Pentiptycene Polymer/Single-Walled Carbon Nanotube Complexes: Applications in Benzene, Toluene, and o-Xylene Detection

Abstract

We report the dispersion of single-walled carbon nanotubes (SWCNTs) using pentiptycene polymers and their use in chemiresistance-based and QCM-D sensors. Poly(p-phenylene ethynylene)s (PPEs) incorporating pentiptycene moieties present a concave surface that promotes π–π interactions and van der Waals interactions with SWCNTs. In contrast to more common polymer-dispersing mechanisms that involve the wrapping of polymers around the SWCNTs, we conclude that the H-shape of pentiptycene groups and the linear rigid-rod structure creates a slot for nanotube binding. UV–vis–NIR, Raman, and fluorescence spectra and TEM images of polymer/SWCNTs support this dispersion model, which shows size selectivity to SWCNTs with diameters of 0.8–0.9 nm. Steric bulk on the channels is problematic, and tert-butylated pentiptycenes do not form stable dispersions with SWCNTs. This result, along with the diameter preference, supports the model in which the SWCNTs are bound to the concave clefts of the pentiptycenes. The binding model suggests that the polymer/SWCNTs complex creates galleries, and we have demonstrated the binding of benzene, toluene, and o-xylene (BTX) vapors as the basis for a robust, sensitive, and selective sensing platform for BTX detection. The utility of our sensors is demonstrated by the detection of benzene at the OSHA short-term exposure limit of 5 ppm in air.

Graphical Abstract

Single-walled carbon nanotubes (SWCNTs) are attractive materials for sensing gaseous analytes as a result of their sensitive resistance changes in response to the binding of molecules.^1-5 The sensing performance in these materials is a consequence of a high surface area-to-volume ratio, molecular adsorption onto their electronically active sidewalls, swelling of the matrix, and/or their restricted conduction pathways.⁵ However, pristine SWCNTs display nonspecific responses to chemical exposures, and covalent or noncovalent functionalization with selectors or receptors is required to produce selectivity to target analytes. Covalent functionalization utilizes reactions that attach chemical groups covalently to the conjugated surfaces or termini of CNTs.^6-9 Covalent functionalization has the advantage that it produces stable anchors of functional groups to CNT. However, covalent attachment of groups to the graphene surface transforms sp² carbons into sites with increased sp³ character that disrupts electronic coherence and decreases carrier mobility. Non-covalent functionalization of SWCNTs is less perturbative to the electronic properties of the nanotubes, and the associated higher carrier mobilities can provide enhanced sensitivity.^10-15

As chemiresistive sensors, SWCNTs have been functionalized noncovalently by physisorption of small aromatic molecules and surfactants through π–π and hydrophobic interactions.^16-25 Physisorbed selector molecules or coatings often have limited stability, are prone to environmentally induced changes in their configuration around the CNT, can undergo phase segregation, and can desorb in solution, leading to unstable dispersions. These changes can produce large conductance changes that give rise to drift and degraded performance in chemiresistive sensors. Polymer wrapping or surface-anchored molecular clips produce stable noncovalent functionalized SWCNTs^26-34 and have also been widely explored to separate and purify^35-47 or orient^48,49 these materials. Conjugated polymers with selectors attached can provide selectivity for the detection of specific analytes and even resolve structural isomers of xylenes.^13,50-53 Simple poly(phenylene ethynylene)s (PPEs) have been shown to disperse SWCNTs,^54-58 and we have been interested to see if PPEs containing pentiptycene, a rigid H-shaped molecule with a cleft of about 10 Å diameter (Figure 1a),⁵⁹ would have strong interactions with SWCNTs of a complementary diameter. Pentiptycene’s structural properties in conjugated oligomers^60,61 and polymers ^62,63 prevent intermolecular π–π interactions to maintain high photoluminescence in solid films. The free volume in these structures creates porous structures that have provided size exclusion for the selective detection of small-molecule analytes.^64-67

Figure 1.

(a) Structural property of pentiptycene. (b) Schematic drawing of the dispersion mechanism of SWCNTs between pentiptycene polymers.

In this report we propose the dispersion method illustrated in a simplified form in Figure 1b, wherein rather than wrapping around a SWCNT, pentiptycene functions as a linear arrangement of clips that cooperatively bind through van der Waals interactions. Central to this model is that the diameter of the cleft defined by the pentiptycene matches SWCNT with a diameter of about 0.9 nm. Commercially available SWCNTs have a size distribution, and we expect pentiptycene polymers to selectively disperse SWCNTs of a complementary size. To this end we designed five pentiptycene-incorporating polymers, P1–P5 (Chart 1), and investigated their dispersions with SWCNTs. All of the polymers contain long alkyl chains to enhance solubility. P3 and P5 also contain bulky tert-butyl groups on the periphery ring of pentiptycene, which has been shown to increase free volume in polymers.⁶⁸ The polymer/SWCNT dispersions and thin films were characterized by UV–vis–NIR absorption, photoluminescence, Raman scattering, and TEM imaging. We further demonstrate that the polymer/SWCNT constructions impart enhanced selective sensing responses to vapors of benzene, toluene, and o-xylene (BTX).

Chart 1.

Chemical Structures of P1–P5

RESULTS AND DISCUSSION

The polymer/SWCNT dispersions were prepared according to the following procedure. The pentiptycene polymer (3 mg) was dissolved in chlorobenzene (4 mL) with 2 mg of CoMoCAT SWCNTs CG100. After 30 min of tip sonication at 63 W, the suspension was centrifuged for 4 h at 30130g. The CG100 SWCNT dispersions with P1, P2, and P4 have high stability and are optically opaque after centrifugation. In contrast, dispersions using tert-butylated pentiptycene polymers, P3 and P5, are unstable and yield clear solutions after centrifugation. The optical absorbance of the supernates obtained directly after the centrifugation was analyzed in a 1 mm short-path quartz cuvette (Figure 2). For P1/CG100, P2/CG100, and P4/CG100, clear interband transitions of the van Hove singularities were observed in the absorption features. Absorption peaks in S₁₁ (830–1600 nm) and S₂₂ (600–800 nm) regions indicate the presence of semiconducting SWCNTs, whereas the broad and featureless absorption background and peaks in the M₁₁ (440–645 nm) region indicate the presence of metallic SWCNTs.⁶⁹ (6,5), (7,6), (7,5), and (8,7) SWCNTs were found to be the major species in the polymer/SWCNT dispersions.^70-72 Notably, P4/CG100 has much sharper absorption peaks than P1/CG100 and P2/CG100, suggesting the superior polymer/SWCNT dispersion quality by P4. The absorption features of P1/CG100 in the S₁₁ (830–1600 nm) region are slightly broader than those of P2/CG100 and P4/CG100, which suggests mild bundling of SWCNTs possibly due to the branched alkyl chains in P1, which could hinder the polymer/SWCNT interaction.⁴⁴ On the contrary, no clear absorption feature was identified for P3/CG100 and P5/CG100, as they failed to yield stable SWCNT dispersions. The fact that the bulky tert-butylated pentiptycene polymers display poor SWCNT dispersion is also consistent with the binding to the SWCNT being mediated by strong van der Waals interactions between the π-systems.

Figure 2.

UV–vis–NIR absorption spectra of supernates of P1/CG100 (red), P2/CG100 (blue), P3/CG100 (purple), P4/CG100 (green), and P5/CG100 (orange). The pink, green, and cyan boxes indicate the locations of optical transitions of metallic (M₁₁) and semiconducting (S₂₂, S₁₁) SWCNTs, respectively. Characteristic absorption peaks are labeled with the assigned chirality.

To support our proposed noncovalent structural model, we collected Raman spectra of the polymer/SWCNT complexes (Figure S3). Thin film samples were prepared by drop-casting polymer/CG100 supernatant dispersions onto silicon wafers. A pristine CG100 SWCNT thin film was also prepared for reference. The spectra were obtained using a 633 nm excitation wavelength and are normalized to the intensity of the G-band, at 1590 cm⁻¹. The D-band, located at around 1300 cm⁻¹, is indicative of the disruption of the sp² network in conjugated nanocarbon systems, and the intensity ratio of the D- to G-bands (I_D/I_G) provides information on the perturbation of the π-system. The I_D/I_G ratios of polymer/CG100 films were 0.05–0.07, which are close to that of pristine CG100 (0.05) and confirm minimal disruption of the nanotubes’ electronic structure.

The need to match the SWCNT diameter with the size of the pentiptycene cleft was evaluated in dispersions using P4 and several commercial SWCNTs with different diameter distributions. Besides CoMoCAT SWCNTs CG100 (0.7–1.3 nm diameter), HiPCO SWCNTs (0.8–1.2 nm diameter), (6,5)-enriched CoMoCAT SWCNTs SG65i (0.7–0.9 nm diameter), and arc-discharge SWCNTs CarboLex (1.2–1.5 nm diameter) were investigated. The UV–vis–NIR spectra of the dispersions are shown in Figure 3, and the photographs of the dispersions are shown in Figure 4. Stable SWCNT dispersions were formed in P4/CG100, P4/HiPCO, and P4/SG65i, as well-resolved absorption peaks were observed in S₁₁, S₂₂, and M₁₁ regions. Similarly, the noncovalent binding of P4 to the SWCNTs is confirmed by the minimal change in I_D/I_G (Figures S6 and S7). For CarboLex, which comprises SWCNTs with diameters larger than the pentiptycene cleft size, the supernatant color is close to that of pure P4, consistent with the UV–vis–NIR spectrum, which indicates that a minimal amount of CarboLex SWCNTs is dispersed.

Figure 3.

UV–vis–NIR absorption spectra of supernates of P4/HiPCO (blue), P4/CG100 (red), P4/SG65i (purple), and P4/CarboLex (green). The pink, green, and cyan boxes indicate the locations of optical transitions of metallic (M₁₁) and semiconducting (S₂₂, S₁₁) SWCNTs, respectively.

Figure 4.

Photographs of supernates of P4/HiPCO, P4/CG100, P4/SG65i, P4/CarboLex, and P4 solution in chlorobenzene (from left to right).

Central to our proposed SWCNT binding model is the diameter selectivity provided by the pentiptycene unit toward a range of SWCNTs. The Raman radial breathing mode (RBM) of the pristine SWCNT and the polymer/SWCNT complexes, which correlates strongly with SWCNT diameter,^73-77 is evaluated in Figure 5. Raman intensities are normalized to the peak at 285 cm⁻¹ to illustrate the relative change in the SWCNT diameter distribution. The estimated SWCNT diameters are indicated by the dotted lines following a reported equation.⁷⁴ As shown in Figure 5, SWCNTs with diameters larger than 1.0 nm or smaller than 0.8 nm are significantly reduced after polymer dispersion. An enrichment of SWCNTs with 0.8–0.9 nm diameters was observed for all dispersions, matching the cleft size of the pentiptycene unit, which facilitates the strong polymer–SWCNT interaction. It should also be noted that the RBM as well as the G-band peaks of the polymer/SWCNT complexes are shifted from their counterparts in pristine SWCNTs (Figures 5 and S8), which suggests effective binding of the polymer on the surface of SWCNTs.^73-77 Moreover, the binding of the polymers to the SWCNTs is also apparent from the excited state electron and/or energy transfer that leads to fluorescence quenching. As shown in Figure 6, the emission intensities of P4/SWCNT complexes were found to be attenuated by 60–80% compared to the pure P4 solution at the same concentration. Emission peaks of P4/SWCNT dispersions agree with that of neat P4 solution, indicating no significant polymer scission from the tip sonication procedure, which is also supported by the UV–vis–NIR absorption comparison (Figure S10).

Figure 5.

Radial breathing mode (RBM) region of the Raman spectra of pristine SWCNTs and P4/SWCNT dispersions normalized to the peak at 285 cm⁻¹ (excitation at 633 nm). Dotted lines indicate the estimated SWCNT diameter.

Figure 6.

Emission spectra (λ_ex = 350 nm) of supernates of P4/CG100 (red), P4/HiPCO (blue), P4/SG65i (purple), and P4 solution (black) in chlorobenzene. The polymer concentration is 0.75 mg/mL in all cases.

TEM images of drop-cast films of P1/SG65i, P2/SG65i, and P4/SG65i are shown in Figure 7. The linear structures observed are consistent with aligned networks of SWCNT/polymer aggregates. The parallel red lines in the inset highlight the periodic fringes of SWCNTs, which have a center-to-center distance of ~3 nm (Figures S11 and S12). This gives an intertube distance of ~2.2 nm, which is significantly larger than the intertube distance in SWCNT bundles determined previously.^78-80 The increased intertube spacing is consistent with the presence of pentiptycene-dispersing polymers between the tubes, as shown schematically in Figure 8. This model is consistent with the 1 nm diameter of pentiptycene clefts and that dispersed SWCNTs will have polymers bound to their surface, consistent with previous reports where PPEs are shown to stack directly on the SWCNT surface by AFM and TEM studies.^55,56 Hence, in the film there are on average two polymer chains between each SWCNT, which creates molecularly defined cavities in the films.

Figure 7.

TEM images of drop-cast films of P1/SG65i (left), P2/SG65i (middle), and P4/SG65i (right). Inset figures are at higher magnification and show periodic fringes of SWCNTs.

Figure 8.

Schematic drawing of pentiptycene polymer/SWCNT composition in drop-cast film.

The porous structures with high aspect ratio galleries created by the polymer/SWCNT complexes present an opportunity to detect volatile organic vapors. Considering the planar nature of the comonomers in the pentiptycene polymers, we have evaluated their efficacy for interacting with the planar aromatic organics: benzene, toluene, and o-xylene. Detection of BTX is of interest in environmental health and safety as a result of their toxicity. The parallel bundles of polymer/SWCNT observed in our TEM studies suggest that these composites may create interstitial galleries that can differentiate vapors based on size exclusion. The diffusion of aromatic vapors into these galleries is expected to swell the polymer/SWCNT structures and thereby induce the changes of conductance. (6,5)-Enriched SWCNTs SG65i were used to fabricate the sensors with revised procedures as a result of their high semiconducting content, which has been shown to aid sensing sensitivity.^51,81 Devices were prepared by drop-casting polymer/SWCNT supernates (1 μL) on gold electrodes with an interelectrode spacing of 1 mm, followed by drying in vacuo. The change in resistance resulting from the exposure to the analyte vapor was converted to the negative normalized change in conductance, −ΔG/G₀, where ΔG and G₀ are the change in conductance and baseline conductance, respectively. Figure 9 summarizes the chemiresistive responses (−ΔG/G₀) of pristine SG65i, P1/SG65i, P2/SG65i, and P4/SG65i with exposures to benzene, toluene, and o-xylene (100 ppm) for 1 min in air. Minor baseline correction was applied to account for the linear drift of the baseline conductance.

Figure 9.

Chemiresistive responses of pristine SG65i (black), P1/SG65i (red), P2/SG65i (green), and P4/SG65i (blue) in response to benzene (left), toluene (middle), and o-xylene (right) in air. Devices were exposed to analyte at 100 ppm in air for 1 min (N ≥ 6).

As shown in Figure 9, films of pristine SG65i show very weak responses to BTX vapor. For P1/SG65i, the averaged response toward 100 ppm benzene vapor is about 0.15% with an excellent signal-to-background ratio and is approximately 2× more sensitive than our previous BTX chemiresistive sensors.⁸² For P2/SG65i and P4/SG65i, the responses to benzene vapor (~0.25%) are higher than that of P1/SG65i, while maintaining high signal-to-noise ratios. Exposure to toluene and o-xylene vapors results in much higher signals across all sensors. The averaged sensor responses of P1/SG65i, P2/SG65i, and P4/SG65i to 100 ppm toluene vapor in air are about 0.8%, 1.2%, and 1.5%, respectively. For exposure of 100 ppm o-xylene vapor in air, the averaged sensor responses of P1/SG65i, P2/SG65i, and P4/SG65i are about 1.1%, 1.7%, and 1.9%, respectively. Notably, the sensing responses toward BTX vapors are highly reversible with small variations across different fabricated devices and are equivalent in dry air or nitrogen (Figure S13). The sensing performance is largely maintained at a relatively high humidity of 50–70% (Figure S14). Together, these results confirm the utility and robustness of these sensory materials in BTX detection.

It is worth noting that the resistance of the devices fabricated from similar amounts of different polymer/SWCNT dispersions varies greatly from ~20 kΩ (P1/SG65i) to ~1 MΩ (P4/SG65i). We believe this is also indicative of the quality of the polymer/SWCNT dispersions, and better dispersions will restrict direct SWCNT/SWCNT interactions and thereby lead to high resistivity. The well-resolved absorption spectrum (Figure 3) and the attenuated emission intensity (Figure 6) of P4/SG65i support that SWCNTs are well dispersed, and tight binding to P4 creates molecular insulation on individual tubes. For all sensors, the responses toward benzene and toluene are extremely rapid with a “step-function” type profile that saturates within 10 s and returns to the baseline within 1 min after the exposure. This indicates that the gas molecules diffuse into and out of binding sites quickly. The conductance profile for o-xylene sensing has more sluggish kinetics, which is consistent with its bulkier structure. It should be noted that the BTX molecules have similar electronic and structural properties. The differences in the temporal responses can potentially be used to differentiate between analytes.

We are particularly interested in detecting benzene, which is a challenging important target in chemical sensing. Figure 10a illustrates the responses of the sensors fabricated from a P4/SG65i dispersion toward benzene vapor at different concentrations in air for 1 min. These results show that we can reproducibly detect benzene down to 5 ppm (Figure 10a, inset), which is at the OSHA short-term exposure limit.⁸³ Similarly, clear signals were observed for all our sensors at 5 ppm of benzene (Figure S15), highlighting the potential utility in real-word benzene monitoring. As summarized in Figure 10b, sensors fabricated from P4/SG65i perform the best with the strongest responses toward benzene at concentrations ranging from 5 to 450 ppm compared to P1/SG65i and P2/SG65i. The change of chemiresistance is proportional to benzene concentration with high linear relationships (R² > 0.997). The limit of detection (LOD) was calculated to be 4, 3, and 3 ppm for P1/SG65i, P2/SG65i, and P4/SG65i, respectively.

Figure 10.

(a) Chemiresistive responses of P4/SG65i at different concentrations of benzene. The enlarged responses to benzene vapor at 5 ppm are shown in the inset. (b) Summary of chemiresistive responses of P1/SG65i (red), P2/SG65i (green), and P4/SG65i (blue) at different concentrations of benzene. Devices were exposed to benzene in dry air for 1 min (N ≥ 6).

Figure 11 summarizes the selectivity of our sensors when challenged against various volatile organic compounds (VOCs) at 100 ppm in dry air. Overall, the sensors exhibit excellent selectivity toward BTX compared to common VOCs that give 0.15% responses at the same concentrations. Notably, the responses toward vapor of cyclohexane, the alkyl analogue to benzene, are only half of the responses toward benzene vapors. Moreover, the SWCNTs, which are naturally p-doped, generally have higher responses to polar molecules, but in this case the nonpolar BTX analytes have higher responses. These observations confirm the importance of the aromatic pentiptycene polymers for selective BTX sensing via structural recognition and π–π interactions.

Figure 11.

Chemiresistive responses of pristine SG65i (white), P1/SG65i (blue), P2/SG65i (orange), and P4/SG65i (green) in response to different volatile organic vapors. Devices were exposed to analyte at 100 ppm in dry air for 1 min (N ≥ 6).

To further access the binding of BTX molecules to the polymer/SWCNT composites, we have performed sensing experiments using a quartz crystal microbalance with dissipation monitoring (QCM-D). Polymer/SWCNT supernates (5 μL) were drop-casted on the gold coating of QCM-D electrodes, followed by drying in vacuo. The sensing responses are represented by the changes in the third overtone of frequency (ΔF₃) and the dissipation factor (ΔD₃) over three cycles of analyte exposures.⁸² Representative sensing profiles of P1/SG65i-coated electrodes toward BTX vapors at 100 ppm in air are shown in Figure 12. As expected, the binding of analyte molecules to the electrode surfaces results in an increase in mass and thus a decrease in frequency. The increase in the dissipation factor indicates the change in the viscoelastic properties induced by the binding of guest molecules to the surface of the P1/SG65i-coated electrode. It should be noted that the structural assemblies of the polymer/SWCNT complexes are found to aid BTX binding as the polymer/SWCNT-coated electrodes exhibit significantly higher change in frequency upon the exposure of BTX vapors than pure polymer-coated or uncoated electrodes (Figures S16-S18). The magnitudes of ΔF₃ and ΔD₃ are in the order of o-xylene > toluene > benzene, which is consistent with the chemiresistance-based sensing results. Moreover, as summarized in Figures 13a, S19a, and S20a, the sensors exhibit a linear response (R² > 0.993) for BTX vapors within the range of 100–1500, 100–1000, and 100–500 ppm, respectively. These results showcase the feasibility of BTX sensing by the polymer/SWCNT complexes with QCM-D.

Figure 12.

Changes in frequency (ΔF₃, 3rd overtone) and dissipation factor (ΔD₃, 3rd overtone) of a P1/SG65i-coated electrode upon three cycles of exposure to BTX vapors at 100 ppm for 1 min in air.

Figure 13.

(a) Summary of changes in frequency (ΔF₃, 3rd overtone) and dissipation factor (ΔD₃, 3rd overtone) of the P1/SG65i-coated electrode upon three cycles of exposure to analyte at different concentrations for 1 min in air. (b) Summary of −ΔF₃/ΔD₃ of the P1/SG65i-coated electrode upon three cycles of exposure to analyte at different concentrations for 1 min in air.

In order to differentiate the signals between BTX vapors, we herein make use of a characteristic value derived from −ΔF₃/ΔD₃,⁸⁴ which has been used to study the contact mechanics of biomolecules,⁸⁵ microbes,^86-88 and nanomaterials^89,90 with substrate surfaces. As illustrated in Figure 13a, the absolute magnitudes of ΔF₃ and ΔD₃ of the P1/SG65i-coated electrode increase with rising analyte concentration. However, the values of −ΔF₃/ΔD₃ shown in Figure 13b remain nearly constant over the dynamic range of the sensor. The values of −ΔF₃/ΔD₃ obtained from the BTX exposures are significantly different, with the order of o-xylene > toluene > benzene; they have been found to correlate with the molecular weight of the analytes.^91,92 Gratifyingly, we have observed the similar BTX differentiation by the values of ΔF₃/ΔD₃ in other sensors made from P2/SG65i and P4/SG65i (Figures S19b and S20b). Therefore, with the ability to differentiate BTX molecules, the QCM-D method can complement chemiresistance-based sensing for the sensitive and selective detection of BTX vapors. This illustrates opportunities for the introduction of additional functionality to the pentiptycene polymer/SWCNT complexes in future sensor design to robustly detect and differentiate more complex analytes.

CONCLUSION

We have developed polymer/SWCNT dispersions and thin film compositions that use the clefts of pentiptycene groups to bind the nanotubes. Strong π–π interactions between polymers and SWCNTs are found to require a complementary “fit” between the nanotube and the pentiptycene. TEM results support a structural model wherein SWCNTs bind polymers coincidently with their long axis, which provides for free volume between the tubes. The quality of dispersions is evidenced by UV–vis–NIR absorption spectra, and the minimal perturbation to the SWCNT electronic structure in the dispersion was confirmed by Raman spectroscopy. The polymer construction allows for the robust and selective detection of BTX vapors in air by monitoring the change of conductance, QCM frequency, and the dissipation factor. Notably, our sensors are sensitive enough to detect benzene at the OSHA short-term exposure limit. In total, we have demonstrated that pentiptycene groups can create defined structural assemblies with SWCNTs and will be working to develop functional materials by extending this supramolecular model.

EXPERIMENTAL SECTION

Materials.

Commercial reagents were purchased from Sigma-Aldrich, Alfa Aesar, and TCI and used as received unless otherwise noted. Deuterated solvents were purchased from Cambridge Isotope Laboratories and used as received. CoMoCAT single-walled carbon nanotubes [Signis CG100, lot no.: MKBP3333V; carbon ≥90%; ≥70% carbon as SWNT; 0.7–1.3 nm diameter], (6,5)-enriched single-walled carbon nanotubes [Signis SG65i, lot no.: MKBZ1159V; (6,5) chirality, ≥93% carbon as SWCNT; 0.7–0.9 nm diameter], and arc discharge single-walled carbon nanotubes [CarboLex AP-grade lot no.: 07826BA; 50–70% carbon basis; 1.2–1.5 nm diameter] were purchased from Sigma-Aldrich and used as received. HiPCO single-walled carbon nanotubes [lot no.: P0261, 0.8–1.2 nm diameter] were purchased from Unidym Inc. and used as received.

Instrument.

NMR spectra were recorded using a Bruker Avance 400 MHz NMR or JEOL 500 MHz spectrometer. Tip sonication was performed with a Qsonica Q125 Sonicator. Absorption spectra were obtained using an Agilent Cary 4000 UV–vis–NIR spectrophotometer. Photoluminescence and excitation spectra were acquired on a Horiba Jobin Yvon Fluorolog-3 spectrofluorometer (model FL-321) equipped with a 450 W xenon lamp as the excitation source and an F-3000 fiber optic mount that allows for fluorescence imaging outside of the sample compartment. The F-3000 couples to the T-box; light is focused from the excitation spectrometer onto the fiber-optic bundle and then directed to the sample. Fluorescence emission from the sample is directed back through the bundle and into the front-face collection port in the sample compartment. Polymer/SWCNT supernates after centrifugation were deposited into a short-path-length cell from Starna for the collection of the absorption spectra. Raman spectra were collected using a Horiba Jobin-Yvon LabRam (model HR 800) Raman confocal microscope with a 633 nm laser (1.4 μm spot size). Laser intensity was set to 10% for the 633 nm excitation wavelength. Analyte gases were generated by a FlexStream FlexBase module with precise temperature (±0.01 °C) and gas flow rate control (±1.5% of the reading). Resistance was measured using an Agilent Keysight 34970A potentiostat equipped with a 34901A 20-channel multiplexer (2/4-wire) module. The potentiostat was connected to the sensing laptop using an Agilent 82357B GPIB-USB interface high-speed USB 2.0 serial cable and controlled using BenchLink Data Logger 3 (available free of charge online). The scan rate was set to 1 scan/second. QCM-D experiments were performed using Q-Sense E1 (Q-Sense, Stockholm, Sweden) with gold-coated AT-cut quartz crystal sensors (QSX 301 Gold, Q-Sense) with a 5 MHz fundamental resonance frequency.

Synthesis of Polymer.

2-Octyldodecan-1-amine⁹³ and pentiptycene diacetylene⁶⁴ were synthesized according to methods previously reported. The syntheses of polymers P2,⁶⁴ P3,⁹⁴ P4,⁹⁵ and P5³ were reported by our group. P1 was synthesized through Sonogashira reaction: under a nitrogen atmosphere, pentiptycene diacetylene (1.0 equiv), dibromonaphthalene diimide (1.0 equiv), Pd(PPh₃)₄ (10 mol %), and CuI (0.5 equiv) were dissolved in a previously degassed mixture of dry toluene and diisopropylamine. The solution was heated at 70 °C for 3 days and then subjected to a CHCl₃/H₂O workup. The combined organic phase was washed with NH₄Cl(aq) and dried by MgSO₄. The solvent was removed in vacuo, and the residue was precipitated in methanol three times. GPC (THF vs PS): M_n = 25100, M_w = 42700, PDI = 1.70.

Preparation of Polymer/SWCNT Dispersion.

Polymer (3 mg) was dissolved in chlorobenzene (4 mL), and the solution was sonicated in a water bath for 10 min. To the polymer solution was added 2 mg of SWCNT, and the resulting mixture was chilled with ice and homogenized for 30 min using a Qsonica Q125 sonicator at 63W. Subsequently, the suspension was centrifuged for 4 h at 30130g. For UV–vis–NIR absorption spectroscopy and photoluminescence spectroscopy, the absorption and the emission spectra of the supernatant were directly recorded in a 1 mm short-path quartz cuvette.

Chemiresistive Device Preparation.

Glass slides (VWR microscope slides) were bath sonicated in acetone for 15 min and then dried with a stream of nitrogen. Using an aluminum mask, chromium (15 nm) followed by gold (50 nm) was deposited using a thermal evaporator (Angstrom Engineering), leaving a 1 mm gap between gold electrodes. For pristine SG65i SWCNTs, a stock solution of SG65i SWCNTs (2 mg) was prepared in o-dichlorobenzene (oDCB) (20 mL) by bath sonication at RT for 30 min. A 1 μL amount of the SG65i SWCNT dispersion was drop-casted in between the gold electrodes and dried at RT under house vacuum in a desiccator or vacuum oven. For polymer/SWCNT dispersions, polymer (10 mg) was dissolved in o-dichlorobenzene (oDCB, 10 mL), and the solution was sonicated in a water bath for 10 min. To the polymer solution was added 1 mg of SG65i SWCNT, and the resulting mixture was chilled with ice and homogenized for 20 min using a Qsonica Q125 sonicator at 63W. Subsequently, the suspension was centrifuged for 30 min at 8000g and allowed to stand overnight undisturbed. A 1 μL amount of the polymer/SWCNT supernate was drop-casted in between the gold electrodes and dried at RT under house vacuum in a desiccator or vacuum oven.

TEM Imaging.

The polymer/SWCNT supernate prepared for sensing was diluted by oDCB 10 times. The solution (~10 μL) was then drop-casted onto the TEM grid (lacey C only) and dried to evaporate all oDCB at room temperature.

Analyte Gas Generation.

A gas generator (FlexStream, Kin-Tek) is used to produce gas vapors from liquid sources. A trace amount of analyte is emitted from a permeation tube diluted in air, which is further diluted with air to adjust the concentration (in ppm) of analyte. BTX and VOCs were calibrated by placing 2–3 mL of the liquid in the oven flow and measuring the mass loss after a known length of time at a constant temperature.

Chemiresistive Gas Sensing Measurements.

The chemiresistive device was enclosed in a homemade Teflon gas flow chamber. The resistance of the device was measured over time (1 scan/sec), with typical procedures including 5 min equilibration time (for the baseline resistance to stabilize) followed by 1 min exposure to analyte in air and then 5 min of recovery. All presented data are given as the numeral average (N ≥ 6) accompanied by the standard deviation.

Quartz Crystal Microbalance with Dissipation Monitoring Gas-Sensing Experiments.

The polymer solution or polymer/SWCNT supernatant (5 μL) was drop-casted on the gold coating of QCM-D electrodes, followed by drying in vacuo. The third overtone of frequency (ΔF₃) and the dissipation factor (ΔD₃) of a film on a QCM sensor were measured by three cycles of exposure of a film to an analyte vapor for 1 min at 23 °C. Typical procedures include 5 min equilibration time followed by 1 min exposure to analyte in air and then 5 min of recovery.