Tarita Biverab,
Francesco Criscitielloa,
Fabio Di Francescoab,
Matteo Minichinoa,
Timothy Swagerc and
Andrea Pucci*ab
aDipartimento di Chimica e Chimica Industriale, Università di Pisa, Via Moruzzi 13, 56124 Pisa, Italy. E-mail: andrea.pucci@unipi.it
bINSTM, Unità di Ricerca di Pisa, Via Moruzzi 13, 56124 Pisa, Italy
cDepartment of Chemistry and Institute for Soldier Nanotechnologies, Massachusetts Institute of Technology, 77 Massachusetts Avenue, Cambridge, MA 02139, USA
First published on 24th July 2015
We report on a new ionic surfactant based on extended polycyclic aromatic perylene bisimides (PBI), namely N,N′-bis(2-(1-piperazino)ethyl)-3,4,9,10-perylenetetracarboxylic acid diimide dichloride (PZPERY), suitable for the exfoliation of multi-walled carbon nanotubes (MWCNTs). The ultrasonication (400 W for 5 min) of MWCNT/PZPERY water mixtures followed by centrifugation (4000 rpm for 30 min) provided disentangled and undamaged MWCNTs, which interact with the PZPERY molecules through π–π stacking interactions. The deposition of MWCNT/PZPERY dispersions on plastic film supporting gold electrodes allowed the fabrication of temperature sensors showing reproducible electrical resistances and typical semiconducting (activated) electrical transport with decreased resistivity when heated from 20 to 40 °C. The resistivity–temperature profile was very reproducible and with a negative temperature coefficient of about −0.02 K−1, a value comparable to that found in common thermistors. Our results suggest that MWCNT/PZPERY based films have utility for the formation of sensitive, stable and reproducible sensors to measure body temperature.
Recently, extended polycyclic aromatic perylene bisimides (PBI) have enabled the dispersion of graphitic materials, such as CNTs, by means of the non-covalent functionalization through strong π–π stacking interactions.16,18–22 PBI was found to constitute an exceptionally well-suited anchor unit to the sp2 carbon network of CNTs and allowed their effective dispersion in solution.20,23,24 In this process, the energy dissipated with sonication disassembles CNT bundles that are immediately surrounded by PBI molecules. The PBIs bind via strong π–π interactions between CNTs and prevent their re-bundling. PBIs are widely used as pigments and dyes, with colours ranging from red to black depending on the particle size and crystal packing of the molecules.25–28 In addition to optical features such as strong absorption in the visible region, high quantum yields and excellent photostability, PBIs also display a low reduction potential,29 which enables their use as electron-acceptors and n-type semiconductors in photoinduced charge–transfer reactions. For these appealing properties, perylene derivatives have been extensively studied for the fabrication of field-effect transistors,30 luminescent solar concentrators,31 dye lasers,32 photovoltaic cells33 and organic light-emitting diodes (OLEDs).34 These properties and the strong associations with CNTs make integrated CNT/perylene materials systems attractive candidates for the generation of nanostructured electronic devices and sensors.
The ability of PZPERY molecules to generate π–π stacking interactions among their aromatic backbones has been previously utilized by our group for the preparation of chromogenic plastic materials as temperature indicators.35 The effective π-associations between PZPERY and MWCNTs suggested to us that these nanocomposites should function as temperature sensors, since both semiconducting and metallic CNTs are characterized by temperature sensitive resistivities.36 Previous investigations have demonstrated the possibility of fabricating flexible, miniaturized MWCNT/polymer temperature sensors in the form of thin films to be deposited on different support materials. This kind of device was reported to find successful application wherever low cost, large surface and low volume of the sensor are fundamental issues. For example, wireless, flexible and disposable temperature sensors could be used in the monitoring of body temperature in the range 20–40 °C, and in particular the bed temperature of chronic wounds.6,37
However, these systems displayed resistance sensitivity to temperature of only −0.004 K−1, a value that is comparable to that found in metals.6,37,38 Such low sensitivity was ascribed to the polymeric dispersants, such as poly(styrene-b-(ethylene-co-butylene)-b-styrene) (SEBS)6 or poly(vinylbenzyl chloride) derivative with triethylamine,37 and the insulating nature of these components limit the thermal and electronic conductivities of the nanocomposite. In the present paper, we explore the possibility that the highly conjugated core of the PZPERY surfactant would be able to produce higher performance (resolution) temperature sensors.
In this work, we report on the preparation and characterization of MWCNTs water dispersions by using the N,N′-bis(2-(1-piperazino)ethyl)-3,4,9,10-perylenetetracarboxylic acid diimide dichloride (PZPERY) as ionic surfactant (Fig. 1).
![]() | ||
Fig. 1 Chemical structure of the N,N′-bis(2-(1-piperazino)ethyl)-3,4,9,10-perylenetetracarboxylic acid diimide dichloride (PZPERY). |
The MWCNT/PZPERY dispersions were obtained by ultra-sonicating their mixtures in water for five minutes followed by centrifugation, and the quantity of dispersed CNTs was quantified by gravimetric analyses. Raman spectroscopy and transmission electron microscopy (TEM) were employed to assess the structural integrity of MWCNTs. The electrical resistance of films, obtained by drop casting aliquots of MWCNT/PZPERY dispersions, was investigated over the range 20–40 °C to explore their potential for the development of sensitive, stable, and reproducible temperature sensors.
FT-IR (KBr): 2926 (νCH aliph.), 1695, 1656, 1593 (νCO imide), 1508, 1441, 1401, 1362 (νCC ring) cm−1.
1H NMR δ: 8.8–8.9 (d, 4H arom., J = 15 Hz), 4.7–4.9 (m, 4H, Nimm–CH2), 4.4–4.5 (m, 4H, Npip–CH2), 4.0 (m, 16H, CH2) ppm.
13C NMR δ: 31.2, 37.8, 45.5, 52.8, 117.5, 120.4, 122.5, 125.4, 129.3, 132.5, 162.5 ppm.
Elemental analysis: calcd (C36H38Cl2N6O4): 62.9% C, 5.3% H, 12.2% N; found: 63.5% C, 5.7% H, 12.4% N.
Thermogravimetric scans were carried out by a Mettler Toledo Starc System (TGA/SDTA851e). Samples were heated from 100 to 850 °C at 10 °C min−1 under a nitrogen flow. Samples were annealed at 100 °C for 10 min before measurement to remove water. The morphologies of the MWCNT/PZPERY films were examined with a Philips CM12 transmission electron microscope (TEM). Raman spectra of solid dispersion obtained by drop casting the supernatant of each sample (after centrifugation) onto glass substrates were measured with a Horiba LabRAM HR Raman Spectrometer at an excitation wavelength of 532 nm. Sensors were fabricated by casting MWCNT/PZPERY dispersions in water onto supports (Cad Line, Pisa, Italy) consisting of a polyimide film (Kapton®, thickness 50 μm) and suitable electrodes. Kapton® was chosen due to its flexibility, chemical inertness, and low permeability to water and vapours, which helps the protection of the sensing film. Copper tracks were prepared by photolithography and then electroplated with nickel and gold for electrodes fabrication (dimensions: length 7 mm, width 1 mm, distance 2 mm; thickness of copper 35 μm, nickel 3.0 μm, gold 1.2 μm). Sensor calibration in the range 20–40 °C was performed by placing the sensors in a temperature controlled hot stage (±0.1 °C) and measuring the electrical resistance with a digital multimeter (KEITHLEY Mod. 2700).
A thermogravimetric analysis permitted the quantification of the amount of MWCNTs included in the MWCNT/PZPERY dispersion (Fig. 2).
The thermal degradation of PZPERY was observed to occur at temperature of about 20 °C higher in the presence of MWCNTs (i.e. 338 instead of 315 °C), suggesting extra stability via the existence of effective interactions between the dye and the graphitic material. A weight percentage of 16.5% was determined for MWCNTs from the difference in weight between the residuals of the same amount of MWCNT/PZPERY and PZPERY at 850 °C, i.e. after the complete degradation of the organic matrix. This allowed us to conclude that the dispersion contained 1.0 mg mL−1 and MWCNT 0.0165 mg mL−1 (PZPERY:
MWCNT ≈ 61) (Fig. S2†).
As concerns PZPERY alone, Fig. S3 of the ESI† shows that the UV-vis spectra recorded with increasing concentrations of the perylene dye did not display new bands or shoulders appearance. This result suggests that PZPERY dye persists as a stable monomeric form over the explored range. Fig. 3a shows a plot of the recorded absorbance vs. CPERY (squares): the linear trend further confirms that no aggregation of the PZPERY (in the absence of MWCNTs) occurs under the explored conditions. To further investigate possible auto-aggregation effects, the Franck–Condon progression can also be used (Abs0–0/Abs0–1), where Abs0–0 is the absorption at λ = 500 nm and Abs0–1 is the absorption at λ = 550 nm. A decrease in the Abs0–0/Abs0–1 value is considered indicative of an aggregation process.11 Fig. 3b (squares) shows a constant (within the experimental error) Abs0–0/Abs0–1 ratio over our concentration range and confirms the absence of auto-aggregation of PZPERY under the explored conditions.
Moving to MWCNT/PZPERY dispersions, Fig. 3a and b (circles) show the relevant absorbance values corrected from scattering effects using a procedure based on double logarithmic plots.39,40 Linearity up to a concentration of 0.4 mg mL−1 and a constant Abs0–0/Abs0–1 ratio are observed, which both indicate that the MWCNT dispersion is efficiently stabilised by PZPERY. The fluorescence emission data suggested similar conclusions. As shown in the ESI (Fig. S4†) the emission characteristics of PZPERY were not significantly affected by the presence of MWCNTs. This confirmed that the interaction of the excess PZPERY with MWCNTs does not result in an aggregation of the dye on the MWCNT surface but had a stabilizing effect on the dispersion. In conditions of a large excess of PZPERY, the interaction with MWCNTs is not reflected in any steady state absorption and emission spectra.
Notably, Raman spectroscopy is a versatile technique used for the structural characterization of CNTs, as well as for studying their interaction with the exfoliating agents. Fig. 4 shows the Raman spectrum of the MWCNT/PZPERY dispersion compared to a previously investigated MWCNT composite with a thermoplastic elastomer, i.e. the poly(styrene-b-(ethylene-co-butylene)-b-styrene) (SEBS), which had been reported to be very effective in CNT exfoliation and dispersion.6,38 The most prominent features in a Raman spectrum of MWCNTs are the G and D bands (Fig. 4). The G-band at 1560 cm−1 is an intrinsic feature of CNTs related to the planar vibration of carbon atoms in most sp2 graphitic materials. Conversely, the disorder-induced D-band at 1320 cm−1 is attributed to the scattering from defects breaking the basic symmetry of the graphene sheet. MWCNTs typically show the highest ratio (ID/IG) between the intensities of these peaks among all carbon nanotubes, since a large number of structural defects are present within the multiple graphite layers.
![]() | ||
Fig. 4 Raman spectra of MWCNT/PZPERY (solid line) and MWCNT/poly(styrene-b-(ethylene-co-butylene)-b-styrene) (SEBS) (dashed line)6 films (λexc. = 532 nm). The wavenumbers (cm−1) of the peaks are reported as insets. |
In the spectrum of MWCNT/PZPERY films, the two lines of PZPERY at 1294 and 1372 cm−1 are well resolved and superimposed on an underlying fluorescent background.41 The D band of MWCNTs is obscured by the large PZPERY scattering, and its low intensity may be also the result of a shielding effect from the dye coating and the CNTs defects.42 It is worth noticing that the G band in the MWCNT/PZPERY dispersion is split in two unresolved peaks, which are blue-shifted of about 15–20 cm−1 compared to the SEBS based nanocomposite. This feature may be a consequence of the effective π-interactions between MWCNTs and the PZPERY molecules, which favour the exfoliation of CNTs by keeping them apart (i.e., isolated double bonds resonate at higher frequencies than in the G band of interacting CNTs).43
T (°C) | R (kΩ) | |||||
---|---|---|---|---|---|---|
Ia | Ib | IIa | IIb | IIIa | IIIb | |
23.7 | 46 | 27 | 24 | 54 | 46 | 43 |
27.0 | 37 | 25 | 22 | 46 | 42 | 38 |
29.0 | 32 | 24 | 21 | 40 | 38 | 34 |
30.2 | 30 | 23 | 20 | 38 | 37 | 34 |
31.5 | 29 | 22 | 19 | 35 | 35 | 33 |
34.0 | 27 | 20 | 18 | 33 | 33 | 31 |
35.6 | 26 | 19 | 18 | 32 | 32 | 30 |
40.0 | 24 | 18 | 16 | 29 | 30 | 28 |
![]() | ||
Fig. 6 Electrical resistance of the MWCNT/PZPERY IIIa sensor as a function of temperature, and exponential fit to the data (y = y0 + Ae−B/T; R2 = 0.98). |
Resistivity variations of more than 35% were recorded over the 23.7 to 40.0 °C temperature interval for all the sensors. Sensors from the dispersion III displayed an almost equivalent response to temperature (Table 1) and were further investigated in successive calibration cycles.
The resistance responses of the MWCNT/PZPERY III sensors were determined over four successive heating cycles (Table 2). Reproducible resistance variations were observed, with maximum amplitudes of 30 kΩ within the temperature interval of 20 °C. These features support the use of the MWCNT/PZPERY system as a resistive sensor for temperature variations within the physiological regime.
1st cycle | 2nd cycle | 3rd cycle | 4th cycle | ||||||||
---|---|---|---|---|---|---|---|---|---|---|---|
T (°C) | R (kΩ) | T (°C) | R (kΩ) | T (°C) | R (kΩ) | T (°C) | R (kΩ) | ||||
IIIa | IIIb | IIIa | IIIb | IIIa | IIIb | IIIa | IIIb | ||||
23.7 | 46 | 43 | 21.5 | 70 | 70 | 21.8 | 77 | 68 | 21.4 | 81 | 72 |
27.0 | 42 | 38 | 28.0 | 56 | 56 | 32.0 | 63 | 54 | 25.0 | 74 | 62 |
29.0 | 38 | 34 | 31.0 | 51 | 51 | 37.0 | 52 | 45 | 27.2 | 69 | 58 |
30.2 | 37 | 34 | 33.5 | 47 | 47 | 38.5 | 50 | 44 | 29.9 | 65 | 56 |
31.5 | 35 | 33 | 34.0 | 46 | 46 | 40.0 | 48 | 43 | 31.3 | 61 | 54 |
34.0 | 33 | 31 | 35.0 | 45 | 44 | 33.4 | 58 | 50 | |||
35.6 | 32 | 30 | 36.5 | 43 | 43 | 37.0 | 54 | 48 | |||
40.0 | 30 | 28 | 38.5 | 42 | 42 | 38.5 | 53 | 46 | |||
39.5 | 41 | 41 | 41.5 | 51 | 44 | ||||||
40.4 | 40 | 40 |
It is worth noting that, plotting the natural logarithm of the resistance variation (R/R0, with R0 the resistance at the lowest temperature) against the inverse of the temperature in K, a linear Arrhenius type behaviour is evident in all the heating cycles (Fig. 7). Moreover, the linear correlation (R2 ≈ 0.99) appears almost identical for all the experiments, thus confirming that the MWCNT/PZPERY system is a highly reproducible temperature sensor. The temperature coefficient of the MWCNT/PZPERY sensor was calculated according to eqn (1) and the values reported in Table 3:
R(T) = R(T0)[1 + α(T − T0)] | (1) |
α (K−1) | |||||||
---|---|---|---|---|---|---|---|
1st cycle | 2nd cycle | 3rd cycle | 4th cycle | ||||
IIIa | IIIb | IIIa | IIIb | IIIa | IIIb | IIIa | IIIb |
−0.021 | −0.021 | −0.023 | −0.023 | −0.023 | −0.021 | −0.019 | −0.020 |
The MWCNT/PZPERY sensor displayed sensitivity corresponding to a negative temperature coefficient of −0.02 K−1, an absolute value that is comparable to the highest values found in common thermistors (0.044 K−1) and an order of magnitude higher than those of metals (0.0037–0.006 K−1; 0.00385 K−1 for a Pt(100) sensor). This higher sensitivity compared to the previously reported MWCNT/polymer systems6,37,38 likely originates from the extended polycyclic aromatic perylene cores, which enable a more efficient electrical and thermal conduction within the sensor device. Notably, α maintains a constant value for every measurement cycle, thus confirming the relevant properties of the prepared temperature sensor.
Footnote |
† Electronic supplementary information (ESI) available. See DOI: 10.1039/c5ra11544b |
This journal is © The Royal Society of Chemistry 2015 |