Acid-Base Control of Valency within Carboranedithiol Self-Assembled Monolayers: Molecules Do the Can-Can

Abstract

We use simple acid-base chemistry to control the valency in self-assembled monolayers of two different carboranedithiol isomers on Au{111}. Monolayer formation proceeds via Au-S bonding, where manipulation of pH prior to or during deposition enables the assembly of dithiolate species, monothiol/monothiolate species, or combination. Scanning tunneling microscopy (STM) images identify two distinct binding modes in each unmodified monolayer, where simultaneous spectroscopic imaging confirms different dipole offsets for each binding mode. Density functional theory calculations and STM image simulations yield detailed understanding of molecular chemisorption modes and their relation with the STM images, including inverted contrast with respect to the geometric differences found for one isomer. Deposition conditions are modified with controlled equivalents of either acid or base, where the coordination of the molecules in the monolayers is controlled by protonating or deprotonating the second thiol/thiolate on each molecule. This control can be exercised during deposition to change the valency of the molecules in the monolayers, a process that we affectionately refer to as the “can-can.” This control enables us to vary the density of molecule-substrate bonds by a factor of two without changing the molecular density of the monolayer.

Graphical Abstract

Self-assembled monolayer (SAM) formation is driven by a combination of substrate-molecule interactions, molecule-molecule interactions, and molecule-environment interactions.^1–3 The most commonly studied SAMs, n-alkanethiolates on Au{111} contain a single thiol group available for substrate binding, have linear backbones, resulting in numerous defects that originate from gauche defects in the alkyl chains, different alkyl tilt orientations, translational and rotational lattice registry offsets, and reconstruction of the underlying substrate.^1,4–9 Monolayers formed from dialkyl disulfides on Au{111} result in identical assemblies as the Au surface cleaves the S-S bond.^10–12 In contrast, in unfunctionalized carboranethiol SAMs, the molecules do not tilt nor can they change their conformation; thus, there are fewer and simpler defects in comparison to SAMs composed of n-alkanethiols.^2–17 Interactions between carboranethiol molecules at both exposed and buried interfaces have been observed. Carboranethiol isomers with nonzero components of their dipoles parallel to the surface exhibit long-range attractive interactions due to dipole alignment.¹⁸ This phenomena was previously inferred from the results of competitive adsorption experiments, where carboranethiol isomers with larger in-plane dipole components outcompeted those with greater out-of-plane components.¹⁹ Mixed assemblies of carboranethiol isomers can be used to tailor the effective metal work function of noble metal surfaces while not changing the wetting properties of the overlying polymers and thus not changing their morphologies.^15,20 These interactions have been observed with submolecular resolution, where correlations between simultaneously acquired scanning tunneling microscopy (STM) topographic and local barrier height (LBH) images enabled the observation of single-molecule orientations within SAM matrices and demonstrated defect-tolerant dipole alignment.¹⁶ The surfactants 9,12-carboranedithiol (9O12) and 1,2-carboranedithiol (1O2) have been shown to be stable isomers, to functionalize noble metal surfaces, and to modify effective metal work functions due to oppositely oriented dipoles originating from the carborane backbone.^21–26 Carboranethiol SAMs can also be used to align overlying liquid crystal orientations.²⁷ Here, both isomers of carboranedithiol studied (1O2 and 9O12) promote formations with higher sulfur-surface coverage and fewer defects due to rigid, nearly spherical backbones.

Typical SAM formation from thiols is governed by simple acid-base reactions, where an acidic thiol group (SH) is deprotonated to a thiolate (S⁻) on reactive surfaces.^4,28 As noted above, disulfides can form SAMs by cleaving the disulfide bonds, again leading to adsorbed thiolates.^10–12 Adsorbed thiolates form stable bonds to Au surfaces (~45 kcal/mol), stronger than typical Au-Au bonds.^3,29–31 The S-Au bond is partially ionic and partially covalent; the effective charge transfer depends on the backbone attached to the S.³² Adsorbed, but relatively labile, thiols (i.e., with protons still in place) have also been observed within SAMs, such as p-carboranethiol, on Au{111} at low conentrations.¹⁷ The ease of formation coupled with tunable defect formations contribute towards thin films with controllable intermolecular interactions and modifiable surface-mediated effects that can be used to advantage to place and to direct single molecules and supramolecular assemblies into controlled environments.^2,3,12,13,33 Acid-base chemistry, at the exposed interface, has shown broad importance in surface wettability, colloid and emulsion stability, biological signal transduction and membrane assembly, and catalysis.^34–37 We sought to manipulate and to control the valency, the molecule-substrate bond density, and subsequent monolayer formation using different isomers of carboranedithiol on Au surfaces, both of which have the potential for one or two bonds to the substrate per molecule. Both 1O2 and 9O12 are stable in their respective dithiol and dianion states, where 1O2 is a stronger acid compared to 9O12.²⁶ The acid-base properties intrinsic to homogenous monolayers composed of either 1O2 or 9O12 can be used as a means of control.

Kitagawa and coworkers assembled adamantanetrithiol on Au{111} and found trivalent interactions that resulted in small chiral cluster structures on the surface.³⁸ In earlier unpublished work, we found that it is important to take into account both bond angles and substrate access in forming multivalent molecule-substrate interactions.^39,40 The lessons learned from those studies resulted in the inclusion of flexible linkers in attaching caltrops and other molecules to surfaces.^41–44

Control of thin-film properties of both exposed and buried interfaces has broad implications, e.g., for molecular devices and lithographic patterning.^{3,29,33,35,45,46} These isomers of o-carboranedithiols serve as ideal test candidates for binding, valency, and surface bond density, where singly bound (monovalent) modes produce monolayers with lower thiolate-Au bond densities in comparison to doubly bound (divalent) modes. We demonstrate, with a variety of surface-sensitive techniques and with density functional theory (DFT) calculations, that homogenous monolayers composed of either 1O2 or 9O12 show two distinct binding states, which are susceptible to manipulation by controlling pH, in solution, prior to deposition, as well as during deposition via exposure to acidic or basic solution; identical nearest-neighbor spacings and lattices are maintained throughout processing, independent of valency. We also show that the difference between the monovalent and divalent modes of these carboranes is not a simple geometric effect, but also critically depends on electronic effects.

Scanning tunneling microscopy is able to probe the topographic landscape of surfaces with sub-Ångström precision.^7,47–50 We have previously used spectroscopic imaging modalities of STM to probe buried interfaces and layers in molecular monolayers of other systems.^7,32,51–55

Results and Discussion

We assembled and measured monolayers composed of either 1O2 or 9O12 carboranedithiols on Au{111}/mica, where, upon imaging, two distinct binding states that differ in apparent height in each homogenous SAM are recorded (Figure 1). Both isomers form hexagonally close-packed monolayers with nearest-neighbor spacings of 7.6 ± 0.5 Å. This formation is best explained by a (√7×√7)R19.12° superstructure that is found for several carboranethiols on Au{111}.^19,56 We measured a bimodal distribution, with tunable coverages, which we attribute to both singly bound (monovalent) and doubly bound (divalent) states. Self-assembled monolayers formed of 1O2 from a neutral solution show 21 ± 8% coverage of the singly bound (greater apparent height in STM images) binding structure and, correspondingly, 79 ± 8% coverage of the doubly bound (lower apparent height) structure. These modes are differentiated as described below. Measurements over monolayers composed of 9O12 show a strong preference (98 ± 1%) for monovalent binding (less protruding, counter-intuitively, as explained below) in comparison to divalent (2 ± 1%, more protruding). Images can be segmented by apparent height in topographic images by applying a gray scale threshold value that is used to compute binding concentrations in image binaries (Figure S1). Monovalent structures in 1O2 monolayers have greater apparent heights (1.5 ± 0.3 Å with the tunneling conditions and tip used to record the data in Figure 1) than the divalent structures under the conditions measured; the monovalent structures in 9O12 SAMs have lower apparent heights (0.4 ± 0.2 Å with the tunneling conditions and probe tip used to record the data in Figure 1) than the divalent structures. The DFT calculations help explain this apparent inverted contrast for monovalent/divalent modes for 9O12.

Figure 1.

Scanning tunneling microscope images of (A,B) 1,2-(HS)₂-1,2-C₂B₁₀H₁₀ (1O2) on Au{111}/mica at two different resolutions (V_sample = −0.1 V, I_tunneling = 100 pA, T = 298 K). Insets depict fast Fourier transforms (FFTs) that indicate hexagonally close-packed arrangements with nearest-neighbor spacings of 7.6 ± 0.5 Å. Two distinct binding states are highlighted in red and black. (C) Binding modes are shown schematically, where 1O2 assembles into both monovalent (black box) and divalent (red box) modes. (D,E) Scanning tunneling microscope images (V_sample = −0.1 V, I_tunneling = 100 pA, T = 298 K) of 9,12-(HS)₂-1,2-C₂B₁₀H₁₀ (9O12) on Au{111}/mica at different resolutions. The inset depicts a FFT showing a hexagonally close-packed arrangement with the same spacing as 1O2. The two binding states are highlighted schematically in red and black boxes. (F) Binding modes for 9O12 are depicted schematically, where both monovalent (black box) and divalent (red box) modes are present.

We performed DFT calculations to gain detailed understanding of molecular chemisorption modes and their related appearances in STM images, and to help identify the adsorbed species. With this aim, we explored all the possible binding sites of singly and doubly dehydrogenated (i.e., monovalent and divalent, respectively) carboranedithiols on the Au{111}: face-centered cubic (fcc), hexagonally close-packed (hcp), over a bridge, and over a single atom (atop). The most stable species that are most closely correlated with the experimental results are shown in Figure 2. The geometric features of 1O2 and 9O12 are almost identical. The preferred binding site for the monovalent molecules is a fcc hollow site, which allows them to maximize the number of Au-S bonds. Divalent species adsorb preferentially on one fcc and one hcp hollow site, although somewhat off center with respect to the bridge site. Overall, S-Au bond-lengths are calculated to be between 2.37 and 3.06 Å. The natural bond order analysis shows that these are predominantly covalent bonds. For the most stable monovalent chemisorption mode, the S-C or S-B bond is simply normal to the surface. If one defines the molecular axis as connecting the center of the cage and the point halfway from one S atom to another, this line then is 28–30° from the surface normal in the monovalent binding mode. The z (surface normal) coordinates of the uppermost atoms (excluding H), also shown in Figure 2, reveal that the monovalent structures in both 1O2 and 9O12 extend further from the surface than their corresponding divalent species, each with a physical height difference of ca. 0.4 Å.

Figure 2.

Preferred structures/binding sites for 1O2 and 9O12 in their respective monovalent and divalent binding modes. The z coordinate of the uppermost atom (without considering H atoms) and the tilt angles of the molecules with respect to the surface normal (α) are shown for each case.

The simulated STM images for these four structures are shown in Figure 3. In agreement with experiments and in line with the geometry, monovalent 1O2 appears more protruding in STM images than divalent 1O2. The calculated apparent height difference in the simulated STM images is 0.6 Å. For STM images of 9O12, however, the situation is different; the monovalent mode appears less protruding, while the divalent structure appears more protruding. The height difference in the simulated images is 0.3 Å. As mentioned above, the z coordinates of the topmost atoms are higher in the monovalent than in the divalent 9O12. These results indicate that electronic effects play important roles in these systems and are critical to explain the differences between images of monovalent and divalent 1O2 and 9O12 on Au{111}.

Figure 3.

Simulated scanning tunneling microscope images for 1O2 and 9O12 in their respective monovalent and divalent binding modes, with the maximum height shown in the insets (in Å). The 10⁻⁴ isovalue for the charge density was used.

In order to understand this “inverted contrast” in the STM images for 9O12, we additionally plot the charge densities within the same energy window captured by the STM (from the Fermi energy (E_f) to (E_f - 0.1 eV). From the images in Figure 4, we observe that the contributions of C and B in these states are quite similar for monovalent and divalent 1O2 structures. Indeed, the relative apparent heights and the relative geometrical heights correlate well. For 9O12, the charge density plots for monovalent and divalent modes are significantly different. In the case of the monovalent structure, the contributions of the carborane cage atoms in the proximity of the Fermi level are significantly smaller, which gives rise to the calculated “inverted contrast”. The density plots in Figure 4 reflect the collective contributions of all electronic states that fall within the experimentally measured energy window (E_F - 0.1 eV). Examination of the electronic states and their energies across the entire Brillouin zone reveal that there are only a few (1-3) of these states for each species. Furthermore, for the monovalent 9O12, there is essentially only a single experimentally accessible state, the least among all considered adsorbed molecules. The projected density of states (PDOS) plots (reported in the Supplemental Information) are consistent with this finding. Since carboranes, in general, have similar delocalized chemical bonding, described by Lipscomb via three-center, two-electron bonds, and obey Wade’s electron-counting rules,⁵⁷ it is expected that the molecular orbitals on all the studied species are similar in structure, and only shift in energy from one species to another. Thus, we suggest that the small number of electronic states detected by STM for the monovalent 9O12 simply has to do with the energies of those states relative to the experimentally probed energy window of (E_F - 0.1 eV).

Figure 4.

Charge density plots from the Fermi energy to E_f - 0.1 eV for (top) 1O2 and (bottom) 9O12 in their respective (left) monovalent and (right) divalent binding modes. The images are plotted with an isodensity value of 0.00002, because most of the charge density is located on the gold surface. “Min” and “Max” stand for the minimum and maximum topographic ranges (in Å), taken from the simulated STM images.

The aggregation of domains in SAMs composed of 1O2 suggests intermolecular interactions between monovalent molecules, which possess larger in-plane dipole components due to binding geometry. The reported dipoles for 1O2 and 9O12 are 3.7 D and 5.5 D, respectively,²⁴ where this component is close the surface normal (8°) if both S are bound (as thiolates), however, the singly bound state is tilted −30° (vide supra), yielding a parallel component (~1.8 D for 1O2 and −2.8 D for 9O12). This dipole component results in increased intermolecular interactions between neighboring singly bound adsorbates¹⁶ and apparently leads to phase separation. This result is confirmed in samples of each monolayer and further enables binding assignments; SAMs composed of 1O2 demonstrate aggregation between higher protrusions, and, in SAMs composed of 9O12, higher intensity protrusions are localized and not phase aggregated. Both isomers form hexagonally close-packed monolayers with nearest-neighbor spacings of 7.6 ± 0.5 Å. This formation is best explained by a (√7×√7)R19.12° superstructure that is found for several carboranethiols on Au{111}.⁵⁶ One of the advantages of working with carboranethiol SAMs is the conservation of surface structure among isomers so that the effects of different interactions can be tested independently of structural variations.^13,19 Both isomers form monolayers with the same spacings and surface structures; however, each contains different concentrations of monovalent and divalent modes. Next, we tested each binding state by coupling STM with scanning tunneling spectroscopy (STS) to monitor both the exposed and buried interfaces.^7,16,54

Multimodal STM, which can simultaneously record the topographic landscape and the dipolar interface, can be used to extract molecular orientations within monolayers.^{7,16,32,54,58–60} Topographic and local barrier height extrema are computed within a defined radial vector (the size of one molecule) and correlations are computed via block-matching,^61,62 to associate symmetric molecular apexes with dipolar extrema.^7,16 Here, local maxima (inverted minima) within SAMs composed of 9O12 can be locally attributed to carbons at the 1- and 2-positions within the cage. Conversely, local maxima within SAMs composed of 1O2 can be attributed to the local dipolar offsets within the boron cage. Correlated topographic maxima to LBH extrema values, shown in Rose plots (Figure 5), indicate that greater apparent protrusions in SAMs composed of 1O2 have larger offsets (3.8 ± 1.0 Å) than lesser protrusions (2.2 ± 0.6 Å). In SAMs composed of 9O12, greater protrusions show slightly smaller offsets (1.9 ± 0.3 Å) in comparison to lesser protrusions (2.6 ± 0.6 Å). For the small areas in the data shown, orientations in each homogenous monolayer (283 ± 39° in 1O2 SAMs, 150 ± 33° in 9O12 SAMs with respect to the fast-scan direction, shown as horizontal, left to right, in these displayed images) suggest charge-separation stabilization and relative dipolar alignment across each two-dimensional landscape. Other orientations are found in other areas of the surfaces. These observations are consistent with our binding model (monovalent and divalent modes) in each homogenous monolayer. Since each binding state maintains the same nearest-neighbor spacing, the fraction of sulfur bound to the Au substrate can be tailored by up to a factor of two if binding can be manipulated. Motivated by the fact that each isomer is stable in both their dithiol and dithiolate states,²⁴ we modify the pH to deposit either the dithiol or the dithiolate selectively. In chemical lift-off lithography,^29–31 the amount of sulfur bound to Au may affect the amount and structure of lifted-off Au from the surface. Controlling the sulfur-surface density would enable tunable amounts of surface-bound Au available for patterning.⁶³

Figure 5.

(A) Scanning tunneling microscope images, recorded at low temperature (V_sample = −0.5 V, I_tunneling = 15 pA, T = 4 K) of 1,2-(HS)₂-1,2-C₂B₁₀H₁₀ (1O2) on Au{111} with local maxima (blue) highlighted. (B) Simultaneously acquired local barrier height (LBH) image, with LBH maxima (red) highlighted. (C) Rose plot (depicting correlated dipole offsets) that is binned by both magnitude (0.5 Å bins) and orientation (4° bins), and a ball-and-stick model of 1O2 showing thiol positions. (D) Scanning tunneling microscope image, recorded at low temperature (V_sample = −0.5 V, I_tunneling = 15 pA, T = 4 K) of 9,12-(HS)₂-1,2-C₂B₁₀H₁₀ (9O12) on Au{111} with local maxima (blue) highlighted. (E) Simultaneously acquired LBH map, with inverted LBH maxima (red) highlighted. (F) Rose plot (depicting correlated dipole offsets) that is binned by both magnitude (0.5 Å bins) and orientation (4° bins), and a ball-and-stick model of 9O12.

We tested the resulting assemblies of each carboranedithiol in both basic (2:1 NaOH:carboranedithiol) and acidic conditions (1:1 HCl:carboranedithiol) with STM. Upon deposition under basic conditions, majority divalent binding is achieved that is attributed to deposition of the molecular dianion (dithiolate) state. Scanning tunneling microscope images depict a concentration change of each phase, and show predominance of the divalent mode in monolayers fabricated under basic conditions (Figure 6). Images are segmented by apparent height to compute percent coverage (Figure S2). The divalent mode is dominant in each single-component monolayer (98 ± 2% for 1O2, 99 ± 1% for 9O12) under basic conditions. The small fractions of the monovalent mode should appear more protruding for 1O2 but less protruding for 9O12. The possible small fractions of monovalent species after deposition from basic solution may be difficult to distinguish from defects and the monolayers, in any case, are within experimental error of being completely divalent.

Figure 6.

(A,B) Scanning tunneling microscope images (V_sample = −0.1 V, I_tunneling = 100 pA, T = 298 K) of 1,2-(HS)₂-1,2-C₂B₁₀H₁₀ (1O2) on Au{111}/mica at two different resolutions after basic deposition conditions (2:1, NaOH:1O2). Insets depict fast Fourier transforms (FFTs) that show hexagonally close-packed arrangements with the same nearest-neighbor spacings as in Figure 1. (C) A majority of the 1O2 molecules have divalent surface attachment, depicted schematically. (D,E) Scanning tunneling topographs (V_sample = −0.1 V, I_tunneling = 100 pA, T = 298 K) of 9,12-(HS)₂-1,2-C₂B₁₀H₁₀ (9O12) on Au{111}/mica at different resolutions after basic deposition (2:1, NaOH:9O12). Inset depicts a FFT showing the same arrangement and spacing as 1O2. (F) A divalently bound 9O12 molecule is shown schematically.

We also performed deposition experiments in acidic conditions and measured the resulting monolayers. Saturating carboranedithiol solutions with excess protons, prior to deposition, enables somewhat higher concentrations of monovalent (thiol/thiolate) molecules, in comparison to their counterparts deposited from neutral solutions. Figure 7 depicts the topographic environment after acidic deposition, where an increase in (greater apparent protrusions) in SAMs composed of 1O2 indicate higher concentrations (31 ± 3%) of monovalent (thiol/thiolate) species; SAMs composed of 9O12 already show an almost total monovalent mode (lesser protrusions) at neutral pH and hence do not show any change for a proton-rich deposition environment (98 ± 1%). In summary, SAMs composed of divalent 1O2 or 9O12 (basic conditions) enable high dithiolate surface attachment, where SAMs composed of monovalent 9O12 (neutral or acidic conditions) permit low sulfur-surface attachment simply due to the controllable binding within structurally equivalent isomeric monolayers.

Figure 7.

Binding assignments measured by scanning tunneling microscopy and scanning tunneling spectroscopy (STS). (Top) Self-assembled monolayers composed of 1,2-(HS)₂-1,2-C₂B₁₀H₁₀ (1O2) form into (left) a thiol/thiolate state (monovalent form) and (right) a dithiolate state (divalent form) that differ in apparent height in scanning tunneling microscope images. (Bottom) Monolayers composed of 9,12-(HS)₂-1,2-C₂B₁₀H₁₀ (9O12) form into (left) an adsorbed monothiol/monothiolate (monovalent) state, under acidic and neutral conditions and (right) a dithiolate (divalent) state that differ in measured apparent height, as shown for the conditions shown in figure 1. Upon deposition under basic conditions, majority dithiolate binding is observed for both isomers, as determined by STM imaging.

In order to monitor the results of SAM formation in each environment at the ensemble scale and to complement the STM measurements, we used both X-ray photoelectron spectroscopy (XPS) and Fourier transform infrared (FTIR) spectroscopy. All measured XPS and FTIR results, and modeled values are detailed in Table 1 and Table S1, using the sulfur and similar B-H vibrational spectra under all conditions also measured with STM. Characteristic orbital energy peaks of each atomic species within the monolayer can be measured with XPS. Our results are consistent with prior XPS measurements in both monolayers deposited at neutral pH, where S 2p orbital peaks show shifts similar to previously reported values.²¹ To test monolayer integrity, we also measured each monolayer after pH manipulation (Figures S4 and S5). Values reported for XPS not only show the retention of S 2p shifts, but also confirm the lack of any measured Cl⁻ or Na⁺ within each monolayer after acid or base treatment. Using FTIR, we specifically tracked the B-H vibrational stretch, at ~2600 cm⁻¹, which is characteristic for carboranes.^19,64 Measured SAMs composed of 1O2 show similar vibrational features under all conditions; however, SAMs composed of 9O12 reveal significant intensity decreases of the peak centered at 2593 cm⁻¹ and loss of the peak at 2559 cm⁻¹ (Figures S6 and S7), which are attributed to a change from majority monovalent to majority divalent binding in SAMs composed of 9O12 after deposition in base. Each experiment is repeated (n > 3) to track functional control at both the local and ensemble scales. Vibrational peaks were also modeled in the gas phase using density functional theory. We attribute measured peaks in SAMs composed of 1O2 to complicated collective modes of B-H stretching vibrations that involve significant contributions from all the boron vertices. In the observed doublet, the peak at higher frequency is composed of modes dominated by stretches of B-H vertices 3, 6, and 4, 5, 7, and 11, while the modes of the lower frequency peak include predominantly the stretches of BH 8, 9, 10, and 12. In SAMs composed of 9O12, a triplet is measured with FTIR that is also modeled (Table S1), with the calculated collective B-H stretching modes differentiated by much stronger dominance of individual vertices or their small groups. The peaks centered at ~2559 cm⁻¹ belong to the stretching modes with major contribution from vertices 8 and 10, peaks at 2595 cm⁻¹ are dominated by stretches at vertices 4, 5, 7, and 11, and peaks at 2633 cm⁻¹ are formed by the modes involving predominantly vertices 3 and 6. Frequency calculations for the gold salts of both 1O2 and 9O12 show relative attenuation of the infrared absorption features connected with the B-H stretching modes with the major contribution from the vertices close to the S atoms upon binding the sulfur atom to gold. Especially in the 9O12 layer, the intensity decrease at ~2595 cm⁻¹ and peak loss at 2595 cm⁻¹ under basic conditions are consistent with our assignments and computational results. The loss of infrared absorption intensity of the stretching modes dominated by the groups of 4, 5, 7, 11 and 8, 10 B-H vertices is attributed to the effect of lateral intermolecular interactions.

Table 1.

X-ray photoelectron spectroscopy energy shifts and Fourier transform spectroscopy frequency values in the B-H region (row). Columns are titled with 1,2-(HS)₂-1,2-C₂B₁₀H₁₀ (1O2) or 9,12-(HS)₂-1,2-C₂B₁₀H₁₀ (9O12) and labeled with neutral (n), basic (b), or acidic (a) deposition conditions.

	1O2 n	9O12 n	1O2 b	9O12 b	1O2 a	9O12 a
Au 4f	84.0 eV	84.0 eV	84.0 eV	84.0 eV	84.0 eV	84.0 eV
B 1s	189.0 eV	189.0 eV	189.0 eV	189.0 eV	189.0 eV	189.0 eV
S 2p	162.2 eV	161.7 eV	162.2 eV	161.7 eV	162.2 eV	161.7 eV
C 1s	285.0 eV	285.0 eV	285.0 eV	285.0 eV	285.0 eV	285.0 eV
Na 1s	-	-	-	-	-	-
Cl 2p	-	-	-	-	-	-
B-H (FTIR)	2589 cm−1 2598 cm−1	2559 cm−1 2595 cm−1 2633 cm−1	2584 cm−1 2596 cm−1	2593 cm−1 2627 cm−1	2600 cm−1 2591 cm−1	2564 cm−1 2598 cm−1 2630 cm−1

Binding configurations are further modeled on Au surfaces with all possible high-symmetry binding sites on the Au{111} surface: face-centered cubic three-fold hollow sites, hexagonally close-packed three-fold hollow sites, bridge sites, and atop sites. The most favorable binding modalities are presented in Table 2, where the most stable 1O2 and 9O12 species are presented schematically in Figures S8 and S9. Both monovalent and bivalent binding modes, for 1O2 and 9O12, are energetically favorable.

We use both STM and STS to test the local valency within monolayers of carboranedithiols on Au{111} after control via pH. This control is monitored by apparent height in STM, dipole offsets in simultaneous STM topography and LBH measurements, infrared spectroscopy, and X-ray photoelectron spectroscopy, and is supported and understood with complementary density functional calculations.

Conclusions and Prospects

We have controlled the valency of binding within SAMs composed of different carboranedithiol isomers. These monolayers retain their two-dimensional lattice structures, independent of their binding configuration. With this level of control, we are able to dictate surface-atom-molecule stoichiometry with simple acid-base chemistry. We anticipate using these and related systems to explore the effects of valency on surface patterning^{29–31,35,36,63} and dynamics.^65,66

Creating monolayers with switchable substrate-molecule valency and bond strengths while preserving lattice constants enables exploration of this important interface and adds to the repertoire of controllable interactions at surface-molecule, molecule-molecule, and molecule-environment interfaces. By simply varying the head group, molecular backbone, and/or the tail group, extraordinary control is attainable.^1–3,6,8,13 Rigid cage molecules, especially carboranethiols, exhibit advantageous properties and provide test beds for independently exploring aspects of self-assembly, such as dipole interactions, molecular orientation, electron transfer, surface polarity, and now valency.^{3,13,15–17,31,67} Here, bifunctional carboranedithiols assemble into well-ordered monolayers on Au surfaces with two distinct binding modes that are confirmed by STM, STS, FTIR, XPS, and DFT. This control is also expected to be of specific use in chemical patterning, where the binding of molecules to substrates and stoichiometry of molecule-to-surface bonds are both critical.³¹

Materials and Methods

Monolayer Preparation

The chemicals 1O2 and 9O12 were synthesized and characterized in accordance with previously published methods.^21,24 Ethanol was used as received (Sigma-Aldrich, St. Louis, MO). The Au{111}/mica substrates (Agilent Technology, Tempe, AZ) were hydrogen-flame annealed prior to SAM formation with 10 passes at a rate of 0.4 Hz. Both unmodified SAMs were prepared by immersion into 1 mM ethanolic solutions and held at room temperature for approximately 24 h. Short deposition times (10 min), in acidic or basic solutions, were employed to decrease the possibility of molecular degradation. After deposition, each sample was rinsed thoroughly with neat ethanol and dried under a stream of ultrahigh purity argon for at least three cycles.

Since carboranes have been reported to be reactive and to degrade upon exposure to concentrated base or acid⁶⁸ we use dilute concentrations of acid and base to prevent side reactions. Hydochloric acid (12 M) and NaOH pellets were used as received (Sigma-Aldrich, St. Louis, MO). Acidic solutions were prepared by mixing 0.5 mL of 2 mM HCl in EtOH and 0.5 mL of 2 mM 1O2 or 9O12 in EtOH in a gasketed v-vial. Basic solutions were prepared by mixing 0.5 mL of 4 mM NaOH in EtOH and 0.5 mL of 2 mM 1O2 or 9O12 in EtOH. Monolayers were prepared by immersing flame-annealed Au{111}/mica substrates into modified solutions for 1 h. Larger ratios of both acid (2:1) and base (4:1) were tested, however, no differences were found.

Scanning Tunneling Microscopy

All STM measurements were performed with either a custom-built Besoke-style scanning tunneling microscope under ambient conditions or a custom-built Besoke-style scanning tunneling microscope held at cryogenic (4 K) and extreme high vacuum (<10⁻¹² torr) conditions.^69,70 Samples were held at a fixed bias (V_sample = −0.5 V) and both topographic and LBH modalities were measured in a constant current fashion (I_t = 15 pA). The tunneling gap distance was modulated above the microscope feedback loop (~3 kHz) with a sinusoidal amplitude (dz ~ 0.1 Å) and dI/dz was measured with a lock-in technique (Stanford Research Systems SR850 DSP, Sunnyvale, CA)^7,57 The well-known lattice of atomic Au{111}, held at 4 K, was measured and used to calibrate all low temperature images, and the known lattice within SAMs of 1-dodecanethiolate were used to calibrate all images obtained at room temperature.

Image Analyses

All STM images were processed with automated routines developed in MATLAB (Mathworks, Natick, MA) to remove high-frequency noise and intensity spikes that may impair reliable interpretation.⁷ Local maxima (minima) for both topography and local barrier height images were chosen as the highest (lowest) intensity pixel within a defined radial vector (the size of one molecule). Dipole offsets were computed using a block-matching approach,^16,61,62 where topographic image patches (size of one molecule) were correlated against larger local barrier height image patches (size of the nearest-neighbor spacing) to obtain a set of points (p and q) that were referenced and plotted. Correlated values (shown in Figure 5) were compared against connecting all points within a defined pixel radius, as a function of size, where correlation yielded the least artifacts (Figures S10 and S11). Gray scale threshold values were chosen based on apparent height differences to produce binary image highlights, which were further used to obtain percent coverages.

Infrared Spectroscopy

All infrared spectra were collected with a Nicolet 6700 FTIR spectrometer (Thermo Electron Corp., Waltam, MA) that was equipped with a mercury-cadmium-telluride detector, held at liquid nitrogen temperatures, and a Seagull variable-angle reflection accessory (Harrick Scientific, Inc., Ossinging, NY). Water and carbon dioxide were removed from the spectrometer by an FTIR purge gas generator (Parker-Balston, Cleveland, OH). A grazing incidence reflection angle (82° with respect to the surface normal) with p-polarized light, a mirror speed of 1.27 cm/s, and a resolution of 2 cm⁻¹. Spectra were averaged over 5120 scans and normalized against spectra of perdeuterated n-dodecanethiolate monolayers on Au{111}.

X-Ray Photoelectron Spectroscopy

All XPS spectra were collected with an AXIS Ultra DLD instrument (Kratos Analytical Inc., Chestnut Ridge, NY). A monochromatic Al K_α X-ray source (20 mA, 15 kV) with a 200 μm circular spot size that was held at ultrahigh vacuum (10⁻⁹ torr) were used for all measurements. Spectra were acquired at a pass energy of 160 eV for survey spectra and 20 eV for high resolution spectra of S 2p, C 1s, B 1s, and Au 4f regions that used a 200 ms dwell time. Different numbers of scans were carried out depending on the amount required for high-resolution spectra, which ranged from 20 scans for C 1s to 75 scans for S 2p. Binding energies were calibrated to the Au 4f peak at 83.98 eV.⁷¹ Spectra were fit using CasaXPS software with Gaussian-Lorentzian lineshapes after Shirley background subtraction. Sulfur regions were fitted by a doublet structure with a 1.18 eV spin-orbit splitting and a defined intensity ratio (2p_3/2:2p_1/2, 2:1).

Computational Modeling

Density functional theory calculations were performed to understand the molecular chemisorption modes and their relation with the STM imaging. Geometry optimizations were performed with the plane-wave periodic DFT VASP program.^72–75 Exchange and correlation effects were described within the generalized gradient approximation (GGA), using the Perdew–Burke–Ernzerhof (PBE) functional⁷⁶ and electron-ion interactions are treated with the projector augmented wave approach.⁷⁷ In order to describe the dispersion interactions, the empirically constructed DFT-D3 method was used.⁷⁸ The calculations were done in a spin-unrestricted fashion when applicable. The gold slab was modeled as a (√7×√7)R19.12° supercell, with four layers along z. The bottom two layers were fixed during the optimization. The calculated lattice constant for Au was 2.95 Å, in good agreement with the experiment.

To avoid spurious interactions between images in the z-direction, a vacuum gap of 10 Å was inserted between the highest atom of the carboranedithiolate molecules, and the repeated image of the slab’s bottom layer. The cutoff energy for the plane-wave basis set was chosen to be 400 eV. A 5×5×1 Monkhorst-Pack k-point grid allowed for the numerical solution of Hamiltonian and overlap matrix elements. For the STM images and the charge density plots a denser k point grid (11×11×1) was used.

The STM images were simulated using the Tersoff-Hamann method⁴⁷ using VASP.

Natural bond order (NBO1.1)⁷⁹ analyses of the molecules on the support were performed. The periodic version of NBO also requires the wave function to be represented in an atom-centered basis, therefore we projected the plane-wave solution onto the Def2-TZVP^80,81 Gaussian-type orbital basis set.

Figure 8.

(A,B) Scanning tunneling microscope images (V_sample = −0.1 V, I_tunneling = 100 pA, T = 298 K) of 1,2-(HS)₂-1,2-C₂B₁₀H₁₀ (1O2) on Au{111}/mica at two different resolutions under acidic deposition conditions (1:1, HCl:1O2). Inset depicts a fast Fourier transform (FFT) that shows a hexagonally close-packed arrangement with the same nearest-neighbor spacings measured in both basic and neutral conditions. (C) A minority push to the monovalent binding is achieved and depicted schematically. (D,E) Scanning tunneling topographs (V_sample = −0.1 V, I_tunneling = 100 pA, T = 298 K) of 9,12-(HS)₂-1,2-C₂B₁₀H₁₀ (9O12) on Au{111}/mica at different resolutions after acidic deposition (1:1, HCl:9O12). Inset depicts a FFT showing the same arrangement and spacing as in neutral conditions. (F) As monolayers composed of 9O12 are already predominately monovalent, no change is measured, in comparison to neutral deposition conditions, for this system that is depicted schematically.