Ethylene-Octene-Copolymer with Embedded Carbon and Organic Conductive Nanostructures for Thermoelectric Applications

Hybrid thermoelectric composites consisting of organic ethylene-octene-copolymer matrices (EOC) and embedded inorganic pristine and functionalized multiwalled carbon nanotubes, carbon nanofibers or organic polyaniline and polypyrrole particles were used to form conductive nanostructures with thermoelectric properties, which at the same time had sufficient strength, elasticity, and stability. Oxygen doping of carbon nanotubes increased the concentration of carboxyl and C–O functional groups on the nanotube surfaces and enhanced the thermoelectric power of the respective composites by up to 150%. A thermocouple assembled from EOC composites generated electric current by heat supplied with a mere short touch of the finger. A practical application of this thermocouple was provided by a self-powered vapor sensor, for operation of which an electric current in the range of microvolts sufficed, and was readily induced by (waste) heat. The heat-induced energy ensured the functioning of this novel sensor device, which converted chemical signals elicited by the presence of heptane vapors to the electrical domain through the resistance changes of the comprising EOC composites.


Introduction
Thermoelectric conductive polymer composites convert thermal energy into electricity when there is a difference in temperatures between the hot and cold junctions of such two dissimilar conductive or semiconductive composites. The conversion is based on a phenomenon called the Seebeck effect. The heat supplied at the hot junction (hot side) of the thermoelectrics causes a flow of electric current to the cold side that can be harnessed as useful voltage.
The classical thermoelectrics are made from inorganic materials such as metals (Al, Cu, Ni), metallic alloys (chromel, alumel) and semiconductors (PbT, Bi 2 Te 3 ), which are thermoelectrically efficient, but at the same time expensive, heavy, and some of them materially in short supply. Consequently, alternative organic thermoelectrics are being developed. These organic thermoelectrics do not produce as much energy as the metal ones. The thermoelectric figure of merit is typically only in the range 0.001-0.01 at room temperature. However, the mechanical flexibility, processability,

Materials and Methods
Purified MWCNTs produced by a chemical vapor deposition of acetylene were supplied by Sun Nanotech Co. Ltd., Jiangxi, China. According to the supplier, the nanotube diameter is 10-30 nm, the length 1-10 µm, the purity~90% and the volume resistivity 0.12 Ωcm. The diameter of individual nanotubes is between 10 and 60 nm (100 measurements) at the average diameter and the standard deviation is 15 ± 6 nm. The nanotube length is from about 0.2 µm up to 3 µm. They consist of 15 to 35 rolled layers of graphene at the interlayer distance of ca 0.35 nm [19]. The pristine nanotubes are denoted further on as MWCNT(Sun)s.
The oxygenated MWCNTs were prepared in a glass reactor with a reflux condenser filled with 250 mL of 0.5 M H 2 SO 4 , into which 5 g KMnO 4 and 2 g MWCNTs were added. Then the dispersion was sonicated at 85 • C for 15 h using a thermostatic ultrasonic bath (Bandelin electronic DT 103H, Merck spol. s r. o., Prague, Czech Republic). Thereafter, the product was filtered, washed with concentrated HCl to remove MnO 2 , thoroughly rinsed with deionized water and dried [19]. Alternatively, the MWCNTs were oxygenated by HNO 3 as follows: 2 g of the MWCNTs were added to 250 mL of HNO 3 (concentrated) and heated at 140 • C for 2 h. After that, the dispersion was cooled and filtered. The sediment was washed by deionized water and dried at 40 • C for 24 h. The corresponding oxygenated MWCNTs are denoted MWCNT(Sun)H 2 SO 4 or MWCNT(Sun)HNO 3 and MWCNT(Bayer)H 2 SO 4 or MWCNT(Sun)HNO 3 .
For preparing PANI emeraldine salt, 0.2 M aniline hydrochloride was mixed with 0.25 M ammonium persulfate (APS) in water, briefly stirred, and left to polymerize for 24 h at room temperature. Then the PANI precipitate was collected on a filter and washed with 0.2 M HCl and acetone. Polyaniline emeraldine particles were in turn dried in air and then under vacuum at 60 • C for 24 h [20].
Pyrrole monomer (Py, purity ≥ 98%, Sigma-Aldrich Inc., St. Louis, MO, USA) was distilled twice under reduced pressure and stored below 4 • C. Polypyrrole particles were prepared via in situ polymerization of the precooled Py in a system containing surfactant CTAB. The precooled initiator APS was added into the system dropwise and the polymerization was allowed to proceed with stirring for 2 h at 0-5 • C. After being washed with water and ethanol, the PPy powder was dried in air and then under vacuum at 60 • C for 24 h. After drying, both types of polymers were gently ground with mortar and pestle.
The carbon allotrope/EOC composites were prepared by an ultrasonication of dispersions of MWCNTs or CNFs in EOC/toluene solution (5% of EOC in toluene). The chosen concentration of the filler in the composites was 30 wt %, which was well above the percolation threshold. The ultrasonication of the dispersion was done in the thermostatic ultrasonic bath (Bandelin electronic DT 103H, Merck spol. s r. o., Prague, Czech Republic) for 3 h at 80 • C. Then the dispersion was poured into acetone to form a precipitate. The final composite sheets were prepared by compression molding at 100 • C [14]. The organic PANI (PPy)/EOC composites were prepared in the same way as the carbon allotrope composites except for filler concentration, which was 70 wt %.
MWCNTs were analyzed by means of transmission electron microscopy (TEM) using the microscope JEOL JEM 2010 (Jeol Ltd., Freising, Germany) at an accelerating voltage of 160 kV. The sample, the MWCNT dispersion in acetone prepared by ultrasonication, was deposited on a 300 mesh copper grid with a carbon film (SPI, Washington, DC, USA) and dried. The structure of CNFs was observed by means of a scanning electron microscope (SEM) Nova NanoSEM 450, FEI, Lincoln, NE, USA, at operating voltage 10 kV. The sample, CNF dispersion in acetone prepared by ultrasonication, was deposited on carbon targets and covered with a thin Au/Pd layer. For the observations, a regime of secondary electrons was used. The same scanning electron microscope was used to observe the morphology of PANI emeraldine salt and of PPy particles.
The X-ray photoelectron spectroscopy (XPS) signals were measured to obtain information on functional groups attached onto the nanotube surfaces. The XPS signals from MWCNT(Sun), MWCNT(Sun)KMnO 4 and MWCNT(Sun)HNO 3 network surfaces were recorded using the Thermo Scientific K-Alpha System TFA XPS (Physical Electronics Instrument, Chanhassen, MN, USA) equipped with a micro-focused, monochromatic A1 Ka X-ray source (1486.6 eV). An X-ray beam of 400 µm size was used at 6 mA × 12 kV [21]. The spectra were acquired in the constant analyzer energy mode with pass energy of 200 eV for the survey. Narrow regions were collected using the snapshot acquisition mode (150 eV pass energy) enabling rapid collection of data (5 s per region). The narrow region data were as post-processed using the Jansson's algorithm to remove the analyzer point spread function, which resulted in an improved resolution of the spectra for the peak deconvolution [22]. The concentration of elements was determined from survey spectra by MultiPak v7.3.1 software (Physical Electronics Inc., Chanhassen, MN, USA).
Fourier-transform infrared (FTIR) analyses of MWCNTs, MWCNT(KMnO 4 )s and MWCNT(HNO 3 )s were performed on the FTIR spectrometer Nicolet 6700 (Thermo Scientific, Waltham, MA, USA). The transmission accessory was used for pristine MWCNT(Sun)s, MWCNT(Sun)KMnO 4 and MWCNT(Sun)HNO 3 samples in powder form prepared by potassium bromide. FTIR analyses of chemical composition of PANI emeraldine and PPy particles were examined by the above mentioned FTIR spectrometer Nicolet 6700 using the attenuated total reflectance technique with a germanium crystal in the range 600-4000 cm −1 at 64 scans per spectrum at 2 cm −1 resolution.
The Hall coefficient, the resistivity, and the conductivity of the samples as well as the charge mobility and the charge carrier concentration were measured by means of the HCS 1 apparatus (Linseis Messgeräte GmbH, Selb, Germany) equipped with static 0.7 T field permanent magnets for bipolar measurement. The disc form samples had diameter 20 mm and thickness 2.65 mm. The sample current was set to 4 mA. The thermoelectric power measurement was carried out for all the samples using the set-up illustrated in Figure 1. The schematic diagram shows that the circular composite sample (diameter 20 mm, thickness 2 mm) was placed between two copper electrodes. The ends of each of the Cu electrodes were immersed in thermostatic silicone oil baths set at different temperatures. The temperature at the copper/composite interfaces was measured by a Pt100 temperature sensor. The arising thermoelectric current was measured by the Keithley 2000 Digital Multimeter (Tektronix, Inc. Beaverton, OR, USA). sample, the MWCNT dispersion in acetone prepared by ultrasonication, was deposited on a 300 mesh copper grid with a carbon film (SPI, Washington, D.C., USA) and dried. The structure of CNFs was observed by means of a scanning electron microscope (SEM) Nova NanoSEM 450, FEI, Lincoln, NE, USA, at operating voltage 10 kV. The sample, CNF dispersion in acetone prepared by ultrasonication, was deposited on carbon targets and covered with a thin Au/Pd layer. For the observations, a regime of secondary electrons was used. The same scanning electron microscope was used to observe the morphology of PANI emeraldine salt and of PPy particles. The X-ray photoelectron spectroscopy (XPS) signals were measured to obtain information on functional groups attached onto the nanotube surfaces. The XPS signals from MWCNT(Sun), MWCNT(Sun)KMnO4 and MWCNT(Sun)HNO3 network surfaces were recorded using the Thermo Scientific K-Alpha System TFA XPS (Physical Electronics Instrument, Chanhassen, MN, USA) equipped with a micro-focused, monochromatic A1 Ka X-ray source (1486.6 eV). An X-ray beam of 400 μm size was used at 6 mA x 12 kV [21]. The spectra were acquired in the constant analyzer energy mode with pass energy of 200 eV for the survey. Narrow regions were collected using the snapshot acquisition mode (150 eV pass energy) enabling rapid collection of data (5 s per region). The narrow region data were as post-processed using the Jansson's algorithm to remove the analyzer point spread function, which resulted in an improved resolution of the spectra for the peak deconvolution [22]. The concentration of elements was determined from survey spectra by MultiPak v7. The Hall coefficient, the resistivity, and the conductivity of the samples as well as the charge mobility and the charge carrier concentration were measured by means of the HCS 1 apparatus (Linseis Messgeräte GmbH, Selb, Germany) equipped with static 0.7 T field permanent magnets for bipolar measurement. The disc form samples had diameter 20 mm and thickness 2.65 mm. The sample current was set to 4 mA. The thermoelectric power measurement was carried out for all the samples using the set-up illustrated in Figure 1. The schematic diagram shows that the circular composite sample (diameter 20 mm, thickness 2 mm) was placed between two copper electrodes. The ends of each of the Cu electrodes were immersed in thermostatic silicone oil baths set at different temperatures. The temperature at the copper/composite interfaces was measured by a Pt100 temperature sensor. The arising thermoelectric current was measured by the Keithley 2000 Digital Multimeter (Tektronix, Inc. Beaverton, OR, USA).

Characterization of Fillers and Composites
The detailed view of individual pristine MWCNT(Bayer)s, their clusters as well as clusters of the oxidized MWCNT(Bayer)KMnO 4 as obtained by the TEM are shown in Figure 2. The wall of the MWCNT(Bayer) consisted of about 15 rolled layers of graphene. The nanotube outer and inner diameter was about 20 nm and 4-10 nm, respectively. There were also defects obstructing nanotube interiors, which were commonly seen in the MWCNT structures. When a cluster of pristine MWCNT(Sun)s was compared with a cluster of wet oxidized nanotubes, a difference in the tube length was visible. The oxidation of MWCNT(Sun)s by KMnO 4 caused shortening of the nanotubes, creation of defect sites, and opened ends. A small amount of amorphous carbon after the KMnO 4 oxidation process can be also expected [23], although another report showed that oxygenation by KMnO 4 in an acidic suspension provides nanotubes free of amorphous carbon [24].
Polymers 2020, 12, x FOR PEER REVIEW 5 of 14 Figure 1. A schematic illustration of the set up for the measurement of electric voltage generated between the hot and cold ends of the Cu/composite/Cu thermoelectric device as a response to a temperature difference.3. Results

Characterization of Fillers and Composites
The detailed view of individual pristine MWCNT(Bayer)s, their clusters as well as clusters of the oxidized MWCNT(Bayer)KMnO4 as obtained by the TEM are shown in Figure 2. The wall of the MWCNT(Bayer) consisted of about 15 rolled layers of graphene. The nanotube outer and inner diameter was about 20 nm and 4-10 nm, respectively. There were also defects obstructing nanotube interiors, which were commonly seen in the MWCNT structures. When a cluster of pristine MWCNT(Sun)s was compared with a cluster of wet oxidized nanotubes, a difference in the tube length was visible. The oxidation of MWCNT(Sun)s by KMnO4 caused shortening of the nanotubes, creation of defect sites, and opened ends. A small amount of amorphous carbon after the KMnO4 oxidation process can be also expected [23], although another report showed that oxygenation by KMnO4 in an acidic suspension provides nanotubes free of amorphous carbon [24]. The micrographs of the surfaces of filler layers-PANI and PPy particles as well as MWCNTs and CNFs-are shown in Figure 3. The layers were formed from the respective aqueous dispersion of the fillers on the surface of the interdigitated electrode by the drop method. The particles were globular with a diameter 0.5-1 μm and the PPy particles had narrower size distribution.
The networks of CNFs, MWCNT(Sun)s and MWCNT(Bayer)s resulted from the filtration of their dispersions through non-woven polyurethane membranes. The dispersion, which consisted of CNFs (0.8 mg), or the same amount of nanotubes, dispersed in 530 mL of water with 15.4 g of the surfactant (sodium dodecyl sulphate) and 8.5 mL of the co-surfactant (1-pentanol), was properly sonicated and the pH adjusted to 10 using an aqueous solution of NaOH. The filtrate layer was washed in situ several times with deionized water and finally with methanol. The micrographs of the surfaces of filler layers-PANI and PPy particles as well as MWCNTs and CNFs-are shown in Figure 3. The layers were formed from the respective aqueous dispersion of the fillers on the surface of the interdigitated electrode by the drop method. The particles were globular with a diameter 0.5-1 µm and the PPy particles had narrower size distribution.
The networks of CNFs, MWCNT(Sun)s and MWCNT(Bayer)s resulted from the filtration of their dispersions through non-woven polyurethane membranes. The dispersion, which consisted of CNFs (0.8 mg), or the same amount of nanotubes, dispersed in 530 mL of water with 15.4 g of the surfactant (sodium dodecyl sulphate) and 8.5 mL of the co-surfactant (1-pentanol), was properly sonicated and the pH adjusted to 10 using an aqueous solution of NaOH. The filtrate layer was washed in situ several times with deionized water and finally with methanol.
The cross-section of the composites presented in Figure 4 showed a uniform distribution of the MWCNT(Sun) and CNF filler in the EOC matrix, which together with the filler concentration 30% ensured the electrical and thermal conductivity of the composites. The cross-section of the composites presented in Figure 4 showed a uniform distribution of the MWCNT(Sun) and CNF filler in the EOC matrix, which together with the filler concentration 30% ensured the electrical and thermal conductivity of the composites.

XPS Data
The main binding energy peak (284.5 eV) in the XPS spectra of pristine MWCNT(Sun)s, Figure 5, was assigned to the C1s-sp2, while the other ones were assigned to the C-O (286.2 eV), C=O (287.1 eV), O-C=O (288.6-289 eV) and C1s-π-π* (291.1-291.5 eV). After the oxidation treatment, the intensities of the peaks corresponding to the oxidized carbon bonds increased as seen in Figure  5. FTIR data of the MWCNTs also confirmed the presence of the C-O, C=O and O-C=O functional groups on their surfaces. It is also stated in other studies that MWCNTs treated with (NH4)2S2O8, H2O2, or O3 have higher concentrations of carbonyl and hydroxyl functional groups, while more aggressive oxidants (HNO3, KMnO4) form higher fractional concentrations of carboxyl groups [25]. Acidic potassium permanganate (KMnO4) is a strong oxidizing agent and produces more surface acidic groups than nitric acid [26]. The increase number of oxygenated functional groups attached on MWCNT(Sun)KMnO4 surfaces significantly increases the contact resistance in the MWCNT junctions of the network structure [27].  The cross-section of the composites presented in Figure 4 showed a uniform distribution of the MWCNT(Sun) and CNF filler in the EOC matrix, which together with the filler concentration 30% ensured the electrical and thermal conductivity of the composites.

XPS Data
The main binding energy peak (284.5 eV) in the XPS spectra of pristine MWCNT(Sun)s, Figure 5, was assigned to the C1s-sp2, while the other ones were assigned to the C-O (286.2 eV), C=O (287.1 eV), O-C=O (288.6-289 eV) and C1s-π-π* (291.1-291.5 eV). After the oxidation treatment, the intensities of the peaks corresponding to the oxidized carbon bonds increased as seen in Figure  5. FTIR data of the MWCNTs also confirmed the presence of the C-O, C=O and O-C=O functional groups on their surfaces. It is also stated in other studies that MWCNTs treated with (NH4)2S2O8, H2O2, or O3 have higher concentrations of carbonyl and hydroxyl functional groups, while more aggressive oxidants (HNO3, KMnO4) form higher fractional concentrations of carboxyl groups [25]. Acidic potassium permanganate (KMnO4) is a strong oxidizing agent and produces more surface acidic groups than nitric acid [26]. The increase number of oxygenated functional groups attached on MWCNT(Sun)KMnO4 surfaces significantly increases the contact resistance in the MWCNT junctions of the network structure [27].

XPS Data
The main binding energy peak (284.5 eV) in the XPS spectra of pristine MWCNT(Sun)s, Figure 5, was assigned to the C1s-sp2, while the other ones were assigned to the C-O (286.2 eV), C=O (287.1 eV), O-C=O (288.6-289 eV) and C1s-π-π* (291.1-291.5 eV). After the oxidation treatment, the intensities of the peaks corresponding to the oxidized carbon bonds increased as seen in Figure 5. FTIR data of the MWCNTs also confirmed the presence of the C-O, C=O and O-C=O functional groups on their surfaces. It is also stated in other studies that MWCNTs treated with (NH 4 ) 2 S 2 O 8 , H 2 O 2 , or O 3 have higher concentrations of carbonyl and hydroxyl functional groups, while more aggressive oxidants (HNO 3 , KMnO 4 ) form higher fractional concentrations of carboxyl groups [25]. Acidic potassium permanganate (KMnO 4 ) is a strong oxidizing agent and produces more surface acidic groups than nitric acid [26]. The increase number of oxygenated functional groups attached on MWCNT(Sun)KMnO 4 surfaces significantly increases the contact resistance in the MWCNT junctions of the network structure [27].  Table 1 and Figure 6 present the frequencies of some of the functional groups in the FTIR spectra of MWCNT networks. FTIR spectra in Figure 6 from MWCNTs showed a broad peak about 3430 cm −1 which was characteristic of the O-H stretch of hydroxyl group. C-H stretching of the pristine MWCNT(Sun) sample was shifted to lower wavelengths for all oxidation treatments. A weak C=O peak at 1705 cm −1 , Figure 6, was observed in the pristine MWCNT(Sun) sample, which showed that there was a carbonyl or carboxylic group on its surface. The reason why the pristine MWCNT(Sun) sample had carbonyl and OH groups could be a partial oxidation of the surfaces of MWCNTs during the purification by the manufacturer [28]. A higher shift in the carbonyl stretching mode was seen for MWCNT(HNO3) than for MWCNT(KMnO4). The reason could be a C=O group or other groups that interacted with the C=O group. FTIR spectra in Figure 6 also indicated that there were probably no anhydride/lactone groups on the surfaces of MWCNTs since these groups are usually observed at around 1750 cm −1 or higher wavenumber [26,29,30]. The peak assigned to the quinone group at 1652 cm −1 in MWCNT(pristine) sample was usually shifted to a higher wavelength in oxidized MWCNTs. Coupling effects (i.e., both of inter-molecular and intra-molecular hydrogen bonding with hydroxyl groups) also might be responsible for the  Table 1 and Figure 6 present the frequencies of some of the functional groups in the FTIR spectra of MWCNT networks. FTIR spectra in Figure 6 from MWCNTs showed a broad peak about 3430 cm −1 which was characteristic of the O-H stretch of hydroxyl group. C-H stretching of the pristine MWCNT(Sun) sample was shifted to lower wavelengths for all oxidation treatments. A weak C=O peak at 1705 cm −1 , Figure 6, was observed in the pristine MWCNT(Sun) sample, which showed that there was a carbonyl or carboxylic group on its surface. The reason why the pristine MWCNT(Sun) sample had carbonyl and OH groups could be a partial oxidation of the surfaces of MWCNTs during the purification by the manufacturer [28]. A higher shift in the carbonyl stretching mode was seen for MWCNT(HNO 3 ) than for MWCNT(KMnO 4 ). The reason could be a C=O group or other groups that interacted with the C=O group. FTIR spectra in Figure 6 also indicated that there were probably no anhydride/lactone groups on the surfaces of MWCNTs since these groups are usually observed at around 1750 cm −1 or higher wavenumber [26,29,30]. The peak assigned to the quinone group at 1652 cm −1 in MWCNT(pristine) sample was usually shifted to a higher wavelength in oxidized MWCNTs. Coupling effects (i.e., both of inter-molecular and intra-molecular hydrogen bonding with hydroxyl groups) also might be responsible for the downshift in the C=O stretching mode, besides the production of surface-bound quinone groups with extended conjugation [31].

FTIR Measurements
The up-shift in the C=C stretching mode of MWCNTs was observed for both oxidized carbon nanotubes. The highest shift was observed for MWCNT(HNO 3 ) compared to MWCNT(KMnO 4 ). This treatment suggested a change in the structure of the MWCNTs [32].
The C-H (CH 2 /CH 3 ) bending at 1460 cm −1 and the peak at 1222 cm −1 for the MWCNT(pristine) sample were shifted to a higher wavelength for all oxidized MWCNTs. There was also observed a new band around 1050 cm −1 in the FTIR spectra of oxidized MWCNTs, which could be assigned to the alcoholic C-O stretching vibration [33]. Overall, the observed changes in the FTIR spectra of the oxidized MWCNTs confirmed the efficiency of the oxidizing process and the formation of the new oxygen-containing functional groups on the surfaces of the carbon nanotubes.
Polymers 2020, 12, x FOR PEER REVIEW 8 of 14 downshift in the C=O stretching mode, besides the production of surface-bound quinone groups with extended conjugation [31]. The up-shift in the C=C stretching mode of MWCNTs was observed for both oxidized carbon nanotubes. The highest shift was observed for MWCNT(HNO3) compared to MWCNT(KMnO4). This treatment suggested a change in the structure of the MWCNTs [32].
The C-H (CH2/CH3) bending at 1460 cm −1 and the peak at 1222 cm −1 for the MWCNT(pristine) sample were shifted to a higher wavelength for all oxidized MWCNTs. There was also observed a new band around 1050 cm −1 in the FTIR spectra of oxidized MWCNTs, which could be assigned to the alcoholic C-O stretching vibration [33]. Overall, the observed changes in the FTIR spectra of the oxidized MWCNTs confirmed the efficiency of the oxidizing process and the formation of the new oxygen-containing functional groups on the surfaces of the carbon nanotubes. FTIR analysis was also carried out to identify whether the prepared powder was indeed PANI emeraldine or PPy (Figure 7). For the FTIR absorption spectroscopy of the prepared PANI emeraldine, the vibration at 1578 cm -1 was attributed to the quinoid ring, while the vibration at 1489 cm -1 depicted the presence of a benzoid ring unit [34]. The peak at 1306 cm -1 is assigned to the C-N stretching of a secondary aromatic amine. Furthermore, the peak characteristic for the electrically conductive form of PANI emeraldine was observed at 1245 cm −1 . This peak was attributed to the stretching of the C-N +• polaron structure [35]. The peak at 1160 cm -1 corresponding to the vibrations associated with the C-H of N=Q=N (Q = quinoid ring) also appeared in the spectrum of PANI emeraldine. The band between 913-680 cm -1 with a maximum at 823 cm -1 was characteristic for an aromatic ring deformation and C-H bond vibrations out of the plane of the ring [36]. These results suggested that the synthesized particles were PANI in the emeraldine state. For the FTIR spectroscopy of the prepared PPy, all the characteristic peaks were in good agreement with the earlier investigations of the same product [37,38]. The vibration peak at 1704 cm -1 was assigned to the C=N bond. Peaks at 1550 and 1477 cm -1 attributed to the in-plane vibrations of the PPy ring, and the peaks at 1178, 1037 and 783 cm -1 attributed to the in-plane bending of the PPy ring were also observed. Finally, the characteristic peak at 900 cm -1 was assigned to the out-of-plane vibration of the Cβ-H group. FTIR analysis was also carried out to identify whether the prepared powder was indeed PANI emeraldine or PPy (Figure 7). For the FTIR absorption spectroscopy of the prepared PANI emeraldine, the vibration at 1578 cm −1 was attributed to the quinoid ring, while the vibration at 1489 cm −1 depicted the presence of a benzoid ring unit [34]. The peak at 1306 cm −1 is assigned to the C-N stretching of a secondary aromatic amine. Furthermore, the peak characteristic for the electrically conductive form of PANI emeraldine was observed at 1245 cm −1 . This peak was attributed to the stretching of the C-N +• polaron structure [35]. The peak at 1160 cm −1 corresponding to the vibrations associated with the C-H of N=Q=N (Q = quinoid ring) also appeared in the spectrum of PANI emeraldine. The band between 913-680 cm −1 with a maximum at 823 cm −1 was characteristic for an aromatic ring deformation and C-H bond vibrations out of the plane of the ring [36]. These results suggested that the synthesized particles were PANI in the emeraldine state. For the FTIR spectroscopy of the prepared PPy, all the characteristic peaks were in good agreement with the earlier investigations of the same product [37,38]. The vibration peak at 1704 cm −1 was assigned to the C=N bond. Peaks at 1550 and 1477 cm −1 attributed to the in-plane vibrations of the PPy ring, and the peaks at 1178, 1037 and 783 cm −1 attributed to the in-plane bending of the PPy ring were also observed. Finally, the characteristic peak at 900 cm −1 was assigned to the out-of-plane vibration of the C β -H group.

Thermoelectric Power Measurement
Values of induced voltage in response to a temperature difference across the measured samples of the investigated thermoelectric composites are presented in Figure 8. The slope of the linear dependence of the resulting voltage VTEV on the temperature difference ΔT defined the thermoelectric power (Seebeck coefficient): which evaluated a potential thermoelectric performance. The corresponding thermoelectric power values are summarized in Figure 9. The thermoelectric power of MWCNT/EOC composites was substantially enhanced by the nanotube oxygenation. When compared to the composite with pristine nanotubes (MWCNT(Bayer)pristine/EOC), the respective composite with oxygenated nanotubes achieved a 150% increase of thermoelectric power. The X-ray photoelectron spectroscopy (XPS) was performed on MWCNTs to ascertain the functional groups attached onto the nanotube surfaces. The oxygen contents (%) of the pristine and oxidized MWCNT samples, as calculated from the XPS spectra, are shown in Figure 9. The comparison of results indicated that all MWCNTs have C-O, C=O and O-C=O groups on their surfaces and that MWCNT(HNO3)s have the maximum percentage of all the oxygen-containing groups. Moreover, the more oxygenated functional groups at the surface of the embedded MWCNTs, the higher the generated electric voltage and the thermoelectric power of the EOC composite as follows from the results in Figure 9. According to the published results on carbon nanotube/thermoplastic polymer composites, positive thermoelectric powers have always been determined [39].

Thermoelectric Power Measurement
Values of induced voltage in response to a temperature difference across the measured samples of the investigated thermoelectric composites are presented in Figure 8. The slope of the linear dependence of the resulting voltage V TEV on the temperature difference ∆T defined the thermoelectric power (Seebeck coefficient): which evaluated a potential thermoelectric performance. The corresponding thermoelectric power values are summarized in Figure 9. The thermoelectric power of MWCNT/EOC composites was substantially enhanced by the nanotube oxygenation. When compared to the composite with pristine nanotubes (MWCNT(Bayer)pristine/EOC), the respective composite with oxygenated nanotubes achieved a 150% increase of thermoelectric power. The X-ray photoelectron spectroscopy (XPS) was performed on MWCNTs to ascertain the functional groups attached onto the nanotube surfaces. The oxygen contents (%) of the pristine and oxidized MWCNT samples, as calculated from the XPS spectra, are shown in Figure 9. The comparison of results indicated that all MWCNTs have C-O, C=O and O-C=O groups on their surfaces and that MWCNT(HNO 3 )s have the maximum percentage of all the oxygen-containing groups. Moreover, the more oxygenated functional groups at the surface of the embedded MWCNTs, the higher the generated electric voltage and the thermoelectric power of the EOC composite as follows from the results in Figure 9. According to the published results on carbon nanotube/thermoplastic polymer composites, positive thermoelectric powers have always been determined [39].

Self-Powered Signaling Sensor of Temperature Change
To demonstrate a possible use of the conductive MWCNT/EOC composites as thermoelectrical materials, a self-powered signaling sensor of temperature change was assembled. The sensor consisted of three conductive strips (thermoelements): one MWCNT(Sun)pristine/EOC and two MWCNT(Sun)KMnO4/EOC composites, Figure 10. The sticky strips were stuck on a PET foil so that the ends overlapped at points A and B. The temperature signal was induced by a finger touch at point A or B, which in turn heated the connection of the strips. The generated electricity was monitored by the Multiplex datalogger 34980A. Even a temperature gradient induced by a mere short touch of a finger sufficed to elicit a detectable signal. The illustrative record of repeated finger touches is presented in Figure 11. Heat transfer from the finger to the hot junction of the sensor modeled a technological situation when such a sensor could monitor for example changes in the generation of waste heat, changes of technological temperature, warming of packaged grocery products, etc.

Self-Powered Signaling Sensor of Temperature Change
To demonstrate a possible use of the conductive MWCNT/EOC composites as thermoelectrical materials, a self-powered signaling sensor of temperature change was assembled. The sensor consisted of three conductive strips (thermoelements): one MWCNT(Sun)pristine/EOC and two MWCNT(Sun)KMnO4/EOC composites, Figure 10. The sticky strips were stuck on a PET foil so that the ends overlapped at points A and B. The temperature signal was induced by a finger touch at point A or B, which in turn heated the connection of the strips. The generated electricity was monitored by the Multiplex datalogger 34980A. Even a temperature gradient induced by a mere short touch of a finger sufficed to elicit a detectable signal. The illustrative record of repeated finger touches is presented in Figure 11. Heat transfer from the finger to the hot junction of the sensor modeled a technological situation when such a sensor could monitor for example changes in the generation of waste heat, changes of technological temperature, warming of packaged grocery products, etc.

Self-Powered Signaling Sensor of Temperature Change
To demonstrate a possible use of the conductive MWCNT/EOC composites as thermoelectrical materials, a self-powered signaling sensor of temperature change was assembled. The sensor consisted of three conductive strips (thermoelements): one MWCNT(Sun)pristine/EOC and two MWCNT(Sun)KMnO 4 /EOC composites, Figure 10. The sticky strips were stuck on a PET foil so that the ends overlapped at points A and B. The temperature signal was induced by a finger touch at point A or B, which in turn heated the connection of the strips. The generated electricity was monitored by the Multiplex datalogger 34980A. Even a temperature gradient induced by a mere short touch of a finger sufficed to elicit a detectable signal. The illustrative record of repeated finger touches is presented in Figure 11. Heat transfer from the finger to the hot junction of the sensor modeled a technological situation when such a sensor could monitor for example changes in the generation of waste heat, changes of technological temperature, warming of packaged grocery products, etc. Polymers 2020, 12, x FOR PEER REVIEW 11 of 14

Self-Powered Vapor Sensor
The induced voltage in the self-powered sensor ( Figure 10) was changed not only by heating, but also by ambient organic vapors, Figure 12. As in our paper [14], the self-powered sensor was exposed to saturated organic solvent vapors. In particular, the sensor was placed in a glass bell, which enclosed a layer of liquid organic solvent and its respective saturated vapor. After a chosen time, the sensor was removed from the bell and the effect of vapor desorption on induced voltage was assessed. Such measurements were conducted in saturated vapors at atmospheric pressure, temperature 22 °C, and relative humidity 60%. The temperature gradient in the sensor was induced by means of the resistive heating of the MWCNT network/epoxy composite unit placed under the thermoelement junction. The heating DC voltage was 13 V and the current 0.09 A. The temperature of the hot end was 40.6 °C and of the cold end ambient room temperature.
The variations of the voltage of the self-powered sensor were caused by the changes of the electric resistance of the forming EOC composites through the chemical signals during the respective adsorption/desorption cycles as is illustrated in Figure 9. When the sensor was subjected to heptane, a larger increase of relative resistance resulted in a larger decrease of induced voltage and vice versa. Thus, the sensor output in terms of variation of voltage indicated a good response to the vapor occurrence. Moreover, considering that the sensor did not require a power supply, but it selfproduced thermoelectricity from a heat source, which could be waste heat of industrial processes, solar energy, or body heat, etc., then the self-powered vapor sensor appears to be an advantageous

Self-Powered Vapor Sensor
The induced voltage in the self-powered sensor ( Figure 10) was changed not only by heating, but also by ambient organic vapors, Figure 12. As in our paper [14], the self-powered sensor was exposed to saturated organic solvent vapors. In particular, the sensor was placed in a glass bell, which enclosed a layer of liquid organic solvent and its respective saturated vapor. After a chosen time, the sensor was removed from the bell and the effect of vapor desorption on induced voltage was assessed. Such measurements were conducted in saturated vapors at atmospheric pressure, temperature 22 °C, and relative humidity 60%. The temperature gradient in the sensor was induced by means of the resistive heating of the MWCNT network/epoxy composite unit placed under the thermoelement junction. The heating DC voltage was 13 V and the current 0.09 A. The temperature of the hot end was 40.6 °C and of the cold end ambient room temperature.
The variations of the voltage of the self-powered sensor were caused by the changes of the electric resistance of the forming EOC composites through the chemical signals during the respective adsorption/desorption cycles as is illustrated in Figure 9. When the sensor was subjected to heptane, a larger increase of relative resistance resulted in a larger decrease of induced voltage and vice versa. Thus, the sensor output in terms of variation of voltage indicated a good response to the vapor occurrence. Moreover, considering that the sensor did not require a power supply, but it selfproduced thermoelectricity from a heat source, which could be waste heat of industrial processes, solar energy, or body heat, etc., then the self-powered vapor sensor appears to be an advantageous

Self-Powered Vapor Sensor
The induced voltage in the self-powered sensor ( Figure 10) was changed not only by heating, but also by ambient organic vapors, Figure 12. As in our paper [14], the self-powered sensor was exposed to saturated organic solvent vapors. In particular, the sensor was placed in a glass bell, which enclosed a layer of liquid organic solvent and its respective saturated vapor. After a chosen time, the sensor was removed from the bell and the effect of vapor desorption on induced voltage was assessed. Such measurements were conducted in saturated vapors at atmospheric pressure, temperature 22 • C, and relative humidity 60%. The temperature gradient in the sensor was induced by means of the resistive heating of the MWCNT network/epoxy composite unit placed under the thermoelement junction. The heating DC voltage was 13 V and the current 0.09 A. The temperature of the hot end was 40.6 • C and of the cold end ambient room temperature.
The variations of the voltage of the self-powered sensor were caused by the changes of the electric resistance of the forming EOC composites through the chemical signals during the respective adsorption/desorption cycles as is illustrated in Figure 9. When the sensor was subjected to heptane, a larger increase of relative resistance resulted in a larger decrease of induced voltage and vice versa. Thus, the sensor output in terms of variation of voltage indicated a good response to the vapor occurrence. Moreover, considering that the sensor did not require a power supply, but it self-produced thermoelectricity from a heat source, which could be waste heat of industrial processes, solar energy, or body heat, etc., then the self-powered vapor sensor appears to be an advantageous alternative to powered vapor sensors. The illustrative thermoelement resistance changes plotted in Figure 10 were quantified by the relative resistance change, where R a represented the initial thermoelement resistance in the air, R g the resistance when the thermoelement was exposed to a vapor, and ∆R denoted the measured resistance change.
Polymers 2020, 12, x FOR PEER REVIEW 12 of 14 alternative to powered vapor sensors. The illustrative thermoelement resistance changes plotted in Figure 10 were quantified by the relative resistance change, where Ra represented the initial thermoelement resistance in the air, Rg the resistance when the thermoelement was exposed to a vapor, and ΔR denoted the measured resistance change.

Discussion
This study presents hybrid thermoelectric composites consisting of organic EOC matrices and embedded conductive inorganic pristine and functionalized multiwalled carbon nanotubes, carbon nanofibers or conductive organic PANI and PPy particles. All these composites induced voltage in response to a temperature difference across the respective samples. This response was affected in the case of the composites with inorganic MWCNT nanostructures by the amount of oxygen containing functional groups attached to their surfaces. The more oxygenated were the functional groups at the surfaces of the embedded MWCNTs, the higher was the thermoelectric power of that EOC composite.
According to the published results about carbon nanotube/thermoplastic polymer composites, positive thermoelectric powers have always been determined indicating p-type composites, in which holes are dominant current carriers. The composites with embedded organic PANI a PPy particles induce compensating negative voltage. According to [38], the composites exhibit n-type conductivity specified by dominating electrons as charge carriers.
The chosen combination of the two p-type EOC composites in the experimental thermoelectric microgenerator, which generated sufficient electric current to make the sensors self-powered, was one of many other possible combinations. Other combinations such as p-type together with n-type thermoelements or two n-type EOC composites could have been chosen as well. An illustration of the practical use of the EOC composites as a thermoelectric generator was provided by the selfpowered signaling sensor of temperature change and the self-powered vapor sensor. The voltage induced in the sensors, which was in the range of microvolts, did not approach by far the magnitude of induced voltage in the classical metal thermocouples, yet it was sufficient to power the novel sensor device, which converted chemical signals to electric ones. In the self-powered vapor sensor, the resistance varied according to the ambient chemical signals and changed the voltage induced by a source of (waste) heat. The EOC thermoelectric composites thus offer a novel and unique set of properties, which are not readily available in any other material.

Discussion and Conclusions
This study presents hybrid thermoelectric composites consisting of organic EOC matrices and embedded conductive inorganic pristine and functionalized multiwalled carbon nanotubes, carbon nanofibers or conductive organic PANI and PPy particles. All these composites induced voltage in response to a temperature difference across the respective samples. This response was affected in the case of the composites with inorganic MWCNT nanostructures by the amount of oxygen containing functional groups attached to their surfaces. The more oxygenated were the functional groups at the surfaces of the embedded MWCNTs, the higher was the thermoelectric power of that EOC composite.
According to the published results about carbon nanotube/thermoplastic polymer composites, positive thermoelectric powers have always been determined indicating p-type composites, in which holes are dominant current carriers. The composites with embedded organic PANI a PPy particles induce compensating negative voltage. According to [38], the composites exhibit n-type conductivity specified by dominating electrons as charge carriers.
The chosen combination of the two p-type EOC composites in the experimental thermoelectric microgenerator, which generated sufficient electric current to make the sensors self-powered, was one of many other possible combinations. Other combinations such as p-type together with n-type thermoelements or two n-type EOC composites could have been chosen as well. An illustration of the practical use of the EOC composites as a thermoelectric generator was provided by the self-powered signaling sensor of temperature change and the self-powered vapor sensor. The voltage induced in the sensors, which was in the range of microvolts, did not approach by far the magnitude of induced voltage in the classical metal thermocouples, yet it was sufficient to power the novel sensor device, which converted chemical signals to electric ones. In the self-powered vapor sensor, the resistance varied according to the ambient chemical signals and changed the voltage induced by a source of (waste) heat. The EOC thermoelectric composites thus offer a novel and unique set of properties, which are not readily available in any other material.