Synthesis of Transition Metal Complexes and Their Effects on Combustion Properties of Semi-Rigid Polyvinyl Chloride

Using introduction of MoO42− and Fe3+, Cu2+, or Zn2+ into amphiphilic polymers (DN) via an ion-exchange reaction, different transition metal complexes, as retardants and smoke suppressants, including (DN)Mo, Fe(DN)Mo, Cu(DN)Mo, and Zn(DN)Mo were synthesized. Combined with the results of X-ray photoelectron spectroscopy (XPS), X-ray diffraction (XRD), and Fourier transform infrared spectroscopy (FTIR), it could be determined that ionic bonding of these ions with DN occurred. Subsequently, the influence of flame-retardant, smoke-suppression, and mechanical properties of (DN)Mo, Fe(DN)Mo, Cu(DN)Mo, and Zn(DN)Mo on polyvinyl Chloride (PVC) were tested. It was demonstrated that transition metal complexes of three metal elements, Fe(DN)Mo, Cu(DN)Mo, and Zn(DN)Mo, showed better flame retardancy, smoke suppression, and thermal stability as confirmed by microcalorimetry, limiting oxygen index (LOI), smoke density, and thermogravimetric analysis (TGA) tests, in which Cu(DN)Mo worked best due to the Lewis acid mechanism and reductive coupling mechanism. Scanning electron microscopy (SEM) showed that the addition of (DN)Mo, Fe(DN)Mo, Cu(DN)Mo, and Zn(DN)Mo promoted the formation of a dense carbon layer on the PVC surface during combustion, which could protect the interior PVC. The addition of these transition metal complexes hardly impaired the mechanical properties of PVC.


Introduction
Polyvinyl chloride (PVC) is one of the most commonly used thermoplastics and is widely applied in the fields of furniture, building materials, textiles, and transportation due to excellent mechanical properties, good chemical resistance, superior electrical insulation, and low price [1][2][3][4]. Although pure PVC has excellent flame retardancy, its smoke-suppression property is rather weak. The inevitable use of plasticizers reduces the flame retardancy of PVC and further profoundly deteriorates the smoke-suppression ability [5]. Therefore, it is urgent to develop efficient flame retardants and smoke suppressants for PVC.
Applications and mechanisms of PVC flame retardancy and smoke suppression have been extensively studied [6,7]. In terms of flame retardancy, common flame retardants are phosphorus-based compounds (i.e., red phosphorus and melamine phosphate), siliconbased compounds (i.e., silicone and polysilsesquioxane), and nitrogen-based compounds (i.e., melamine cyanurate) [8][9][10]. In particular, nitrogen-based flame retardants tend to produce NH 3 , N 2 , N x O y , and H 2 O during combustion, which can reduce the concentration of oxygen and the surface temperature of polymers, and exhibit superior flame

Synthesis of Transition Metal Complexes (A(DN)Mo)
The (DN)Mo was soaked in A + Cl − solution (1 mol/L, A + =Fe 3+ , Cu 2+ , or Zn 2+ ), ensuring that the ion exchange for A + to bond with C6H5O3S − was in balance. The synthetic route is shown in Figure 1; the three products were recorded as Fe(DN)Mo, Cu(DN)Mo, and Zn(DN)Mo, respectively.

Preparation of PVC Composites
The basic formula of the PVC composites was 100 parts per hundred of resin (phr) PVC, 1 phr organic tin stabilizer, 20 phr DOP, and 0.4 phr calcium stearate. The addition of (DN)Mo, Fe(DN)Mo, Cu(DN)Mo, or Zn(DN)Mo was 8 phr. These components were melt-mixed with a two-roll mill (XH-401AE, Dongguan, China) at 160 °C for 5 min. After that, the samples were compressed into a sheet at 160 °C and 10 MPa for 8 min, and cooled to room temperature. The sheet was cut into the desired shapes and sizes for further testing.

X-ray Photoelectron Spectroscopy Test
The synthesized samples were characterized by X-ray photoelectron spectroscopy (XPS, PHI5000, JEOL, Beijing, China), using Al Kα (1486.6 eV) as the X-ray source with a high voltage of 12 kV, an emission current of 20 mA, and a fluence energy of 20 eV. Before processing the XPS spectra, the standard peak spectra of C1s were used to correct.

Synthesis of Transition Metal Complexes (A(DN)Mo)
The (DN)Mo was soaked in A + Cl − solution (1 mol/L, A + =Fe 3+ , Cu 2+ , or Zn 2+ ), ensuring that the ion exchange for A + to bond with C 6 H 5 O 3 S − was in balance. The synthetic route is shown in Figure 1; the three products were recorded as Fe(DN)Mo, Cu(DN)Mo, and Zn(DN)Mo, respectively.

Preparation of PVC Composites
The basic formula of the PVC composites was 100 parts per hundred of resin (phr) PVC, 1 phr organic tin stabilizer, 20 phr DOP, and 0.4 phr calcium stearate. The addition of (DN)Mo, Fe(DN)Mo, Cu(DN)Mo, or Zn(DN)Mo was 8 phr. These components were melt-mixed with a two-roll mill (XH-401AE, Dongguan, China) at 160 • C for 5 min. After that, the samples were compressed into a sheet at 160 • C and 10 MPa for 8 min, and cooled to room temperature. The sheet was cut into the desired shapes and sizes for further testing.

X-ray Photoelectron Spectroscopy Test
The synthesized samples were characterized by X-ray photoelectron spectroscopy (XPS, PHI5000, JEOL, Beijing, China), using Al Kα (1486.6 eV) as the X-ray source with a high voltage of 12 kV, an emission current of 20 mA, and a fluence energy of 20 eV. Before processing the XPS spectra, the standard peak spectra of C1s were used to correct.

Fourier Transform Infrared Spectroscopy Test
The synthesized samples were examined with a spectrometer (FTIR, Vertex 70, Bruker, Karlsruhe, Germany) using KBr pellets.

Thermogravimetric Analyses
Thermogravimetric analyses (TGA) of the synthesized samples and PVC composite samples were performed using a TGA Q5000 thermal gravimetric analyzer (TA instruments, New Castle, DE, USA). The weight of all samples was about 5 mg. Each sample was tested in an Al 2 O 3 pan, and the experimental temperature range was from 20 • C to 750 • C at a heating rate of 20 • C /min in a nitrogen atmosphere.

Scanning Electron Microscopy
The morphologies of the synthesized samples, burned carbon layers, and PVC composites were all characterized by field-emission scanning electron microscopy (SEM, SU8010, Hitachi, Chiyoda City, Japan) with an accelerating voltage of 5 kV. The synthesized sample particles were distributed on conductive adhesive tape for SEM observation. The surfaces and cross-sections were obtained in the burned carbon layers which were derived from the samples after a limiting oxygen index test. PVC composites were freeze-fractured in liquid nitrogen. All surfaces to be tested were sputtered-coated with a layer of gold (deposition time 20 s) before observation.

Smoke Density Test
The smoke densities (Ds) of the PVC composites were measured in a smoke density chamber (FTT 0064, Earth Products China Limited, Hong Kong, China) following the procedure of the ISO 5959-2: 2006 standard with an external heat flux of 25 KW/m 2 and flameless combustion. The dimensions of the tested specimens were 75 mm × 75 mm × 1 mm. According to the standard, Ds 600 is the Ds value at 600s and Ds max is the maximum Ds value during the 600 s of combustion.

Limiting Oxygen Index
The limiting oxygen index (LOI) of the PVC composites was measured by oxygen index meter (HC-2, Jiangning Testing Equipment, Nanjing, China) in sheet dimensions of 120 mm × 10 mm × 4 mm following GB/T 2406.2-2009 standards.

Microcalorimetry Test
The actual combustion of the PVC composites was measured by microcalorimetry (MCC, FTT 0001, Earth Products China Limited, Hong Kong, China), with samples weighing about 5 mg. The heating temperature was raised to 750 • C at a heating rate of 2 • C/s in a stream of oxygen flowing at 20 mL/min. The combustion temperature was set at 900 • C, and the oxygen/ nitrogen flow rate was set at 20 mL/80 mL.

Mechanical Properties
Impact strength of samples in the dimensions of 80 mm × 10 mm × 4 mm was tested according to the GB/T1043.1-2008 standard using an electronic impact tester (GT-7045-MDL, GTS systems co., Ltd., Dongguan, China). Tensile strength of dumbbell-shaped PVC composites samples in the dimensions of 75 mm × 4 mm × 2 mm was tested using an electronic tensile tester (GMT6130, MTS systems Co., Ltd., Eden Prairie, MN, USA) according to the procedure of the GB/T1040.2-2006 standard. Tests were conducted at an extension rate of 5.0 mm/min at 25 • C. Flexural strength of PVC composites samples in the dimensions of 80 mm × 10 mm × 4 mm was tested using an electronic tensile tester (GMT6130, MTS systems Co., Ltd., Eden Prairie, MN, USA) according to the procedure of the GB/T 9341-2008 standard. For each result, an average value from a minimum of five test specimens were report together with standard deviations. Figure 2 shows the XPS spectra of (DN)Mo, Fe(DN)Mo, Cu(DN)Mo, and Zn(DN)Mo. As shown, XPS signals at 164 eV, 285 eV, 402 eV, and 530.6 eV corresponded to S 2p, C 1s, N 1s, and O 1s of the organic matrix in flame retardants and smoke suppressants, respectively, as reported in previous works [18,19]. The XPS signals at 232.1 eV (corresponding to Mo 3d), 706.7 eV (corresponding to Fe 2p3), 932.5 eV (corresponding to Cu 2p3), and 1021.8 eV (corresponding to Zn 2p3) indicated that Mo, Fe, Cu, and Zn were introduced into the DN. extension rate of 5.0 mm/min at 25 °C. Flexural strength of PVC composites samples in the dimensions of 80 mm × 10 mm × 4 mm was tested using an electronic tensile tester (GMT6130, MTS systems Co., Ltd., Eden Prairie, MN, USA) according to the procedure of the GB/T 9341-2008 standard. For each result, an average value from a minimum of five test specimens were report together with standard deviations. Figure 2 shows the XPS spectra of (DN)Mo, Fe(DN)Mo, Cu(DN)Mo, and Zn(DN)Mo. As shown, XPS signals at 164 eV, 285 eV, 402 eV, and 530.6 eV corresponded to S 2p, C 1s, N 1s, and O 1s of the organic matrix in flame retardants and smoke suppressants, respectively, as reported in previous works [18,19]. The XPS signals at 232.1 eV (corresponding to Mo 3d), 706.7 eV (corresponding to Fe 2p3), 932.5 eV (corresponding to Cu 2p3), and 1021.8 eV (corresponding to Zn 2p3) indicated that Mo, Fe, Cu, and Zn were introduced into the DN.   [20]. Combined with the XPS results, it could be concluded that the ionic bonding took place between MoO4 2− , Fe 3+ , Cu 2+ , or Zn 2+ and DN, rather than DN simply physically adsorbing inorganic metal salts such as FeCl3, CuCl3, and ZnCl2, proving the occurrence of the desired ion-exchange reactions.   [20]. Combined with the XPS results, it could be concluded that the ionic bonding took place between MoO 4 2− , Fe 3+ , Cu 2+ , or Zn 2+ and DN, rather than DN simply physically adsorbing inorganic metal salts such as FeCl 3 , CuCl 3 , and ZnCl 2 , proving the occurrence of the desired ion-exchange reactions.  Figure 4. It can be seen that the absorption peaks between 1400 cm −1 and 3500 cm −1 were basically similar when comparing DN with the four transition metal ion complexes. From the IR spectrum of DN, it was found that the asymmetric absorption peak of the main chain methylene C-H appeared near 2926 cm −1 , and the stretched vibration peak of the quaternary ammonium cation unit ester bond C=O appeared at 1731 cm −1 , which were basically similar when comparing DN with the four transition metal ion complexes. From the IR spectrum of DN, it was found that the asymmetric absorption peak of the main chain methylene C-H appeared near 2926 cm −1 , and the stretched vibration peak of the quaternary ammonium cation unit ester bond C=O appeared at 1731 cm −1 , which indicated that DN had been produced by a polymerization reaction. Broad absorption peaks at 3444 cm −1 and 1636 cm −1 , corresponding to -OH bonds, probably were due to crystalline water or adsorbed water in the DN [21]. Then, the change of functional group before and after the ion-exchange reaction were analyzed. The C-N stretching vibration peak appeared near 1400 cm −1 , and the C-N stretching vibration peak of four transition metal complexes shifted toward high wavenumbers after the reaction with MoO 4 2− . In addition, the four transition metal complexes had a strong absorption peak at 840 cm −1 , 839 cm −1 , 840 cm −1 , and 838 cm −1 , respectively, while the absorption peak of the Mo=O bond was located around 964 cm −1 . This indicated that the quaternary ammonium group had interacted with MoO 4 2− , which proved the success of the ion-exchange reaction. The stretched vibrational peak of the -S=O group in the sulfonic acid group was located at 1186 cm −1 . After the reaction with Fe 3+ , Cu 2+ , and Zn 2+ , the stretched vibrational peak of the -S=O group for Fe(DN)Mo, Cu(DN)Mo, and Zn(DN)Mo shifted toward low wavenumbers, which indicated that the metal ions interacted with the -S=O group of the sulfonic acid group [22].   Figure 5. It can be seen that the thermal degradation process consisted of two stages. The first decomposition stage occurred from 30 °C to 280 °C, and the weight loss was about 10%, which was attributed to the evaporation of water vapor [23,24]. The second stage proceeded from 300 °C to 480 °C, which corresponded to the degradation of the organic matter matrix and the production of volatile gases such as CO, CO2, and NH3. The residual was mainly composed of transition metal compounds and carbon. It was noteworthy that (DN)Mo, Fe(DN)Mo, Cu(DN)Mo, and Zn(DN)Mo showed residual content at 700 °C of up to about 53.1 wt %, 59.3 wt %, 64.2 wt %, and 64.1 wt %, respectively, which indicated these transition metal complexes had sufficient thermal stability to meet the processing temperature of plastics.  Figure 5. It can be seen that the thermal degradation process consisted of two stages. The first decomposition stage occurred from 30 • C to 280 • C, and the weight loss was about 10%, which was attributed to the evaporation of water vapor [23,24]. The second stage proceeded from 300 • C to 480 • C, which corresponded to the degradation of the organic matter matrix and the production of volatile gases such as CO, CO 2 , and NH 3 . The residual was mainly composed of transition metal compounds and carbon. It was noteworthy that (DN)Mo, Fe(DN)Mo, Cu(DN)Mo, and Zn(DN)Mo showed residual content at 700 • C of up to about 53.1 wt %, 59.3 wt %, 64.2 wt %, and 64.1 wt %, respectively, which indicated these transition metal complexes had sufficient thermal stability to meet the processing temperature of plastics. ond stage proceeded from 300 °C to 480 °C, which corresponded to the degradation of the organic matter matrix and the production of volatile gases such as CO, CO2, and NH3. The residual was mainly composed of transition metal compounds and carbon. It was noteworthy that (DN)Mo, Fe(DN)Mo, Cu(DN)Mo, and Zn(DN)Mo showed residual content at 700 °C of up to about 53.1 wt %, 59.3 wt %, 64.2 wt %, and 64.1 wt %, respectively, which indicated these transition metal complexes had sufficient thermal stability to meet the processing temperature of plastics.  This attribute is critical to maintaining the mechanical properties of PVC, which will be discussed later.

Characterization of Smoke Density for PVC Composites
The ( Figure 7, and the corresponding Dsmax and Ds600 are summarized in Table 1. Compared with pure PVC, the addition of (DN)Mo significantly decreased the Dsmax and Ds600 of PVC, which indicated that (DN)Mo could improve the smoke-suppression performance of the PVC. Mo could react with the HCl decomposed from PVC during combustion, which tends to form cross-linking of polyene chains catalyzed by strong Lewis acid [6]. This behavior can promote charring of PVC, and reduce smoke emission. Notably, compared with PVC/(DN)Mo, PVC/Fe(DN)Mo, PVC/Cu(DN)Mo, and PVC/Zn(DN)Mo showed lower Dsmax and Ds600, indicating that two metal elements of Fe-Mo, Cu-Mo or Zn-Mo exhibited more outstanding smoke-suppression performance. This was due to the fact that Fe and Zn have a mutual promotion effect with Mo by the Lewis acid mechanism [25]. PVC/Cu(DN)Mo also showed the optimum effect of smoke suppression due to the fact that Cu works in smoke suppression mainly through a reductive coupling mechanism, and by a synergistic contribution of the Lewis acid mechanism [12,26]. In conclusion, all transition metal ion complexes had excellent smoke-suppression efficiency.  Figure 7, and the corresponding Ds max and Ds 600 are summarized in Table 1. Compared with pure PVC, the addition of (DN)Mo significantly decreased the Ds max and Ds 600 of PVC, which indicated that (DN)Mo could improve the smokesuppression performance of the PVC. Mo could react with the HCl decomposed from PVC during combustion, which tends to form cross-linking of polyene chains catalyzed by strong Lewis acid [6]. This behavior can promote charring of PVC, and reduce smoke emission. Notably, compared with PVC/(DN)Mo, PVC/Fe(DN)Mo, PVC/Cu(DN)Mo, and PVC/Zn(DN)Mo showed lower Ds max and Ds 600 , indicating that two metal elements of Fe-Mo, Cu-Mo or Zn-Mo exhibited more outstanding smoke-suppression performance. This was due to the fact that Fe and Zn have a mutual promotion effect with Mo by the Lewis acid mechanism [25]. PVC/Cu(DN)Mo also showed the optimum effect of smoke suppression due to the fact that Cu works in smoke suppression mainly through a reductive coupling mechanism, and by a synergistic contribution of the Lewis acid mechanism [12,26]. In conclusion, all transition metal ion complexes had excellent smoke-suppression efficiency.
lyzed by strong Lewis acid [6]. This behavior can promote charring of PVC, and reduce smoke emission. Notably, compared with PVC/(DN)Mo, PVC/Fe(DN)Mo, PVC/Cu(DN)Mo, and PVC/Zn(DN)Mo showed lower Dsmax and Ds600, indicating that two metal elements of Fe-Mo, Cu-Mo or Zn-Mo exhibited more outstanding smoke-suppression performance. This was due to the fact that Fe and Zn have a mutual promotion effect with Mo by the Lewis acid mechanism [25]. PVC/Cu(DN)Mo also showed the optimum effect of smoke suppression due to the fact that Cu works in smoke suppression mainly through a reductive coupling mechanism, and by a synergistic contribution of the Lewis acid mechanism [12,26]. In conclusion, all transition metal ion complexes had excellent smoke-suppression efficiency.

Characterization of Flame Retardancy for PVC Composites
LOI is an important measurement of the flame-retardant properties of a material as a reference value. The combustion behavior of the composites is often characterized by MCC, which is widely used to evaluate the flammability of polymer materials [27]. Several important combustion parameters, including heat-release rate (HRR), peak value of heat release (pHRR), time at pHRR (T p ), and total heat release (THR) can be obtained from MCC. The curves of HRR and THR versus time for pure PVC and the PVC composites are shown in Figure 8.

Characterization of Flame Retardancy for PVC Composites
LOI is an important measurement of the flame-retardant properties of a material as a reference value. The combustion behavior of the composites is often characterized by MCC, which is widely used to evaluate the flammability of polymer materials [27]. Several important combustion parameters, including heat-release rate (HRR), peak value of heat release (pHRR), time at pHRR (Tp), and total heat release (THR) can be obtained from MCC. The curves of HRR and THR versus time for pure PVC and the PVC composites are shown in Figure 8. As shown in Figure 8, the HRR curves had two peaks, which was consistent with the reported results [28]. The first peak was mainly related to HCl emission and DOP degradation, while the second peak was the main chain cracking of PVC and production of some low molecular weight gaseous compounds [29]. It was reported that the HCl re- As shown in Figure 8, the HRR curves had two peaks, which was consistent with the reported results [28]. The first peak was mainly related to HCl emission and DOP degradation, while the second peak was the main chain cracking of PVC and production of some low molecular weight gaseous compounds [29]. It was reported that the HCl removal reaction, crosslinking of PVC, and carbon production can by catalyzed by a metal element based on a strong Lewis acid mechanism [30].
The combustion parameters are summarized in Table 2. The two stages of the pHRR and THR of the PVC composites was much lower than those of pure PVC, as shown in Table 2. This is because the formation of stable carbon on the PVC surface during combustion could effectively protect the interior from burning and heat, which could reduce the heat released throughout the combustion process. Notably, the pHRR in the second peak and THR of PVC/Cu(DN)Mo had the lowest values of 36.45% and 48.89%, respectively, compared with that of PVC/(DN)Mo, PVC/Fe(DN)Mo, and PVC/Zn(DN)Mo. This was because Cu and Mo have a strong synergistic effect in smoke suppression [12].

Thermal Stability of PVC Composites
The mass losses of PVC, PVC/(DN)Mo, PVC/Fe(DN)Mo, PVC/Cu(DN)Mo, and PVC/Zn(DN)Mo are presented in Figure 9, and the corresponding thermogravimetric parameters are illustrated in Table 3. The thermal degradation of the samples showed two stages, as shown in Figure 9. In the first stage, the side chains of PVC decomposed to produce HCl. In the second stage, the PVC main chain was cleaved to produce low molecular weight gases [31].

Thermal Stability of PVC Composites
The mass losses of PVC, PVC/(DN)Mo, PVC/Fe(DN)Mo, PVC/Cu(DN)Mo, and PVC/Zn(DN)Mo are presented in Figure 9, and the corresponding thermogravimetric parameters are illustrated in Table 3. The thermal degradation of the samples showed two stages, as shown in Figure 9. In the first stage, the side chains of PVC decomposed to produce HCl. In the second stage, the PVC main chain was cleaved to produce low molecular weight gases [31]. As presented in Table 3, the temperature at 5% mass loss (T5%) of PVC was 273.71 °C. In the first stage, the mass loss was 66.05%, and the corresponding temperature of maximum mass loss rate (Tmax) was 319.02 °C. In the second stage, the mass loss was 27.12% and the Tmax was 468.66 °C, with a final char residue of 5.94%. Compared with PVC, the T5% and mass loss of PVC/(DN)Mo in the second stage decreased, and the final solid residue increased significantly. This indicated that (DN)Mo had a superior charcoal-forming  As presented in Table 3, the temperature at 5% mass loss (T 5% ) of PVC was 273.71 • C. In the first stage, the mass loss was 66.05%, and the corresponding temperature of maximum mass loss rate (T max ) was 319.02 • C. In the second stage, the mass loss was 27.12% and the T max was 468.66 • C, with a final char residue of 5.94%. Compared with PVC, the T 5% and mass loss of PVC/(DN)Mo in the second stage decreased, and the final solid residue increased significantly. This indicated that (DN)Mo had a superior charcoalforming ability, which facilitated the formation of stable carbon slag and improvement of the thermal stability for PVC at a high temperature. It can be seen that the T 5% and mass losses of the PVC/Cu(DN)Mo, PVC/Fe(DN)Mo, and PVC/Zn(DN)Mo composites were reduced compared with that of PVC/(DN)Mo. This indicated that the PVC/Fe(DN)Mo, PVC/Cu(DN)Mo, and PVC/Zn(DN)Mo composites had good thermal-stability performance. Since Fe 3+ and Zn 2+ could react with HCl to produce FeCl 3 and ZnCl 2 , FeCl 3 and ZnCl 2 as a Lewis acid in condensed matter could catalyze the removal of HCl from PVC and promote earlier cross-linking of PVC, leading to rapid carbonization. The PVC/Cu(DN)Mo composite had the best thermal-stability performance. This was because the Cu in PVC/Cu(DN)Mo catalyzed the elimination of allyl chloride groups through a reductive coupling reaction, which inhibited the removal HCl reaction of "zipper-type" chain and slow down the thermal decomposition of PVC. Nitrogen-based flame retardants tend to produce NH 3 , N 2 , N x O y , and H 2 O during combustion, which can reduce the concentration of oxygen and the surface temperature of a polymer, contributing to flame retardancy [11].

Morphology Analysis of Char Residue
The morphology of surfaces and cross-sections for the char residues after the LOI test is shown in Figure 10. As shown in Figure 10(a1), many small holes existed on the surface of the PVC, which may have resulted in less protection for the matrix during combustion. This was because the dense carbon layer had the ability to insulate from heat and flames, while the presence of holes made the carbon layer less protective. As shown in Figure 10(a2,a3), numerous large holes existed in the cross-section, which confirmed the inadequacy of the porous carbon layer on the surface to protect the internal matrix. The surface of PVC/(DN)Mo was relatively flat and dense without holes, as shown in Figure 10(b1). This was due to the fact that the addition of (DN)Mo promoted the crosslinking of the PVC and the formation of a dense carbon layer, which was in agreement with the results of a higher char residue in TGA. The flat and dense characteristics were beneficial to isolating the contact between the external flame or heat and the internal matrix, which could be confirmed by the smaller and fewer holes shown in Figure 10   were slightly increased by 6%, 5%, 1%, and 4%, respectively. This was likely due to the fact that the transition metal complexes could passivate the front end of a crack, providing a slight toughening effect [16]. As shown in Figure 11b, a slight decrease in tensile strength and elongation at break appeared due to a good compatibility between transition metal complexes and the PVC, which prevented a remarkable decrease for adding so much fillers. As shown in Figure 11c, the bending strength and modulus of the four transition metal  were slightly increased by 6%, 5%, 1%, and 4%, respectively. This was likely due to the fact that the transition metal complexes could passivate the front end of a crack, providing a slight toughening effect [16]. As shown in Figure 11b, a slight decrease in tensile strength and elongation at break appeared due to a good compatibility between transition metal complexes and the PVC, which prevented a remarkable decrease for adding so much fillers.

Mechanical Properties of PVC Composites
As shown in Figure 11c, the bending strength and modulus of the four transition metal ion complexes showed satisfactory improvement. Combining the impact, tensile properties, and bending properties, it was demonstrated that the addition of the transition metal ion complexes could slightly strengthen and embrittle the PVC composites. On the other hand, the addition of transition metal ion complexes had little adverse effect on the mechanical properties of the PVC matrix due to good compatibility between the transition metal ion complexes and the PVC matrix. ion complexes showed satisfactory improvement. Combining the impact, tensile properties, and bending properties, it was demonstrated that the addition of the transition metal ion complexes could slightly strengthen and embrittle the PVC composites. On the other hand, the addition of transition metal ion complexes had little adverse effect on the mechanical properties of the PVC matrix due to good compatibility between the transition metal ion complexes and the PVC matrix.  Figure  11a,c. The surfaces of PVC/Cu(DN)Mo and PVC/Zn(DN)Mo were particle-free and consistent with the surface characteristics of the PVC, which indicated their good compatibility with the PVC matrix.

Conclusions
In this work, the four transition metal complexes of (DN)Mo, Fe(DN)Mo, Cu(DN)Mo, and Zn(DN)Mo were synthesized via photocatalytic copolymerization and an ion-exchange reaction. The results of XPS, XRD, and FTIR showed that the MoO4 2− , Fe 3+ , Cu 2+ , and Zn 2+ were ionically bonding with DN. The TGA and SEM results showed that these transition metal complexes possessed sufficient thermal stability and good dispersion. The MCC and LOI tests showed that Fe(DN)Mo, Cu(DN)Mo, and Zn(DN)Mo possessed better flame retardancy. The smoke density test showed the superior smoke suppression of Fe(DN)Mo, Cu(DN)Mo, and Zn(DN)Mo, and the TG test showed that these transition metal complexes could endow the PVC matrix with better stability. Cu(CN)Mo exhibited an overall better flame retardancy and smoke-suppression effect. It was also found that the addition of transition metal complexes promoted the formation of a dense carbon layer on the PVC surface during combustion, which could protect the interior PVC matrix. Owing to good compatibility of the four transition meatal complexes with the PVC matrix, they hardly decreased the mechanical properties of PVC.