Al13− and B@Al12− superatoms on a molecularly decorated substrate

Aluminum nanoclusters (Aln NCs), particularly Al13− (n = 13), exhibit superatomic behavior with interplay between electron shell closure and geometrical packing in an anionic state. To fabricate superatom (SA) assemblies, substrates decorated with organic molecules can facilitate the optimization of cluster–surface interactions, because the molecularly local interactions for SAs govern the electronic properties via molecular complexation. In this study, Aln NCs are soft-landed on organic substrates pre-deposited with n-type fullerene (C60) and p-type hexa-tert-butyl-hexa-peri-hexabenzocoronene (HB-HBC, C66H66), and the electronic states of Aln are characterized by X-ray photoelectron spectroscopy and chemical oxidative measurements. On the C60 substrate, Aln is fixed to be cationic but highly oxidative; however, on the HB-HBC substrate, they are stably fixed as anionic Aln− without any oxidations. The results reveal that the careful selection of organic molecules controls the design of assembled materials containing both Al13− and boron-doped B@Al12− SAs through optimizing the cluster–surface interactions.


Supplementary Note 2 Synthesis and deposition of Aln and AlnBm NCs.
The mass-selected Aln  or AlnBm  NCs were deposited on organic substrates, in which C60 and HB-HBC molecules were pre-decorated prior to the NCs deposition 2,3,4 . A substrate of highly oriented pyrolytic graphite (HOPG) was cleaved in air and was heated at ~700 K for 50 h in an ultrahigh vacuum (UHV) condition (<3  10 8 Pa) to remove surface impurities. The C60 (Sigma Aldrich, sublimed, 99.9%) and HB-HBC (see Supplementary Note 1) powders separately loaded in quartz effusion cells were degassed with heating in the UHV system before evaporations. During the evaporations, the thickness of C60 and HB-HBC was monitored by a quartz microbalance (INFICON, IC5) to obtain the amounts of 2 ML (C60) and 5 ML (HB-HBC), respectively.
Aln or AlnBm NCs were generated in the MSP system (Ayabo Corp. nanojima-NAP-01), 25 in which the pure Al or mixed Al-B targets (Rare Metallic. Co., LTD.) were sputtered with Ar + ions in the aggregation cell filled with a cooled (77 K) He buffer gas. Formed Aln (or AlnBm) NC ions were guided through a radio frequency octupole ion guide to an ion bender for charge selection. The negatively charged species (i.e. Aln  or AlnBm  NC ions) bended were introduced into a quadrupole mass filter (Extrel CMS; MAX-16000). Since the production of Aln  /AlnBm  NCs sensitively depends on the conditions of DC power to the MSP and the flow rates of sputtering Ar and buffer He gases, they were optimized to maximize the ion intensities at the targeted m/z ratios by monitoring the mass spectra of Aln  /AlnBm  NCs (see Supplementary Figs. 2 and 8).
The mass-selected Aln  (or AlnBm  ) NCs were deposited on the prepared C60 and HB-HBC substrates with the mass resolution of m/m ~70 during the deposition which was enough to exclude neighboring minor products (see Supplementary Figs. 2 and 8). To achieve a "soft-landing condition" in the NC deposition, a positive bias voltage of +5 V was applied to the substrates to avoid decomposing NCs at the surface deposition. The number of deposited Aln  (or AlnBm  ) ions was counted through an ion current, in which a typical ion current of 300 pA (~1.9 × 10 9 NC ions per second) was generated. The sample temperature during the deposition was kept at 300 K.
Supplementary Fig. 2 Mass spectrum for the Aln anions (n = 8-50). With an Al disk target, anionic Aln − nanoclusters (NCs) were formed by magnetron sputtering, and the mass distributions were measured using a quadrupole mass spectrometer. Aln − NCs are formed with ion currents of a few hundred picoamperes (pA) for n = 7-25, and a specific size of Aln − was mass-selectively deposited on a molecularly decorated substrate.  Fig. 1). This means that the HB-HBC film is grown in a flat conformation on the HOPG substrate. Bright dots correspond to the deposited Lu@Si16, showing the successful immobilization of nanoclusters in a monodisperse manner. Note that Lu@Si16 (67 e − valence electrons) is a halogen-like superatom, 5,6 as is Al13, owing to the lack of a single electron to close the electron shell. The imaging conditions of the tip voltage and the tunneling current are −2.0 V and 0.2 pA for (a), and −2.0 V and 0.5 pA for (b), respectively.
Supplementary Note 3 Evaluation of peak shifts of C 1s for Al13/C60 and Al13/HB-HBC.
To discuss the XPS results more quantitatively, peak deconvolutions were performed as follows: (1) The C 1s peak of the highly oriented pyrolytic graphite (HOPG), which was used as a base substrate and exhibits an asymmetric structure, is fitted with a Doniach−Šunjić line shape 7 , and its line shape and energy are fixed as "G." (2) The C 1s peak of C60 or HB-HBC is fitted by the "G" component and an additional peak component of C60 ("F") or HB-HBC ("H"), where the fitting parameters for "F" and "H" are given by the Voigt function. (3) In the deconvolution of the C 1s peaks for Al13/C60 and Al13/HB-HBC, a new peak component, originating from the C atoms in C60 ("N") or HB-HBC ("P") that interact with the deposited Al13, is added using the same contribution ratio of "G" to "F" or "H" obtained in (2). The above deconvoluted components are schematically illustrated in Supplementary Fig. 4. The binding energy (BE) of the extracted component "N" is 0.33 eV (284.75 eV) lower than that of "F," while the BE of the extracted component "P" is 0.50 eV (284.70 eV) higher than that of "H".
Supplementary Fig. 4 XPS C 1s peaks for Al13/C60 and Al13/HB-HBC. (a) XPS C 1s peak for Al13/C60/HOPG, which is deconvoluted into three peak components for the HOPG base substrate (G, gray area), the non-interacted C60 (F, green area), and the interacted C60 with Aln oxide (N, blue area); the binding energy (BE) of the interacted C60 peak is 0.33 eV lower than that of the non-interacted C60; the shift amount of 0.33 eV corresponds to the formation of C60 − as described in the literature 8 . (b) XPS C 1s peak for Al13/HB-HBC/HOPG, which is also deconvoluted into three peak components for HOPG (G), the noninteracted HB-HBC (H, green area), and the interacted HB-HBC with Al13 (P, orange area); the BE of the interacted HB-HBC peak is 0.50 eV higher than that of the non-interacted HB-HBC. The shift corresponds to the formation of a cationic HB-HBC + state, suggesting that Al13 forms a CT complex, such as Al13 − /HB-HBC + .      Supplementary Fig. 9 Oxidative behaviors in the Al 2p XPS spectra for B@Al12 on HB-HBC. With an increasing O2 exposure, the intensity of the zerovalent component Al 0 peak decreases, while that of the oxidized component Al 3+ increases accordingly. The oxidative reactivity can be evaluated by the dependence of the O2 exposure amount from 0 L to 1 × 10 4 L. At the highest exposure of 5 × 10 10 L, the Al 2p peak shifts to a lower BE, likely owing to a structural change relevant to the phase transition of aluminum oxide (see the manuscript text for further details).

Supplementary Note 5 Energetics for charge transfer complexation.
For the molecular complexation between Al13/B@Al12 and organic molecules, the energy balance toward a charge transfer (CT) is evaluated from the ionization energies (Ei) and the electron affinities (EA) of Al13/B@Al12, C60, and HB-HBC. Using these values, the endothermic dissociation limits of the corresponding cations and anions, E, can be calculated when one electron is transferred between neutral Al13/B@Al12 and organic molecules 10 , as shown schematically in Supplementary Fig. 12.
The experimental EA of Al13/B@Al12 has been previously measured using anion photoelectron spectroscopy in the gas phase 11,12 . However, no accurate experimental Ei for B@Al12 has been reported, with only a rough estimation being available 13 ; the Ei for B@Al12 is less than 7.90 eV (157 nm, F2 laser) and more than 6.42 eV (193 nm, ArF laser). In contrast, both the EA and Ei values have been theoretically obtained using DFT calculations 1417 . Although there are some discrepancies between the experimental and calculated values, the calculated Ei and experimental EA were adopted for the estimation for Al13/B@Al12. These values are tabulated in Supplementary Table 2.
The experimentally determined values of Ei and EA of C60 have been reported to 7.57 eV 18 and 2.683 eV 19 , respectively. However, the corresponding values of HB-HBC molecules have not been reported experimentally or theoretically. Thus, they were evaluated from the HOMO and LUMO energy levels, which were experimentally observed from ultraviolet photoelectron spectroscopy (UPS) and two-photon photoelectron spectroscopy (2PPE) for the HB-HBC thin film. In combination with an approximate value of the polarization energy of 1.7 eV obtained for many polycyclic aromatic hydrocarbons (PAHs) 20 , the Ei and EA of the HB-HBC molecules were evaluated. Supplementary Fig. 10 shows the UPS spectrum for an HB-HBC thin film, wherein the HOMO level was observed at 5.9 eV from the vacuum level. On the other hand, Supplementary Fig. 11 shows the 2PPE spectra for the HB-HBC thin film, wherein the LUMO level was observed at 2.9 eV from the vacuum level. Since the polarization energy, i.e., the stabilization energy of ions by the surrounding molecules in the thin film, is 1.7 eV for typical PAHs, the Ei and EA values for an HB-HBC molecule were determined to be 7.6 eV and 1.  Table 3); the values used for this estimation are underlined in Supplementary Table 2. On C60, therefore, it is favorable for both Al13 and B@Al12 to become cations combined with C60 anions, because their E values are smaller than those for the opposite charge combinations. On HB-HBC, however, the anions of Al13 and B@Al12 more favorably combine with the HB-HBC cations due to the smaller E values resulting from these interactions. The HB-HBC molecule exhibits an opposite behavior to that of C60 because the Ei value of HB-HBC is smaller than that of C60, and because the EA value of HB-HBC is 1 eV lower than that of C60.
By comparing the absolute values of E, it can be seen that the E for B@Al12 on HB-HBC is larger (4.47 eV) than the other values, because CT complexation overcomes the endothermic energy gap even between B@Al12 on HB-HBC. In other words, the stabilization due to CT between B@Al12 and HB-HBC is considered larger than 4.47 eV. Considering a point charge model, when the positive charge and the negative charge approach to a distance (r) of 3 Å, the stabilization energy (= e 2 /((40)r)) is calculated to be 4.8 eV, which is actually larger than the endothermic gap between B@Al12 and HB-HBC. Moreover, the CT complex on a C60 or HB-HBC substrate is further stabilized by the polarization of the C60 or HB-HBC surroundings. On the C60/HB-HBC substrate, the polarization energy of charged C60/HB-HBC on the topmost is estimated to be ~0.85 eV, which is half of the polarization energy of the corresponding negative or positive polaron in the organic film (1.7 eV). Furthermore, similar to well-known CT complexes 21,22 , the presence of covalent bonds due to molecular orbital overlapping should also contribute to the interactions between Al13/B@Al12 and C60/HB-HBC, thereby allowing the endothermic gap to be seemingly overcome.
This study shows that by covering the surface with organic molecules, the molecularly localized interactions with the nanoclusters are emphasized, and the cluster-surface interactions can be designed by tuning of the physical properties of the deposited organic molecule. Although it is essential to select a suitable substrate, designing the interaction by the appropriate selection of an organic molecule is a more strategic means to fabricate nanoscale heterointerfaces and nanocluster-based assemblies on a surface. Supplementary Fig. 12 Schematic potential curves for the complexation between Al13/B@Al12 and C60/HB-HBC. When B@Al12 is deposited onto C60 (left-hand side) or HB-HBC (right-hand side) molecular layers, charge transfer molecularly takes place between them, where the charged C60 or HB-HBC is further stabilized by the polarization of the surroundings. The energy gap, E, represents the endothermic dissociation limits of the corresponding cations and anions against the neutral B@Al12 and C60 or HB-HBC. Instead of B@Al12, these schematic energetic representations can also be applied to Al13.
Supplementary Table 2 Experimental and calculated ionization energies (Ei) and electron affinities (EA) of Al13, B@Al12, C60, and HB-HBC in eV. The underlined values are used for estimation purposes.