Tailoring of the chlorine sensing properties of substituted metal phthalocyanines non-covalently anchored on single-walled carbon nanotubes

To investigate how central metal tunes the synergetic interactions between substituted metallo-phthalocyanine and single-walled carbon nanotubes in enhancing the gas sensing properties, a comparative study has been performed by varying the central metal ion in fluorinated metal phthalocyanines and single-walled carbon nanotube hybrid. Hybrids of metal(ii)-1,2,3,4,8,9,10,11,15,16,17,18-24,25-hexa-decafluoro-29H,31H-phthalocyanine/single-walled carbon nanotube (F16MPc/SWCNTs–COOH, where M = Co, Zn) have been synthesized through π–π stacking interactions using the solution route. Spectroscopic (FT-IR, UV-vis, XPS and Raman), electron microscopic (TEM and FE-SEM) and TGA investigations have confirmed the successful functionalization and interaction of SWCNTs–COOH with F16MPc. Parts per billion (ppb) level Cl2-selective chemiresistive gas sensors have been fabricated on glass substrates with precoated gold electrodes by using these hybrids. The responses of various F16MPc/SWCNTs–COOH sensors have demonstrated the central metal ion-dependence in the sensitivity of Cl2.


Introduction
Chlorine (Cl 2 ) is commonly used in water purication, pharmaceuticals, textiles, plastics, agrochemicals and household cleaning products etc. Despite being a toxic gas with an occupational exposure limit (OEL) of 500 ppb for a time-weighted average of over eight hours, it can cause distress in the respiratory system and severely affect the environment and mankind. 1,2 The very precise monitoring of Cl 2 at the parts per billion (ppb) or parts per trillion (ppt) level has led to the development of economical, exible, compact and low power consuming sensors. Different materials like metal oxides, 3-5 organic semiconductors 6,7 and carbon-based nanomaterials 8,9 have been widely explored for the fabrication of gas sensors. Metal oxide-based (particularly SnO 2 , ZnO, WO 3 , TiO 2 , V 2 O 5 ) chemiresistive gas sensors have been investigated for the detection of various toxic gases. 10,11 In organic semiconductors, polycyclic aromatic hydrocarbons, phthalocyanines, porphyrin derivatives and polymers have been used as excellent sensing materials for the detection of various harmful gases. 12,13 Carbon-based nanomaterials, fullerenes, graphene, and carbon nanotubes (CNTs), have been demonstrated as promising gas sensing materials. 14,15 However, the gas sensing characteristics of CNT hybrids with noble metal nanoparticles, 16 metal oxides 17 and organic semiconductors 18,19 are better in comparison to pristine CNTs due to the better charge transfer between the hybrid and gas analytes.
Among the organic semiconductors, metallophthalocyanines (MPcs) are attractive choices for the noncovalent functionalization of CNTs because of the synergic interaction of MPcs with CNTs due to p-p interactions. 20,21 MPcs have emerged as outstanding sensing materials in highly selective, sensitive and reversible chemiresistive gas sensors to detect various toxic gases due to their conjugated macrocyclic units. 22,23 We have previously reported the nanostructured growth of substituted MPcs for ppb level Cl 2 gas sensors with detection limits as low as 5 ppb. 24,25 Monllau et al. 26 reported highly sensitive multiwalled carbon nanotubes and epoxy resinbased amperometric sensors for the detection of free chlorine in water, at concentrations as low as 20 mg L À1 . Wang et al. 27 fabricated lead phthalocyanine modied CNTs with enhanced NH 3 sensing performance as compared to pristine CNTs. Liang et al. 20 developed substituted metal(II) phthalocyanine/multiwalled carbon nanotube hybrid (TFPMPc/MWCNT, M ¼ Co, Zn, Cu, Pb, Pd, and Ni) sensors where the central metal atoms play an important role in the high sensitivity and selectivity of the sensor towards NH 3 . The response of the TFPMPc/MWCNT hybrid sensor to ammonia vapour is in the order of Co > Zn > Cu > Pb > Pd $ Ni, which has been attributed to the binding energies of the MPc-NH 3 system. 20 Recently, we fabricated Cl 2 sensors using hybrids of carboxylic functionalized multi-walled carbon nanotubes with hexadecauorinated metal phthalocyanines, (F 16 MPc, M ¼ Cu, Zn, Co) and the response of the sensors to Cl 2 lies in the order of Co > Cu > Zn. [28][29][30] Kaya et al. 31 have shown that the response of the sensor to ammonia vapor in the concentration range 20-50 ppm is of the order CuPc-py > CoPc-py > H 2 Pc-py. It is worth mentioning that singlewalled carbon nanotubes (SWCNTs) have certain superior features compared to the MWCNTs due to the comparatively smaller size, stronger inter-tube attraction, and larger specic area of SWCNTs that will enhance the gas adsorption capability and will enhance the gas sensing parameters of the sensor. In our previous study, we have shown that the F 16 CuPc/SWCNTs-COOH hybrid sensor seems to be a signicantly better candidate for gas sensing applications in comparison to the F 16 CuPc/ MWCNTs-COOH hybrid sensor. 28 Taking these facts into consideration, in order to tune the Cl 2 sensing properties of the MPc/CNTs hybrid and to verify the effect of the central metal in the phthalocyanine molecule, we have synthesized hybrids of SWCNTs-COOH with metal(II) 1,2,3, 4,8,9,10,11,15,16,17,18-24,25- Fig. 1 shows the schematic diagram of the F 16 MPcs molecules used in this work. Further, due to molecular functionalization, the molecular orbitals come closer to the Fermi level, leading to an increase in the ionization potential and electron affinity, which result in the preferred acceptor behaviour. 34

Experimental
The SWCNTs and F 16 MPcs (where M ¼ Co, Zn) samples were commercially procured from Sigma-Aldrich. The acidication of SWCNTs consisting of the acidic group (-COOH) was performed through the established multi-step acid treatment procedures. 35 It is worth mentioning that the carboxyl group imparts negative charges and results in the long-term stability of the CNTs dispersion. 35,36 Varying amounts of F 16 MPcs (0.1 to 0.5 wt%) were dissolved in 5 mL of dimethylformamide (DMF), and subsequently subjected to stirring to give the F 16 MPcs/DMF solution. The above solution was then successively and cautiously added dropwise to the SWCNTs-COOH (30 mg) suspensions in DMF and then sonicated at room temperature (25 C) for 3 h and stirred in the dark for 6 h at 100 C. Aer stirring and ltration through a PTEF lter (0.22 mm, Millipore), the product was washed thoroughly with DMF to eradicate the excess F 16 MPcs derivative, followed by washing with ethanol numerous times, then nally drying to acquire F 16 CoPc/ SWCNTs-COOH (S 1 ) and F 16 ZnPc/SWCNTs-COOH (S 2 ) hybrids.
Raman spectra were obtained using a Renishaw inVia micro-Raman spectrometer. Fourier transform infrared (FT-IR) and ultraviolet-visible (UV-vis) spectra of S 1 and S 2 hybrids were obtained on a Perkin Elmer Frontier FT-IR spectrometer and UV-2450PC (Shimadzu, Japan) spectrophotometer, respectively. The morphologies of S 1 and S 2 hybrids were determined by eld emission scanning electron microscopy (FE-SEM, Carl Zeiss, supra 55) and transmission electron microscopy (Jeol, TEM-2100). Thermogravimetric analysis (TGA) was performed using a thermogravimetric analyzer (Hitachi STA 7200) under a nitrogen atmosphere from 40 to 900 C at a scan rate of 10 C min À1 . X-ray photoelectron spectroscopy (XPS) was conducted using a Mg K a X-ray beam as the excitation source (1253.6 eV) and a MAC2 electron analyzer system attached to an MBE machine (EVA-32 Riber, France). The binding energy scale was calibrated to the Au 4f 7/2 line of 84.0 eV.
The gas sensing studies were carried out using a homemade gas handling test chamber (1000 mL) containing a sample holder geometry as shown in Fig. 2. To prepare gas sensors of S 1 and S 2 hybrids, 2 mg of the as-prepared F 16 MPcs/SWCNTs hybrids were dispersed in 1 mL of DMF and then multiple sensors with effective area of 3 mm Â 1 mm were fabricated by drop casting 30 mL of hybrid solution onto a glass substrate with two precoated gold electrodes (3 mm Â 3 mm at a spacing of 1 mm). Silver wires were attached to the gold electrodes using silver paste. Sensor resistance was recorded continuously by applying a constant bias of 3 V during both dosing and purging cycles as a function of time, using a computer interfaced Keithley electrometer 6517A. The desired concentrations of (NO 2 , NO, Cl 2 , H 2 S, C 2 H 5 OH, CO and NH 3 ) gases in the test chamber were achieved by injecting a known quantity of gas using a micro-syringe; once steady-state was achieved aer exposure, sensor resistance was recovered by opening the lid of the test chamber.
The response of the gas sensor was calculated using eqn (1): where R a and R g represent the sensor resistance in air and gas environments, respectively. The response time is dened as the time needed for the sensor resistance to reach 90% of its nal value aer the gas is introduced into the test chamber, and recovery time is dened as the time required for the sensor resistance to regain 90% of its original value aer the purging out of the gas. XPS studies of exposed samples were conducted by the ex situ exposure of gases to the samples in the gas sensing set up (Fig. 2) and transferring them to the XPS analysis chamber. The saturated aqueous solutions of LiCl, MgCl 2 , K 2 CO 3 , NaBr, KI, NaCl, KCl and K 2 So 4 at an ambient temperature of 25 C were used for maintaining homogeneous and stable environments with relative humidities of nearly 11.5%, 32.6%, 43.4%, 57.3%, 68.4%, 75.7%, 84.5% and 97.1 respectively. 37 A hygrometer (Keithley 6517 A) was used to independently monitor the relative humidity (RH) levels. Impedance spectroscopy studies of the S 1 and S 2 hybrids were carried out using a frequency response analyser (FRA) attached with a potentiostat (Autolab) in the frequency range of 10 Hz to 1 MHz. In accordance with the sensing study, 0.3 wt% of S 1 and S 2 hybrid sensors were chosen for further study.

Material characterization of the F 16 MPc/SWCNTs-COOH hybrid
In order to explore the interactions between phthalocyanine molecules and SWCNTs-COOH, Raman and FTIR spectroscopic measurements of all the samples were conducted. The Raman spectra ( Fig. 3) of SWCNTs-COOH contain the characteristic Gband due to the bond stretching of sp 2 atoms at around 1593 cm À1 and the D band at around 1360 cm À1 due to the breathing mode of sp 2 atoms. 38,39 Moreover, a characteristic peak at 164 cm À1 is for the radial breathing mode (RBM) of SWCNTs-COOH, which signies the distribution of diameters in the SWCNTs-COOH sample. 40 16 ZnPc are due to the vibrations of isoindole moieties. 41 The characteristic peaks between 1200 and 1600 cm À1 are due to pyrrole groups. Moreover, bands at 1544 and 1509 cm À1 correspond to cobalt and zinc ions, in agreement with earlier studies. [41][42][43] It is worth noting that a combination of peaks of both the F 16 MPcs and SWCNTs-COOH were found in the Raman spectra of S 1 and S 2 hybrids. Moreover, D and G bands were found to be marginally broadened due to the superimposition with F 16 MPc peaks. Fig. 3(b) shows enlarged parts of the spectra from 100 to 1300 cm À1 , in which there is a change in the peak position and intensity of the characteristic Raman peak of the phthalocyanine macrocycle by interaction with SWCNTs-COOH. The relative intensity ratio (I D /I G ) was determined to be 0.3, 0.19 and 0.24 for SWCNTs-COOH, S 1 and S 2 samples, respectively. 28   frequency shi in the S 1 hybrid as compared to the S 2 hybrid indicates that the adsorption of F 16 CoPc induces a more signicant shi in comparison to F 16 ZnPc, due to the enhanced F 16 CoPc molecule-SWCNT-COOH interaction. 44 FTIR spectra (Fig. 4) of SWCNTs-COOH show the C-O stretching vibration peak at 1021 cm À1 , the O-H stretching vibration peak at 3440 cm À1 due to the carboxylic group 36 and a peak at 1637 cm À1 due to the C]C stretching vibration. 36 The peaks at 2855 and 2921 cm À1 correspond to asymmetric and symmetric CH 2 stretching vibrations. 46 47 It is worth mentioning that the peak appearing at 1385 cm À1 in both the S 1 and S 2 hybrids is due to the C-N-C vibration, 48 which conrms the interaction between phthalocyanine and CNTs. The characteristic peaks of the phthalocyanine molecule in the IR spectra of S 1 and S 2 hybrids are found to be slightly red shied in wavenumber in comparison to their individual peaks. Nevertheless, a higher shi in wavenumber in the S 1 hybrid in comparison to S 2 reveals that adsorption of F 16 CoPc induces a more signicant shi due to the enhanced electron delocalization via p-p stacking interactions between the F 16 CoPc molecule and SWCNT-COOH, which is concomitant with Raman spectroscopic studies. 49 Fig . 5 depicts the UV-vis absorption spectra of SWCNTs-COOH, F 16 MPcs, S 1 and S 2 hybrids. The UV-vis absorption spectrum of SWCNTs-COOH was observed to be featureless 50 and the spectra of F 16 MPcs exhibit two strong absorption bands, one broad B band in the wavelength range 305-371 nm due to the electronic transitions from the HOMO a 2 u to the LUMO e g level and the Q band in the visible range at 632-672 nm is because of the electronic transitions from the HOMO a 1 u level to the LUMO e g level. 51 However, in the case of S 1 and S 2 hybrids, the Q-band was found to be comparatively broadened with a decrease in absorption in the dispersions containing F 16 MPc/SWCNTs-COOH and their maxima were red-shied by 19 and 16 nm, respectively, as compared to that of the individual F 16 MPc spectrum. It is worth mentioning that the expanded macrocyclic conjugated structure of F 16 MPc and the reduced energy difference between the HOMO and the LUMO facilitates charge transfer between F 16 MPc and SWCNTs-COOH. In addition, the higher red-shi in the S 1 hybrid as compared to S 2 conrms the signicant p-p interaction and the charge transfer between F 16 CoPc and SWCNTs-COOH. 40 TEM (Fig. 6(a and b)) images of S 1 and S 2 hybrids demonstrate the exohedral anchoring of phthalocyanine molecules on the walls of SWCNTs-COOH with a mean diameter of about 36 and 29 nm in comparison to SWCNTs-COOH (inset view) with diameter of about 10 nm. Additionally, scanning electron microscopy images (Fig. 6(c and d)) of S 1 and S 2 hybrids also highlight that phthalocyanine molecules are anchored on the surface of the SWCNTs-COOH matrix, making a thicker SWCNTs-COOH surface in contrast to individual SWCNTs-COOH. The weight loss as a function of temperature for SWCNTs-COOH, F 16 ZnPc, F 16 CoPc, S 1 and S 2 hybrid materials has been investigated using TGA plots (Fig. 7)).
A loss of weight of about 54.60% and 49.05% up to 900 C was observed for F 16 ZnPc and F 16 CoPc (Fig. 7(a and b)), comprised of major weight losses in steps from 200 to 330 C and 366 to 604 C due to the desorption of adsorbed water and the decomposition of F 16 MPc, respectively. 45,52 TGA plots of SWCNTs-COOH (Fig. 7(e)) exhibited a weight loss of about 4.21% due to the destruction of the residual carbon and decarboxylation of oxidized species. 40 In contrast, S 1 and S 2 (Fig. 7(c and d)) had weight losses of 21.83% and 22.79%, respectively, on heating the hybrid to 900 C, corresponding to the decomposition of the F 16 MPc on the SWCNTs-COOH  surface. 45 The amount of F 16 MPc molecules adsorbed on the SWCNTs-COOH was calculated using the ratio of the difference in weight loss between SWCNTs-COOH and the F 16 MPc/ SWCNTs-COOH hybrid to the weight loss for F 16 MPc and was found to be 35.92% and 34.02% for the S 1 and S 2 hybrids, respectively.

Gas sensing measurements
To demonstrate the gas sensing properties of prepared S 1 and S 2 hybrid sensors, we recorded the response curves (change in resistance of the lm as a function of time) of the sensors to 500 ppb of different test gases at room temperature (25 C). From the selectivity histogram (Fig. 8), it can be seen that for the tested gases at room temperature, 0.3 wt% of S 1 and S 2 hybrid sensors exhibited the best response towards Cl 2 among all the prepared sensors, with sensitivity values of $40% and 30%, respectively, in comparison to the pristine SWCNTs sensor with a sensitivity value of $1%. 28 This indicates that the F 16 MPcs molecule enhances the sensor response because of the synergic interaction of MPcs with CNTs due to p-p interactions. As such, 0.3 wt% of S 1 and S 2 hybrid sensors was chosen for further sensing studies. The sensitivity values for all other tested gases were <4%. Moreover, at room temperature, the sensors showed irreversible behaviour, as they did not recover to the baseline resistance even aer a long interval of time. It was observed that heating improves the recovery characteristics of the sensors; the operating temperature was optimized in order to make the sensors reversible. Here, both sensors S 1 and S 2 were exposed to 500 ppb of Cl 2 at different operating temperatures ranging from 25 C to 200 C. A plot of sensor response for 500 ppb of Cl 2 as a function of temperature is shown in Fig. 9(a). The response for  Cl 2 was rapidly enhanced with increasing temperature and the maximum responses of $59% and 46% were obtained for sensors S 1 and S 2 , respectively, at 150 C. Furthermore, the response of the sensors decreases beyond 150 C, due to desorption of Cl 2 from the surface of the sensors. Fig. 9(b) shows the resistance variation in S 1 and S 2 sensors as a function of time for different concentrations of Cl 2 (40-2000 ppb) at 150 C. Upon exposure to Cl 2 , the sensor resistance decreases and it becomes saturated aer some time; aer purging with air, it again starts approaching its initial baseline value, indicating good reversibility. Fig. 9(c) demonstrates the response behaviour of S 1 and S 2 sensors to 40-2000 ppb concentrations of Cl 2 . The response value of the sensors increased with increasing the concentration of Cl 2 . The responses of S 1 and S 2 sensors were found to lie in the range of $18-82% and $8-72%, respectively. It is worth mentioning that the response of F 16 MPc/SWCNTs-COOH hybrids towards chlorine is greater than that of F 16 MPc/MWCNTs-COOH hybrids. 28,29 Further, due to certain superior features of SWCNTs such as smaller size, stronger inter-tube attraction and larger specic surface area compared to the MWCNTs, the gas sensing parameters of the SWCNTs-based sensors are enhanced. The response of SWCNTs hybrids decreases in the order F 16 CoPc > F 16 ZnPc > F 16 CuPc, 28 which has been claried in terms of the central ion size; i.e., the larger ionic radius and especially the interaction effects between Cl 2 and different central ions. It was found that the interactions increase with the corresponding increase in the atomic size of the atoms/ions for a given Fig. 8 (a and b) Selectivity histogram of S 1 and S 2 sensors for 500 ppb of NO 2 , NO, Cl 2 , C 2 H 5 OH, H 2 S, NH 3 and CO at room temperature. separation distance because larger atoms are more easily polarizable and provide more electrons to polarize, which results in strong van der Waals forces, indicating that the central metal size plays an important role in the sensitivity of Cl 2. This is in agreement with charge transfer and the number of Pc molecules adsorbed onto the SWCNTs wall, as estimated by TGA and Cl 2 interactions with the sensor as observed in X-ray photoelectron and EIS studies discussed in Section 3.3. 53 Fig. 9(d) shows the response curves of S 1 and S 2 sensors for successive exposures to Cl 2 . The sensors showed no signicant changes in response and recovery characteristics aer repeated gas exposure, demonstrating the reproducible and stable sensing characteristics of the sensors. Fig. 10 represents the variation in the Cl 2 response of S 1 and S 2 sensors with relative humidity (11-98%) for 2000 ppb of Cl 2 at room temperature. It was observed that both sensors showed only small variations (2.4% for S 1 and 3.26% for S 2 sensor) in their Cl 2 response as the humidity level was varied from 11% to 98%, indicating that humidity has a negligible effect on the Cl 2 response of these sensors.
The response variation with the gas concentration was studied using eqn (2): 54,55 where a is the coefficient for the adsorption capacity and b is the strength of adsorption, 55 which are obtained by curve tting of Fig. 10 Variation in the Cl 2 response of S 1 and S 2 sensors with humidity for 2 ppm of Cl 2 at room temperature. Fig. 11 Raman spectra of the S 1 sensor recorded (a) before exposure (black curve), after exposure (red curve) to 25 ppm of Cl 2 and after full recovery (blue curve); (b) magnified view in the range 30-1300 cm À1 . Fig. 12 Raman spectra of the S 2 sensor recorded (a) before exposure (black curve); after exposure (red curve) to 25 ppm of Cl 2 and after full recovery (blue curve) (b) in the range 30-1300 cm À1 (magnified view).
the response curve. It is interesting to note that a smaller value of b, lying between 0 and 1 for normal adsorption, signies the greater heterogeneity of the sample. The values of a and b were calculated to be 6.94 and 0.32 for S 1 and 2.62 and 0.43 for S 2 hybrids. A b value less than one indicates the normal mode of adsorption on the heterogeneous surface of the sensor. 54 In the above experimental study, the lowest detectable concentration is limited due to the experimental set up used. Nevertheless, the limit of detection (LOD) of the sensor was derived from the signal-to-noise ratio (S/N), which is dened as DR/s, where DR is the maximum resistance change with respect to R a (baseline resistance) and s represents the root mean square (rms) noise of the baseline in air. 56 The LOD was calculated by using the eqn (3): 57 The signal-to-noise ratios of sensors S 1 and S 2 are 3000 and 2400, respectively, with the corresponding detection limits of 0.04 ppb and 0.05 ppb, respectively. Thus, the higher sensitivity, reversibility and reproducibility of the F 16 MPc/SWCNTs hybrid sensor in comparison to other CNTs-based Cl 2 sensors reported in literature 1,58 make these sensors favourable candidates for ppb level Cl 2 detection.

Gas sensing mechanism
Raman, XPS and impedance spectroscopic measurements of S 1 and S 2 hybrid sensors have been performed both in air and aer purging in Cl 2 in order to explore the sensing mechanism of the sensors. On exposure of the S 1 sensor to Cl 2 , the Raman peak (Fig. 11) corresponding to the cobalt-nitrogen bond 59 (171 cm À1 ) is shied by 8 cm À1 and macro-cyclic vibration 60 peaks (283, 586, 738, 820, 1196 cm À1 ) are shied by 6 cm À1 , whereas D and G bands corresponding to SWCNTs-COOH (1341 and 1595 cm À1 ) are shied by 1 cm À1 . 38 In contrast, in the Raman spectra (Fig. 12) of sensor S 2 , the peak corresponding to the zinc-nitrogen bond 59 (168 cm À1 ) is shied by 7 cm À1 and macro-cyclic vibration 60 peaks (119, 466, 586, 746 cm À1 ) are shied by 5 cm À1 , whereas D and G bands corresponding to SWCNTs-COOH (1342 and 1593 cm À1 ) are shied by 2 and 1 cm À1 , respectively. 38 The major shi of 10 cm À1 in S 1 and 9 cm À1 in S 2 (1544 cm À1 and 1509 cm À1 ) corresponds to the displacement of the C-N-C bridge bond, closely linked to the cobalt and zinc ions of the phthalocyanine molecule. 43 The higher shi in the S 1 hybrid as compared to the S 2 hybrid indicates the predominant interaction of Cl 2 with the cobalt ions of the hybrid sensor. The Raman spectra of the hybrid sensors recorded aer purging Cl 2 showed identical peaks to those of fresh samples, which reects the excellent reversibility of these sensors.
The interactions between Cl 2 and the S 1 /S 2 hybrid sensors were further conrmed by observing the shis in binding energy in the XPS spectra of unexposed and Cl 2 exposed samples. The XPS spectrum (Fig. 13) of the fresh S 1 hybrid shows characteristic peaks at 284.7, 532.1, 399.0, 686.9, 780.0, 795.3 eV corresponding to C-1s, O-1s, N-1s, F-1s, Co-2p 3/2 and Co-2p 1/2 levels, respectively. 40 Once the sample was exposed to Cl 2, there was a peak shi of 0.2 eV in the spectrum of the core level C-1s, a shi of 0.1 eV in the spectrum of O-1s and F-1s, a shi of 0.3 eV in the spectrum of N-1s and a prominent peak shi of 0.8 eV in the core level spectrum of Co-2p. In contrast, the XPS spectrum (Fig. 14) of the fresh S 2 hybrid showed characteristic peaks at 284.8, 532.6, 399.0, 687.1, 1019.6, 1042.6 eV corresponding to C-1s, O-1s, N-1s, F-1s, Zn-2p 3/2 and Zn-2p 1/2 levels 24 and aer Cl 2 exposure, there was a peak shi of 0.2 eV in spectrum of core level F-1s, a shi of 0.1 eV in the spectra of C-1s and O-1s, a shi of 0.3 eV in the spectrum of N-1s and a prominent peak shi of 0.7 eV in the core level spectrum of Zn-2p. The prominent shi of 0.8 eV towards the higher BE side in the Co-2p core level in the S 1 hybrid, a shi of 0.7 eV towards the higher BE side in the Zn-2p core level in the S 2 hybrid and a shi of 0.5 eV towards the higher BE side in Cu-2p 28 conrm that charge transfer interactions occur upon adsorption of strong electron acceptor Cl 2 molecules to the hybrid, leading to a decrease in electron density due to the transfer of electrons from the hybrid to Cl 2 . 29,61 Thus, the analysis of the Raman and XPS spectroscopic observations is concomitant with the higher sensing response of the S 1 hybrid because greater charge transfer takes place between Cl 2 and the S 1 hybrid through the central metal ion with the adsorption of Cl 2 . Nevertheless, charge can favourably travel from CNTs to F 16 MPcs, which leads to an increase in the hole concentration in CNTs and results in the fast variation in resistance as observed in Fig. 9(b). 61,62 It is worth mentioning that there was no shiing of peak position in the XPS spectrum aer recovery, and the absence of any chlorine signal conrms that the sensing process is highly reproducible.
The interaction between the F 16 MPc/SWCNTs-COOH sensor and Cl 2 has also been further studied using impedance spectroscopy tools, providing information about F 16 MPc/SWCNTs-COOH grains and the respective grain boundaries in accordance with morphological studies. Fig. 15(a and b) shows the impedance spectra of the F 16 MPc/SWCNTs-COOH sensor, obtained in air and under exposure to 500 ppb of Cl 2 , i.e., the Cole-Cole plot. 53,63 With an equivalent circuit (in the inset in Fig. 15) Paper consisting of the RC network in series with a resistor R 0 , a single semi-circle was observed before and aer exposure to Cl 2 . The intercept of the arc at high frequency with the real axis gives the grain resistance (R 0 ). The resistance across the grain boundary (R 1 ) was found from the diameter of the arc in Fig. 15, whereas the capacitance across the grain boundary (C 1 ) was estimated from the relation.
Interestingly, the parameter R 0 remained the same in S 1 and S 2 sensors in air and on exposure to Cl 2 , but R 1 decreased and C 1 increased in the presence of Cl 2 . This result is also supported by the fact that R 1 changes across the grain boundary in the order of S 1 > S 2 > H 1 (ref. 28) for hybrid sensors, indicating that incoming Cl 2 molecules were adsorbed onto the outer surfaces of the grains and resulted in increased hole conductivity because of the charge transfer between phthalocyanines and CNTs, as explained in XPS investigations (Table 1). 28,53

Conclusions
We have fabricated F 16 CoPc/SWCNTs-COOH and F 16 ZnPc/ SWCNTs-COOH hybrid sensors using the solution assembly route, through p-p stacking interactions between F 16 MPc and SWCNTs-COOH for chlorine sensing applications. The results demonstrate that the F 16 CoPc/SWCNTs-COOH sensor exhibits high sensitivity ($82% for 2 ppm with LOD of 0.04 ppb), excellent reproducibility and selectivity towards chlorine with response decreasing in the order of Co > Zn > Cu, indicating that the central metal ions play an important role in the sensitivity of Cl 2 , and this is in good agreement with the central ion size: the larger the ionic radius, the greater the charge transfer and Cl 2 interaction with the sensor as observed from Xray photoelectron, Raman and electrochemical impedance spectroscopic studies. Such effectiveness of the sensor originates from the synergetic interaction between F 16 MPc and SWCNTs-COOH. Strong response and good selectivity underline the signicant potential of these hybrid materials in designing a new low-cost Cl 2 sensor.

Conflicts of interest
There are no conicts to declare.  15 Impedance spectra of fresh and Cl 2 exposed S 1 and S 2 sensors (the insets show the equivalent circuits used for the analysis of data obtained from the S 1 and S 2 sensors).