Solution Deposition of Phenylphosphinic Acid Leads to Highly

Jun 8, 2017 - *E-mail: [email protected]. ... molecular vibrations—the P–H stretch vibration and the symmetric and antisymmetric OPO stret...
2 downloads 8 Views 6MB Size
Article pubs.acs.org/JPCC

Solution Deposition of Phenylphosphinic Acid Leads to Highly Ordered, Covalently Bound Monolayers on TiO2 (110) Without Annealing Erik S. Skibinski, William J. I. DeBenedetti, and Melissa A. Hines* Department of Chemistry and Chemical Biology, Cornell University, Ithaca, New York 14853, United States S Supporting Information *

ABSTRACT: Solution-deposited phosphonic acids, OP(OH)2R, have been used to impart new functionality to a variety of metal oxide surfaces for applications ranging from organic field-effect transistors to biocompatible coatings on implants. Interestingly, the as-deposited monolayers are easily rinsed off, becoming robust and strongly adherent only after a long, low-temperature thermal anneal (e.g., 18 h at 120 °C). The need for this thermal treatment has raised questions about the nature of the bonding of the as-deposited monolayer. Is it merely physisorbed, requiring heat treatment for covalent bonding? To understand the first stages of monolayer formation, we have studied the reactivity and molecular bonding geometry of a prototypical, solution-deposited phosphinic acid, OPH(OH)R, on the prototypical metal oxide surface rutile (110). We show that solution deposition produces near ideal, dense phenylphosphinate monolayers covalently bound in a bridged bidentate geometry. Three nearly orthogonal molecular vibrationsthe P−H stretch vibration and the symmetric and antisymmetric OPO stretch vibrations provided an unambiguous signature of the three-dimensional structure and binding of the adsorbed monolayer; scanning tunneling microscopy provided information on long-range order and intermolecular conformation; and X-ray photoemission spectroscopy provided coverage quantification. Despite their covalent bidentate attachment and significantly higher binding energy than the corresponding carboxylic acid, a H2O rinse removed most of the phosphinate monolayer, demonstrating that hydrolytic stability does not result from covalent attachment alone. The H2O rinse also oxidized ∼25% of the phosphinate monolayer to the corresponding phosphonate, producing a species that was somewhat more resistant to H2O rinsing.



INTRODUCTION The formation of organic self-assembled monolayers (SAMs) on metal oxide surfaces is important to many applications. Selfassembled monolayers offer a simple and effective method to alter the chemical, physical, or electronic properties of a surface, opening new applications for materials or optimizing existing properties. For example, monolayers formed from phosphonic acids, OP(OH)2R, can be used to improve the performance of solar cells,1 for biocompatible coatings on titanium orthopedic implants,2,3 to improve adhesion, and for corrosion inhibition of metal surfaces.4,5 Additionally, phosphonate monolayers show promise in the modification of interfacial electronic properties of transparent conducting oxides, such as TiO2, indium tin oxide (ITO), and zinc oxide,6,7 with applications ranging from organic field-effect transistors8 to dye-sensitized solar cells.9 Many of these applications require monolayers that are stable under ambient conditions, such as humid air or solution, and often under sunlight. In general, phosphonate SAMs on metal oxides have been found to be more stable toward hydrolysis than other options, such as carboxylate SAMs or siloxane SAMs.9−12 Gawalt et al.13 showed that the stability of © XXXX American Chemical Society

octadecylphosphonic acid deposited from solution onto the native oxide of titanium is greatly improved by a long, lowtemperature thermal annealing step (e.g., 18 h at 120 °C). After annealing, the deposited film was robust enough to survive a mechanical peel test using adhesive tape. On the basis of this behavior, Gawalt et al. suggested that the octadecylphosphonic acid molecules are only weakly physisorbed after solution deposition and that thermal annealing provides the energy necessary for the formation of covalent attachments. Others have suggested that the increased stability after annealing is due to the inherently high binding energy of phosphonates.5 Carboxylic acids are well-known to readily attach to rutile (110) in a bridged bidentate geometry;14,15 however, the binding of phosphonic acids has been the subject of significantly less research, at least on well-controlled, singlecrystal substrates. There is still controversy over the energetically favored bonding geometry and whether it is mono-, bi-, or tridentate on TiO2.16 The formation, properties, and stability of Received: May 2, 2017 Revised: June 7, 2017 Published: June 8, 2017 A

DOI: 10.1021/acs.jpcc.7b04167 J. Phys. Chem. C XXXX, XXX, XXX−XXX

Article

The Journal of Physical Chemistry C

carboxylates on rutile (110). Despite the covalent bidentate attachment and significantly higher binding energy than the corresponding carboxylic acid (−2.62 eV calculated for phenylphosphinate vs −1.94 eV for benzoate), a quick H2O rinse removed most of the phosphinate monolayer, demonstrating that hydrolytic stability does not result from covalent attachment alone. The H2O rinse also oxidized ∼25% of the phenylphosphinate monolayer to phenylphosphonate, producing a species that was much more resistant to H2O rinsing. Our ability to produce phosphinate monolayers from entirely solution-based processes on atomically flat, single-crystal TiO2 creates an ideal platform for studying monolayer formation, stability, and reactivity. This insight will be used to tailor monolayer properties for specific applications.

self-assembled monolayers largely result from substrate− molecule interactions and intermolecular interactions. Understanding the nature of these interactions will enable rational design of higher performance monolayers. To understand the first stages of monolayer formation as well as the molecular bonding geometry on the prototypical metal oxide surface, rutile (110), we have studied the reactivity of a prototypical phosphinic acid, OPH(OH)R, which has one less hydroxyl group than the corresponding phosphonic acid, OP(OH)2R. As shown in Figure 1, the rutile (110) surface is



EXPERIMENTAL AND COMPUTATIONAL Experimental Section. Rutile samples for analysis in the multiple-internal-reflection (MIR) geometry were beveled at 45° on parallel edges from 10 × 10 × 0.5 mm die. Samples for XPS or STM analysis were thermally reduced at 700 °C for 5 min in vacuum to induce conductivity, resulting in a light blue color to the crystals. All labware was cleaned prior to use in a basic peroxide bath of 1:1:5 28% NH4OH (aq, BDH, ACS grade):30% H2O2 (aq, J.T. Baker, CMOS grade):H2O (MilliQ) by volume for 10 min at 80 °C. When preparing a sample for analysis in ultrahigh vacuum (UHV), labware was additionally cleaned in an acidic peroxide bath of 1:1:5 37% HCl (aq, BDH, ACS grade): H2O2 (aq, J.T. Baker, CMOS grade):H2O (Milli-Q). Atomically flat, bicarbonate-terminated rutile (110) (MTI or Crystek) samples were prepared by etching in a basic peroxide bath of 1:1:2 NH4OH:H2O2:H2O for 10 min at 80 °C.17,18 Samples for UHV were additionally cleaned in an acidic peroxide bath of 1:1:2 of HCl:H2O2:H2O at 80 °C and then another basic peroxide bath of 1:1:2 NH4OH:H2O2:H2O at 80 °C. Samples were thoroughly rinsed in ultrapure water between each bath and as a final step. Phenylphosphinate monolayers were typically prepared by immersing cleaned rutile crystals in a boiling 13 mM aqueous solution of phenylphosphinic acid (Sigma, >99%) for 2 min; however, tests on samples prepared in room temperature solutions were spectroscopically indistinguishable. The resulting hydrophobic crystal was removed from solution and either loaded to an UHV chamber through an oil-free load lock or a dry-air-purged Fourier transform infrared spectrometer (Nicolet 670). Rinsing experiments were performed by immersing phenylphosphinate monolayers in water for 2 min before analysis. Benzoate monolayers for use as a spectroscopic reference were prepared by immersing cleaned crystals in a boiling 16 mM aqueous solution of benzoic acid for 2 min. Polarized infrared spectra were collected with a mercury− cadmium−telluride detector and ZnSe wire grid polarizer (Molectron). Multiple-internal-reflection spectra were collected on separate crystals beveled in orthogonal orientations, so that light propagated along either the [001] or [11̅0] directions. Spectra taken in MIR were then computationally transformed to a Cartesian reference frame.19,20 Infrared spectra in the reflection geometry were collected on 10 × 10 × 0.5 mm crystals with the plane of incidence along either the [001] or [110̅ ] directions. STM images were collected using recrystallized,21,22 electrochemically etched tungsten tips. X-ray photoelectron spectra were collected using an unmonochromated Mg Kα source. A Tougaard baseline was subtracted from each

Figure 1. Rutile (110) surface. (a) STM image of chemically etched, bicarbonate-terminated TiO2 (110) showing the production of near ideal, atomically flat terraces (+1.7 V, 500 pA). (b) STM image of UHV-prepared TiO2 (110) showing rows of undersaturated Ti atoms, which image as elliptical protrusions, separated by bridging O rows, which image as depressions (+1.3 V, 100 pA). (c) Molecular model of the rutile (110) surface in which Ti and O atoms are represented by blue and red spheres, respectively.

terminated by rows of undersaturated Ti atoms aligned with the [110] direction and separated by rows of bridging O atoms. Here we study the phenyl-substituted phosphinic acid, where −R = −C6H5. As we shall show, three nearly orthogonal molecular vibrationsthe P−H stretch vibration and the symmetric and antisymmetric OPO stretch vibrationsprovide an unambiguous signature of the three-dimensional structure and binding of the adsorbed monolayer, whereas scanning tunneling microscopy (STM) provides information on longrange order and intermolecular conformation. Using these experimental indicators in combination with absolute molecular densities from X-ray photoemission spectroscopy (XPS) as well as density functional theory (DFT) calculations, we show that solution deposition creates near-ideal phosphinate monolayers essentially instantaneously (2 min) on atomically flat rutile (110). These highly ordered, covalently bound monolayers attach in a bridged bidentate geometry similar to that of B

DOI: 10.1021/acs.jpcc.7b04167 J. Phys. Chem. C XXXX, XXX, XXX−XXX

Article

The Journal of Physical Chemistry C spectrum, and a small energy correction (∼0.05 eV) was applied to offset band bending effects.23 Computational. Density functional theory (DFT) was used to model the structure of phenylphosphinate and benzoate monolayers on 2 × 1 and 4 × 1 periodically repeating slabs consisting of five TiO2 trilayers separated by a 12.5 Å vacuum spacing with autocompensated surfaces, as shown by the structures in the Supporting Information. The 2 × 1 supercell contained two unsaturated Ti atoms capable of adsorbing one phenylphosphinate or one benzoate molecule and one H atom produced by the dissociation of the corresponding acid. The benzoate adsorption structure was determined by comparison to experiment in previous studies.15,20 During optimization, the positions of the atoms in the bottommost TiO2 layer and its terminating bridging O rows were held fixed. Calculations were performed within the generalized gradient approximation24 (GGA) with the Perdew, Burke, and Ernzerhof (PBE) exchange-correlation functional,25 as implemented in the Vienna ab initio simulation package (VASP).26−29 The functional was corrected for long-range dispersion interactions using the zero-damping DFT-D3 method.30 Electron−ion interactions were described using the projector-augmented wave (PAW) method.31,32 Electronic states were expanded in plane waves with a kinetic energy cutoff of 400 eV and a 6 × 6 × 1 Monkhorst−Pack grid of k points with the 2 × 1 supercell. Larger supercells used proportionately fewer k points. Brillouinzone integration was performed using Gaussian smearing. Vibrational energies were calculated using density-functional perturbation theory without dispersion interactions.33,34 Calculated vibrational energies were multiplied by a constant scaling factor (1.024) determined by the best fit to previously reported molecular vibrational energies, as discussed in the Supporting Information.35 STM images were modeled within the Tersoff− Hamann approximation36 as isosurfaces of constant local density of states in an energy band from 0.33 eV below the highest occupied band to the Fermi energy.

Figure 2. (a−c) Experimental STM images of phenylphosphinateterminated rutile (110) surfaces at different magnifications (a: +1.9 V, 200 pA; b: +2.1 V 200 pA; c: +2.0 V, 200 pA). Simulated STM images of the (d) aligned and (e) alternating conformers of the phenylphosphinate monolayer. All images have the same crystallographic orientation as shown in (a).



RESULTS The initial surface for all experiments was a freshly etched rutile (110) surface, which previous experiments have shown to be atomically flat17 over hundreds of nanometers, as shown in Figure 1, and terminated by a monolayer of bicarbonate produced by the catalytic reaction of ambient H2O and CO2.18 By eye, these surfaces were markedly hydrophilic, remaining wet as they were pulled from a H2O rinse. After immersion of the freshly etched surface in boiling or room-temperature phenylphosphinic acid for 2 min, the crystal emerged from the solution dry and notably hydrophobicthe first evidence of the creation of a phenylphosphinate monolayer and displacement of the bicarbonate monolayer. This surface was analyzed without further rinsing as described below. Immersion of the phenylphosphinate-terminated surface in H2O for as little as 30 s resulted in a notably hydrophilic surfacethe first evidence that the phenylphosphinate SAM is easily, albeit partially, removed by H2O rinsing. STM Analysis. Figure 2 shows STM images of the phenylphosphinate-terminated rutile (110) surface at a variety of magnifications. Under most imaging conditions, the majority of the surface was terminated by roughly spherical protrusions with a characteristic separation of 0.60 nm measured along the [001] directiona spacing that is twice that of the Ti−Ti spacing and consistent with bridged bidentate bonding. The protrusions displayed two different packings along the [001]

direction. In some regions of the surface, the protrusions formed “aligned” rows parallel to the [001] direction, as marked in Figure 2(b). In other regions, the protrusions formed an “alternating” pattern, with adjacent protrusions along the [001] direction being displaced by ∼0.3 nm in the [11̅0] direction. Under optimal imaging conditions, elliptical protrusions were observed, with the long axis of the ellipses being roughly ±45° from the [001] direction, as shown in Figure 2(c). These observations can be explained by a very simple model of phenylphosphinate bonding, which will be confirmed by additional experiments below. First, each phenylphosphinate molecule binds to two Ti atoms in a bridged bidentate geometry.37 This bonding gives rise to two distinct conformers with the phenyl group pointing either to the left (L) or right (R) of the Ti rows (i.e., pointing toward either the [11̅0] or [1̅10] directions). If adjacent molecules along a Ti row had the same configuration (e.g., LLLL or RRRR), the molecules appeared “aligned” in STM images. If adjacent molecules had an alternating configuration (e.g., LRLR), the molecules were “alternating” in STM images. Second, steric interactions between H atoms on the adjacent molecules prevented adjacent phenyl groups from adopting a coplanar configuration. As a result, the phenyl groups rotated out of the coplanar geometry by ∼25° about the P−C bond. The elliptical features observed C

DOI: 10.1021/acs.jpcc.7b04167 J. Phys. Chem. C XXXX, XXX, XXX−XXX

Article

The Journal of Physical Chemistry C

respectively. For the limiting case in which three mutually orthogonal infrared active modes of equal oscillator strength aligned with the x, y, and z axes (i.e., μx, μy, and μz) are probed by s- or p-polarized radiation incident in the x−z plane where z is the surface normal, the relative absorbances of the three modes as a function of reflectance angle are shown in Figure 3(b).38 In this geometry, s-polarized light is sensitive to μy, whereas p-polarized radiation is sensitive to μx and μz. For all incident angles, light absorption by μx and μz has the opposite effect on reflectivity: one mode increases reflectivity, whereas the other decreases reflectivity. For the 80° incidence angle used in these experiments, parallel modes probed with s radiation and perpendicular modes probed with p radiation lead to negative reflected absorbances (i.e., they increase reflectivity), whereas parallel modes probed with p radiation lead to positive reflected absorbances. Importantly, although Figure 3(b) seems to suggest that p-polarized radiation is very sensitive to surface absorption at incidence angles near Brewster’s angle, this effect is offset by the extremely low reflectivity of p-polarized radiation near Brewster’s angle, as shown in Figure 3(a). Optimizing the reflection geometry for surface sensitivity requires careful consideration of a number of factors, including the crystal size.39 Infrared spectra obtained in the MIR geometry showed that the P−H bond was roughly perpendicular to the Ti rows (i.e., parallel to the [110̅ ] direction), consistent with the bridged bidentate geometry suggested by STM images. Three-dimensional information on bonding geometry was obtained by measuring spectra in two orthogonal geometries with both sand p-polarized radiation, as sketched in the top panel of Figure 4, and then mathematically transforming the four spectra to the Cartesian basis defined by the surface normal (z), the Ti row direction (x or [001]), and the cross-row direction (y or [11̅0]). In this spectral region, the infrared spectrum was dominated by the intense P−H stretch mode at 2376 cm−1, which was assigned with the aid of DFT calculations (vide inf ra) as tabulated in Table 1. As shown in Figure 4, Cartesian analysis showed that the P−H stretch mode was highly aligned with the cross-row or y axis. In addition, the modes at 3017, 3060, and 3078 cm−1 were assigned to aromatic C−H stretch vibrations on the basis of their energies. DFT calculations significantly overestimated the energies of the C−H stretch modes, which may be due to our neglect of vibrational anharmonicity.40 Infrared spectra obtained in the reflectance geometry confirmed that the OPO moiety in phenylphosphinate is bound parallel to the Ti rows in a nominally vertical configuration, consistent with the bridged bidentate geometry suggested by the STM images. Reflectance spectra of phenylphosphinate-terminated rutile (110) referenced to benzoate-terminated rutile (110) obtained in the four geometries sketched in Figure 4 are shown in Figure 5. The modes at 1427 and 1489 cm−1 were assigned to the symmetric and asymmetric OCO stretch vibrations, respectively, on the basis of previous experiments41 as well as DFT calculations (vide inf ra). The symmetric OCO stretch mode was aligned with the surface normal (z) and thus had a positive absorbance in a background spectrum, as predicted by Figure 3. Similarly, the asymmetric mode was aligned with the Ti rows (x) and had a negative (positive) absorbance in p (s) polarization. In analogy with the OCO stretch vibrations, the modes at 1049 and 1092 cm−1 were assigned to the symmetric and asymmetric OPO stretch vibrations, respectively. These modes had the same

in STM correspond to the elevated edge of the phenyl ring as shown by the simulated STM images (vide inf ra) in Figure 2. Infrared Spectroscopy. Infrared spectroscopy of metal oxide surfaces is complicated by the existence of intense phonon and multiphonon absorption bands which, in the case of TiO2, block infrared transmission below ∼2000 cm−1. For energies greater than this cutoff, surface spectroscopy is most conveniently performed in the high sensitivity, multiple internal reflection (MIR) geometry, which allows the infrared radiation to make multiple interactions with the surface. For example, light passing through a 10 mm long, 0.5 mm thick crystal beveled at 45° from the surface normal interacts with the surface ∼19 times, leading to a substantial increase in sensitivity over single-pass transmission. In the MIR geometry, infrared absorption by the monolayer leads to a decrease in the transmitted light and a corresponding positive absorbance A ≡ log(I0/I), where I and I0 are the transmitted intensities with and without the absorbing species, respectively. If spectra are obtained in two orthogonal geometries with both s- and ppolarized radiation, the Cartesian components of the absorbing modes, and thus the orientation of the vibrating bonds, can be extracted.19,20 The situation is more complicated for modes below ∼2000 cm−1 where the infrared spectrum must be measured in the reflectance geometry. On semiconductor surfaces, such measurements are usually performed at grazing incidence to increase the amount of reflected light, as illustrated by the reflectance curves for TiO2 in Figure 3(a). Depending on the incidence angle of the radiation and the polarization of the mode, infrared absorption by the monolayer can either increase or decrease the amount of reflected light, which complicates analysis. In analogy with the MIR geometry, we define the reflected absorbance A ≡ log(R0/R) where R and R0 are the reflected intensities with and without the absorbing species,

Figure 3. Important characteristics of the reflectance geometry for TiO2 substrates. (a) The reflectance of s- and p-polarized radiation as a function of incidence angle measured from the surface normal and (b) the relative absorbance of infrared active modes aligned with the x, y, and z axes, denoted by μx, μy, and μz, respectively, and probed by the in-plane (||) and out-of-plane (⊥) components of s- and p-radiation. The vertical dashed line denotes the 80° incidence angle used here. Both panels assume a substrate dielectric constant of 2.2, which is appropriate for rutile near 1400 cm−1. D

DOI: 10.1021/acs.jpcc.7b04167 J. Phys. Chem. C XXXX, XXX, XXX−XXX

Article

The Journal of Physical Chemistry C

Figure 5. Reflectance spectra of phenylphosphinate-terminated rutile (110) referenced to benzoate-terminated rutile (110) obtained in the four geometries defined in the top panel of Figure 4. Transitions in the orange-shaded region are phenylphosphinate modes, whereas transitions in the purple-shaded region are benzoate modes.

orientation as the analogous OCO stretch modes, but the opposite signs of absorbance as the phenylphosphinate spectrum was the “signal (R)” not the “background (R0)” spectrum. The modes at 1022 and 1145 cm−1 had the same polarization dependence as the symmetric OPO stretch vibration, which implied they were also nominally polarized in the z direction. The mode at 1145 cm−1 was assigned to the P−C stretch mode on the basis of DFT calculations. The mode at 1022 cm−1 was tentatively assigned to a P−H wag mode; however, the identity of this mode is uncertain. The energy of the mode was in good agreement with the calculated energy of the in-plane P−H wag mode; however, the polarization was consistent with the out-of-plane P−H wag. The assignments of all modes and their polarizations are summarized in Table 1. In summary, the three mutually orthogonal vibrational modes νstr(CH), νasym(OPO), and νsym(OPO) were consistent with the vast majority of the adsorbed molecules being in the bridged bidentate bonding geometry sketched in Figure 6. XPS Analysis. The chemical identity and coverage of the monolayer were confirmed using XPS analysis, as shown in Figure 7. High-resolution scans of the P 2p region on the as-deposited monolayer displayed a prominent transition centered at 133.6 eV, which we assigned to phenylphosphinate. The relative intensities of the P 2p and Ti 2p transitions were consistent with the formation of a near-complete monolayer, as shown by the data in Table 2 and described in the Supporting Information. Rinsing the monolayer for 2 min in H2O removed 75−80% of the P atoms in the monolayer, as shown in Table 2. Rinsing also caused a chemical transformation of the remaining surface-bound molecules, as shown by the 0.55 eV increased binding energy of the P 2p photoelectrons. This binding energy shift was indicative of increased oxidation of the P atom, which

Figure 4. Cartesian analysis of infrared spectra of a phenylphosphinate monolayer on rutile (110) referenced to a bicarbonate monolayer. (Top) Experimental geometry, (left) four experimental spectra, and (right) Cartesian components of the spectra. The P−H stretch mode at 2376 cm−1 is highly aligned with the y axis, which implies that the P−H bond points across the Ti rows.

Table 1. Comparison of Experimental and Calculated Vibrational Energiesa molecule

mode

exptl (cm−1)

DFTb (cm−1)

error

PhPH PhPH PhPH PhPH

vstr vstr vstr vstr

(CH) (CH) (CH) (PH)

3078 3060 3017 2376

3224 3207 3200 2434

2.4%

Benz Benz PhPH PhPH PhPH PhPH PhPH

vasym (OCO) vsym (OCO) vstr (PC) vasym (OPO) vsym (OPO) δ∥ (PH), wag δ⊥ (PH), wag

1489 1427 1145 1092 1049 1022

1516 1454 1135 1072 1036 994 952

1.8% 1.9% −0.9% −1.8% −1.2% −2.7% −6.8%

pol.

across row along row z z along row z z

a

Neglect of vibrational anharmonicity in the calculations made assignment of the C−H stretch modes uncertain, so specific errors are not reported for these modes. bAll DFT energies increased by 2.8% to correct for systematic error. PhPH = phenylphosphinate; Benz = benzoate; pol. = polarization.

E

DOI: 10.1021/acs.jpcc.7b04167 J. Phys. Chem. C XXXX, XXX, XXX−XXX

Article

The Journal of Physical Chemistry C

Table 2. Surface Coverages of the Phenylphosphinate Monolayer before and after a 2 min H2O Rinse As Quantified from the Integrated Areas of the P 2p and Ti 2p XPS Transitions Collected at Two Different Ejection Angles ejection angle θ monolayers as prepared monolayers after rinse fraction remaining

Figure 6. Orientations of the three mutually orthogonal vibrational modesνstr(PH), νasym(OPO), and νsym(OPO)relative to the molecular bonding geometry.

46°

71°

0.78 0.17 0.22

1.04 0.23 0.22

the P 2p binding energy upon oxidation of the phosphinate to the phosphonate. Further evidence of the partial removal of the monolayer and oxidation of the remaining surface-bound molecules came from high-resolution scans of the O 1s region, as also seen in Figure 7. This spectrum was dominated by bulk O atoms; however, the rinsing process caused two notable changes in the high binding energy region of the O 1s photoelectron spectrum. First, rinsing caused a decrease in photoelectrons with energies near 531.9 eV, as most easily seen by the blue difference spectrum in Figure 7. We assigned these photoelectrons to O atoms in the phenylphosphinate layer, attributing the decrease to the removal of ∼75% of the monolayer. In addition, rinsing led to the development of a weak shoulder at higher binding energy, ∼533.3 eV, which we assigned to O atoms in the hydroxyl group attached to phenylphosphonate. Finally, high-resolution scans of the C 1s region provided further confirmation of this chemical transformation. The asdeposited monolayer displayed two well-resolved transitions consistent with a phenylphosphinate monolayer: a transition at 285.6 eV which we assigned to C atoms in the phenyl ring and an affiliated π−π* satellite at 292.3 eV. Rinsing reduced the intensity of these transitions substantially, consistent with the removal of ∼75% of the phenyl groups. In addition, rinsing led to the appearance of a new transition at 289.3 eV, which we previously assigned to the formation of surface bicarbonate (HCO3) from the catalytic conversion of CO2 and H2O.18 The bicarbonate molecules terminate sites vacated by phenylphosphinate molecules during rinsing. In summary, XPS analysis of the surface is consistent with the net reaction 2 min H2O rinse

100% PhPH(ads) + CO2 (g) ⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯→ 25% PhPOH(ads) + ∼75% HCO3(ads) + PhPHOOH(aq)

where PhPH is phenylphosphinate, and PhPOH is phenylphosphonate. DFT Simulations. To further understand the adsorption process, DFT calculations were performed on 2 × 1 and 4 × 1 periodically repeating slabs containing one and two phenylphosphinate or phenylphosphonate molecules, respectively. The relaxed geometries of phenylphosphinate and phenylphosphonate adsorbed in a close-packed monolayer are shown in Figure 8. In both cases, the phenyl rings were twisted from the coplanar geometry about the P−C bond to relieve steric interactions between adjacent molecules. The phenylphosphinate molecule adsorbs in an upright geometry, with the H−P− C bisector being nominally perpendicular to the surface. In contrast, phenylphosphonate molecules rotate about the [001] direction by approximately 15° to allow the H atom on the hydroxyl group to interact with an O atom in the adjacent bridging O row. As shown by the data in Table 3, this H-

Figure 7. X-ray photoemission spectra of phenylphosphinate monolayers on rutile (110) (black) as prepared and (red) after a 2 min H2O rinse showing the (top) P 2p, (middle) O 1s, and (bottom) C 1s regions. The blue curve in the middle panel represents the difference spectrum magnified vertically by a factor of 2.

we attributed to the replacement of the P−H moiety in phenylphosphinate with a P−OH group. In other words, the rinsing process converted phenylphosphinate to phenylphosphonate. This assignment was consistent with DFT simulations (vide inf ra) of phenylphosphinate and phenylphosphonate monolayers, which predicted a 0.44 eV increase in F

DOI: 10.1021/acs.jpcc.7b04167 J. Phys. Chem. C XXXX, XXX, XXX−XXX

Article

The Journal of Physical Chemistry C

where ΔGrxn and ΔErxn are the change in Gibbs free energy and energy of eq 1, respectively; kB is Boltzmann’s constant; and T is the deposition temperature. From eq 2 and the density of alternating isomers, we estimate ΔErxn = 0.04 eV, which is in good agreement with the ΔErxn = 0.02 eV/molecule estimated with DFT calculations. These simulations and the results of previous studies on benzoic acid adsorption15,20 were used to calculate the binding energies of phenylphosphinic acid, phenylphosphonic acid, and benzoic acid on rutile (110) as shown in Table 3. In spite of the structural similarities in their binding, the phenylphosphinic acid bound 0.7 eV more strongly to rutile (110) than benzoic acid. The stronger binding of the phosphonate group is consistent with experiments that measured the relative affinities of carboxylic and phosphonic acids toward nanocrystalline TiO2 surfaces,42,43 as well as with experiments showing that carboxyalkylphosphonic acids selectively bond to nanocrystalline TiO2 through the phosphonate group.44 Similar trends in the relative adsorption energies of carbonic and phosphonic acids have also been predicted previously.45−47 These DFT simulations were also used to calculate the relative energies of the P 2p photoelectrons emitted by phenylphosphinate and phenylphosphonate monolayers. As shown in the Supporting Information, the photoelectron energies measured with respect to the vacuum energy of a series of phosphorus-containing molecules were calculated using initial and final state approximations48 and then compared to previously reported experimental results.49 The initial state approximation produced the best agreement with experiment and was used for subsequent calculations. The energies of P 2p photoelectrons emitted from protonated and deprotonated phenylphosphonate monolayers were calculated to be 0.44 and 0.03 eV higher in energy than those emitted from phenylphosphinate monolayers, with the former energy being in reasonable agreement with the 0.55 eV shift observed experimentally after rinsing (see Figure 7). This agreement suggests that most phenylphosphonate molecules were protonated.

Figure 8. Lowest-energy structures from DFT simulations of (left) phenylphosphinate and (right) phenylphosphonate monolayers adsorbed to rutile (110). The bisector of the H−P−C bond in phenylphosphinate is nearly aligned with the surface normal, whereas the phenylphosphonate molecule tilts by ∼15° toward the adjacent bridging O atom to enable H bonding. Because of this tilt, phenylphosphonate will image ∼0.5 Å taller than phenylphosphinate in STM. The Ti, O, P, and H atoms are represented by blue, red, green, and white spheres, respectively.

Table 3. Calculated Adsorption Energies of Three Organic Acids to Rutile (110) organic acid

adsorption energy (eV)

benzoic acid phenylphosphinic acid phenylphosphonic acid

−1.94 −2.62 −2.96



DISCUSSION Solution deposition of phenylphosphinic acid from roomtemperature or boiling solutions led to the development of highly oriented, bridged bidentate-bound phenylphosphinate monolayers on rutile (110). No additional heat treatments were needed to orient the monolayer or induce covalent attachment. The fact that phenylphosphinate is strongly bound and highly oriented suggests that phosphonatesspecies capable of three bonding interactionswill be similarly bound after solution deposition. Experiments on the solution-phase deposition of phenylphosphonate monolayers are in progress. If phosphonate monolayers are strongly bound and highly oriented after solution deposition, why is a long, lowtemperature thermal annealing step (e.g., 18 h at 120 °C) needed to induce high adhesion? We speculate that the primary effect of the annealing treatment is not on the monolayer per se but rather on the surface structure. STM experiments have shown that Ti interstitials, common defects in TiO2, migrate to the surface at temperatures as low as 120 °C.50 On clean surfaces in ultrahigh vacuum, this migration leads to the formation of Ti islands49 or, in the presence of O2, added TiO2 rows.51 On a phosphonate-terminated surface, we speculate that the Ti interstitials will react with both the bridging O atoms and the hydroxyl group on the phosphonate, leading to

bonding interaction increased the binding energy of phenylphosphonic acid by 0.34 eV relative to phenylphosphinic acid. This rotation also led to an ∼0.5 Å height increase of the phenylphosphonate molecule with respect to phenylphosphinate, which may explain some of the height variations seen in Figure 2. The deprotonated isomer of phenylphosphonate, in which the hydroxyl proton was transferred to the bridging O row, was essentially isoenergetic with the protonated isomer; however, the P 2p XPS spectrum (vide inf ra) was more consistent with the protonated isomer. The existence of “aligned” and “alternating” isomers on the phenylphosphinate monolayer was consistent with DFT simulations, which showed that the two packing geometries were nearly isoenergetic. Quantitative analysis of STM images showed that 22% of the molecules were in the alternating configuration. If the surface structure was in thermal equilibrium with the deposition solution, then aligned (ads) ⇋ alternating (ads)

(1)

and the relative densities of the two isomers should then be given by the equilibrium constant K = e−ΔGrxn / kBT ≈ e−ΔErxn / kBT

(2) G

DOI: 10.1021/acs.jpcc.7b04167 J. Phys. Chem. C XXXX, XXX, XXX−XXX

Article

The Journal of Physical Chemistry C

Research Shared Facilities supported through the NSF MRSEC program (DMR-1120296).

the formation of a phosphorus SAM bound through three Ti− O bonds. The addition of a third Ti−O bond would strengthen the interaction of the monolayer with the surface, leading to increased adhesion. Experiments to test this hypothesis are in progress. These experiments also show that phenylphosphonate is significantly more resistant to H2O rinsing than phenylphosphinate, an effect that we attribute to both entropic and energetic effects. As shown by the DFT models, the interaction between the −OH moiety on phenylphosphonate and an adjacent bridging O atom leads to a ∼0.3 eV stabilization of the phosphonate. In addition, phenylphosphonate has bonding interactions with three speciestwo Ti atoms and a bridging O atomwhereas phenylphosphinate only interacts with two Ti atoms. This higher multiplicity of interactions will entropically stabilize the phosphonate through the chelate effect.



(1) Li, X.; Dar, M. I.; Yi, C.; Luo, J.; Tschumi, M.; Zakeeruddin, S. M.; Nazeeruddin, M. K.; Han, H.; Grätzel, M. Improved Performance and Stability of Perovskite Solar Cells by Crystal Crosslinking with Alkylphosphonic Acid Ω-Ammonium Chlorides. Nat. Chem. 2015, 7, 703−711. (2) Amalric, J.; Mutin, P. H.; Guerrero, G.; Ponche, A.; Sotto, A.; Lavigne, J.-P. Phosphonate Monolayers Functionalized by Silver Thiolate Species as Antibacterial Nanocoatings on Titanium and Stainless Steel. J. Mater. Chem. 2009, 19, 141−149. (3) Auernheimer, J.; Zukowski, D.; Dahmen, C.; Kantlehner, M.; Enderle, A.; Goodman, S. L.; Kessler, H. Titanium Implant Materials with Improved Biocompatibility through Coating with PhosphonateAnchored Cyclic RGD Peptides. ChemBioChem 2005, 6, 2034−2040. (4) Jaehne, E.; Oberoi, S.; Adler, H.-J. P. Ultra Thin Layers as New Concepts for Corrosion Inhibition and Adhesion Promotion. Prog. Org. Coat. 2008, 61, 211−223. (5) Queffélec, C.; Petit, M.; Janvier, P.; Knight, D. A.; Bujoli, B. Surface Modification Using Phosphonic Acids and Esters. Chem. Rev. 2012, 112, 3777−3807. (6) Paniagua, S. A.; Giordano, A. J.; Smith, O. L.; Barlow, S.; Li, H.; Armstrong, N. R.; Pemberton, J. E.; Brédas, J.-L.; Ginger, D.; Marder, S. R. Phosphonic Acids for Interfacial Engineering of Transparent Conductive Oxides. Chem. Rev. 2016, 116, 7117−7158. (7) Li, H.; Winget, P.; Brédas, J.-L. Transparent Conducting Oxides of Relevance to Organic Electronics: Electronic Structures of Their Interfaces with Organic Layers. Chem. Mater. 2014, 26, 631−646. (8) Wöbkenberg, P. H.; Ball, J.; Kooistra, F. B.; Hummelen, J. C.; de Leeuw, D. M.; Bradley, D. D. C.; Anthopoulos, T. D. Low-Voltage Organic Transistors Based on Solution Processed Semiconductors and Self-Assembled Monolayer Gate Dielectrics. Appl. Phys. Lett. 2008, 93, 13303. (9) Bae, E.; Choi, W.; Park, J.; Shin, H. S.; Kim, S. B.; Lee, J. S. Effects of Surface Anchoring Groups (Carboxylate vs Phosphonate) in Ruthenium-Complex-Sensitized TiO2 on Visible Light Reactivity in Aqueous Suspensions. J. Phys. Chem. B 2004, 108, 14093−14101. (10) Silverman, B. M.; Wieghaus, K. A.; Schwartz, J. Comparative Properties of Siloxane vs Phosphonate Monolayers on a Key Titanium Alloy. Langmuir 2005, 21, 225−228. (11) Marcinko, S.; Fadeev, A. Y. Hydrolytic Stability of Organic Monolayers Supported on TiO2 and ZrO2. Langmuir 2004, 20, 2270− 2273. (12) Gao, W.; Dickinson, L.; Grozinger, C.; Morin, F. G.; Reven, L. Self-Assembled Monolayers of Alkylphosphonic Acids on Metal Oxides. Langmuir 1996, 12, 6429−6435. (13) Gawalt, E. S.; Avaltroni, M. J.; Koch, N.; Schwartz, J. SelfAssembly and Bonding of Alkanephosphonic Acids on the Native Oxide Surface of Titanium. Langmuir 2001, 17, 5736−5738. (14) Pang, C. L.; Lindsay, R.; Thornton, G. Chemical Reactions on Rutile TiO2(110). Chem. Soc. Rev. 2008, 37, 2328. (15) Skibinski, E. S.; Song, A.; DeBenedetti, W. J. I.; Ortoll-Bloch, A. G.; Hines, M. A. Solution Deposition of Self-Assembled Benzoate Monolayers on Rutile (110): Effect of π−π Interactions on Monolayer Structure. J. Phys. Chem. C 2016, 120, 11581−11589. (16) Boissezon, R.; Muller, J.; Beaugeard, V.; Monge, S.; Robin, J.-J. Organophosphonates as Anchoring Agents onto Metal Oxide-Based Materials: Synthesis and Applications. RSC Adv. 2014, 4, 35690− 35707. (17) Song, A.; Jing, D.; Hines, M. A. Rutile Surface Reactivity Provides Insight Into the Structure-Directing Role of Peroxide in TiO2 Polymorph Control. J. Phys. Chem. C 2014, 118, 27343−27352. (18) Song, A.; Skibinski, E. S.; DeBenedetti, W. J. I.; Ortoll-Bloch, A. G.; Hines, M. A. Nanoscale Solvation Leads to Spontaneous Formation of a Bicarbonate Monolayer on Rutile (110) Under Ambient Conditions: Implications for CO2 Photoreduction. J. Phys. Chem. C 2016, 120, 9326−9333.



CONCLUSIONS Solution deposition of phenylphosphinic acid creates near-ideal phenylphosphinate monolayers on atomically flat rutile (110) surfaces. These monolayers are highly ordered and covalently bound in a bridged bidentate geometry, as evidenced by XPS measurements of absolute monolayer density, STM images, and the orientations of three nearly orthogonal molecular vibrationsthe P−H stretch vibration and the symmetric and antisymmetric OPO stretch vibrations. Despite the covalent bidentate attachment and significantly higher binding energy than the corresponding carboxylic acid (−2.62 eV calculated for phenylphosphinate vs −1.94 eV for benzoate), a quick H2O rinse removed most of the phosphinate monolayer, demonstrating that hydrolytic stability does not result from covalent attachment alone. The H2O rinse also oxidized ∼25% of the phenylphosphinate monolayer to phenylphosphonate, producing a species that was more resistant to H2O rinsing.



ASSOCIATED CONTENT

S Supporting Information *

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acs.jpcc.7b04167. XPS quantification of phosphorus coverage, calibration of systematic errors in DFT calculations of vibrational energy, comparison of approximations for P 2p XPS calculations, and structures used in DFT calculations (PDF)



REFERENCES

AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected]. E-mail: +1-607-255-3040. ORCID

Melissa A. Hines: 0000-0002-7960-8208 Notes

The authors declare no competing financial interest.



ACKNOWLEDGMENTS This work was supported by the National Science Foundation (NSF) under Award CHE-1303998. ESS was supported by the NSF IGERT program (DGE-0903653). This research used resources of the National Energy Research Scientific Computing Center, a DOE Office of Science User Facility supported by the Office of Science of the U.S. Department of Energy (DEAC02-05CH11231), as well as the Cornell Center for Materials H

DOI: 10.1021/acs.jpcc.7b04167 J. Phys. Chem. C XXXX, XXX, XXX−XXX

Article

The Journal of Physical Chemistry C

(39) Chabal, Y. J. Surface Infrared Spectroscopy. Surf. Sci. Rep. 1988, 8, 211−357. (40) Li, Y.; Galli, G. Vibrational Properties of Alkyl Monolayers on Si(111) Surfaces: Predictions From Ab-Initio Calculations. Appl. Phys. Lett. 2012, 100, 071605. (41) Buchholz, M.; Xu, M.; Noei, H.; Wiedler, P.; Nefedov, A.; Fink, K.; Wang, Y.; Wöll, C. Interaction of Carboxylic Acids with Rutile TiO2(110): IR Investigations of Terephthalic and Benzoic Acid Adsorbed on a Single Crystal Substrate. Surf. Sci. 2016, 643, 117−123. (42) Péchy, P.; Rotzinger, F. P.; Nazeeruddin, M. K.; Kohle, O.; Zakeeruddin, S. M.; Humphry-Baker, R.; Grätzel, M. Preparation of Phosphonated Polypyridyl Ligands to Anchor Transition-Metal Complexes on Oxide Surfaces: Application for the Conversion of Light to Electricity With Nanocrystalline TiO2 Films. J. Chem. Soc., Chem. Commun. 1995, 1, 65−66. (43) Gillaizeau-Gauthier, I.; Odobel, F.; Alebbi, M.; Argazzi, R.; Costa, E.; Bignozzi, C. A.; Qu, P.; Meyer, G. J. Phosphonate-Based Bypyridine Dyes for Stable Photovoltaic Devices. Inorg. Chem. 2001, 40, 6073−6079. (44) Pawsey, S.; McCormick, M.; De Paul, S.; Graf, R.; Lee, Y. S.; Reven, L.; Spiess, H. W. 1H Fast MAS NMR Studies of HydrogenBonding Interactions in Self-Assembled Monolayers. J. Am. Chem. Soc. 2003, 125, 4174−4184. (45) Nilsing, M.; Persson, P.; Lunell, S.; Ojamäe, L. Dye-Sensitization of the TiO2 Rutile (110) Surface by Perylene Dyes: QuantumChemical Periodic B3LYP Computation. J. Phys. Chem. C 2007, 111, 12116−12123. (46) Luschtinetz, R.; Frenzel, J.; Milek, T.; Seifert, G. Adsorption of Phosphonic Acid at the TiO2 Anatase (101) and Rutile (110) Surfaces. J. Phys. Chem. C 2009, 113, 5730−5740. (47) Luschtinetz, R.; Gemming, S.; Seifert, G. Anchoring Functional Molecules on TiO2 Surfaces: A Comparison Between the Carboxylic and the Phosphonic Acid Group. Eur. Phys. J. Plus 2011, 126, 98. (48) Köhler, L.; Kresse, G. Density Functional Study of CO on Rh(111). Phys. Rev. B: Condens. Matter Mater. Phys. 2004, 70, 165405. (49) Sodhi, R. N.; Cavell, R. G. KLL Auger and Core-Level (1s and 2p) Photoelectron Shifts in a Series of Gaseous Phosphorous Compounds. J. Electron Spectrosc. Relat. Phenom. 1983, 32, 283−312. (50) Wendt, S.; Sprunger, P. T.; Lira, E.; Madsen, G. K. H.; Li, Z.; Hansen, J. Ø.; Matthiesen, J.; Blekinge-Rasmussen, A.; Lægsgaard, E.; Hammer, B.; et al. The Role of Interstitial Sites in the Ti 3d Defect State in the Band Gap of Titania. Science 2008, 320, 1755−1759. (51) Park, K. T.; Pan, M.; Meunier, V.; Plummer, E. W. Reoxidation of TiO2(110) via Ti Interstitials and Line Defects. Phys. Rev. B: Condens. Matter Mater. Phys. 2007, 75, 245415.

(19) Clark, I. T.; Aldinger, B. S.; Gupta, A.; Hines, M. A. Extracting Maximum Information From Polarized Surface Vibrational Spectra: Application to Etched, H-Terminated Si(110) Surfaces. J. Chem. Phys. 2008, 128, 144711. (20) DeBenedetti, W. J. I.; Skibinski, E. S.; Hinckley, J. A.; Nedessa, S. B.; Hines, M. A. Cartesian Decomposition of Infrared Spectra Reveals the Structure of Solution-Deposited, Self-Assembled Benzoate and Alkanoate Monolayers on Rutile (110). J. Phys. Chem. C 2016, 120, 24866−24876. (21) Greiner, M.; Kruse, P. Recrystallization of Tungsten Wire for Fabrication of Sharp and Stable Nanoprobe and Field-Emitter Tips. Rev. Sci. Instrum. 2007, 78, 026104. (22) Schmucker, S. W.; Kumar, N.; Abelson, J. R.; Daly, S. R.; Girolami, G. S.; Bischof, M. R.; Jaeger, D. L.; Reidy, R. F.; Gorman, B. P.; Alexander, J.; et al. Field-Directed Sputter Sharpening for Tailored Probe Materials and Atomic-Scale Lithography. Nat. Commun. 2012, 3, 935. (23) Hugenschmidt, M. B.; Gamble, L.; Campbell, C. T. The Interaction of H2O with a TiO2(110) Surface. Surf. Sci. 1994, 302, 329−340. (24) Kohn, W.; Sham, L. J. Self-Consistent Equations Including Exchange and Correlation Effects. Phys. Rev. 1965, 140, A1133− A1138. (25) Perdew, J. P.; Burke, K.; Ernzerhof, M. Generalized Gradient Approximation Made Simple. Phys. Rev. Lett. 1996, 77, 3865−3868. (26) Kresse, G.; Hafner, J. Ab Initio Molecular Dynamics for Liquid Metals. Phys. Rev. B: Condens. Matter Mater. Phys. 1993, 47, 558−561. (27) Kresse, G.; Hafner, J. Ab Initio Molecular-Dynamics Simulation of the Liquid-Metal-Amorphous-Semiconductor Transition in Germanium. Phys. Rev. B: Condens. Matter Mater. Phys. 1994, 49, 14251− 14269. (28) Kresse, G.; Furthmüller, J. Efficiency of Ab-Initio Total Energy Calculations for Metals and Semiconductors Using a Plane-Wave Basis Set. Comput. Mater. Sci. 1996, 6, 15−50. (29) Kresse, G.; Furthmüller, J. Efficient Iterative Schemes for Ab Initio Total-Energy Calculations Using a Plane-Wave Basis Set. Phys. Rev. B: Condens. Matter Mater. Phys. 1996, 54, 11169−11186. (30) Grimme, S.; Antony, J.; Ehrlich, S.; Krieg, H. A Consistent and Accurate Ab Initio Parameterization of Density Functional Dispersion Correction (DFT-D) for the 94 Elements H-Pu. J. Chem. Phys. 2010, 132, 154104. (31) Blöchl, P. E. Projector Augmented-Wave Method. Phys. Rev. B: Condens. Matter Mater. Phys. 1994, 50, 17953−17979. (32) Kresse, G.; Joubert, D. From Ultrasoft Pseudopotentials to the Projector Augmented-Wave Method. Phys. Rev. B: Condens. Matter Mater. Phys. 1999, 59, 1758−1775. (33) Kowalski, P. M.; Meyer, B.; Marx, D. Composition, Structure, and Stability of the Rutile TiO2(110) Surface: Oxygen Depletion, Hydroxylation, Hydrogen Migration, and Water Adsorption. Phys. Rev. B: Condens. Matter Mater. Phys. 2009, 79, 115410. (34) Camellone, M. F.; Kowalski, P. M.; Marx, D. Ideal, Defective, and Gold-Promoted Rutile TiO2(110) Surfaces Interacting with CO, H2, and H2O: Structures, Energies, Thermodynamics, and Dynamics From PBE+U. Phys. Rev. B: Condens. Matter Mater. Phys. 2011, 84, 035413. (35) Porezag, D.; Pederson, M. R. Infrared Intensities and RamanScattering Activities Within Density-Functional Theory. Phys. Rev. B: Condens. Matter Mater. Phys. 1996, 54, 7830−7836. (36) Tersoff, J.; Hamann, D. R. Theory of the Scanning Tunneling Microscope. Phys. Rev. B: Condens. Matter Mater. Phys. 1985, 31, 805− 813. (37) Pang, C. L.; Watkins, M.; Cabailh, G.; Ferrero, S.; Ngo, L. T.; Chen, Q.; Humphrey, D. S.; Shluger, A. L.; Thornton, G. Bonding of Methyl Phosphonate to TiO2(110). J. Phys. Chem. C 2010, 114, 16983−16988. (38) Mielczarski, J. A.; Yoon, R. H. Fourier Transform Infrared External Reflection Study of Molecular Orientation in Spontaneously Adsorbed Layers on Low-Absorption Substrates. J. Phys. Chem. 1989, 93, 2034−2038. I

DOI: 10.1021/acs.jpcc.7b04167 J. Phys. Chem. C XXXX, XXX, XXX−XXX