Near-Edge X-ray Absorption Fine Structure Studies of Electrospun Poly(dimethylsiloxane)/Poly (methyl methacrylate)/Multiwall Carbon Nanotube Composites

Abstract

This work describes the near conduction band edge structure of electrospun mats of MWCNT-PDMS-PMMA by near edge X-Ray absorption fine structure (NEXAFS) spectroscopy. Effects of adding nanofillers of different sizes were addressed. Despite observed morphological variations and inhomogeneous carbon nanotube distribution, spun mats appeared homogeneous under NEXAFS analysis. Spectra revealed differences in emissions from glancing and normal spectra; which may evidence phase separation within the bulk of the micron-size fibers. Further, dichroic ratios show polymer chains did not align, even in the presence of nanofillers. Addition of nanofillers affected emissions in the C-H, C=O and C-C regimes, suggesting their involvement in interfacial matrix-carbon nanotube bonding. Spectral differences at glancing angles between pristine and composite mats suggest that geometric conformational configurations are taking place between polymeric chains and carbon nanotubes. These differences appear to be carbon nanotube-dimension dependent, and are promoted upon room temperature mixing and shear flow during electrospinning. CH-π bonding between polymer chains and graphitic walls, as well as H-bonds between impurities in the as-grown CNTs and polymer pendant groups are proposed bonding mechanisms promoting matrix conformation.

Introduction

The discovery of CNTs ¹ has presented the possibility of innovative applications.² In particular, physical properties of polymers are greatly modified by the addition of carbon nanotubes (CNTs) as fillers,³ yielding enhanced mechanical behavior through mechanisms heavily influenced by the nanotube-polymer interface.⁴ In this scenario, CNTs are often chemically treated to first, promote dispersion, and further, enhance load transfer. Chemical processes usually aim, in first instance, at eliminating residual carbonaceous species that decorate outer walls. Further chemical treatments can include covalent functionalization, that contribute to the creation of local defects that act as nucleation sites to yield latching sites for various oxygenated groups.⁵ It is precisely through those groups where bonding to molecules in a matrix and, therefore, mechanical properties derive.⁶ However, high interfacial shear stress (in excess of 140 MPa) has been found in untreated MWCNTs-polymer composites, suggesting strong electrostatic and van der Waals forces suffice to promote polymer adhesion, with additional assistance from irregularities along the graphitic walls by way of a “nano-mechanical interlocking”.⁷

It is therefore critical to understand existing surface states of CNTs prior to any modification or further polymer embedding. It is through those surface states that physical properties will nucleate. Although much work has aimed at exploring a variety of surface modification protocols, little work has been done in examining interfacial bonding of as-grown CNT-polymer composites.⁸ Further on the processing end, electrospinning has shown value in the process of CNT alignment,⁹ a parameter of importance to produce both structurally sound ⁴ and responsive composites.¹⁰ However, interfacial interactions promoted during electrospinning have not been discussed.

PDMS and PMMA are polymers of high technological relevance, finding their way in multiple applications, despite their well-known phase separation.¹¹ Tribology properties of PDMS and PMMA have been extensively studied towards microfluidic and micro/nano-electromechanical systems (MEMS/NEMS). Indeed, stiction in silicon is a strong concern during computer hard-disk operation and consumer electronic devices featuring accelerometers, where stiction leads to failure.¹² The technological importance of PDMS drives interest towards fiber production by electrospinning, and PMMA is an effective carrier polymer to electrospin PDMS; providing chain entanglement needed for fiber production.¹³ In the context of biology, electrospun PDMS-PMMA fibers have been assembled into nanofibrous membranes; enabling protein growth.

In this work, we examine the effects of as-grown CNT bonding on a PDMS-PMMA polymer matrix extruded by electrospinning and collected on a rotating drum. Surface states of composites with untreated CNTs of different dimensions are studied and compared with pristine polymeric blends. The technique of choice is Near Edge X-Ray Absorption Fine Spectroscopy (NEXAFS), which provides excellent energy resolution along with chemical bonding and molecular orientation information.¹⁴ This property ensemble is of great importance in the study of CNT composites, where sophisticated conformational bonding has been reported by way of “polymer-wrapping”, ⁸ and supramolecular chemistry provides multiple bonding possibilities at the nanotube-polymer interface.^{8, 15}

Interfacial interactions of polymer composites have been studied priorly by a combination of NEXAFS and X-Ray Microscopy.¹⁶ Bianco et al.¹⁷ reported the first assessment on filler alignment on electrospun polyamide by Infared techniques. To our knowledge, this could be one of the earliest reports of NEXAFS studies on electrospun nanocomposites. This discussion will first address NEXAFS spectra of constituents: pristine CNTs, and PMDS/PMMA electrospun mats, followed by a discussion of signals from composites, photon induced damage during NEXAFS acquisition, spectral homogeneity of electrospun mats, and molecular conformation as evidenced by NEXAFS spectra. Finally, a model for CNT-polymer bonding will be provided.

Experimental Section

Materials and Reagents

The precursor solution for electrospinning contained of Poly(methyl methacrylate) (PMMA) purchased from Sigma Aldrich with an average of the molecular weight at 996,000 and Polydimethylsiloxane (PDMS) Sylgard 184 silicone elastometer from Dow Corning, with a molecular weight of ~60,000 (Figure 1). The solvents were Dimethylformamide (DMF) and Tetrahydrofuran (THF), both from Sigma Aldrich. Two types of MWCNTs were purchased from Nanostructured & Amorphous Materials, Inc. Filler 1 (30–50 nm diameter, 0.5–2 μm length) and Filler 2 (50–80 nm diameter, 5–15 μm length). Both were synthesized by catalytic CVD growth, yielding CNTs with 95 wt% purity upon acid treatment.

Figure 1.

Schematic of PDMS and PMMA monomers, as well as MWCNT with as-grown impurities

Preparation of PDMS/PMMA/CNT solutions

A solution prepared from 2.04 grams of PMMA, 10 ml DMF, 10 ml THF, which were mixed at 50°C for 1 hour under constant magnetic stirring. Upon cooling, 2.02 ml of Sylgard 184 and 0.2 ml PDMS curing agent were added to the mixture.¹⁸ Finally, two types of unmodified MWCNT were added to produce two composite samples. Polymer blends (1:1 ratio) produced Composite 1 (0.6 wt% Filler 1) and Composite 2 (0.2 wt% Filler 2), which were kept under magnetic stirring at room temperature.

Electrospinning

Electrospinning (aided by a syringe pump and a power supply) was conducted at room temperature (8kV, 5 cm emitter/collector distance) onto a rotating drum under a white lamp to assist residual solvent evaporation.¹⁹ As a result, dry mats were easily peeled from the collector, avoiding stiction. Following this procedure, three mats were electrospun, producing: Composite 1 and 2, as well as a pristine Matrix (Table I). Solutions of CNTs on methanol were drop-casted on Si wafers and air dried prior to inspection.

Table I.

Studied samples names and description.

Name	Sample description
Matrix	Electrospun PDMS-PMMA mat
Filler 1	MWCNTs with diameters 30–50nm
Filler 2	MWCNTs with diameters 60–100nm
Composite 1	PDMS-PMMA-Filler 1 (0.6 wt%)
Composite 2	PDMS-PMMA-Filler 2 (0.2 wt%)

Scanning and Transmission Electron Microscopy

SEM images were acquired at 5 keV on a FEI-235 Strata on SE mode. Some mats were Aucoated to prevent electron charge build up, and Filler 2 was imaged without coating. Focus Ion Beam milling facilitated cross-section inspection. TEM images were acquired in a JEOL JEM-2010F field-emission operated at 200 kV, and a JEOL JEM-ARM 200F electron microscope. STEM images were simultaneously recorded in both the High Angle Anular Dark Filed (HAADF) and Bright Field (BF) modes at 80 kV. Probe correction was performed with a CEOS corrector obtaining a twelve-fold Ronchigram with a flat area of ~40 mrad. Images were registered with a condenser lens aperture of 30 microns (convergence angle 25 mrad), and HAADF collection angle ranged from 45 to 180 mrad. Spot size used was ~35 pA.

NEXAFS

Carbon K-edge emissions were collected at U7A (NSLS-BNL) in partial electron yield (PEY) mode using a horizontally polarised beam. Spectra acquired in PEY mode accounted for the transition probability of each resonance while scanning the monochromator. Choice of 600 lines/mm monochromator grating with slits at 30 Nm × 30 Nm, provided an incident beam spot of 2 mm in diameter. A floodgun set at 60 NA was used to prevent surface charging. Samples were mounted on a bar for measurements with deposited fiber direction parallel and perpendicular to the beam polarisation vector. Macros were used to control the position of the sample bar, subsequently adjusting heights and angles towards automatic spectra acquisition at room temperature. Three incidence angles were used in these experiments: normal (90°), magic (55°), and glancing (30°). To examine homogeneity of samples, spectra from three different locations were acquired. Those locations were freshly irradiated upon spectra collection, minimizing beam damage effects.

Results and discussion

The SEM image in Figure 2a shows curvy and undispersed Filler 2, whose surface is covered by an a residual carbonaceous layer (Figure 2b), in agreement with recent aberration-corrected TEM reports of CVD-grown CNT.²⁰ Diameters of electrospun fibers ranged in the tens of microns, with shapes varying between cylindrical fibers and flatter ribbons (Figure 2c and d). Dispersion and alignment of CNTs was not homogeneous at a micron-sale; with some areas showing well aligned and dispersed fibers, both surficially, along the fiber and visible through the cross-section (Figure 2c). Other areas, however, showed reduced contrast from CNTs (Figure 2d).

Figure 2.

SEM and TEM images of Filler 2 (a and b respectively). SEM side views of Composite 2 show different areas with either well-aligned CNTs along the surface of a fiber (c), or a local absence of CNTs (d). Cross-sectional SEM image in Composite 2 reveals phase separation upon Focus Ion Beam milling (e). Electrospun Composite 2 mat is shown in (f), whose perimeter has been traced by a dashed white line.

Notably, phase separation, as well as porosity is observed on fibers cross-section (Figure 2e). Despite the dominant secondary electron-contrast in Figure 2e, backscattered electrons emitted within the solid angle of the detector, will be collected and contribute to the overall contrast. Given the larger atomic weight of the repeating unit on PMDS, brighter areas on the cross-sectioned fiber are attributed to PDMS, yielding increased backscattered contrast than PMMA.²¹ Electrospun mats (Figure 2f) were dissected mindfully of the emitter’s direction, E as seen in Figure 2f, which favored NEXAFS analysis of mats in parallel and perpendicular modes with respect to the polarization of the X-ray beam.

CNT spectra

Emission assignment in NEXAFS is inherently challenging due to the multiplicity of factors affecting electronic transitions.²² However, analysis of MWCNTs spectra reported here is in good agreement with previous findings.^{14, 23, 24} Spectra from pristine MWCNTs (Figure 3 top) confirm π* C=C and σ* C-C antibonding transitions in MWCNTs at 285 and 292 eV, respectively, and a series of oxidized states in the 287–290 eV region¹⁴, with 287.6 eV having been assigned to π* C=O transition and 288.2 eV assigned to σ * C–O transition during ozone-purifying processing. Emissions around 287 eV range are also associated to σ * C–H.¹⁴

Figure 3.

NEXAFS C-K edge spectra of Filler 1 and 2 (top a and b, respectively) acquired at normal (90°), magic (55°), and glancing (30°) incidence respectively. Reported PDMS²⁵ and PMMA²⁶ spectra at magic and normal incidence respectively (center a and b). NEXAFS spectra from Matrix in parallel and perpendicular (bottom a and b, respectively) orientations. Ranges of emissions have been adapted from Flemming ²⁷ and Hemraj-Benny ¹⁴ for CNTs and from Dhez,²⁶ and Efimenko ²⁵ for polymer emissions.

NEXAFS spectra shown in Figure 3 (top) reveals intense π* and σ* C1s emissions from both Filler 1 and 2, where the π* to σ* C-C ratio is larger in Filler 1. Filler 1 shows some distinct angular dependence on NEXAFS emissions, unlike Filler 2 (Figure 3 (top)). CNTs were drop-casted for NEXAFS analysis. The bundled configuration of Filler 2 has promoted vertical arrangements, at least partially, of the tubes ensemble (Figure 2a). The propensity of Filler 1 to readily disperse in methanol (owing to a decreased length) is likely to have promoted a preferential in-plane arrangement. In this configuration, increased π* C=C signals would correlate with increased incident angles, in good agreement with our findings (Figure 3 (top)).¹⁴ Albeit, despite Filler 1 preferentially lying on the substrate, there is no further alignment or symmetry within the substrate, hence the σ* C-C is not following a decreasing trend with angular dependence. Despite intense oxygen-carbon states in both cases,²³ intensities of the first σ* C-C exciton (292 eV) are comparable in both Filler 1 and 2, indicative of similar short-range order. Indeed, most of the functionalized and impurity groups are believed to be located at the caps of the CNTs (as seen in Figure 1), where surface curvature is higher.²³ This is consistent with higher presence of Oxygen groups in Filler 2 than in Filler 1, i.e., there are more accessible sites on Filler 2, with larger diameter (Table I).

PDMS and PMMA spectra

Standard and calibrated spectra of pristine PDMS²⁵ and PMMA²⁶ have been reported at magic incidence (Figure 3 center), with distinctive emissions from σ* C-Si in PDMS centered at 291 eV and π* C=O in PMMA at 288.5 eV. In addition, σ* C-C in PMMA was also centered at 291eV. Common σ* C-H was reported in both cases in the vicinity of 287.4 eV.

Matrix spectra

NEXAFS spectra collected at normal incidence from matrix in the parallel (Figure 3a (below)) and perpendicular (Figure 3b (below)) configurations, shows three distinctive emissions at 284, 287.5, and 291 eV. Assignment of emissions has been conducted taking into account the nominal emissions in pristine PDMS and PMMA discussed above. Literature search has identified possible contributors to the 291 eV emission as σ* C-C and σ* C-Si from PMMA and PDMS respectively.^{25, 26}

Further, a likely contributor to the 287.5 eV emission observed in our spectra is σ* C-H from both PMMA and PDMS.^{25, 26} Pristine PMMA presents a π* C=O at 288.5 eV, possibly also contributing to this emission. The origin of the signal at 284 eV is likely related to unintentional surface contamination combined with photon-induced damage and will be discussed in further detail in the next section.

Similar spectral signatures are observed at magic incident angles in both orientations. However, notable spectral differences appear at glancing incident angles, as seen in Figure 3 (below). Incidence at glancing angles show emissions π* C=C, π* C=O, and σ* C=C at 285 eV, 288.5 eV, and 293 eV respectively; as previously reported.²⁶

Multiple scenarios could cause such different spectral signatures at glancing angles. First, the signal is originating in a phase-separated volume, where PMMA is the matrix bulk component, and PDMS is the segregating phase. Spectra collection at higher angles could be altering the scape depth profile of the collected electrons, therefore modifying the measured signal. In this scenario, acquisition at glancing angles would evidence spectroscopy emissions from PDMS.²²

Second, it is known that different molecular environments in the vicinity of C-C, C-O, and C-H bonds in the polymeric chain could have a large impact on spectral intensities acquired under different angles.²² A third scenario would suggest that spectra at normal and magic incidence is dominated by methyl groups in PMMA,²⁸ as predicted by Gross et al. This would explain the large intensity of the 287 eV emission at normal and magic incidences.

We mentioned earlier orientation is a determining factor in NEXAFS, since absorption cross-section probabilities as calculated by Fermi’s Golden Rule,²⁹ show strong angular dependence. As a result, intensities in the emission spectra are strongly dependent on orbital orientation.²² Conversely, spectral signatures at different angles offer a window into molecular conformation. Indeed, discrimination of molecular orientation by NEXAFS is often applied to simple systems with strong molecular orientation,³⁰ with NEXAFS spectra revealing strong interplay of different emissions at different angles. Gross et al. also predicted a preferred orientation of the C=O bond.²⁸ Our findings suggest that π* C=O is less accessible at glancing angles. This observation will be implemented in a proposed polymer-CNT conformational model, later on in the discussion.

Photon-induced damage

The origin of the signal at 284 eV in Figure 3 (below) is unclear, and has not been reported during calibrated acquisition of PMMA films in a different beam line at BNL.²⁶ An emission band below the C-K absorption edge has been found upon UV irradiation and ozone exposure of PDMS,²⁵ which was attributed to oxygen second harmonics. Processing of the films under study, however, only involved exposure to lamp irradiation in air to promote solvent evaporation. Intensities registered below the C-K absorption edge could correspond to double carbon bonds. The only orbitals suitable for sp² hybridization on the polymeric matrix correspond to π* C=O bond in PMMA. However, that emission is typically observed at 288.5 eV, as discussed earlier. Ultimately, unintended incorporation of aromatic chemical species could be responsible for this emission; which is observed both in polymer blends and in polymer composites reported here. On one hand, molecular residual contaminants could yield a pervasive and random deposition of π* C=C aromatic molecules suitable to further polymerization upon ion beam exposure. Interestingly, π* C=C emissions in polymeric systems have been reported to increase and shift as a result of X-Ray damage,³¹ however, although PMMA resulted in a slight mass decrease, no additional intensities at 285 eV were observed.

Albeit, the polymeric nature of these samples presents the possibility of decreased signals due to sample preparation, e-beam, and X-ray beam damage. ^{31, 32} In particular, π* C=O is extremely sensitive to irradiation damage.³¹ Since π* C=C bonds are absent in the polymeric matrix, data suggests either the unintended deposition of a contamination layer, rich on aromatic groups, or the generation of C=C bonds resulting from damage-driven cross-linking. Given the higher intensity of aromatic emissions at glancing angles, those groups are likely adsorbed/cross-linked at an angle along the surface.

Spectroscopic identification and spectral homogeneity of electrospun composites

Addition of CNTs to the polymeric matrix has little impact on spectral fingerprints at normal and magic incident angles in either orientation (Figure 4). Spectral differences in composites also arise at glancing incidences, as seen previously on pristine matrices, and will be the point of discussion in the following section.

Figure 4.

Spectra at the C-K edge collected from Matrix (a), Composite 1 (b) and Composite 2 (c) at normal (above) magic (middle) and glancing (below) incidence angles, on the parallel (solid line) and perpendicular (dashed line) configurations.

Analysis of spectra at normal and magic incidence of all three samples, in all orientations, reveals a lack of additional emissions in all three locations. Ratios between emissions at 287.5 and 291 eV remain largely unmodified with mat orientation, and independently of whether CNTs are present in all locations. In fact, calibrated spectral signature of PMMA ²⁶ nearly matches both emission energies (with slight shift of π*C=O bond reported at 288.5 and σ* C-H at 287.5 eV) and intensities (π*C=O / σ*C-C ratio is 1.8 in calibrated spectra and 1.5 in the polymeric matrix).

NEXAFS spectrum of Composite 2 acquired at glancing incidence (Figure 4 bottom) shows a clear increase in intensity of 1s→π*_C=C transition compared to Composite 1. This may be explained in terms of projection of the electric field E onto π* orbitals. It is helpful to consider highly ordered pyrolytic graphite (HOPG) as all π* orbitals are oriented perpendicular to the substrate. At glancing incidence, where the electric field is nearly perpendicular to the substrate, there will be maximum projection of E onto the π* orbitals, resulting in higher NEXAFS intensity.¹⁴ MWCNTs can be thought of as HOPG rolled into a tube, and a similar trend is expected. As curvature increases with decreasing CNT diameter, Composite 2, with larger diameter CNTs, will show higher 1s→π*_C=C intensity than Composite 1.

Phase separation in dissimilar components yield undesirable properties in polymeric matrices.³³ Phase separation in PMMA and PDMS (or dissimilar polymers alike) is often reported in casted and spin-coated films,³⁴ and would be of great importance in electrospun fibers towards practical device application.³⁵ Cross-section SEM analysis revealed complex – and habitual-phase separation of both polymers, as seen in Figure 2e. Contrast from CNTs is often not apparent in cross-sectional views, and the larger molecular mass of Si on PDMS attributes brighter regions to the thermoset component of the polymeric matrix.²¹ The bright areas are confined to island-like regions with sizes varying between the micron and the sub-micron scale. The 10 μm-wide fiber confers PMMA a dominant matrix role, where PDMS phase-separates. A network of internal pores coexists within hydrophobically-dewetted PDMS; a phenomenon associated with non-equilibrium thermodynamics in un-annealed films.³⁴

The SEM image in Figure 2e, shows high-brightness contrast around the fiber perimeter due to a deposited Au layer to prevent charging during electron beam analysis. The observed dark contrast beneath the coating layer corresponds to PMMA. This is consistent with reported behavior of PMMA in polystyrene (PS) blends,³⁴ where a surficial PMMA rim enclosed precipitative segregation of PS upon spin-coating on a Si wafer.

We conclude that NEXAFS signals from PDMS droplets in the vicinity of the PMMA rim are likely to be detected. However, it is reasonable to assume that PMMA is the main contributor to the NEXAFS signals reported in this work. Indeed, X-Ray irradiation of uncoated composites has irradiated mats top down; where fibers lay mostly flat. The spectral signature of PMMA-C60 composites as reported by Pozdnyakov³⁶ bears strong resemblance with the signature of PDMS-PMMA-CNT composites reported here, further confirming PMMA as the main contributor to the NEXAFS spectra in most configurations (i.e. composite, normal/magic incidence, and relative orientation).

Finally, in electrospun composites, the question arises per spatial homogeneity of spun mats, where CNT dispersion, alignment, and intrinsic electrospinning parameters themselves are influential variables³⁷ determining chemistry and alignment of spun fibers. Spectra were acquired in all three samples with mats parallel and perpendicular to the direction of the polarization of the electric field, as shown in Figure 4. To address spectral homogeneity on electrospun mats, spectra were subsequently acquired in different locations for comparison. The initial scan was followed by two additional scans, situated 2 mms above and below (data not shown). Data acquired at the different locations showed no variations, further confirming spectral homogeneity in electrospun mats. This newly acquired knowledge on spectral homogeneity could be of great relevance towards the use of electrospun composites in electronic applications.

With all, spectra at normal and magic incidences are matrix-dominated at both parallel and perpendicular orientations, and mats appear uniform under the X-Ray beam. Spectra at glancing angles are indicative of bonding, and will be discussed in the next section.

Molecular conformation and dichroism in electrospun nanocomposites as evidenced by NEXAFS

Spectra of composites and blend mats were acquired in parallel and perpendicular configurations and showed little variations (Figure 4). Lack of intensity variation at all angles and orientation suggests non-linear arrangement of polymeric chains within the blend. Indeed, despite collection on a rotating drum, combined effects of extrusion and electric field on the fly to the collector, are insufficient to disentangle polymeric chains into aligned macromolecules. This lack of molecular orientation is consistent with findings in electrospinning,^{38, 39} where fiber collection onto charged plates, impose an additional electric field interaction conducive to molecular alignment, reflected on dicroic ratios measured by IR techniques.

Large variations in emissions and intensities however are observed, at glancing incidences in the regions of C-H, C=O and C-C/C-Si bonding (Figure 4). Progressive decrease of the π* C=O signal is observed between Matrix, Composite 1 and Composite 2 at glancing incidence. The decreased intensity from π* C=O in PMMA seems to point at a specific orientation of the pendant group with respect to the chain; i.e. the π* C=O bonds are most likely emerging from the plane of the sample with the π* C=O at a specific angle to the surface, offering maximum overlap with the irradiating field at normal and magic incidences (Figure 5). Spectra at normal and magic incidences show identical emissions at parallel and perpendicular orientations due to symmetry of the bond.²⁹ Slight differences in intensity between parallel and perpendicular orientations in Composite 2 at glancing incidence could indicate that conformational effects are possibly more efficient on Filler 2.

Figure 5.

π* resonances enhanced at normal incidence and σ* resonances enhanced at grazing incidence for a molecule positioned normal to the plane as overlap of electric field and orbitals is maximized (left, adapted from²⁹); proposed π*C=O bond orientation offering maximum beam accessibility at normal and magic incidences and minimum accessibility at glancing incidence, resulting in decreased NEXAFS intensity (right).

However, π* C=O is probably not perfectly orthogonal to the surface, it is possibly forming some angle with the plane normal to the surface in order to be equally accessible at magic incidence.³⁰ Schematics in Figure 5 illustrate the dependence of electric field incidence with bond configuration and subjacent atomic distribution in a molecule. This schematic illustrates the possibility of large π* C=O contributions at normal and magic incidence and reduced signals at glancing angles.

Compared with the pristine matrix, addition of small CNTs is reflected in glancing spectrum as a slight decrease of the π* C=O bond and a slight increase on the σ* C-C. Decreased π* C=O intensities seem to point at a variation in the polymer conformation, hinting at polymer wrapping.³⁰ Baskaran et al.²³ had established earlier the presence of noncovalent molecular interactions between polymers and MWCNTs of nonspecific nature; setting the frame for a universal effect. These findings suggest that although the composites described here were not shear-mixed, those CNTs whose van der Walls interactions were modified, therefore affecting their ability to agglomerate, could have done so by allowing a CH-π wrapping effect. In addition, the increased σ* C-C signal could be related to additional contributions by surficial CNTs. Two arguments further support this assertion. First, the σ* C-C signal from Composite 2 show a small variation between parallel and perpendicular orientation, possibly due to aligned CNTs. Second, SEM images have confirmed some alignment of CNTs in the vicinity of the fiber’s surface that would render them available for spectroscopic probing. This would enhance a CNT alignment-driven dichroism. In depth study of this phenomenon would require an in-plane rotating stage, unavailable at U7a at this time.

Spectra strongly suggest conformational effects of polymer chains, shown by modified spectral signatures (relative to pristine matrix) at glancing angles; possibly as a consequence of intermolecular interactions between polymeric chains and untreated graphitic walls in CNTs. Indeed, notwithstanding the geometrical configuration revealed at high incidence angles, variations of intensities at the π* C=C, π* C=O, and σ* C-H are observed first moderately in Composite 1, but to a large extent in Composite 2. Their variations will be the topic of discussion in the next section.

Proposed CNT-matrix conformational bonding models

X-Ray absorption spectroscopies are amenable to description of either covalent or noncovalent bonding,⁴⁰ making NEXAFS highly relevant in the study of supramolecular chemistry. Scans from Composite 2, show higher degree of interaction, as shown in Figure 6. The degree of CNT-polymer interaction can be evidenced in NEXAFS as progressive differences between composite and matrix spectra, following a building block model. NEXAFS of Composite 1 is only slightly dissimilar to the matrix, suggesting a small degree of interaction. In contrast, spectrum of Composite 2 is significantly different to the matrix, pointing to higher degree of CNT-polymer interaction, reflected by changes in the π* C=O and σ* C-H environments.

Figure 6.

C-K edge NEXAFS spectra of matrix, Composite 1, and Composite 2 acquired at glancing incidence (above), where the region has been adapted from Kocharova.⁴³ Proposed model for polymer-CNT interaction through CH-π bonding (below).

Indeed, Filler 2 having greater surface area leads to more –COOH groups (Figure 3b (above)) and other impurities through which polymeric chains can append, yielding a higher interaction for Composite 2. This is observed as higher NEXAFS intensities in the region of 287–291eV for Filler 2 than for Filler1. Additionally, the reduced curvature means π systems are less distorted. Increased curvature leads to increased π-orbital misalignment and pyramidalization, resulting in enhanced reactivity of CNTs towards covalent functionalization.⁴¹ However, non-covalent interactions involving aromaticity will favour pristine π systems, suggesting filler in Composite 2 could be more suitable for interaction with polymeric chains.⁴² Conversely, non-covalent bonding is often preferred to covalent surface-treated CNTs, as electronic structure of pristine CNTs is preserved.⁴¹ Given the absence of any CNT surface treatment, we propose two intermolecular communication mechanisms, establishing CH-π and Hydrogen bonds.

Scans at magic and normal incidences of composites and pristine samples do not show large differences (Figure 4). However, spectra at glancing incidence of Composite 1 do show different intensities at 287 to 294 eV (Figure 6 above) encompassing the σ* C-H, π*C=O, and σ* C-C/C-Si regimes. Those differences are magnified in Composite 2, featuring an additional variation on the π*C=C intensities. Development of strong steric repulsion between alkyl groups as a consequence of conformational adsorption of polymers to CNTs has been reported earlier ¹⁵ and might be influencing the C-H, C-C, and C-Si environments of mostly, Composite 2, as observed in Figure 4.

Effects of CH-π interactions have been identified by NEXAFS on the 287–290 eV range.⁴³ Indeed, CH-π bonding to graphitic walls is a possible scenario where CH groups from polymer side chains interact with π bonds perpendicular to CNTs surfaces. Moreover, CH-π bonding has been confirmed in a polymer non-specific scenario,⁸ independently of polymer size, upon shear mixing at temperatures above Tg. Later, Llanes-Pallas and coworkers confirmed CH-π bonding in the context of supramolecular polymer-CNT composites in the absence of CNT surface modification.¹⁵ In that scenario, matrix and filler were mixed by conventional stirring methods at room temperature, as reported here. Moreover, wrapping of supramolecular polymers onto untreated CNTs was confirmed by TEM and AFM, and molecular dynamics simulations have also shown the viability of polymeric chains wrapping conformationally onto CNTs through CH-π bonding.

Albeit, it seems apparent that CH-π interactions are viable upon magnetic stirring at room temperature. The proposed model describing the conformational arrangement of matrix and fillers is depicted in Figure 6 (below). We hypothesize that extrusion forces during electrospinning may be favoring CH-π latching initiation between unmodified CNT and, preferentially, PMMA,⁸ whose frequent CH pendant groups solubilize (to some degree) untreated CNTs rather than PDMS, with lower molecular weight. In this scenario, PMMA polymer chains would latch, at least partially, to CNT surfaces, promoting either partial wrapping or full coating. Higher intensity of NEXAFS signals in Composite 2 suggests higher interaction probability when larger CNTs are involved.

The second model proposed involves H-bonding through CNT impurities. Indeed, common impurities in CVD-grown CNTs include an approximate 1 mol% carboxylic acid groups (COOH: OH-C=O),^{8, 44} as determined by XPS and tritation methods. Approaches in impurity management (all leading to improved solubility) champion acidic treatments for their removal, often at the expense of causing additional damage,⁴⁵ distorting pristine properties by creating defect sites towards functionalization.⁴⁴ With all, matrix bonding through impurities on untreated MWCNTs often go undiscussed,^{8, 15} despite feasibility of inorganic nanoparticles been attached to SWCNTs.⁴⁶

Figure 7 plots NEXAFS intensities acquired at glancing incidence on the C-K edge of CNTs, matrix and composites. Some variation is seen on the spectra of Composite 1 compared with the matrix at the C=O emission (Figure 7a). However, it is in Composite 2 where larger differences are apparent, as seen in Figure 7b, i.e. increased signals at the π* C=C, σ* C-H, π* C=O, and σ* C-C regimes. The increased availability of impurity groups along Filler 2 is indeed likely to promote interactions with the H groups from the polymeric chains. Figure 7c shows a schematic where H bonding configurations are possible with both PDMS and PMMA.

Figure 7.

C-K edge NEXAFS spectra of Composite 1 (a) and Composite 2 (b), plotted along with spectra of matrix and filler, all acquired at glancing incidence. Schematic of proposed H-bonding model (c).

Conclusions

We have measured NEXAFS spectroscopy from electrospun PDMS-PMMA-MWCNT mats. Results reveal a complex interplay of intensities in phase-separated composites, with strong orientational effects from unmodified MWCNTs. However, spectra showed no variability when acquired in different regions, suggesting a homogeneous near edge spectra of electrospun composites. Although PMMA was first introduced as an electrospinning carrier, it is the principal actor in this study, preferentially detected by NEXAFS possibly due to phase separation, and to CNTs linking to PMMA chains through CH groups. Variations of intensities at glancing incidences from σ* C-H and π* C=O emissions have promoted two conformational non-covalent bonding models. Both CH-π and Hydrogen bonding models evidence increased interaction with larger MWCNTs, with increased availability of π systems as well as impurity-related oxidized states. The combined action of magnetic stirring and extrusion forces involved in electrospinning could have favored CH-π and Hydrogen bonding. Transient compressive radial stress through the spinneret during electrospinning could enhance CNT-polymer interactions and through local CNT deformations promote mechanical interlocking. Effects of the specific role of electrospinning on CNT-polymer interactions are the subject of on-going research.