], the fits provide consistent σ values yielding a mean cluster radius R̅ of ≈2.2 ± 0.2 nm. Because the same cluster beam was used in all experiments, we assume that this represents the mean cluster radius of the Bz clusters implemented in these pickup experiments.

Assuming the benzene density ρ = 876 kg m–3, the sphere with an estimated radius of 2.2 nm would contain ∼300 Bz molecules. We can compare this value to the estimate from classical molecular dynamics simulations of clusters containing 200 and 300 Bz molecules. We perform the simulations at 150 K (the cluster temperature is discussed below) employing the OPLS-AA force field,[34,35] which is recommended for condensed phase benzene.[36]Figure shows a normalized distribution of carbon and hydrogen atoms as a function of the distance from the cluster center of mass. The shape of the curves corresponding to the C and H atom is similar, which indicates the preference of a flat or just slightly tilted orientation of the benzene molecules on the cluster surface. The curves drop to half of the cluster density at 1.8 nm for 200 molecules and 2.1 nm for 300 molecules. The latter value is in good agreement with the mean cluster radius derived from the experiment; thus, we can conclude that our neutral clusters are composed of ∼300 Bz molecules.

Figure 3

Radial distributions of C and H atoms from the center of mass of Bz clusters with 200 and 300 Bz molecules. The distributions were normalized on R2.

It should be mentioned that the cluster pickup cross section does not necessarily coincide with the geometrical cross section of the sphere representing the cluster. This has been discussed in detail in our previous work on water clusters, where the measured pickup cross sections were larger than the geometrical cross sections due to the attractive interactions between polar and polarizable molecules and clusters.[37,38] In our case, the clusters are composed of nonpolar benzene molecules; therefore, we do not expect the polarization forces to play such an important role.

The measured pickup cross section involves the probability of the collision of the cluster with the molecule as well as the probability of the molecule sticking to the cluster. To evaluate the pickup cross section from the Poisson distribution, we have assumed that the sticking probability equals 1. If this is not the case, the evaluation procedure is still valid, but the evaluated quantity is the product of the pickup cross section and the sticking coefficient.

The complete coagulation of the adsorbed molecules on Bz clusters is assumed to justify the observed Poisson distribution. This requires the molecules to be highly mobile on Bz and coagulate to clusters within the 3 ms flight time from the pickup chamber to the ionization region. This is interesting to compare with the pickup of different molecules on other clusters, e.g., Ar and (H2O).[39−41] We observed coagulation of different adsorbed molecules on Ar, while the same molecules adsorbed on ice (H2O) nanoparticles were less mobile and did not coagulate. Most recently, these observations have also been supported by a combined experimental and theoretical investigation of H2O2 molecules on Ar and (H2O).[42] Nevertheless, we have observed coagulation of some molecules on ice nanoparticles, as well.[43,44]

In this respect, the benzene clusters are similar to argon rather than water ice nanoparticles; the embedded molecules coagulate on benzene readily as they do on argon clusters. This can be justified by different binding strengths of the picked up molecule with the cluster constituents and between the picked up molecules (M···M). The binding energies are summarized in Table . The M··· Ar interaction is much weaker than that between the hydrogen bonding molecules themselves. Thus, the hydrogen bonding molecules move on the Ar cluster surface until they encounter another hydrogen bonding partner. On the contrary, the alcohols picked up by water clusters are stabilized by two to three hydrogen bonds,[45−47] which corresponds to ∼40 kJ mol–1.[48] Therefore, they are not likely to move and find other alcohol molecules, even though the dissociation energies of (MeOH···H2O) and (EtOH···H2O) heterodimers are slightly lower than those of the corresponding alcohol homodimers.[49]

The differences in binding strengths of the (M···Bz) dimers and (M···M) dimers are not as pronounced as for the (M···Ar) case. Upon comparison of the available theoretical estimates, it is important to take into account the method, the size of the basis set, and whether thermal effects were included.[55] Therefore, we have calculated all of the (M···Bz) binding strengths in the third column of Table by the same method that was used in ref (49). The selected values (10.6, 12.9, and 13.1 kJ mol–1) show that the (M···Bz) binding strength is similar for all molecules M and increases in the following order: H2O < MeOH < EtOH. The binding of the heterodimer is 8–9 kJ mol–1 weaker than the (M···M) binding.

The coagulation of the molecules M on Bz clusters is also supported by the tendency of M to form hydrogen-bonded chains or cycles attached to the aromatic ring.[53,57,58] In these structures, each of the M molecules forms two hydrogen bonds, not just one as in the dimer case.

From the point of view of the uptake process, the important question is the state of the benzene clusters: are they liquid or solid? This is indeed dependent on their temperature. In general, clusters in molecular beams generated by supersonic expansions attain low temperatures, e.g., in extreme 0.37 K for large superfluid He nanodroplets,[59] ∼40 K for large Ar clusters,[60,61] 50–180 K for water ice nanoparticles,[62] 90–135 K for methanol clusters,[63] etc. However, the cluster temperature is difficult to determine experimentally, and various models were proposed. A simple estimate is outlined in the Supporting Information. Earlier electron diffraction studies of benzene clusters in supersonic jets revealed a cluster vibrational temperature on the order of 100–150 K.[64] Electron diffraction experiments by Heenan et al. suggest that the benzene molecules in the clusters are not organized into regular arrays and the local order is similar but not identical to that in the solid, resembling that expected for a supercooled liquid. Stace et al. addressed the question of the cluster state by infrared spectroscopy of relatively small clusters (n < 15) and concluded that these clusters are best described in terms of a liquid-like rather than a solid-like state.[65]

Our simulations shown in Figure have been performed at 150 K and agree with a noncrystalline disordered benzene cluster structure. We could speculate about whether the maxima in the radial distribution of atoms might correspond to certain crystalline-like layers or shell closures as known, e.g., for argon clusters. However, the molecules at the cluster surface are not organized into regular arrays. Many clusters, e.g., water, have a crystalline core covered by a quasi-liquid layer (QLL).[66,67] Recent simulations together with our pickup experiments[42] have demonstrated that the adsorbed molecules on large argon clusters can submerge into the surface layer but remain very mobile, finding each other quickly and forming clusters in <1 μs. In this respect, our experiments suggest that the molecules behave similarly on a benzene cluster; i.e., the benzene clusters are liquid-like or possess a QLL, and the embedded molecules are highly mobile and coagulate to form the hydrogen-bonded molecular clusters.

Several conclusions can be drawn from these experiments and calculations.

(1) The cluster ion fragment abundance follows the theoretical Poisson distribution, which allows us to evaluate the nanoparticle radius R of ≈2.2 Å. This radius is consistent with the Bz clusters composed of about N̅ ≈ 300 benzene molecules according to our theoretical simulations.

(2) All adsorbed molecules, water (H2O), methanol (CH3OH), and ethanol (CH3CH2OH), are highly mobile on Bz and generate clusters within <3 ms.

(3) The relative uptake for water, methanol, and ethanol can be directly compared using the Rup values evaluated from the mass spectra: Rup = 0.21, 0.30, and 0.29 for H2O, MeOH, and EtOH, respectively. They are discussed in terms of different binding energies and radii of the molecules, and the qualitative trend can be substantiated with water exhibiting the lowest relative pickup probability.

The measurements of the relative uptake probability for different molecules on benzene clusters presented here represent the very first experimental estimates of this quantity. The investigated pickup and coagulation processes might be important for the condensation of water on PAH aerosols in the atmosphere. Additionally, it can be related to the initial steps of the growth of ice mantles on PAH particles in the interstellar medium. The probability of uptake of water molecules is lower than that of methanol. Thus, methanol ice mantles can be generated in some regions despite methanol being less abundant than water molecules. The experiments presented here pave the road to future investigations with nanoparticles of more complex PAHs and pickup of other atmospherically and astrochemically relevant molecules.