ARTICLE pubs.acs.org/Langmuir
Metastable Patterning of Plasma Nanocomposite Films by Incorporating Cellulose Nanowhiskers P. Samyn and M.-P. Laborie* Institute for Forest Utilization and Works Science, Freiburg University, Werthmannstrasse 6, 79085 Freiburg, Germany
A. P. Mathew Department of Applied Physics and Mechanical Engineering, Division of Manufacturing and Design of Wood and Bionanocomposites, Lulea University of Technology, SE-97187 Lulea, Sweden
A. Airoudj, H. Haidara, and V. Roucoules Institut de Science des Materiaux de Mulhouse 15, Universite de Haute-Alsace, Rue Jean Starcky BP2488, 68057 Mulhouse Cedex, France
bS Supporting Information ABSTRACT: A new method is presented for developing patterned, thin nanocomposite films by introducing cellulose nanowhiskers during the pulsed plasma polymerization of maleic anhydride. Metastable film structures develop as a combination of dewetting and buckling phenomena. By controlling the maleic anhydride monomer to cellulose nanowhisker weight ratio, the whiskers can be incorporated into a homogeneously covering patterned polymer film. Excess nanowhiskers are required to prevent complete dewetting and deposit dimensionally stable films. The formation of anchoring points is assumed to stabilize the film through a “pinning” effect to the substrate. The latter control the in-plane film stresses, similar to the effects of surface inhomogeneities such as artificial scratches. The different morphologies are evaluated by optical microscopy, AFM, contact angle measurements, and ellipsometry. Further analysis by infrared spectroscopy and XPS suggests esterification between the maleic anhydride and cellulose moieties.
1. INTRODUCTION Lateral surface patterns in micro- and nanoscale ranges are relevant in microsystems13 or biological devices4 because adhesion, wettability, biocompatibility, friction, and optical characteristics, among others, can be controlled. Structural sizes down to 100 nm can be produced by photolithography5 or nanolithography,6 whereas smaller features of 1 to 10 nm develop by molecular assembly7,8 or a combination of both.9,10 The chemical self-assembly relies on interactions between a monolayer and a substrate,11 whereas physical self-assembly mechanisms are driven by mechanical stress,12 capillary forces,13 dispersion forces,14 or crystallization and dewetting.15 Thin polymer films are often patterned by controlled phase separation on surfaces with regions of different surface tension16,17 by the arrangement of sequences in block copolymers.18 However, many of the above patterning techniques require synthetic polymers and the use/ disposal of solvents to deposit thin films by solvent casting or spin coating. As the trend for sustainable material design continues, new approaches to creating polymer films should include alternative solvent-free conditions as offered by plasma modification. Similar to natural phenomena such as human skin deformation,19,20 complex film patterns may develop by spontaneous wrinkling or buckling mechanisms under stress,21,22 which r 2011 American Chemical Society
mainly result from property gradients between the substrate and its surface layer.23 What might have initially been considered to be an undesirable defect24 is nowadays exploited as an opportunity to control the surface topography.2528 This is not only important for reproducing functional surfaces29 or templating30 but also for better balancing the adhesion and delamination properties of thin films, providing insight into the interfacial fracture toughness31 and determining mechanical properties.32,33 Indeed, the modulus of thin polymer films can be estimated from the buckling pattern.34 The compression and relaxation of stresses in buckled films are better understood from both analytical models35,36 and finite-element analysis.37 The film stresses are often generated by artificial stretching,38 curing,39 annealing above the glass-transition temperature,40 or swelling in a solvent,41 sometimes in combination with soft mold imprints.42 The buckling-induced waves generally develop for rigid films on a soft substrate43 or for elastic films on viscoelastic substrates44,45 and soft polymer films on rigid substrates.46,47 Received: July 2, 2011 Revised: October 19, 2011 Published: November 07, 2011 1427
dx.doi.org/10.1021/la202503h | Langmuir 2012, 28, 1427–1438
Langmuir
ARTICLE
Figure 1. Schematic plasma reactor setup with the inlet of monomer mixtures and substrate positioning (not to scale).
Bulk samples and films of synthetic polymers such as poly(dimethylsiloxane) can be directly patterned by oxidizing plasma48 or argon plasma.49 The patterns generally occur through the oxidation of the polymer surface, producing a silica-rich surface layer or a cross-linked network,50 and may control local wetting properties.51 In those situations, the top layer is under lateral compressive stress after cooling because of different thermal contraction relative to the underlying polymer. In many experiments, buckling during plasma treatment depends on the substrate thickness52 and is better controlled by the relaxation of a prestrained elastomeric substrate in uni- or biaxial mode.53 Another way to control buckling patterns in elastomeric substrates includes focused ion beam irradiation54,55 or metal implantation.56 Thus, local stress concentrations around incorporated defects generally provide a versatile bottom-up way to create and control discrete topographical morphologies. The specific aim of this work is to design patterned thin films under the codeposition of crystalline cellulose nanowhiskers or CNW5760 during the plasma polymerization of maleic anhydride or MA.61 The MA monomers contain a reactive double bond with additional functional anhydride groups and have been used in various applications,62,63 especially for biosurface engineering.64 For this precursor system, low-energy inputs are required to polymerize MA because the anhydride group may dissociate,65 making the pulsed plasma process particularly adequate.66 In a trial to compatibilize hydrophilic CNW with hydrophobic polymers, we aimed to derivatize the hydroxyl groups of CNW under plasma polymerization. However, we observed that the CNW rather mixes into a polymer film and allows for surface patterning. A further study of this system is extremely interesting because it allows the incorporation of moderate amounts of renewable polymer into engineered surface layers. At present, cellulose nanofibers are applied as a reinforcement in polymeric composites67,68 or for the formation of model films,6974 but their ability to structure plasma films topographically has not been explored to date. In this report, we demonstrate that incorporating CNW into plasma-polymerized nanocomposite films allows for surface patterning by buckling.
2. EXPERIMENTAL SECTION 2.1. Materials. Cellulose nanowhiskers (CNWs) were produced by the 63% sulfuric acid hydrolysis of microcrystalline cellulose (MCC, VIVAPUR 105 from JRS Pharma) using the procedure reported by
Bondeson et al.75 After the characterization of the aqueous suspension and the CNW morphology, the CNWs were freeze-dried and stored in the open air until further use. Maleic anhydride (MA) monomer powder (Prolabo 99.5% purity, used as received) was mixed with crushed CNW in different weight ratios, including pure MA, MA + 50 wt % CNW, MA + 200 wt % CNW, and pure CNW. These mixtures were plasma deposited onto single-side-polished Æ100æ silicon wafers bearing a native oxide layer (∼2 nm) or gold-coated wafer substrates for better detection of the film during reflection-mode infrared spectroscopy. 2.2. Procedures. Thin films were deposited by pulsed plasma polymerization. The experiments were done in an electrodeless cylindrical glass reactor enclosed in a Faraday cage (Figure 1), as used elsewhere.76,77 The reactor chamber was connected to a monomer gas inlet on one side and a Pirani pressure gauge on the other side in series with a two-stage rotary pump connected to a liquid nitrogen cold trap. The substrates were placed onto a glass support and centered along the externally wound copper coil (16 cm from the gas inlet). The latter was connected to an L-C matching network and radio frequency generator providing an output frequency of 13.56 MHz. The shape of the electrical pulses was monitored with an oscilloscope, and the average power P delivered to the system was calculated according to eq 1, with Pp being the average continuous wave power output, DC being the duty cycle, ton being the pulse-on time, and toff being the pulse-off time: P ¼ Pp DC ¼ Pp
ton ton þ tof f
ð1Þ
Before each experiment, the reactor was cleaned for 30 min with a highpower air plasma (P = 60 W) treatment. The monomer reactor tubes with loadings of pure MA, pure CNWs, or MA + CNW powder mixtures were subsequently connected to the reaction chamber. The monomer vapor was introduced into the reactor by sublimation under evacuation at constant reactor pressures of 0.1 to 0.2 mbar and a constant flow rate of approximately 1.6 109 kg s1. The plasma was initiated through a high-frequency generator and was applied for a total time of 30 min at an optimized wave power output of Pp = 20 W, a reflected power of 2 W maximum, a pulse frequency 816 Hz, and a duty cycle of DC = 2% (ton = 25 μs, toff = 1200 μs). These parameters were obtained after the preliminary optimization of the plasma power for Pp = 10 to 60 W and duty cycles DC = 2, 25, 50, and 100% (where the DC was varied by appropriately increasing ton), as presented elsewhere.78 Upon the completion of deposition, the radio frequency generator was switched off while the monomer continued to flow for about 2 min prior to venting to atmospheric pressure and unloading the samples. This step allows us to prevent the reaction of residual free radicals in the plasma polymer thin film with undesirable atmospheric atoms. 1428
dx.doi.org/10.1021/la202503h |Langmuir 2012, 28, 1427–1438
Langmuir 2.3. Characterization. The morphologies of CNWs and deposited thin films were studied under atmospheric conditions (23 C, 50% RH) by optical light microscopy (Leitz Metallux II, Wetzlar, Germany) and AFM measurements in tapping mode (Nanoscope IV, Veeco (now Bruker Nano)). Tips with a stiffness of k = 48 N 3 m1 and a resonance frequency of fo = 190 kHz were used (Nanoworld, type NCLR). The images were captured at an amplitude ratio of 1.2, a scanning rate of 0.88 Hz, and 512 data points per line. The data were further processed with WSxM 4.0 software to determine the roughness parameters and Fourier transform after a first-order plane fit to compensate for the sample tilt. The reported values for wavelengths and amplitudes are averages of triplicates with the standard deviation as error intervals. The thickness d and refractive index n of the plasma films were both estimated by ellipsometry (Multiskop M-033k001, Physik Instruments). The cross-section of the HeNe laser beam (632.8 nm) has a spot size of about 1 mm2, and measurements were averaged over three places on the sample. Similar features were observed for the films deposited onto silicon or gold-coated silicon substrates. Polarization-modulated infrared reflection absorption spectroscopy (PM-IRRAS) was carried out on the films deposited onto gold-coated silicon wafers using a Bruker IFS 66/S infrared spectrometer equipped with a PMA 37 polarization modulation mode and a nitrogen-cooled MCT detector. The infrared beam was p polarized with a ZnSe wire grid polarizer (Specac) before passing through a photoelastic modulator (Hinds Instruments). The spectra were averaged from 200 scans at a resolution of 4 cm1, recorded at an angle of incidence of 82.5 relative to the surface normal. The CNWs were characterized on the same equipment in attenuated total reflection mode (ATR) with a germanium crystal. A monoreflection technique was used with a germanium MIRacle from Pike Technologies Inc. (Madison, WI, U.S.). X-ray photoelectron spectroscopy (XPS) spectra were recorded with a VG SCIENTA SES-200 spectrometer equipped with a concentric hemispherical analyzer. The incident radiation used was generated by a monochromatic Al Kα X-ray source (1486.6 eV) operating at 420 W (14 kV, 30 mA). Photoemitted electrons were collected at a takeoff angle of 90 from the substrate, with electron detection in constant analyzer energy mode. The survey spectrum signal was recorded with a pass energy of 500 eV, and for high-resolution spectra (C 1s and O 1s), the pass energy was set to 100 eV. The analyzed surface area was approximately 3 mm2, and the base pressure in the analysis chamber during experimentation was about 109 mbar. Peak fitting was carried out with mixed GaussianLorentzian (30%) components with equal full-widthat-half-maximum (fwhm) values using CASAXPS software. The binding energy of the CHx component in the C 1s region was set to 285.0 eV and used for referencing. The surface composition expressed in atom % was determined using integrated peak areas of each component and taking into account the transmission factor of the spectrometer, the mean free path, and the Scofield sensitivity factors of each atom (C 1s, 1.00; O 1s, 2.93). Static contact angle measurements were made on a DSA 100 system (Kr€uss, Hamburg Germany), depositing 2 μL high-purity water drops. The droplets were fitted with a tangent method, and three independent measurements were made on films deposited on silicon substrates.
3. RESULTS 3.1. Cellulose Nanowhisker Preparation. The structure and size distribution of the CNWs are analyzed by AFM in Figure 2. The height (Figure 2a) and amplitude (Figure 2b) images were obtained by placing a droplet of the aqueous suspension (0.5 wt % CNW) on a mica surface and air drying, showing well-isolated, dispersed whiskers in the nanometer-scale range. To avoid tipbroadening effects, the diameters were determined from height measurements on a height image. They are in the range of 4 to
ARTICLE
Figure 2. Characterization of synthesized CNW components: (a) 3 3 μm2 AFM height image, (b) 3 3 μm2 AFM amplitude image, and (c) birefringence image.
12 nm, in good agreement with earlier reports of diameters of 5 to 10 nm based on TEM studies of CNWs produced by this procedure.79 To obtain further evidence of the existence of wellindividualized CNW in the aqueous medium, the flow birefringence of the CNWs in an aqueous medium was evaluated under cross-polarized light. Therefore, suspensions of the samples (0.5 wt % CNW) were placed with stirring between 90 crossedpolarizing filters, and visible light was allowed to pass through them. The flow birefringence pattern illustrated in Figure 2c was observed under stirring (not under static conditions), even after several months of storage. This is supporting evidence for the existence of CNWs and the stability of the suspensions. The surface charge of the CNW determined by conductometric titration was 259 ( 9 μmol/g. The whiskers were in the acid form with negative surface charges. 3.2. Deposition of Nanocomposite Plasma Films with Different Monomer Feeds. The monomer feeds for the deposition of nanocomposite plasma films subsequently contained pure MA, MA + 50 wt % CNW, MA + 200 wt % CNW, and pure CNW. The weight of MA in the monomer tube was kept constant at 0.5 g, and the quantity of CNWs was increased appropriately. The surface patterns of the resulting films with various compositions on silica substrates are presented in Figure 3 under constant plasma conditions of Pp = 20 W and DC = 2%. • As a reference, the pure MA pulsed plasma films (Figure 3a) homogeneously cover the entire substrate and contain a nanoscale roughness pattern with an average roughness of Sa = 3.1 nm, a root-mean-square roughness of Sz = 3.5 nm, and a peak-to-peak height of St = 17.2 nm on a standard AFM image of 10 10 μm2 (inset in Figure 3a). For comparison, an uncoated Si wafer is atomically flat with roughness Sa = 0.6 nm, Sz = 0.7 nm, and St = 5.6 nm (data not shown). The homogeneous MA film has a thickness of d = 20 ( 5 nm as estimated from ellipsometry. The static water contact angle on a pure MA film equals 45. Here, the film is thermodynamically stable with a positive spreading coefficient. • The MA + 50 wt % CNW films (Figure 3b) are discontinuously distributed over the surface as separate spots. The films are destabilized and behave as if they were metastable. The central droplets are surrounded by an irregular pattern of residual materials, suggesting that the plasma film forms by dewetting and shrinking over the substrate. Similar results were observed on the gold-coated substrates. This pattern was observed immediately after plasma deposition, suggesting that dewetting occurs during the condensation of the plasma polymer film onto the substrate. Under the same deposition conditions compared to those of pure MA, a low CNW concentration destabilizes the nanocomposite plasma film. 1429
dx.doi.org/10.1021/la202503h |Langmuir 2012, 28, 1427–1438
Langmuir
ARTICLE
Figure 3. Morphology of MA/CNW nanocomposite films with different compositions for constant plasma parameters (Pp = 20 W, DC = 2%): (a) pure MA by optical microscopy and a 10 10 μm2 AFM inset, (b) MA + 50 wt % CNW by optical microscopy, (c) MA + 200 wt % CNW by optical microscopy, and (d) pure CNW by optical microscopy. All substrates in ac are silicon wafers, and d is a gold-coated silicon wafer.
• The MA + 200 wt % CNW films (Figure 3c) form a homogeneous film over the entire silica and gold-coated silica surfaces (larger than 1 1 cm2). The films exhibit a regular undulation pattern except for a few “defect” points, which seem to prevent complete dewetting and shrinkage and provide better dimensional stability than do low CNW compositions. This film morphology was easily reproduced in multiple deposition experiments. The static water contact angle measured on MA + CNW nanocomposite films is unstable: it varies from 10 to 20 and is strongly sensitive to the presence of hydrophilic CNWs and/or the film topography. Indeed, contact angles for polymer films including CNWs range from 20 to 40,80 and the present morphology may amplify the hydrophilic nature. • In contrast, no film is observed when depositing pure CNWs without MA monomer. On a silicon wafer, aggregates of CNWs were randomly deposited over the substrate. On a gold-coated silicon wafer, aggregates of CNWs together with abrasive scratches along the monomeric gas flow direction were seen (Figure 3d). These observations suggest that the CNWs do not become dispersed in the gas flow in the absence of MA and that they locally wear the substrate because of their extremely high strength and stiffness. In conclusion, the CNW/MA nanocomposite films develop at moderate concentrations of CNW. Being highly crystalline and polymeric in nature, the CNWs have no opportunity to sublime during the plasma process but can be favorably carried by a precursor gas flow. As a result, however, the exact concentration of CNW in the polymerized film cannot be controlled at present.
Nevertheless, this process shows interesting possibilities for surface patterning by the formation of metastable structures reminiscent of dewetting and shrinkage phenomena, depending on the monomer composition. Specifically, the buckling patterns obtained at a composition of MA/CNW = 1/200 are of particular interest and are thoroughly investigated in the next analyses. 3.3. Buckling Pattern of Nanocomposite Plasma Films. From the detailed AFM analysis of the buckling pattern in Figure 4, the periodicity outside of the defect points is evidenced. The film has a roughness Sa = 22.3 nm, Sz = 27.7 nm, and St = 83 nm (measured on a 10 10 μm2 AFM image) or Sa = 24.5 nm, Sz = 30.3 nm, and St = 89 nm (measured on a 100 100 μm2 AFM image), according to Figure 4a. It is known that the roughness parameters generally increase on higher sampling areas,81 but the relatively small changes for the present morphology indicate that the film is very homogeneously distributed. The periodicity and random orientation of the buckles obviously result from the interference of different waves that develop under equibiaxial in-plane stresses. The wrinkled pattern has an average wavelength of λ = 2 ( 0.2 μm and an amplitude of A = 100 ( 10 nm (Figure 4b). In good agreement with the amplitude A and peak-to-peak height St, the film has an average thickness of d = 93.5 nm and a refractive index of n = 1.62 as determined by ellipsometry. Note that the refractive index is considerably higher than literature values of
