Comparative Studies on Thermal, Mechanical, and Flame Retardant Properties of PBT Nanocomposites via Different Oxidation State Phosphorus-Containing Agents Modified Amino-CNTs

High-performance poly(1,4-butylene terephthalate) (PBT) nanocomposites have been developed via the consideration of phosphorus-containing agents and amino-carbon nanotube (A-CNT). One-pot functionalization method has been adopted to prepare functionalized CNTs via the reaction between A-CNT and different oxidation state phosphorus-containing agents, including chlorodiphenylphosphine (DPP-Cl), diphenylphosphinic chloride (DPP(O)-Cl), and diphenyl phosphoryl chloride (DPP(O3)-Cl). These functionalized CNTs, DPP(Ox)-A-CNTs (x = 0, 1, 3), were, respectively, mixed with PBT to obtain the CNT-based polymer nanocomposites through a melt blending method. Scanning electron microscope observations demonstrated that DPP(Ox)-A-CNT nanoadditives were homogeneously distributed within PBT matrix compared to A-CNT. The incorporation of DPP(Ox)-A-CNT improved the thermal stability of PBT. Moreover, PBT/DPP(O3)-A-CNT showed the highest crystallization temperature and tensile strength, due to the superior dispersion and interfacial interactions between DPP(O3)-A-CNT and PBT. PBT/DPP(O)-A-CNT exhibited the best flame retardancy resulting from the excellent carbonization effect. The radicals generated from decomposed polymer were effectively trapped by DPP(O)-A-CNT, leading to the reduction of heat release rate, smoke production rate, carbon dioxide and carbon monoxide release during cone calorimeter tests.


Introduction
Owing to the superior dimensional stability and heat resistance, poly(1,4-butylene terephthalate) (PBT) has been extensively applied in the electrical and electronics industries [1,2]. Nevertheless, PBT resins without flame retarded modification are easily ignitable either by an electric spark or

Preparation of Functionalized CNTs
Functionalized CNTs was prepared based on the method reported in our study [19]. Typically

Preparation of Functionalized CNTs
Functionalized CNTs was prepared based on the method reported in our study [19]. Typically, A-CNT (100 mg), DMF (50 mL), and suitable amount of TEA (catalyst) were charged into a three-necked flask. DPP-Cl (75 mg), DPP(O)-Cl (80 mg) and DPP(O3)-Cl (90 mg) were then dissolved in DMF (10 mL), and added dropwise into the suspension of A-CNT and TEA. The mixture was stirred in an ice-water bath for 1 h under dry nitrogen condition, subsequently heated up to 80 °C and kept at this temperature for 24 h. The mixture was filtered, washed with DMF and dried in a vacuum oven at 80 °C to a constant weight. These functionalized CNTs are denoted as DPP-A-CNT, DPP(O)-A-CNT and DPP(O3)-A-CNT. The preparation route of the functionalized carbon nanotubes is depicted in Figure 1.

Preparation of PBT/Functionalized CNTs Nanocomposites
Before melt processing, PBT, A-CNT, DPP-A-CNT, DPP(O)-A-CNT and DPP(O3)-A-CNT were continuously dried at 80 °C for 12 h. In a typical experiment, 0.5 g A-CNT was blended with 49.5 g PBT to prepare the nanocomposites using a torque rheometer (RTOI-55/20, POTOP Co. Ltd., Guangzhou, China) at 235 °C at a constant rotation speed of 100 rpm for 10 min. These mixtures were then molded through a hot press or microinjection molding machine at 240-250 °C to obtain the samples with different sizes for further measurements. Other samples were fabricated according

Preparation of PBT/Functionalized CNTs Nanocomposites
Before melt processing, PBT, A-CNT, DPP-A-CNT, DPP(O)-A-CNT and DPP(O 3 )-A-CNT were continuously dried at 80 • C for 12 h. In a typical experiment, 0.5 g A-CNT was blended with 49.5 g PBT to prepare the nanocomposites using a torque rheometer (RTOI-55/20, POTOP Co. Ltd., Guangzhou, China) at 235 • C at a constant rotation speed of 100 rpm for 10 min. These mixtures were then molded through a hot press or microinjection molding machine at 240-250 • C to obtain the samples with different sizes for further measurements. Other samples were fabricated according to the same procedure. These nanocomposites were designated as PBT/A-CNT, PBT/DPP-A-CNT, PBT/DPP(O)-A-CNT and PBT/DPP(O 3 )-A-CNT at the same loading of nanoadditives (1 wt %).

Instruments and Measurements
Fourier transform infrared spectroscopy (FTIR) spectra were recorded on a FTIR spectrophotometer (Nicolet 6700, Thermo Fisher Scientific Inc., Waltham, MA, USA). Raman spectra were obtained from a DXR Smart Raman Spectrometer (Thermo Fisher Scientific Inc.), in the wavenumber range of 500-2000 cm −1 . Thermal decomposition behaviors of samples under nitrogen atmosphere were investigated by a Q5000 IR TGA (TA Instruments, New Castle, DE, USA) from 25 to 700 • C at a heating rate of 20 • C/min. The dispersion and morphology of nano-additives were observed via scanning electron microscope (SEM) on a Hitachi SU8010 SEM (Tokyo, Japan) with the acceleration voltage of 10 kV. The samples were obtained by immersing PBT nanocomposites into liquid nitrogen, and a conductive gold layer was coated on the fractured surface prior to SEM observations. Thermal behaviors were studied by a Perkin Elmer Diamond DSC under nitrogen condition. Samples were heated from 50 to 300 • C at a heating rate of 10 • C/min and kept at 300 • C for 5 min to eliminate any thermal history, and they were then cooled to 50 • C at a rate of 10 • C/min. Finally, these samples were maintained at 50 • C for 5 min and heated to 300 • C. They were held at 300 • C for 5 min, and subsequently cooled to 50 • C at 10 • C/min. Tensile properties were evaluated by a WD-20D universal testing machine according to the standard ASTM D-638. The width and thickness of specimens were 4.0 ± 0.1 mm and 2.0 ± 0.1 mm, respectively. The crosshead speed was set as 20 mm/min. Five runs for each sample were measured, and the average value was recorded.
Flame retardant properties were investigated on a FTT cone calorimeter (FTT, Derby, UK) based on the ISO 5660-1 standard. The sample size was 100 mm × 100 mm × 3.0 mm. All samples were wrapped by a layer of aluminum foil, and they were then irradiated under a heat flux of 35 kW/m 2 . Residues were analyzed by SEM coupled with energy dispersive X-ray (EDX). The surface elements were attained from EDX on an EMAX energy spectroscopy (HORIBA, Ltd., Kyoto, Japan).  Figure 2b. Two main absorption peaks at 1340 and 1571 cm −1 are attributed to the D-band and G-band, respectively [21]. The G-band is attributed to the first-order scattering of the sp 2 carbon atoms of CNTs, while the D-band corresponds to the disorder-induced or sp 3 carbon atoms of CNTs [22][23][24]. A new band D' at higher wavenumber close to G-band corresponds to the functionalized CNTs [24,25].       Figure 4 shows the schematic pattern of the hydrogen-bond interaction between DPP(Ox)-A-CNT and PBT. Because the electronegativity of nitrogen (N) atom is higher than that of phosphorus (P) atom, the electronic density around N atom in DPP-A-CNT increases, leading to the reduction of electropositivity of hydrogen (H) atom in secondary amine. Therefore, the hydrogen-bond interaction between DPP-A-CNT and PBT is weak. As the quantity of oxygen atom increases, the electronic density in N atom decreases in DPP(O)-A-CNT and DPP(O3)-A-CNT, resulting in the enhancement of electropositivity of H atom in secondary amine. As a result, the hydrogen-bond interaction between DPP(O3)-A-CNT and PBT is highest. Owing to the strong hydrogen-bond interaction, the PBT macromolecular chains can firmly envelope the DPP(O)-A-CNT and DPP(O3)-A-CNT nanoparticles. Because of the poor dispersion characteristic of A-CNT in PBT, the material properties of PBT/A-CNT will not be evaluated in the current study.

Thermal Properties
The melt and non-isothermal crystallization behaviors of neat PBT and PBT/DPP(Ox)-A-CNT nanocomposites were characterized by DSC. Figure 5 depicts the thermal behavior curves recorded for all samples at the heating and cooling scan of 10 °C/min. The thermal parameters obtained from the thermograms are summarized in Table 1. In the heating scan, the influence of DPP(Ox)-A-CNT on melting temperature (Tm) of PBT can be seen to be negligible. The multiple melt behavior observed for neat PBT (Figure 5a) is caused by the fusion of a certain amount of original crystals, followed by the recrystallization and final melting of more perfect crystals, partly formed during primary crystallization and through the recrystallization process occurring during the heating scan [30][31][32]. However, the overlapping of two peaks of PBT mix forms a new peak with the introduction of DPP(Ox)-A-CNT nanoparticles. This indicates that the presence of DPP(Ox)-A-CNT nanoparticles improve the crystallization process, leading to the formation of more perfect and stable crystals, which becomes melt at higher temperature [32][33][34]. Therefore, the first melting peak shifts to high temperature region, resulting in the reduction of multiple melt behavior.
As shown in Figure 5b, all crystallization temperatures (Tc) of the PBT nanocomposites are significantly improved with the addition of DPP(Ox)-A-CNT nanoparticles. In comparison with that of neat PBT, the Tc value of PBT/DPP(O3)-A-CNT evaluated by DSC is increased by 34 °C, higher than that of PBT/carboxylated-CNT reported in our previous work [9,35]. Functionalized CNTs show more significant heterogeneous nucleation effect on the crystallization process of PBT. On the other hand, the Tc value of the three nanocomposites increases with the increasing quantity of oxygen atom, due to the different hydrogen-bond interaction: The strong hydrogen-bond interaction promotes the PBT macromolecular chains to envelope the CNT nanoparticles thereby resulting in the formation of non-covalent crosslinking points, which can strongly restrict the segmental motion of PBT chains [27][28][29], thus accelerating its crystallization process. When the crystallization begins at higher temperature, more perfect and stable crystals will be formed, which is beneficial to the improvement of mechanical strength as well as the reduction of multiple melt behavior.

Thermal Properties
The melt and non-isothermal crystallization behaviors of neat PBT and PBT/DPP(O x )-A-CNT nanocomposites were characterized by DSC. Figure 5 depicts the thermal behavior curves recorded for all samples at the heating and cooling scan of 10 • C/min. The thermal parameters obtained from the thermograms are summarized in Table 1. In the heating scan, the influence of DPP(O x )-A-CNT on melting temperature (T m ) of PBT can be seen to be negligible. The multiple melt behavior observed for neat PBT (Figure 5a) is caused by the fusion of a certain amount of original crystals, followed by the recrystallization and final melting of more perfect crystals, partly formed during primary crystallization and through the recrystallization process occurring during the heating scan [30][31][32]. However, the overlapping of two peaks of PBT mix forms a new peak with the introduction of DPP(O x )-A-CNT nanoparticles. This indicates that the presence of DPP(O x )-A-CNT nanoparticles improve the crystallization process, leading to the formation of more perfect and stable crystals, which becomes melt at higher temperature [32][33][34]. Therefore, the first melting peak shifts to high temperature region, resulting in the reduction of multiple melt behavior.
As shown in Figure 5b, all crystallization temperatures (T c ) of the PBT nanocomposites are significantly improved with the addition of DPP(O x )-A-CNT nanoparticles. In comparison with that of neat PBT, the T c value of PBT/DPP(O 3 )-A-CNT evaluated by DSC is increased by 34 • C, higher than that of PBT/carboxylated-CNT reported in our previous work [9,35]. Functionalized CNTs show more significant heterogeneous nucleation effect on the crystallization process of PBT. On the other hand, the T c value of the three nanocomposites increases with the increasing quantity of oxygen atom, due to the different hydrogen-bond interaction: (PBT/DPP(O 3 )-A-CNT > PBT/DPP(O)-A-CNT > PBT/DPP-A-CNT). The strong hydrogen-bond interaction promotes the PBT macromolecular chains to envelope the CNT nanoparticles thereby resulting in the formation of non-covalent crosslinking points, which can strongly restrict the segmental motion of PBT chains [27][28][29], thus accelerating its crystallization process. When the crystallization begins at higher temperature, more perfect and stable crystals will be formed, which is beneficial to the improvement of mechanical strength as well as the reduction of multiple melt behavior.  Table 1. Calorimetric data of the melting and non-isothermal crystallization processes for each sample (Tm, melting peak temperature of 2nd heating; Tc, crystallization peak temperature of 2nd cooling).

Sample No.
Tm

Thermal Decomposition Behaviors
The thermal decomposition behaviors of neat PBT and its nanocomposites under nitrogen condition are shown in Figure 6, and the corresponding data are summarized in Table 2. In Figure 6a, it is seen that the thermal decomposition behavior of each sample exhibits a one-stage degradation process. Neat PBT leaves only 2.7 wt % char residue at 700 °C. In the thermal decomposition process, the main volatiles, composed of butadiene, carbon dioxide, tetrahydrofuran, benzoic acid and ester derivatives, are released, leaving small solid residues with acidic and anhydride structures [8,9]. The addition of DPP(Ox)-A-CNT nanoparticles results in the improvement of thermal stability and char yields of PBT, and PBT/DPP(O)-A-CNT is seen to be more efficient than the other two functionalized CNTs. For example, the T−5% value is increased from 367 °C for neat PBT to 388 °C for PBT/DPP(O)-A-CNT, and the Tmax value is improved from 408 to 419 °C. PBT/DPP(O3)-A-CNT shows inferior thermal stability due to the unstable organic phosphate grafted on the surface of DPP(O3)-A-CNT, while the thermal stability is still higher than that of neat PBT. The results show that the presence of DPP(Ox)-A-CNT nanoparticles can significantly enhance the thermal stability of the PBT nanocomposites, which aligns with findings in our previous studies [9,35]. The improved thermal stability is attributed to the excellent thermal conductivity and homogeneous dispersion of functionalized CNTs [36,37].
PBT/DPP(O)-A-CNT has a high residual weight (6.5 wt %), more than that of PBT/DPP-A-CNT (5.2 wt %) and PBT/DPP(O3)-A-CNT (4.5 wt %), as shown in Figure 6a and Table 2. DPP(O3)-A-CNT with high oxidation state phosphorus-containing groups shows weak carbonization effect in PBT during the thermal decomposition. The phosphorus-containing compound with low oxidation state may preferably promote the cross-linking reaction between the organophosphorus-based pyrolysis products and ester derivatives decomposed from PBT chains. From the SEM observations of PBT/DPP(Ox)-A-CNT fracture surfaces, it is known that the PBT macromolecular chains compactly envelope the DPP(O)-A-CNT nanoparticles due to the strong hydrogen-bond interaction, which is beneficial for cross-linking reaction. In combination with the barrier effect of CNTs, more organic phosphate-based derivatives are formed in the condensed phase, enhancing the strength and thermal stability of the char layer. Figure 6b reveals that the addition of DPP(Ox)-A-CNT  Table 1. Calorimetric data of the melting and non-isothermal crystallization processes for each sample (T m , melting peak temperature of 2nd heating; T c , crystallization peak temperature of 2nd cooling).

Sample No.
T

Thermal Decomposition Behaviors
The thermal decomposition behaviors of neat PBT and its nanocomposites under nitrogen condition are shown in Figure 6, and the corresponding data are summarized in Table 2. In Figure 6a, it is seen that the thermal decomposition behavior of each sample exhibits a one-stage degradation process. Neat PBT leaves only 2.7 wt % char residue at 700 • C. In the thermal decomposition process, the main volatiles, composed of butadiene, carbon dioxide, tetrahydrofuran, benzoic acid and ester derivatives, are released, leaving small solid residues with acidic and anhydride structures [8,9]. The addition of DPP(O x )-A-CNT nanoparticles results in the improvement of thermal stability and char yields of PBT, and PBT/DPP(O)-A-CNT is seen to be more efficient than the other two functionalized CNTs. For example, the T −5% value is increased from 367 • C for neat PBT to 388 • C for PBT/DPP(O)-A-CNT, and the T max value is improved from 408 to 419 • C. PBT/DPP(O 3 )-A-CNT shows inferior thermal stability due to the unstable organic phosphate grafted on the surface of DPP(O 3 )-A-CNT, while the thermal stability is still higher than that of neat PBT. The results show that the presence of DPP(O x )-A-CNT nanoparticles can significantly enhance the thermal stability of the PBT nanocomposites, which aligns with findings in our previous studies [9,35]. The improved thermal stability is attributed to the excellent thermal conductivity and homogeneous dispersion of functionalized CNTs [36,37].
PBT/DPP(O)-A-CNT has a high residual weight (6.5 wt %), more than that of PBT/DPP-A-CNT (5.2 wt %) and PBT/DPP(O 3 )-A-CNT (4.5 wt %), as shown in Figure 6a and Table 2. DPP(O 3 )-A-CNT with high oxidation state phosphorus-containing groups shows weak carbonization effect in PBT during the thermal decomposition. The phosphorus-containing compound with low oxidation state may preferably promote the cross-linking reaction between the organophosphorus-based pyrolysis products and ester derivatives decomposed from PBT chains. From the SEM observations of PBT/DPP(O x )-A-CNT fracture surfaces, it is known that the PBT macromolecular chains compactly envelope the DPP(O)-A-CNT nanoparticles due to the strong hydrogen-bond interaction, which is beneficial for cross-linking reaction. In combination with the barrier effect of CNTs, more organic phosphate-based derivatives are formed in the condensed phase, enhancing the strength and thermal stability of the char layer. Figure 6b reveals that the addition of DPP(O x )-A-CNT nanoparticles improve the T max value of PBT, but there is no substantial influence on the maximum mass loss rates (MMLR). These results demonstrate that the introduction of DPP(O x )-A-CNT can improve the thermal stability and char yields, but does not alter the decomposition pathway of PBT. nanoparticles improve the Tmax value of PBT, but there is no substantial influence on the maximum mass loss rates (MMLR). These results demonstrate that the introduction of DPP(Ox)-A-CNT can improve the thermal stability and char yields, but does not alter the decomposition pathway of PBT.

Tensile Properties
The tensile properties of neat PBT and its nanocomposites are illustrated in Figure 7. The corresponding data are summarized in Table 3. From the stress-strain curves, it is seen that the introduction of the three DPP(Ox)-A-CNT nanoparticles into PBT improves the tensile strength due to nano-reinforcing effect of CNTs with ultra-high aspect surface area [27,38] (Figure 4), is favorable to load transfer from the polymer matrix to the CNTs. As shown in Figure 7 and Table 3

, the elongation at break for the PBT nanocomposites decreases with the introduction of DPP(Ox)-A-CNT nanoparticles (PBT/DPP-A-CNT > PBT/DPP(O)-A-CNT > PBT/DPP(O3)-A-CNT), while the tensile strength values
show the opposite trend. PBT nanocomposites become brittle in comparison with neat PBT, because of the increased stiffness of the PBT nanocomposites and the micro-voids formed around the nanotubes during the tensile measurement [27,39]. The elongation at break for PBT nanocomposites is in relation with the interfacial interaction between CNTs and PBT matrix. The stronger interfacial adhesion, the more difficult segmental stretching and motion of PBT chains, the lower elongation at break [39].

Tensile Properties
The tensile properties of neat PBT and its nanocomposites are illustrated in Figure 7. The corresponding data are summarized in Table 3. From the stress-strain curves, it is seen that the introduction of the three DPP(O x )-A-CNT nanoparticles into PBT improves the tensile strength due to nano-reinforcing effect of CNTs with ultra-high aspect surface area [27,38] (Figure 4), is favorable to load transfer from the polymer matrix to the CNTs. As shown in Figure 7 and Table 3 [27,39]. The elongation at break for PBT nanocomposites is in relation with the interfacial interaction between CNTs and PBT matrix. The stronger interfacial adhesion, the more difficult segmental stretching and motion of PBT chains, the lower elongation at break [39].

Flame Retardancy
Cone calorimeter tests [40,41] were performed to measure the heat release rate (HRR), total heat release (THR), smoke production rate (SPR), and CO2 and CO productions of PBT and its nanocomposites. The HRR and THR curves under a heat flux of 35 kW/m 2 are shown in Figure 8a,b, respectively, and the related data are listed in Table 4. Neat PBT burns at 120 s (TTI value) with a high PHRR value (944 kW/m 2 ). The incorporation of DPP(Ox)-A-CNT nanoparticles lead to the slightly reduced TTI values and increased full width at half maximum for the HRR curves. The PHRR value decreases from 944 kW/m 2 for neat PBT to 759 kW/m 2 for PBT/DPP-A-CNT, 668 kW/m 2 for PBT/DPP(O)-A-CNT, and 710 kW/m 2 for PBT/DPP(O3)-A-CNT, with reductions of 20%, 29%, and 25%, respectively. From the THR curves shown in Figure 8b, it can be observed that PBT/DPP(O)-A-CNT demonstrates the lowest THR value, which is reduced by 5% compared to neat PBT. The results show that the introduction of DPP(Ox)-A-CNT nanoparticles inhibits the heat release through the barrier effect of the char residues and nanoparticle networks, of which DPP(O)-A-CNT exhibits the highest reduction on HRR and THR.
PBT is a kind of aromatic polymers which releases lots of smoke and toxic gases during burning process [42][43][44]. The reduction of smoke and toxic products during the combustion process is a very important consideration in view of the tenability condition for occupants in enclosed environments. Figure 8c,d exhibit the SPR and total smoke production (TSP) curves of neat PBT and its nanocomposites. The corresponding data are summarized in Table 4

Flame Retardancy
Cone calorimeter tests [40,41] were performed to measure the heat release rate (HRR), total heat release (THR), smoke production rate (SPR), and CO 2 and CO productions of PBT and its nanocomposites. The HRR and THR curves under a heat flux of 35 kW/m 2 are shown in Figure 8a,b, respectively, and the related data are listed in Table 4. Neat PBT burns at 120 s (TTI value) with a high PHRR value (944 kW/m 2 ). The incorporation of DPP(O x )-A-CNT nanoparticles lead to the slightly reduced TTI values and increased full width at half maximum for the HRR curves. The PHRR value decreases from 944 kW/m 2 for neat PBT to 759 kW/m 2 for PBT/DPP-A-CNT, 668 kW/m 2 for PBT/DPP(O)-A-CNT, and 710 kW/m 2 for PBT/DPP(O 3 )-A-CNT, with reductions of 20%, 29%, and 25%, respectively. From the THR curves shown in Figure 8b, it can be observed that PBT/DPP(O)-A-CNT demonstrates the lowest THR value, which is reduced by 5% compared to neat PBT. The results show that the introduction of DPP(O x )-A-CNT nanoparticles inhibits the heat release through the barrier effect of the char residues and nanoparticle networks, of which DPP(O)-A-CNT exhibits the highest reduction on HRR and THR.
PBT is a kind of aromatic polymers which releases lots of smoke and toxic gases during burning process [42][43][44]. The reduction of smoke and toxic products during the combustion process is a very important consideration in view of the tenability condition for occupants in enclosed environments. Figure 8c,d exhibit the SPR and total smoke production (TSP) curves of neat PBT and its nanocomposites. The corresponding data are summarized in Table 4  In a fire scenario, CO and CO2 are the main toxic gases generated from the burning of polymers [45][46][47], which can lead to asphyxiation or suffocation of occupants in enclosed environments. Figure  8e,f represents the CO2 and CO production curves of pure PBT and its nanocomposites. The related parameters are listed in Table 4  In a fire scenario, CO and CO 2 are the main toxic gases generated from the burning of polymers [45][46][47], which can lead to asphyxiation or suffocation of occupants in enclosed environments. Figure 8e,f represents the CO 2 and CO production curves of pure PBT and its nanocomposites. The related parameters are listed in Table 4 Table 4. Cone calorimeter data for each sample at 35 kW/m 2 . (TTI: time to ignition; PHRR: peak heat release rate; THR: total heat release; PSPR: peak smoke production rate; TSP: total smoke production; PCO 2 P: peak CO 2 production; PCOP: peak CO production).   To investigate the flame retardant mechanism, the structures of these residues were characterized by SEM coupled with EDX analyzer. The SEM images of the residues for PBT/DPP(Ox)-A-CNT nanocomposites are shown in Figure 10. From the SEM images, there appeared to be more solid chars left behind on the surface of PBT/DPP(O)-A-CNT, compared to PBT/DPP-A-CNT and PBT/DPP(O3)-A-CNT. This could be explained by the better carbonization effect of DPP(O)-A-CNT on PBT matrix, which corresponds well with the TGA analysis and residue results from cone calorimeter tests. The EDX results of the residues for PBT/DPP(Ox)-A-CNT nanocomposites are shown in Figure 11, and the related data are summarized in Table 5. All the residue samples are primarily composed of C, O, and P element (element content: C > O > P). In the case of PBT/DPP(O3)-A-CNT, the proportion of P element is only 0.13%. The phosphorus-containing groups decomposed from DPP(O3)-A-CNT participates in the gas-phase flame retardant action. Compared to PBT/DPP-A-CNT, more O and P elements are left in the char layers of PBT/DPP(O)-A-CNT. The improved O and P elements contents contribute to the reduction of HRR, CO2 and CO production as well as smoke emission. Moreover, the phosphinic-based groups decomposed from DPP(O)-A-CNT may efficiently trap radicals to participate in the carbonization reaction. The thermally stable chars act as an effective barrier to reduce the exposure of PBT nanocomposites to an external heat source [48][49][50][51][52][53][54], which is beneficial to reduce the fire hazards. To investigate the flame retardant mechanism, the structures of these residues were characterized by SEM coupled with EDX analyzer. The SEM images of the residues for PBT/DPP(O x )-A-CNT nanocomposites are shown in Figure 10. The improved O and P elements contents contribute to the reduction of HRR, CO 2 and CO production as well as smoke emission. Moreover, the phosphinic-based groups decomposed from DPP(O)-A-CNT may efficiently trap radicals to participate in the carbonization reaction. The thermally stable chars act as an effective barrier to reduce the exposure of PBT nanocomposites to an external heat source [48][49][50][51][52][53][54], which is beneficial to reduce the fire hazards.

Conclusions
With the aim of improving the fire performance of PBT, three covalently functionalized CNTs, DPP(Ox)-A-CNTs (x = 0, 1, 3), were successfully prepared and mixed with PBT using one-pot functionalization method via the reaction between different oxidation state phosphorus-containing agents and amino-carbon nanotube (A-CNT). These covalently functionalized CNTs were embedded with PBT through the consideration of a melt blending method. SEM observations revealed that the DPP(Ox)-A-CNT nano-fillers were found to be more homogeneously distributed within the PBT matrix compared with A-CNT alone. The incorporation of the three DPP(Ox)-A-CNT nanoparticles significantly improved the thermal stability of PBT. PBT/DPP(O3)-A-CNT showed the highest crystallization temperature and tensile strength, resulting from the good dispersion and interfacial interactions between DPP(O3)-A-CNT and PBT matrix. PBT/DPP(O)-A-CNT exhibited the best flame retardant properties due to the excellent carbonization effect. The decomposed polymer radicals can be effectively trapped by DPP(O)-A-CNT, leading to the reduction of PHRR, SPR, PCO2P and PCOP in cone calorimeter tests. This simple method to prepare functionalized CNTs in the current work can be extended to the surface functionalization of other nanoadditives. Functionalized nano-additives will enhance the dispersion and interfacial interaction within polymer hosts, resulting in the superior properties of polymeric materials, which shows the promising industrial application.

Conclusions
With the aim of improving the fire performance of PBT, three covalently functionalized CNTs, DPP(O x )-A-CNTs (x = 0, 1, 3), were successfully prepared and mixed with PBT using one-pot functionalization method via the reaction between different oxidation state phosphorus-containing agents and amino-carbon nanotube (A-CNT). These covalently functionalized CNTs were embedded with PBT through the consideration of a melt blending method. SEM observations revealed that the DPP(O x )-A-CNT nano-fillers were found to be more homogeneously distributed within the PBT matrix compared with A-CNT alone. The incorporation of the three DPP(O x )-A-CNT nanoparticles significantly improved the thermal stability of PBT. PBT/DPP(O 3 )-A-CNT showed the highest crystallization temperature and tensile strength, resulting from the good dispersion and interfacial interactions between DPP(O 3 )-A-CNT and PBT matrix. PBT/DPP(O)-A-CNT exhibited the best flame retardant properties due to the excellent carbonization effect. The decomposed polymer radicals can be effectively trapped by DPP(O)-A-CNT, leading to the reduction of PHRR, SPR, PCO 2 P and PCOP in cone calorimeter tests. This simple method to prepare functionalized CNTs in the current work can be extended to the surface functionalization of other nanoadditives. Functionalized nano-additives will enhance the dispersion and interfacial interaction within polymer hosts, resulting in the superior properties of polymeric materials, which shows the promising industrial application.