Article pubs.acs.org/Langmuir
Efficient Chemisorption of Organophosphorous Redox Probes on Indium Tin Oxide Surfaces under Mild Conditions Amélie Forget, Benoît Limoges,* and Véronique Balland* Laboratoire d’Electrochimie Moléculaire, UMR CNRS 7591, Université Paris Diderot, Sorbonne Paris Cité, 15 rue Jean-Antoine de Baïf, F-75205 Paris, Cedex 13, France S Supporting Information *
ABSTRACT: We report a mild and straightforward one-step chemical surface functionalization of indium tin oxide (ITO) electrodes by redox-active molecules bearing an organophosphoryl anchoring group (i.e., alkyl phosphate or alkyl phosphonate group). The method takes advantage of simple passive adsorption in an aqueous solution at room temperature. We show that organophosphorus compounds can adsorb much more strongly and stably on an ITO surface than analogous redox-active molecules bearing a carboxylate or a boronate moiety. We provide evidence, through quantitative electrochemical characterization (i.e., by cyclic voltammetry) of the adsorbed organophosphoryl redox-active molecules, of the occurrence of three different adsorbate fractions on ITO, exhibiting different stabilities on the surface. Among these three fractions, one is observed to be strongly chemisorbed, exhibiting high stability and resistance to desorption/hydrolysis in a freeredox probe aqueous buffer. We attribute this remarkable stability to the formation of chemical bonds between the organophosphorus anchoring group and the metal oxide surface, likely occurring through a heterocondensation reaction in water. From XPS analysis, we also demonstrate that the surface coverage of the chemisorbed molecules is highly affected by the degree of surface hydroxylation, a parameter that can be tuned by simply preconditioning the freshly cleaned ITO surfaces in water. The lower the relative surface hydroxide density on ITO, the higher was the surface coverage of the chemisorbed species. This behavior is in line with a chemisorption mechanism involving coordination of a deprotonated phosphoryl oxygen atom to the non-hydroxylated acidic metal sites of ITO.
■
INTRODUCTION The surface functionalization of transparent conductive oxide materials (e.g., tin-doped indium oxide, ITO, or fluoride-doped tin oxide, FTO) by well-controlled layers of organic molecules is a subject of increasing interest, mostly driven by the wide range of potential applications in optoelectronic devices, photovoltaic cells, and sensors or biosensors.1,2 The emergence of functionalized nanostructured transparent conductive oxide films and their applications in solar cells, liquid crystal displays, light-emitting diodes,3−5 and advanced spectroelectrochemical approaches6−8 has further stimulated the development of new and improved methods of surface modification of these materials. Functionalization of metal oxide surfaces by organic monolayers can be achieved with a variety of functional groups such as carboxylic acids, thiols, alcohols, organosilanes, and phosphonic acids. Specifically, phosphonic acids are believed to form the most robust modification layers with an increased resistance to desorption by hydrolysis as compared to those created from organosilanes or carboxylic acids.9−13 The enhanced stability is attributed to a surface condensation reaction between the phosphonic acid moiety and some specific binding sites on the metal oxide surface, leading thus to chemisorbed layers firmly anchored to the surface through P− © 2015 American Chemical Society
O bonds. Depending on the Lewis acidity of the surface metal ions, two binding mechanisms were proposed:11,14,15 (i) binding where the coordination of the phosphoryl oxygen primarily occurs at a Lewis acidic site on the surface (i.e., to non-hydroxylated surface metal sites) followed by heterocondensation with adjacent surface hydroxyl groups and (ii) binding involving two successive heterocondensation reactions at low Lewis acidic metal sites (i.e., at hydroxylated sites) with adjacent surface hydroxyl groups. Spectroscopic studies and theoretical calculations have suggested that the surface condensation reaction would result preferentially from a bidentate or tridentate binding of the phosphonic acid moiety to the surface.16−18 While there have been reports in which the initial surface hydroxylation state of the metal oxide does not appear to influence the maximal coverage of adsorbed molecules,19,20 it is usually recognized that the surface −OH density remains a key parameter of the functionalization process.20−22 For instance, the chemisorption rate of organophosphonic acids on aluminum oxide has been observed to be accelerated by high surface −OH content.20 The adsorption in Received: September 19, 2014 Revised: January 20, 2015 Published: January 22, 2015 1931
DOI: 10.1021/la503760x Langmuir 2015, 31, 1931−1940
Article
Langmuir low polar organic solvents23 complemented by an annealing step of the modified surfaces22,24 has also been described to increase the extent of heterocondensation of organophosphorus compounds to metal oxide surfaces and to promote the covalent bond formation via mono-, bi-, or even tridentate bonding interactions to the surface. A two-step modification procedure is therefore frequently proposed, including a first step of passive adsorption in an organic solvent followed by a second step of thermal annealing of the modified surface.11,12,16,25,26 The main disadvantage of this process is the long time required (from a few hours to days), even if it could be reduced by performing the functionalization under heating in a low-humidity environment.27,28 There is thus still a need for a surface functionalization method that would produce efficient chemical modification of ITO surfaces with reduced processing times. It would also be advantageous to combine this within a single-step approach operating under mild conditions (i.e., in aqueous solvent and at ambient temperature), therefore avoiding the use of harmful organic solvents and offering improved compatibility with the immobilization of delicate water-soluble biomolecules.29 Although a few studies report on the preparation of modified ITO electrodes by passive adsorption of organophosphorus compounds in water (exclusively achieved with alkylphosphonic compounds),23,28,29 no attempt has been done to quantitatively differentiate the amount of molecules that can be truly chemisorbed from those that can be simply physisorbed, nor to determine to which extent the resulting modified surfaces could be resistant to hydrolysis/desorption in an aqueous buffer. The objective of the present work is thus to fill this gap and to examine the key parameters that may control the adsorption/desorption process in water as well as to determine the maximal surface coverage that can be reached under such conditions. A series of hydrosoluble redox-active molecules bearing a variety of functional groups (i.e., carboxylate, boronate, phosphate, and phosphonate groups) were thus selected to chemically modify commercially available ITO-coated glass electrodes by simple passive adsorption in water at room temperature. Because of the conductive properties of ITO and the presence of a redox-active function on the selected molecules, the surface coverage of adsorbed molecules could be assessed electrochemically, by cyclic voltammetry (CV), and monitored as a function of time for different ITO surface pretreatments and conditioning. A key advantage offered by direct electrochemical detection in contrast to indirect techniques (e.g., indirect monitoring of nonelectroactive adsorbates by impedance spectroscopy)32 is that it allows for quantitative and selective distinction of excess surface adsorbate from unrelated adsorbed species (e.g., solvents, ions, etc.) with a high degree of sensitivity and accuracy. From the present results, we demonstrate the coexistence of three different adsorbate fractions with dissimilar reactivities and stabilities on ITO, including one very strongly chemisorbed. The data also show that the surface coverage of chemisorbed molecules is strongly dependent on the degree of surface hydroxylation and on the acidity of the organophosphorous anchoring moiety. On the basis of this behavior, a possible heterocondensation mechanism is proposed and discussed.
■
ferrocenecarboxylic acid and ferroceneboronic acid, riboflavin (Rb), and riboflavin 5′-monophosphate sodium salt (i.e., flavin mononucleotide FMN)) were purchased from Sigma-Aldrich or Strem Chemicals. The Fc(CH2)2PO(OH)2 and Fc(CH2)2O2P(OH)2 were synthesized from ferrocenylmethyltrimethylammonium iodide following published procedures.33−35 XPS. An ULTRA spectrometer (Kratos Analytical) was used with base pressure in the analytical chamber lower than 4 × 10−8 Pa. A monochromatic Al Kα source (hν = 1486.6 eV) was run at a power of 210 W. The analysis spot was 300 × 700 μm. The resolution of the instrument is 0.55 eV for Ag 3d and 0.70 eV for Au 4f peaks. The photoelectron takeoff angle was 0° with respect to the surface normal. The survey scans were collected for binding energy spanning from 1100 to 0 eV with an analyzer pass energy of 160 eV and a step of 0.4 eV. For the high-resolution spectra the pass energy was 20 eV with a step of 0.1 eV. Since samples were electrically conductive, no charge neutralizer was required. The data are presented as collected without calibration for the binding energy. Instrument software Vision-2 was used for data processing. Spectra were analyzed by using the CasaXPS software. Atomic ratios were estimated by using a Shirley background and instrument relative sensitivity factors (RSF). Major photoelectron lines were used for calculating the composition from the highresolution spectra. When only one peak of the doublet spectra was used (2p3/2 or 3d5/2), then the RSF value was adjusted accordingly. Uncertainties were calculated using standard deviation area and RSF values. The C(1s) and O(1s) peaks were fitted using Gaussian/ Lorentzian GL(30) products and fixed fwhm values, whereas the peak position and intensity were unconstrained. Accordingly, the relative atomic composition of the surface is given with an uncertainty of 10%. Uncertainty on the atomic ratios is less than 5%. Electrochemistry. The as-received ITO-coated glass substrates were cleaned by subsequent immersion of the electrodes in trichloroethylene, acetone, and ethanol for 10 min each at 50 °C. After cleaning, a portion of the ITO surface was delimited by a nail vanish insulating layer in such a way to have a square working electrode area of 0.25 ± 0.05 cm2. The ITO electrodes were functionalized by dipping the active electrode area in a pure aqueous solution containing the desired redox probe (unless otherwise stated, a 100 μM solution of the redox probe in Milli-Q water was used). After ∼15 h immersion in the adsorption solution, the electrodes were thoroughly rinsed first with Milli-Q water and then with a heated (50 °C) buffer solution (45 mM Hepes, 0.32 M KCl, pH 7). The electrodes were then ready for their characterization by cyclic voltammetry. The electrochemical experiments were carried out with an Autolab potentiostat (PGSTAT-12) under ohmic drop compensation in a thermostated double-jacket electrochemical cell (20 °C), under air for the ferrocene-modified electrodes, and under inert atmosphere (argon) for the flavin-modified electrodes. The reference electrode is a saturated calomel electrode (SCE), and the counter electrode is a platinum wire.
■
RESULTS XPS Characterization of the ITO Electrodes. The ITO electrodes were cleaned in organic solvents as described above. This chemical cleaning procedure was reported to be efficient for removing the organic contaminants adsorbed on ITO surfaces.36 Moreover, it has the advantage to avoid ultrasonic cleaning, which is advised if one has to transpose the procedure to nanostructured TCO electrodes. The surface chemical composition of the as-received ITO electrodes (sample I) as well as of ITO electrodes that were beforehand chemically cleaned (sample II-0) or chemically cleaned and then conditioned in Milli-Q water for 1 or 24 h (samples II-1 and II-24, respectively) were all analyzed by XPS. The atomic composition deduced from the survey spectra is given in Table S1. All spectra are characterized by a constant Sn/In intensity ratio of 0.14, indicating that the ITO stoichiometry is not affected by the chemical cleaning procedure.36 Taking into
MATERIALS AND METHODS
Chemical Reagents. ITO-coated glass substrate (8−12 Ω/□; ITO film thickness: 120−160 nm) and chemicals (solvents, 1932
DOI: 10.1021/la503760x Langmuir 2015, 31, 1931−1940
Article
Langmuir
low-energy component while the surface atomic ratios of the others carbon components remains almost constant (%C decreases from 50.5 to 21.1%). A hydration step immediately after the cleaning step results in a slight increase of the two lowenergy XPS components of C(1s). A marked dependence on the hydration time suggests a slight back-contamination of the ITO surface, likely by some organic traces present in the water rather than by dissolved CO2 or carbonates species. To better understand the oxygen composition of the metal oxide surface, the high-resolution O(1s) peak was fit with four components for all ITO samples (Figure 1 and Table S2). According to the literature, the two peaks OA (∼530.4 eV, ΔBE = 85.7 eV vs the main In 3d5/2 peak component in all cases) and OB (∼531.5 eV) are assigned to bulk lattice In2O3-like oxygen, the second one being specifically assigned to oxygen atoms immediately adjacent to oxygen deficiency sites, characteristic of ITO relative to undoped indium oxide.37,38 The OB/(Sn + In) atomic ratio is thus a direct measurement of the oxide defect density at the ITO surface (Table 1). The total (OA + OB)/(Sn + In) atomic ratio is expected to be 1.56 for the mean bulk composition (In2O3)0.78(SnO2)0.22 determined above. The lower experimental values obtained (
Efficient Chemisorption of Organophosphorous Redox Probes on Indium Tin Oxide Surfaces under Mild Conditions Amélie Forget, Benoît Limoges,* and Véronique Balland* Laboratoire d’Electrochimie Moléculaire, UMR CNRS 7591, Université Paris Diderot, Sorbonne Paris Cité, 15 rue Jean-Antoine de Baïf, F-75205 Paris, Cedex 13, France S Supporting Information *
ABSTRACT: We report a mild and straightforward one-step chemical surface functionalization of indium tin oxide (ITO) electrodes by redox-active molecules bearing an organophosphoryl anchoring group (i.e., alkyl phosphate or alkyl phosphonate group). The method takes advantage of simple passive adsorption in an aqueous solution at room temperature. We show that organophosphorus compounds can adsorb much more strongly and stably on an ITO surface than analogous redox-active molecules bearing a carboxylate or a boronate moiety. We provide evidence, through quantitative electrochemical characterization (i.e., by cyclic voltammetry) of the adsorbed organophosphoryl redox-active molecules, of the occurrence of three different adsorbate fractions on ITO, exhibiting different stabilities on the surface. Among these three fractions, one is observed to be strongly chemisorbed, exhibiting high stability and resistance to desorption/hydrolysis in a freeredox probe aqueous buffer. We attribute this remarkable stability to the formation of chemical bonds between the organophosphorus anchoring group and the metal oxide surface, likely occurring through a heterocondensation reaction in water. From XPS analysis, we also demonstrate that the surface coverage of the chemisorbed molecules is highly affected by the degree of surface hydroxylation, a parameter that can be tuned by simply preconditioning the freshly cleaned ITO surfaces in water. The lower the relative surface hydroxide density on ITO, the higher was the surface coverage of the chemisorbed species. This behavior is in line with a chemisorption mechanism involving coordination of a deprotonated phosphoryl oxygen atom to the non-hydroxylated acidic metal sites of ITO.
■
INTRODUCTION The surface functionalization of transparent conductive oxide materials (e.g., tin-doped indium oxide, ITO, or fluoride-doped tin oxide, FTO) by well-controlled layers of organic molecules is a subject of increasing interest, mostly driven by the wide range of potential applications in optoelectronic devices, photovoltaic cells, and sensors or biosensors.1,2 The emergence of functionalized nanostructured transparent conductive oxide films and their applications in solar cells, liquid crystal displays, light-emitting diodes,3−5 and advanced spectroelectrochemical approaches6−8 has further stimulated the development of new and improved methods of surface modification of these materials. Functionalization of metal oxide surfaces by organic monolayers can be achieved with a variety of functional groups such as carboxylic acids, thiols, alcohols, organosilanes, and phosphonic acids. Specifically, phosphonic acids are believed to form the most robust modification layers with an increased resistance to desorption by hydrolysis as compared to those created from organosilanes or carboxylic acids.9−13 The enhanced stability is attributed to a surface condensation reaction between the phosphonic acid moiety and some specific binding sites on the metal oxide surface, leading thus to chemisorbed layers firmly anchored to the surface through P− © 2015 American Chemical Society
O bonds. Depending on the Lewis acidity of the surface metal ions, two binding mechanisms were proposed:11,14,15 (i) binding where the coordination of the phosphoryl oxygen primarily occurs at a Lewis acidic site on the surface (i.e., to non-hydroxylated surface metal sites) followed by heterocondensation with adjacent surface hydroxyl groups and (ii) binding involving two successive heterocondensation reactions at low Lewis acidic metal sites (i.e., at hydroxylated sites) with adjacent surface hydroxyl groups. Spectroscopic studies and theoretical calculations have suggested that the surface condensation reaction would result preferentially from a bidentate or tridentate binding of the phosphonic acid moiety to the surface.16−18 While there have been reports in which the initial surface hydroxylation state of the metal oxide does not appear to influence the maximal coverage of adsorbed molecules,19,20 it is usually recognized that the surface −OH density remains a key parameter of the functionalization process.20−22 For instance, the chemisorption rate of organophosphonic acids on aluminum oxide has been observed to be accelerated by high surface −OH content.20 The adsorption in Received: September 19, 2014 Revised: January 20, 2015 Published: January 22, 2015 1931
DOI: 10.1021/la503760x Langmuir 2015, 31, 1931−1940
Article
Langmuir low polar organic solvents23 complemented by an annealing step of the modified surfaces22,24 has also been described to increase the extent of heterocondensation of organophosphorus compounds to metal oxide surfaces and to promote the covalent bond formation via mono-, bi-, or even tridentate bonding interactions to the surface. A two-step modification procedure is therefore frequently proposed, including a first step of passive adsorption in an organic solvent followed by a second step of thermal annealing of the modified surface.11,12,16,25,26 The main disadvantage of this process is the long time required (from a few hours to days), even if it could be reduced by performing the functionalization under heating in a low-humidity environment.27,28 There is thus still a need for a surface functionalization method that would produce efficient chemical modification of ITO surfaces with reduced processing times. It would also be advantageous to combine this within a single-step approach operating under mild conditions (i.e., in aqueous solvent and at ambient temperature), therefore avoiding the use of harmful organic solvents and offering improved compatibility with the immobilization of delicate water-soluble biomolecules.29 Although a few studies report on the preparation of modified ITO electrodes by passive adsorption of organophosphorus compounds in water (exclusively achieved with alkylphosphonic compounds),23,28,29 no attempt has been done to quantitatively differentiate the amount of molecules that can be truly chemisorbed from those that can be simply physisorbed, nor to determine to which extent the resulting modified surfaces could be resistant to hydrolysis/desorption in an aqueous buffer. The objective of the present work is thus to fill this gap and to examine the key parameters that may control the adsorption/desorption process in water as well as to determine the maximal surface coverage that can be reached under such conditions. A series of hydrosoluble redox-active molecules bearing a variety of functional groups (i.e., carboxylate, boronate, phosphate, and phosphonate groups) were thus selected to chemically modify commercially available ITO-coated glass electrodes by simple passive adsorption in water at room temperature. Because of the conductive properties of ITO and the presence of a redox-active function on the selected molecules, the surface coverage of adsorbed molecules could be assessed electrochemically, by cyclic voltammetry (CV), and monitored as a function of time for different ITO surface pretreatments and conditioning. A key advantage offered by direct electrochemical detection in contrast to indirect techniques (e.g., indirect monitoring of nonelectroactive adsorbates by impedance spectroscopy)32 is that it allows for quantitative and selective distinction of excess surface adsorbate from unrelated adsorbed species (e.g., solvents, ions, etc.) with a high degree of sensitivity and accuracy. From the present results, we demonstrate the coexistence of three different adsorbate fractions with dissimilar reactivities and stabilities on ITO, including one very strongly chemisorbed. The data also show that the surface coverage of chemisorbed molecules is strongly dependent on the degree of surface hydroxylation and on the acidity of the organophosphorous anchoring moiety. On the basis of this behavior, a possible heterocondensation mechanism is proposed and discussed.
■
ferrocenecarboxylic acid and ferroceneboronic acid, riboflavin (Rb), and riboflavin 5′-monophosphate sodium salt (i.e., flavin mononucleotide FMN)) were purchased from Sigma-Aldrich or Strem Chemicals. The Fc(CH2)2PO(OH)2 and Fc(CH2)2O2P(OH)2 were synthesized from ferrocenylmethyltrimethylammonium iodide following published procedures.33−35 XPS. An ULTRA spectrometer (Kratos Analytical) was used with base pressure in the analytical chamber lower than 4 × 10−8 Pa. A monochromatic Al Kα source (hν = 1486.6 eV) was run at a power of 210 W. The analysis spot was 300 × 700 μm. The resolution of the instrument is 0.55 eV for Ag 3d and 0.70 eV for Au 4f peaks. The photoelectron takeoff angle was 0° with respect to the surface normal. The survey scans were collected for binding energy spanning from 1100 to 0 eV with an analyzer pass energy of 160 eV and a step of 0.4 eV. For the high-resolution spectra the pass energy was 20 eV with a step of 0.1 eV. Since samples were electrically conductive, no charge neutralizer was required. The data are presented as collected without calibration for the binding energy. Instrument software Vision-2 was used for data processing. Spectra were analyzed by using the CasaXPS software. Atomic ratios were estimated by using a Shirley background and instrument relative sensitivity factors (RSF). Major photoelectron lines were used for calculating the composition from the highresolution spectra. When only one peak of the doublet spectra was used (2p3/2 or 3d5/2), then the RSF value was adjusted accordingly. Uncertainties were calculated using standard deviation area and RSF values. The C(1s) and O(1s) peaks were fitted using Gaussian/ Lorentzian GL(30) products and fixed fwhm values, whereas the peak position and intensity were unconstrained. Accordingly, the relative atomic composition of the surface is given with an uncertainty of 10%. Uncertainty on the atomic ratios is less than 5%. Electrochemistry. The as-received ITO-coated glass substrates were cleaned by subsequent immersion of the electrodes in trichloroethylene, acetone, and ethanol for 10 min each at 50 °C. After cleaning, a portion of the ITO surface was delimited by a nail vanish insulating layer in such a way to have a square working electrode area of 0.25 ± 0.05 cm2. The ITO electrodes were functionalized by dipping the active electrode area in a pure aqueous solution containing the desired redox probe (unless otherwise stated, a 100 μM solution of the redox probe in Milli-Q water was used). After ∼15 h immersion in the adsorption solution, the electrodes were thoroughly rinsed first with Milli-Q water and then with a heated (50 °C) buffer solution (45 mM Hepes, 0.32 M KCl, pH 7). The electrodes were then ready for their characterization by cyclic voltammetry. The electrochemical experiments were carried out with an Autolab potentiostat (PGSTAT-12) under ohmic drop compensation in a thermostated double-jacket electrochemical cell (20 °C), under air for the ferrocene-modified electrodes, and under inert atmosphere (argon) for the flavin-modified electrodes. The reference electrode is a saturated calomel electrode (SCE), and the counter electrode is a platinum wire.
■
RESULTS XPS Characterization of the ITO Electrodes. The ITO electrodes were cleaned in organic solvents as described above. This chemical cleaning procedure was reported to be efficient for removing the organic contaminants adsorbed on ITO surfaces.36 Moreover, it has the advantage to avoid ultrasonic cleaning, which is advised if one has to transpose the procedure to nanostructured TCO electrodes. The surface chemical composition of the as-received ITO electrodes (sample I) as well as of ITO electrodes that were beforehand chemically cleaned (sample II-0) or chemically cleaned and then conditioned in Milli-Q water for 1 or 24 h (samples II-1 and II-24, respectively) were all analyzed by XPS. The atomic composition deduced from the survey spectra is given in Table S1. All spectra are characterized by a constant Sn/In intensity ratio of 0.14, indicating that the ITO stoichiometry is not affected by the chemical cleaning procedure.36 Taking into
MATERIALS AND METHODS
Chemical Reagents. ITO-coated glass substrate (8−12 Ω/□; ITO film thickness: 120−160 nm) and chemicals (solvents, 1932
DOI: 10.1021/la503760x Langmuir 2015, 31, 1931−1940
Article
Langmuir
low-energy component while the surface atomic ratios of the others carbon components remains almost constant (%C decreases from 50.5 to 21.1%). A hydration step immediately after the cleaning step results in a slight increase of the two lowenergy XPS components of C(1s). A marked dependence on the hydration time suggests a slight back-contamination of the ITO surface, likely by some organic traces present in the water rather than by dissolved CO2 or carbonates species. To better understand the oxygen composition of the metal oxide surface, the high-resolution O(1s) peak was fit with four components for all ITO samples (Figure 1 and Table S2). According to the literature, the two peaks OA (∼530.4 eV, ΔBE = 85.7 eV vs the main In 3d5/2 peak component in all cases) and OB (∼531.5 eV) are assigned to bulk lattice In2O3-like oxygen, the second one being specifically assigned to oxygen atoms immediately adjacent to oxygen deficiency sites, characteristic of ITO relative to undoped indium oxide.37,38 The OB/(Sn + In) atomic ratio is thus a direct measurement of the oxide defect density at the ITO surface (Table 1). The total (OA + OB)/(Sn + In) atomic ratio is expected to be 1.56 for the mean bulk composition (In2O3)0.78(SnO2)0.22 determined above. The lower experimental values obtained (
