ARTICLE pubs.acs.org/Langmuir
Directing Alkyl Chain Ordering of Functional Phosphorus Coupling Agents on ZrO2 Christoph J. Lomoschitz,† Bernhard Feichtenschlager,† Norbert Moszner,‡ Michael Puchberger,† Klaus M€uller,§ Matthias Abele,§,|| and Guido Kickelbick*,^ †
Vienna University of Technology, Institute of Materials Chemistry, Getreidemarkt 9/165-AC, A-1060 Vienna, Austria Ivoclar Vivadent AG, Bendererstrasse 2, FL-9494 Schaan, Principality of Liechtenstein § University of Trento, Department of Materials Engineering and Industrial Technologies, Via Mesiano 77, I-38123 Trento, Italy University of Stuttgart, Institute of Physical Chemistry, Pfaffenwaldring 55, D-70569 Stuttgart, Germany ^ Saarland University, Inorganic Solid State Chemistry, Am Markt Zeile 3, D-66125 Saarbr€ucken, Germany
)
‡
bS Supporting Information ABSTRACT: ZrO2 powder (6.6 m2/g) was modified using polymerizable phosphorus-based coupling agents (P-CAs) (i.e., phosphonic acid, phosphoric acid, and bis-phosphonic acid), resulting in densely grafted layers as determined by thermogravimetry and elemental analysis (up to 4.2 molecules/nm2). The applied P-CAs contained a methacrylate group, which led to the covalent incorporation of a polymerizable moiety into the grafted layer. To direct the ordering of the alkyl chains in the layer, three different approaches were evaluated with respect to their structuredirecting ability by means of FT-IR and nitrogen sorption at 77 K: (i) variation of the chain length, (ii) variation of the anchoring group and (iii) comodification with a defined amount of a nonfunctional phosphonic acid (variation of the functional/nonfunctional acid ratio). It was shown that the chain length and anchoring group size have significant effects on the alkyl chain ordering and morphology of the layer.
1. INTRODUCTION The surface modification of metal oxides is of great interest considering their significance in biomaterials,1 catalysts,2 pigments,3 inorganic-organic hybrid materials,4 and membranes.5 A major approach for the surface modification of metal oxides uses the covalent attachment of organic coupling molecules, which may form monolayers on the surfaces of these materials. Over the last few decades, organosilanes of the type RnSiX3 - n (X = H, Cl, alkoxy) have received much attention as coupling agents in metal oxide hybrid materials. On the basis of well-known Si chemistry and the apolar character and hydrolytic stability of the Si-C bond, a large variety of different organic functionalities were developed. The mentioned organosilanes can react with surface OH groups of the oxides via condensation reactions, resulting in the formation of a covalent bond between the coupling agent and oxide substrate.6,7 An example of an important coupling agent is 3-(trimethoxysilyl)propyl methacrylate (γ-MPS),8 which is used in a variety of applications to promote the adhesion of polymers to inorganic surfaces such as glass or metal oxide particles. However, Fadeev et al. have shown that the stabilization of R-SiX3 coupling molecules on metal oxide surfaces (ZrO2 and TiO2) is mainly caused by the lateral cross condensation of the silanes, generating Si-O-Si bonds (homocondensation), whereas modification with R3-SiX r 2011 American Chemical Society
(hetercondensation-generation of Ti/Zr-O-Si bonds) results in less-stable layers9 because of the sensitivity of the Ti/Zr-OSi bonds toward hydrolysis.10 Phosphorus coupling agents (P-CAs), such as phosphonic and phosphoric acid derivates, are superior in many cases because they (i) do not undergo homocondensation but (ii) interact exclusively with surface hydroxyls or coordinatively unsaturated surface metal atoms of metal oxides such as TiO2,11,12 ZrO2,11 Ta2O5,13 Y2O3,14 Fe3O4,15 lanthanide oxides,16 and perowskites (BaTiO3)17 to form P-O-M bonds. A variety of experiments proved the high stability of the P-O-M (M = Ti, Zr) bond under various conditions such as UV irradiation,12 thermal treatment,18 and hydrolysis at different pH values.9 Despite the high stability of the P-O-M bond observed, results published in literature about the binding modes of the phosphorus-coupling molecule headgroup are not unambiguous. Although the first experiments using FT-IR spectroscopy suggested a tridentate mode,5 later studies using 31P or 17O MAS NMR,11,19 ToFSIMS,20 and XPS21 suggested that a mixture of monodentate and
Received: November 14, 2010 Revised: January 6, 2011 Published: February 24, 2011 3534
dx.doi.org/10.1021/la104527b | Langmuir 2011, 27, 3534–3540
Langmuir Scheme 1. Investigated Mechanisms to Direct the Alkyl Chain Ordering of Functional P-CA Layers
bidentate bonding modes is preferred. Bridging bidentate modes seem to be favored over chelating bidentate bonds.19,21 In this study, we discuss the covalent grafting of P-CA molecules carrying a polymerizable methacrylate moiety on ZrO2, generating highly functional surfaces. We mainly focused on investigations of different methods to direct the ordering of functional alkyl chains that may be an effective way to overcome the interdigitation of the layers (zipper effect) at surfaces with high curvatures, such as particles in the entire nanometer size regime,22,23 or at surfaces with low curvatures, on which bilayer formation is promoted by hydrophobic interactions over a large surface area.24 Factors that direct orientation and alkyl chain ordering have been touched in the literature for the ZrO2phosphonate system with respect to the variation in temperature for different chain lengths25 as well as for simple functional groups.26 For Au-thiol systems fundamental studies of comodification on planar surfaces have been published, showing that this approach leads to layers for which the ordering and morphology may be accurately tailored.27,28 Recent work in our group focused on the comodification of methyl-terminated alkylphosphonic acids with other nonfunctional phosphonic acids, showing the observations of the Au-thiol system to be applicable to the ZrO2-phosphonate system.29 However, until now, no literature has been published that elucidates the influence of different approaches to direct alkyl chain ordering in monolayers on ZrO2 that contain a complex functionality (e.g., methacrylate). Therefore, three different mechanisms were investigated that offer the possibility to tailor the properties of functional surfaces accurately: (i) the variation of the chain length between the polymerizable moiety and the anchoring group, (ii) the introduction of an additional anchoring group, creating a bulky anchoring moiety, and (iii) the concurrent reaction of a nonfunctional P-CA with a polymerizable P-CA, disturbing the alkyl chain ordering of the layers (Scheme 1).
2. EXPERIMENTAL SECTION 2.1. Materials and Methods. All chemicals used for the syntheses of molecular compounds were obtained by commercial suppliers (Aldrich, Fluka, ABCR, and Acros) and used as received. 10-Methacryloyloxydecyl dihydrogen phosphate (MC10-OPO3H2) and yttrium -stabilized ZrO2 (tetragonal, 6.6 m2/g) were provided by Ivoclar Vivadent AG.
ARTICLE
Dichloromethane, THF, toluene, and methanol were dried in a commercial system (PureSolv). Manipulations under an inert gas atmosphere were carried out by applying standard Schlenk techniques or using a glovebox. Fourier transform infrared (FT-IR) spectroscopic measurements were performed on a Bruker Tensor 27 spectrometer under ambient conditions (at least 32 scans at a resolution of 4 cm-1) using an ATR unit (ZnSe optics). Liquid NMR spectra were recorded on a Bruker AVANCE 250 (250.13 MHz 1H, 62.89 MHz 13C, 101.25 MHz 31P) spectrometer equipped with a 5 mm broadband probe head and a z-gradient unit. Solid-state NMR spectra were recorded on a Bruker AVANCE DPX300 (Vienna) or CXP 300 (Trento) equipped with a 4 mm broadband MAS probe head (299.87 MHz 1H, 75.40 MHz 13C, 121.39 MHz 31P). Spectra were acquired using magic angle spinning (MAS), cross-polarization (CP), and high-power proton decoupling at a rotor spinning rate of 4-9 kHz. Thermogravimetric analyses were performed on a Netzsch Iris TG 209 C in a platinum crucible at a heating rate of 10 K/min under synthetic air. The calculation of the grafting density for P-CAs on the ZrO2 surface was carried out using the mass loss between 200 and 700 C (Supporting Information). Elemental analyses were carried out at the Microanalytical Laboratory of the University of Vienna. Nitrogen sorption measurements were carried out using a Micromeritics ASAP 2020 or ASAP 2010 instrument. The samples were degassed for at least 5 h at 40 C prior to the measurement. The surface area and CBET were calculated using the adsorption model by Brunauer, Emmet, and Teller (BET)30 in the relative pressure region between 0.05 and 0.2 with five to seven points. The reported CBET refers to correlation coefficients of the respective linear fits of at least 0.9999. Random control samples showed excellent reproducibility with an error of less than 2%. Scanning electron microscopy (SEM) was performed on a JEOL6400 instrument using a tungsten filament. Gold vapor was deposited on the samples prior to analysis. 2.2. Syntheses. Syntheses for the polymerizable phosphonic,31-33 bis-phosphonic,34,35 and phosphoric acids36 were carried out according to literature-known procedures or by their adaptation. Synthesis details and spectroscopic data for the free acids as well as for the intermediate products are given in the Supporting Information. 2.3. Modification of ZrO2. ZrO2 was modified by adapting a literature-known procedure.12 In a typical experiment, 1.00 g of ZrO2 was dispersed in 5 mL of distilled water using an ultrasonic finger. At the same time, a defined amount of the respective P-CA was dissolved in 15 mL of methanol, which was calculated to be the 5-fold amount necessary to give a complete monolayer, assuming a required area of 0.24 nm2 per phosphoric/phosphonic acid group.5 Consequently, bis-phosphonic acids were assumed to require at least 0.48 nm2/molecule. For the comodification approach, the total amount of P-CAs corresponded to the 5-fold excess used to form a theoretically dense monolayer. The methanolic solution was filtered and mixed with the aqueous particle dispersion. The mixture was brought to a pH of between 2 and 3 using either conc. HCl or NaOH and reacted by vigorous stirring for 3 days at ambient temperature. Then, the particles were separated by centrifugation. The supernatant was discarded, and the particles were dispersed in 20 mL of ethanol using a vortex shaker before being centrifuged again. This procedure was repeated again in ethanol and in water. After the last washing cycle, the particles were dried at ambient temperature in vacuo in a desiccator over P4O10. Typical yields ranged from 90 to 95% because of losses in the washing procedure. Grafting densities were calculated from mass losses in TGA studies assuming complete decomposition of the organic part of the P-CAs between 200 and 700 C (Supporting Information). TGA studies were 3535
dx.doi.org/10.1021/la104527b |Langmuir 2011, 27, 3534–3540
Langmuir
ARTICLE
Figure 1. Adsorption/desorption branch of nitrogen sorption at 77 K with the BJH pore size distribution inserted (left) and a SEM image of ZrO2 used (right).
Scheme 2. P-CAs Used in Modification Experiments (Left to Right: MCnPO3H2, MCnOPO3H2, and MCnbis-PO3H2)
Table 1. Grafting Density of ZrO2 Samples Modified with Different P-CAs grafting density (CA/nm2)
confirmed by elemental analyses. In both cases, it was assumed that the remaining phosphorus on the particle surface was phosphate.18
3. RESULTS AND DISCUSSION ZrO2 (tetragonal crystal phase) was modified with a variety of P-CAs containing polymerizable groups at the end of an alkyl chain spacer. Nitrogen sorption measurements of the unmodified powder at 77 K exhibited a type II isotherm with H3 hysteresis. Such sorption isotherms are typical of macroporous systems, such as aggregated particles in the micrometer size regime (Figure 1 left).37 An evaluation of the Barret-Joyner-Halenda (BJH) pore size distribution of the adsorption and desorption branch revealed that pores of at least 10 nm diameter were present and microporosity could be ruled out. The surface area was measured to be 6.6 m2/g according to the BET method.30 Electron microscopy investigations confirmed the results found with nitrogen sorption experiments (Figure 1 right). The P-CAs used for surface functionalization possessed a tridentate grafting group (PO3), a polymerizable bulky tail (methacrylate group), and an alkyl spacer of variable chain length (CH2)n (n = 2, 6, 11) for phosphonic acid and O-(CH2)n (n = 2, 5, 10) for phosphoric acid, respectively (Scheme 2). In addition, coupling agents were prepared that contained two phosphonic acid groups, which were potentially able to bind in a hexadentate manner. The ZrO2 powder was modified by simply mixing an aqueous dispersion of ZrO2 powder with a methanolic solution of P-CA, resulting in a methanol/water ratio of 3:1, and stirring at room temperature for 3 days at pH 2 to 3. The methanolic solution contained a 5-fold excess of P-CA to form a theoretically dense monolayer. The interaction of the P-CAs with the ZrO2 particles was evidenced using FT-IR and 31P MAS NMR spectroscopy. It has to be emphasized that the comparison of the results obtained 3.1.
Bonding
toward
ZrO2. Yttrium-stabilized
elemental analysis
TGA
MC2-OPO3
1.4
1.5
MC5-OPO3
2.8
3.4
MC10-OPO3
3.2
3.4
MC2-PO3
2.5
2.7
MC6-PO3
3.2
3.3
MC11-PO3
4.0
4.3
MC6-bisPO3
2.3
2.6
MC11-bisPO3
2.2
2.3
with the published literature in this field suggested that a variety of different bonding states on the ZrO2 surface were present.11,12,17,19 On the basis of the spectral data, the predominant bonding state for monophosphonic acids was bidentate, whereas phosphoric acid-based P-CAs seemed to bond in a predominantly tridentate manner. Bis-phosphonic acids were found to attach by both anchoring groups, resulting in a predominant bis-bidentate binding mode. It has to be noted that phosphoric acid-based P-CAs formed considerable amounts of a bulk salt phase during the modification of the ZrO2 surface, similar to reports in the literature.11,12 For the interested reader, a detailed discussion of surface attachment with the corresponding spectra can be found in the Supporting Information (Figures SI 1-5 and Table SI 1). The grafting density of the P-CAs was investigated using TGA and elemental analysis (Table 1). The values were determined by assuming complete decomposition of the alkyl chain, and the anchoring group remained on the surface as phosphate groups.18 The results of the two methods correlated well with each other, and deviations of only 0.2 molecules/nm2 were observed. The values obtained were near a theoretically dense monolayer (∼4.2 P-CA/nm2), given the assumption of a required bonding area of 0.24 nm2 per phosphonate/phosphate group.5 It was apparent that the grafting density could be related to the length of the alkyl spacer in the P-CA. It was also observed that in all cases the phosphate coupling molecules gave lower grafting densities than the respective phosphonate coupling molecules, which is 3536
dx.doi.org/10.1021/la104527b |Langmuir 2011, 27, 3534–3540
Langmuir
ARTICLE
Table 2. Comparison of C-H Stretching Vibrations of ZrO2 Samples Modified with Different P-CAs Determined with FT-IR C-H stretching vibration (cm-1) asymmetric
symmetric
alkane (liq)
2928
2856
alkane (cryst)38
2920
2850
2940 2926
2860 2854
MC6-PO3
2931
2860
MC11-PO3
2923
2852
MC6-bisPO3
2935
2861
MC11-bisPO3
2928
2857
38
MC2-OPO3 MC5-OPO3 MC10-OPO3 MC2-PO3
explained by the replacement of a methylene unit by an oxygen atom (vide infra). In addition, differences in the thermal stability were noticed. Whereas grafted phosphates of short and medium chain length started to decompose between 250 and 260 C, the corresponding phosphonates started to decompose between 260 and 270 C, which was most likely caused by the higher stability of the P-C bond. The difference was even larger between the long-chain coupling molecules: the thermally induced decomposition of MC10-OPO3-grafted layers took place between 250 and 260 C, whereas MC11-PO3-grafted layers degraded at approximately 280 C. This behavior was attributed to (i) the superior thermal stability of the P-C bond compared to that of the P-O-C bond and (ii) the higher grafting density of phosphonates resulting in a higher onset temperature of the decomposition. Grafting densities of bis-phosphonic acids were significantly lower than those of the respective monophosphonic acids, showing that the second anchoring group had a strong influence on bonding to the ZrO2 substrate. 3.2. Tailoring of Alkyl Chain Ordering by Variation of the Chain Length and Headgroup. The C-H stretching region holds information about the conformation of the alkyl chain spacer. It is known that crystalline alkanes give signals at lower wavenumbers than do liquid alkanes as a result of van der Waals interactions between the chains and the increasingly predominant all-trans conformation.38 It has been well established to use this information for the characterization of self-assembled monolayers (SAMs).39 Table 2 shows values of the symmetric and asymmetric C-H stretching vibrations, indicating that with increasing chain length the ordering of the alkyl chains increases as well. The comparison of the systems with long alkyl spacers (MC10-OPO3 and MC11-PO3) with the systems with medium alkyl spacers (MC5-OPO3 and MC6-PO3) indicated that ordering was strongly dependent on the alkyl spacer length.25 Lower wavenumbers of the C-H stretching signals clearly indicated that longer alkyl chains induced higher ordering of the molecules on the surface as a result of van der Waals interactions because it is known that the ordering of alkyl chains leads to a shift to lower wavenumbers in FT-IR spectroscopy.38 Regarding the C-H
Figure 2. Photograph of the phase distribution of differently modified ZrO2 particles between toluene and water after vigorous shaking.
stretching vibrations, it seemed that P-CAs with a phosphonate headgroup tended to give layers of higher alkyl chain ordering than did the phosphates. However, the ordering of the presented layers was still somewhat lower than that of nonfunctional methyl-terminated phosphonate SAMs, which is most likely to be explained by the bulkier methacrylate group.9,11 The results for the C2 derivates are not presented in Table 2 because their chains were too short to present all-trans configurations. It was observed that alkyl chain ordering on ZrO2 surfaces modified with P-CAs with a bulky functionality at the chain end is still possible to some extent using chain lengths of at least 10 CH2 units. C-H stretching vibrations indicated that the headgroup of the coupling molecules had a significant influence on the ordering of the resulting layer. No direct comparison of phosphate and phosphonate groups has been carried out in the literature. An evaluation of the work of Gawalt et al.,40 who used octadecyl phosphonic acid (νa 2914 and νs 2846 cm-1), and Spori et al.,21 who used octadecyl phosphoric acid (νa 2919 and νs 2850 cm-1), showed that on flat titania/oxidized titanium substrates the phosphonic acid comprised higher ordering than did the phosphoric acid. Our results confirm that even the replacement of a CH2 unit by oxygen next to the anchor group on the surface decreased the grafting density and thus the ordering significantly, as can be seen from the comparison of phosphonate with phosphate headgroups. These observations became even more striking if the C-H stretching vibrations of bis-phosphonic acid-modified ZrO2 were evaluated. The bulky anchoring groups seemed to have a strong influence on the ordering of the alkyl chains for long-chain molecules, most likely because a dense packing of the chains was not possible as a result of the larger space between the chains based on the large-area anchoring groups. For lower chain lengths, this effect became less apparent becauase ordering was intrinsically less as a result of lower van der Waals interactions. These results are well related to the conclusions drawn from the thermogravimetric analyses, where a higher grafting density of long-chain P-CAs was found, which suggested higher ordering of the alkyl chains. Long-chain P-CAs are amphiphilic molecules preferably oriented at the interfaces of liquids. Thus, the surface polarity of ZrO2 intrinsically changes upon modification with such molecules. This became striking when the modified particles were mixed with water. Under these conditions, the particles tended to agglomerate and float on the aqueous phase (Figure 2). The change in surface polarity was also monitored using nitrogen sorption at 77 K and by evaluating the obtained isotherm with the BET model.30 Besides the specific surface 3537
dx.doi.org/10.1021/la104527b |Langmuir 2011, 27, 3534–3540
Langmuir
ARTICLE
area, the BET model holds information about the absorption properties of the substrate toward the adsorbed gas, which is expressed by the CBET value. The latter is related to the adsorption enthalpy of the adsorbant on the respective adsorbate.30 Because physisorption occurs via polarization processes, the CBET using nitrogen is usually high (100-300) for polarizing surfaces (e.g., metal oxides), whereas it is low (
