Effect of Ammonium Polyphosphate to Aluminum Hydroxide Mass Ratio on the Properties of Wood-Flour/Polypropylene Composites

Two halogen-free inorganic flame retardants, ammonium polyphosphate (APP) and aluminum hydroxide (ATH) were added to wood-flour/polypropylene composites (WPCs) at different APP to ATH mass ratios (APP/ATH ratios), with a constant total loading of 30 wt % (30% by mass). Water soaking tests indicated a low hygroscopicity and/or solubility of ATH as compared to APP. Mechanical property tests showed that the flexural properties were not significantly affected by the APP/ATH ratio, while the impact strength appeared to increase with the increasing ATH/APP ratio. Cone calorimetry indicated that APP appeared to be more effective than ATH in reducing the peak of heat release rate (PHRR). However, when compared to the neat WPCs, total smoke release decreased with the addition of ATH but increased with the addition of APP. Noticeably, WPCs containing the combination of 20 wt % APP and 10 wt % ATH (WPC/APP-20/ATH-10) showed the lowest PHRR and total heat release in all of the formulations. WPCs combustion residues were analyzed by scanning electron microscopy, laser Raman spectroscopy, and Fourier transform infrared spectroscopy (FTIR). Thermogravimetric analysis coupled with FTIR spectroscopy was used to identify the organic volatiles that were produced during the thermal decomposition of WPCs. WPC/APP-20/ATH-10 showed the most compact carbonaceous residue with the highest degree of graphitization.


Introduction
Wood-plastic composites (WPCs), consisting of wood processing residue (as fibers or finely ground particles) and thermoplastic polymer, have been showing a rapid growing usage in the last decades due to their significant advantages, such as high durability, low maintenance and cost, relatively high strength and stiffness, sustainability, and wood-like material appearance and properties [1][2][3][4]. WPCs have been widely used in both exterior and interior applications, including decking, fencing, cladding, window and door frames, landscaping timbers, and automotive components, etc. [5][6][7][8]. However, the high flammability of WPCs may limit their application. Several researchers have focused on the method to improve the flame retardancy of WPCs [9][10][11][12][13][14][15].
The most widely used method is to incorporate flame retardants (FRs) directly into WPCs during the melt processing [16][17][18][19][20]. Halogenated compounds based on chlorine or bromine are effective FRs Table 1. Compositions of the wood-plastic composites (WPCs). The test specimens were immersed into distilled water at (25 ± 2) • C for 15 days. Then the water absorption (WA) values were calculated based on the weight changes after 6 h, 24 h, and thereafter at 48 h intervals by removing the surface water with a dry cloth. Six replicates (50 mm × 50 mm × 3 mm) were tested for each group. The values of WA as percentages were calculated with the following equation:

Samples PP (wt %) WF (wt %) PP-g-MAH (wt %) APP (wt %) ATH (wt %)
where W 0 is the weight of oven-dried sample at 60 • C for 12 h before immersion and W t is the weight of the sample at a given immersion time t.

Three-Point Bending Test
Three-point tests were performed at a cross-head speed of 5 mm/min in accordance with the Chinese standard GB/T 9341-2008 (equivalent to ISO 178-2001). Six replicates (60 mm × 25 mm × 3 mm) were tested for each group.

Impact Strength
Unnotched Izod impact tests were carried out in accordance with the Chinese standard GB/T 1843-2008 (equivalent to ISO 180-2000). Six unnotched replicates (80 mm × 10 mm × 3 mm) were tested for each group.

Thermogravimetric Analysis (TGA)
The thermal stability of WPCs was evaluated by TGA (Q50 TGA analyzer, TA Instruments Ltd., New Castle, DE, USA). Each formulation (about 7 mg) was tested under a nitrogen flow (40 mL/min) at a linear heating rate of 20 • C/min from room temperature up to 700 • C.

Cone Calorimetry
Cone calorimetry (Fire Testing Technology Ltd., East Grinstead, UK) was performed at an incident heat flux of 50 kW/m 2 in accordance with ISO 5660 standard procedures. Each specimen (100 mm × 100 mm × 3 mm) was wrapped in aluminum foil. Three replicates were tested for each group.

Scanning Electron Microscopy (SEM)
The surface morphology of WPCs combustion residues was investigated by a field-emission SU8010 SEM analyzer (Hitachi Ltd., Tokyo, Japan) with an acceleration voltage of 3.0 kV. All of the samples were sputter-coated with gold layer.

Laser Raman Spectroscopy (LRS)
LRS spectra of WPCs combustion residues at the end of cone calorimetry tests were investigated by a SPEX-1403 spectrometer (SPEX SamplePrep Ltd., Metuchen, NJ, USA). The spectra were recorded under an Ar-ion laser (514.5 nm) at room temperature. Each spectrum was subjected to peak fitting using the software Origin 9.0/Multiple Peak Fit Module to resolve the curve into two Lorentz bands.

Fourier Transform Infrared Spectroscopy (FTIR)
FTIR spectra of WPCs combustion residues at the end of cone calorimetry tests were recorded by a Nicolet 6700 FTIR spectrometer (Thermo Fisher Scientific Ltd., Waltham, MA, USA) in attenuated total reflection mode. Before testing, all the samples were ground into powders and oven dried. Spectra were signal averaged over 32 scans at a resolution of 4 cm −1 in the (4000-650) cm −1 wavenumber region.

Thermogravimetric Analysis-Infrared Spectrometry (TGA-IR)
TGA-IR spectra were recorded using a Q 5000 TGA analyzer (TA Instruments Ltd., New Castle, DE, USA), which was interfaced to a Nicolet 6700 FTIR spectrophotometer (Thermo Fisher Scientific Ltd., Waltham, MA, USA). Samples (about 15 mg) were tested in an alumina crucible at a 45 mL/min nitrogen flow and at a heating rate of 20 • C/min from room temperature up to 700 • C.

Physical-Mechanical Performance of WPCs
As can be seen in Figure 1, WA of the neat WPCs (no FR fillers) increased at a constant rate in the investigated time range. This was attributed to the slow diffusion of water molecules into the WF/polymer interface, likely promoting by hydrogen bonding between the water molecules and the free -OH groups in the WF [48]. When 30 wt % APP was added, WA value of the composites (WPC/APP-30) dramatically increased because of the high hygroscopicity of APP. However, the WA value showed a maximum at the ninth day, and then decreased significantly, likely due to partial APP dissolution in water [31]. When 30 wt % ATH was added, WA value of the composites (WPC/ATH-30) slightly but obviously decreased as compared to the neat WPC throughout the entire tested time range. The WPC/ATH-30 showed the lowest WA value, even lower than the neat WPCs. When compared with the WPC/APP-30, WA value of the composites containing ATH decreased with the increase of ATH loading in the matrix. The maximum in WA occurred at the 11th day of immersion for WPC/APP-20/ATH-10, two days after the maximum observed for WPC/APP-30. It should be noted that the WA maxima were not observed for the WPC/APP-15/ATH-15, WPC/APP-10/ATH-20, and WPC/ATH-30 within the tested 15 days. According to the results, APP increased the hygroscopicity and limited the hydrolytic stability of the composites. A significant amount of APP was solubilized and extracted from the composites over the tested 15 days. On the other side, ATH appeared to decrease the hygroscopicity of WPCs and increase the hydrolytic stability of the composites as compared to APP. APP dissolution in water [31]. When 30 wt % ATH was added, WA value of the composites (WPC/ATH-30) slightly but obviously decreased as compared to the neat WPC throughout the entire tested time range. The WPC/ATH-30 showed the lowest WA value, even lower than the neat WPCs. When compared with the WPC/APP-30, WA value of the composites containing ATH decreased with the increase of ATH loading in the matrix. The maximum in WA occurred at the 11th day of immersion for WPC/APP-20/ATH-10, two days after the maximum observed for WPC/APP-30. It should be noted that the WA maxima were not observed for the WPC/APP-15/ATH-15, WPC/APP-10/ATH-20, and WPC/ATH-30 within the tested 15 days. According to the results, APP increased the hygroscopicity and limited the hydrolytic stability of the composites. A significant amount of APP was solubilized and extracted from the composites over the tested 15 days. On the other side, ATH appeared to decrease the hygroscopicity of WPCs and increase the hydrolytic stability of the composites as compared to APP. When the composites are used as load bearing structures in furniture, automotive, or building industries, their flexural strength (MOR) and flexural modulus (MOE) are the critical performance metrics [49]. Figure 2 shows that the incorporation of 30 wt % APP significantly decreased the MOR of the composites from (72.6 ± 0.4) MPa to (63.7 ± 1.8) MPa (≈12% average decrease), likely due to the poor APP/matrix compatibility, and increased the MOE from (2.8 ± 0.1) GPa to (3.5 ± 0.1) GPa (≈26% average increase) due to the intrinsic high MOE of APP. These results were consistent with the previous reports [31,34,50,51]. ATH had a similar effect on the MOR and MOE. The incorporation of 30 wt % ATH decreased the MOR of composites to (61.9 ± 2.6) MPa (≈15% average decrease) and increased the MOE of composites to (3.4 ± 0.1) GPa (≈22% average increase). The MOR and MOE values were not significantly affected by the APP to ATH mass ratio (see WPC/APP-20/ATH-10, WPC/APP-15/ATH-15, WPC/APP-10/ATH-20 in Figure 2). When the composites are used as load bearing structures in furniture, automotive, or building industries, their flexural strength (MOR) and flexural modulus (MOE) are the critical performance metrics [49]. Figure 2 shows that the incorporation of 30 wt % APP significantly decreased the MOR of the composites from (72.6 ± 0.4) MPa to (63.7 ± 1.8) MPa (≈12% average decrease), likely due to the poor APP/matrix compatibility, and increased the MOE from (2.8 ± 0.1) GPa to (3.5 ± 0.1) GPa (≈26% average increase) due to the intrinsic high MOE of APP. These results were consistent with the previous reports [31,34,50,51]. ATH had a similar effect on the MOR and MOE. The incorporation of 30 wt % ATH decreased the MOR of composites to (61.9 ± 2.6) MPa (≈15% average decrease) and increased the MOE of composites to (3.4 ± 0.1) GPa (≈22% average increase). The MOR and MOE values were not significantly affected by the APP to ATH mass ratio (see WPC/APP-20/ATH-10, WPC/APP-15/ATH-15, WPC/APP-10/ATH-20 in Figure 2). The results of impact strength tests are shown in Figure 3. As compared to the neat WPCs without FR fillers, the impact strength decreased from (13.5 ± 0.7) KJ/m 2 to (11.9 ± 0.7) KJ/m 2 (about 12% average decrease) with the addition of 30 wt % APP. This might be due to the presence of residual agglomerates of APP, which act as micro-crack initiators [31,52]. As more APP was replaced by ATH in the composites (see WPC/APP-20/ATH-10, WPC/APP-15/ATH-15, and WPC/APP-10/ATH-20 in Figure 3), a gradual increase in impact strength was observed. The impact strength increased up to (14.6 ± 0.9) KJ/m 2 with the addition of 30 wt % ATH. The improvement in impact strength achieved by replacing APP with ATH could be due to several reasons, such as, a stronger particle-matrix interface, better dispersion, and a variation in the PP crystallinity.  The results of impact strength tests are shown in Figure 3. As compared to the neat WPCs without FR fillers, the impact strength decreased from (13.5 ± 0.7) KJ/m 2 to (11.9 ± 0.7) KJ/m 2 (about 12% average decrease) with the addition of 30 wt % APP. This might be due to the presence of residual agglomerates of APP, which act as micro-crack initiators [31,52]. As more APP was replaced by ATH in the composites (see WPC/APP-20/ATH-10, WPC/APP-15/ATH-15, and WPC/APP-10/ATH-20 in Figure 3), a gradual increase in impact strength was observed. The impact strength increased up to (14.6 ± 0.9) KJ/m 2 with the addition of 30 wt % ATH. The improvement in impact strength achieved by replacing APP with ATH could be due to several reasons, such as, a stronger particle-matrix interface, better dispersion, and a variation in the PP crystallinity. The results of impact strength tests are shown in Figure 3. As compared to the neat WPCs without FR fillers, the impact strength decreased from (13.5 ± 0.7) KJ/m 2 to (11.9 ± 0.7) KJ/m 2 (about 12% average decrease) with the addition of 30 wt % APP. This might be due to the presence of residual agglomerates of APP, which act as micro-crack initiators [31,52]. As more APP was replaced by ATH in the composites (see WPC/APP-20/ATH-10, WPC/APP-15/ATH-15, and WPC/APP-10/ATH-20 in Figure 3), a gradual increase in impact strength was observed. The impact strength increased up to (14.6 ± 0.9) KJ/m 2 with the addition of 30 wt % ATH. The improvement in impact strength achieved by replacing APP with ATH could be due to several reasons, such as, a stronger particle-matrix interface, better dispersion, and a variation in the PP crystallinity.

Thermal Degradation Analysis
TGA and derivative thermogravimetry (DTG) curves of the WPCs are displayed in Figure 4a,b, respectively, and the related data are summarized in Table 2. The thermal degradation process of the neat WPCs showed two main degradation steps (DSs), with only 10.8 wt % residue being left at 700 • C. The temperatures at which the derivative weight peaks are achieved for the first DS, the second DS, and the third DS are indicated in Table 2 as T max1 , T max2 , and T max3 , respectively. For the neat WPCs, T max1 , and T max2 were 330 • C and 462 • C, respectively. The WPCs containing APP or a combination of APP and ATH showed lower T max1 and higher T max2 than the neat WPCs without APP. The lower T max1 indicated that the presence of APP accelerated the thermal decomposition of the composites. WPCs containing APP (WPC/APP-30, WPC/APP-20/ATH-10, WPC/APP-15/ATH-15, and WPC/APP-10/ATH-20) also showed a higher residue at 700 • C ( Table 2) than the neat WPCs due to polymer charring (promoted by APP) and thermally stable aluminum-orthophosphate and aluminum metaphosphate (generated by the reaction between ATH and APP) [42]. Noticeably, WPC/APP-20/ATH-10 was the sample with the highest residue. Unlike other formulations, WPC/ATH-30 showed three different DSs. The first DS was mainly due to the thermal degradation of ATH [43]. The other two DSs were similar to those of the neat WPCs. ATH decomposed without significantly affecting the decomposition of the WPCs and without promoting charring reactions. As a result, WPC/ATH-30 generated less residue than the other WPCs containing FR fillers.

Thermal Degradation Analysis
TGA and derivative thermogravimetry (DTG) curves of the WPCs are displayed in Figure 4a,b, respectively, and the related data are summarized in Table 2. The thermal degradation process of the neat WPCs showed two main degradation steps (DSs), with only 10.8 wt % residue being left at 700 °C. The temperatures at which the derivative weight peaks are achieved for the first DS, the second DS, and the third DS are indicated in Table 2 as Tmax1, Tmax2, and Tmax3, respectively. For the neat WPCs, Tmax1, and Tmax2 were 330 °C and 462 °C, respectively. The WPCs containing APP or a combination of APP and ATH showed lower Tmax1 and higher Tmax2 than the neat WPCs without APP. The lower Tmax1 indicated that the presence of APP accelerated the thermal decomposition of the composites. WPCs containing APP (WPC/APP-30, WPC/APP-20/ATH-10, WPC/APP-15/ATH-15, and WPC/APP-10/ATH-20) also showed a higher residue at 700 °C (Table 2) than the neat WPCs due to polymer charring (promoted by APP) and thermally stable aluminum-orthophosphate and aluminum metaphosphate (generated by the reaction between ATH and APP) [42]. Noticeably, WPC/APP-20/ATH-10 was the sample with the highest residue. Unlike other formulations, WPC/ATH-30 showed three different DSs. The first DS was mainly due to the thermal degradation of ATH [43]. The other two DSs were similar to those of the neat WPCs. ATH decomposed without significantly affecting the decomposition of the WPCs and without promoting charring reactions. As a result, WPC/ATH-30 generated less residue than the other WPCs containing FR fillers.

Cone Calorimetry
Cone calorimetry was used to assess the forced combustion behavior of WPCs. Table 3 lists the corresponding data, including the ignition time (IT), peak heat release rate (PHRR), average heat release rate (AHRR), total heat release (THR), and residue. Heat release rate (HRR) is the most important parameter for evaluating the flame retardancy [53]. The HRR curves of WPCs are

Cone Calorimetry
Cone calorimetry was used to assess the forced combustion behavior of WPCs. Table 3 lists the corresponding data, including the ignition time (IT), peak heat release rate (PHRR), average heat release rate (AHRR), total heat release (THR), and residue. Heat release rate (HRR) is the most important parameter for evaluating the flame retardancy [53]. The HRR curves of WPCs are displayed in Figure 5. As can be observed, the neat WPCs without FR fillers showed the highest HRR with a prolonged PHRR of about 1000 kW/m 2 between 60 s and 120 s. It yielded a short IT of (15 ± 2) s, a THR of (170 ± 3) MJ/m 2 , and a residue of (6 ± 1) wt %. The addition of 30 wt % APP significantly delayed the IT to (23 ± 2) s, decreased the PHRR, AHRR, and THR to (560 ± 3) kW/m 2 , (164 ± 2) kW/m 2 , and (124 ± 2) MJ/m 2 , respectively, and increased the residue at the end of the tests to (19 ± 1) wt %. These results agreed with previously published reports [28,31,32]. When 10 wt % APP was replaced by ATH, a further improvement in performance was observed. The IT increased to (29 ± 2) s, the PHRR and THR decreased to (498 ± 3) kW/m 2 and (115 ± 2) MJ/m 2 , respectively, and the residue increased to (27 ± 2) wt %. This high residue (consistent with the TGA results) could act as an effective barrier layer on the sample surface as a result of the reaction between ATH and APP to generate a thermally stable aluminum-orthophosphate and aluminum-metaphosphate [42]. WPC/APP-20/ATH-10 showed the best flame-retardant performance between the WPCs investigated in this study. A decrease in performance was observed when more APP was replaced by ATH. displayed in Figure 5. As can be observed, the neat WPCs without FR fillers showed the highest HRR with a prolonged PHRR of about 1000 kW/m 2 between 60 s and 120 s. It yielded a short IT of (15 ± 2) s, a THR of (170 ± 3) MJ/m 2 , and a residue of (6 ± 1) wt %. The addition of 30 wt % APP significantly delayed the IT to (23 ± 2) s, decreased the PHRR, AHRR, and THR to (560 ± 3) kW/m 2 , (164 ± 2) kW/m 2 , and (124 ± 2) MJ/m 2 , respectively, and increased the residue at the end of the tests to (19 ± 1) wt %. These results agreed with previously published reports [28,31,32]. When 10 wt % APP was replaced by ATH, a further improvement in performance was observed. The IT increased to (29 ± 2) s, the PHRR and THR decreased to (498 ± 3) kW/m 2 and (115 ± 2) MJ/m 2 , respectively, and the residue increased to (27 ± 2) wt %. This high residue (consistent with the TGA results) could act as an effective barrier layer on the sample surface as a result of the reaction between ATH and APP to generate a thermally stable aluminum-orthophosphate and aluminum-metaphosphate [42]. WPC/APP-20/ATH-10 showed the best flame-retardant performance between the WPCs investigated in this study. A decrease in performance was observed when more APP was replaced by ATH.  The evolution of smoke is considered to be another important parameter for flame-retardant materials. Figure 6 presents the total smoke release (TSR) curves of WPCs. The related smoke property data are summarized in Table 4, including the yield of carbon monoxide (Yco), yield of carbon dioxide (Yco2), CO/CO2 weight ratio, and TSR. The higher this ratio, the greater the toxicity of the produced combustion products [54]. The incorporation of 30 wt % APP into WPCs significantly increased the TSR and CO/CO2 weight ratio by 57% and 387%, respectively. These two parameters continued to reduce when APP was replaced by more and more ATH. Noticeably, WPC/ATH-30 had an even lower TSR and CO/CO2 weight ratio than the neat WPCs without FR fillers. This is due to The evolution of smoke is considered to be another important parameter for flame-retardant materials. Figure 6 presents the total smoke release (TSR) curves of WPCs. The related smoke property data are summarized in Table 4, including the yield of carbon monoxide (Yco), yield of carbon dioxide (Yco 2 ), CO/CO 2 weight ratio, and TSR. The higher this ratio, the greater the toxicity of the produced combustion products [54]. The incorporation of 30 wt % APP into WPCs significantly increased the TSR and CO/CO 2 weight ratio by 57% and 387%, respectively. These two parameters continued to reduce when APP was replaced by more and more ATH. Noticeably, WPC/ATH-30 had an even lower TSR and CO/CO 2 weight ratio than the neat WPCs without FR fillers. This is due to the release of water vapor (generated by ATH decomposition) that diluted the combustion products and inhibited soot formation in the gas phase [55]. the release of water vapor (generated by ATH decomposition) that diluted the combustion products and inhibited soot formation in the gas phase [55].   Figure 7 shows the optical microscopy and SEM micrographs of the residues after cone calorimetry. The white color observed in the neat WPCs indicated the formation of ash and the absence of substantial carbonaceous residue on the surface of the sample. The WPC/APP-30 residue showed an expanded honeycomb-like structure with multiple cracks and holes on the surface. This loose char structure was not effective enough to act as a heat insulating barrier at high temperatures [56]. In comparison, WPC/APP-20/ATH-10 had a more compact and smoother carbonaceous residue (less holes and cracks on the surface) that can likely decrease the burning rate by generating a more effective heat and mass transfer on the surface of the sample. Therefore, WPC/APP-20/ATH-10 showed a lower PHRR and THR values than WPC/APP-30, as discussed above (Table 3). When more APP was replaced by ATH, the carbonaceous residues of WPCs became looser and less compact, with more holes and cracks on the surface. Finally, as indicated by the white color of the residue, WPC/ATH-30 showed the generation of an effective inorganic barrier without significant charring (as expected due to the absence of APP acting as charring agent).  Figure 7 shows the optical microscopy and SEM micrographs of the residues after cone calorimetry. The white color observed in the neat WPCs indicated the formation of ash and the absence of substantial carbonaceous residue on the surface of the sample. The WPC/APP-30 residue showed an expanded honeycomb-like structure with multiple cracks and holes on the surface. This loose char structure was not effective enough to act as a heat insulating barrier at high temperatures [56]. In comparison, WPC/APP-20/ATH-10 had a more compact and smoother carbonaceous residue (less holes and cracks on the surface) that can likely decrease the burning rate by generating a more effective heat and mass transfer on the surface of the sample. Therefore, WPC/APP-20/ATH-10 showed a lower PHRR and THR values than WPC/APP-30, as discussed above (Table 3). When more APP was replaced by ATH, the carbonaceous residues of WPCs became looser and less compact, with more holes and cracks on the surface. Finally, as indicated by the white color of the residue, WPC/ATH-30 showed the generation of an effective inorganic barrier without significant charring (as expected due to the absence of APP acting as charring agent). Raman spectroscopy has been applied to study the char structure and to further investigate the flame-retardant mechanism [57,58]. The D band (at around 1585 cm −1 ) is associated to the presence of amorphous carbon, whereas the G band (at around 1360 cm −1 ) is associated to the presence of graphitic carbon [59,60]. The integrated area ratio of the D band and G band (ID/IG) can be used to estimate the graphitization degree of the carbonaceous char (ID/IG ratio is inversely proportional to the degree of graphitization). Graphitic carbon is more thermally stable than amorphous carbon, so the higher the degree of graphitization, the more durable and effective the barrier effect (i.e., effectively prevent the heat and mass transfer between the burning matrix surface and the underlying material) of the char [58,61]. Figure 8 presents the Raman spectra of WPCs combustion residues after cone calorimetry. As can be seen in Figure 8, the ID/IG ratio values are 2.13, 2.43, 2.45, 2.82 for WPC/APP-20/ATH-10, WPC/APP-15/ATH-15, WPC/APP-30, WPC/APP-10/ATH-20, respectively. These data indicated that the combination of 20 wt % APP and 10 wt % ATH provided the WPCs with the highest degree of graphitization and the most thermally stable char structure. These results agree with the TGA and cone calorimetry tests, where WPC/APP-20/ATH-10 showed the highest residue, and might suggest that the level of graphitization plays a significant role in the flame-retardant mechanism. Raman spectroscopy has been applied to study the char structure and to further investigate the flame-retardant mechanism [57,58]. The D band (at around 1585 cm −1 ) is associated to the presence of amorphous carbon, whereas the G band (at around 1360 cm −1 ) is associated to the presence of graphitic carbon [59,60]. The integrated area ratio of the D band and G band (I D /I G ) can be used to estimate the graphitization degree of the carbonaceous char (I D /I G ratio is inversely proportional to the degree of graphitization). Graphitic carbon is more thermally stable than amorphous carbon, so the higher the degree of graphitization, the more durable and effective the barrier effect (i.e., effectively prevent the heat and mass transfer between the burning matrix surface and the underlying material) of the char [58,61]. Figure 8 presents the Raman spectra of WPCs combustion residues after cone calorimetry. As can be seen in Figure 8 To further understand the flame-retardant mechanism of APP and ATH in WPCs, the cone calorimeter residues were investigated by FTIR (Figure 9). The major absorption peaks of the neat WPCs were at 1795 cm −1 (C=O stretching), 1410 cm −1 (O-H bending), and the region between 1120 cm −1 to 650 cm −1 (C-H bending) [32]. The residues of the flame-retardant WPCs showed three additional features in the FTIR spectra. The first one is the presence of an absorption peak for WPC/APP-30 and WPC/APP-20/ATH-10 at 1288 cm −1 , which was attributed to the P-O stretching of decomposed APP. The second one is the presence of an absorption peak for WPC/APP-20/ATH-10 and WPC/ATH-30 at 1425 cm −1 , which was likely due to the Al2O3 produced by ATH decomposition. The last one is the absorption peak at about 715 cm −1 , which is a characteristic peak of aluminum metaphosphate [62]. This peak is obvious in WPC/APP-20/ATH-10, indicating that APP and ATH underwent a chemical reaction during the thermal decomposition process to generate aluminum metaphosphate. The formed aluminum metaphosphate had an ultra-high thermal stability [43] that could enhance the residue and barrier effect during the combustion process.  To further understand the flame-retardant mechanism of APP and ATH in WPCs, the cone calorimeter residues were investigated by FTIR (Figure 9). The major absorption peaks of the neat WPCs were at 1795 cm −1 (C=O stretching), 1410 cm −1 (O-H bending), and the region between 1120 cm −1 to 650 cm −1 (C-H bending) [32]. The residues of the flame-retardant WPCs showed three additional features in the FTIR spectra. The first one is the presence of an absorption peak for WPC/APP-30 and WPC/APP-20/ATH-10 at 1288 cm −1 , which was attributed to the P-O stretching of decomposed APP. The second one is the presence of an absorption peak for WPC/APP-20/ATH-10 and WPC/ATH-30 at 1425 cm −1 , which was likely due to the Al 2 O 3 produced by ATH decomposition. The last one is the absorption peak at about 715 cm −1 , which is a characteristic peak of aluminum metaphosphate [62]. This peak is obvious in WPC/APP-20/ATH-10, indicating that APP and ATH underwent a chemical reaction during the thermal decomposition process to generate aluminum metaphosphate. The formed aluminum metaphosphate had an ultra-high thermal stability [43] that could enhance the residue and barrier effect during the combustion process. To further understand the flame-retardant mechanism of APP and ATH in WPCs, the cone calorimeter residues were investigated by FTIR (Figure 9). The major absorption peaks of the neat WPCs were at 1795 cm −1 (C=O stretching), 1410 cm −1 (O-H bending), and the region between 1120 cm −1 to 650 cm −1 (C-H bending) [32]. The residues of the flame-retardant WPCs showed three additional features in the FTIR spectra. The first one is the presence of an absorption peak for WPC/APP-30 and WPC/APP-20/ATH-10 at 1288 cm −1 , which was attributed to the P-O stretching of decomposed APP. The second one is the presence of an absorption peak for WPC/APP-20/ATH-10 and WPC/ATH-30 at 1425 cm −1 , which was likely due to the Al2O3 produced by ATH decomposition. The last one is the absorption peak at about 715 cm −1 , which is a characteristic peak of aluminum metaphosphate [62]. This peak is obvious in WPC/APP-20/ATH-10, indicating that APP and ATH underwent a chemical reaction during the thermal decomposition process to generate aluminum metaphosphate. The formed aluminum metaphosphate had an ultra-high thermal stability [43] that could enhance the residue and barrier effect during the combustion process.

Evolved Gas Analysis
The volatiles produced during the anaerobic pyrolysis of WPCs were investigated by TGA-IR. Figure 10a displays the FTIR spectra at the maximum of mass loss rate. The FTIR characteristic peaks of the volatilized gases in the spectra agree with previously reported data and are assigned to the generation of H 2 O (3800 cm −1 to 3200 cm −1 ), hydrocarbons (2800 cm −1 to 3000 cm −1 ), CO 2 (2340 cm −1 ), and different carbonyl compounds (1750 cm −1 to 650 cm −1 ) [61]. The intensities of the characteristic absorption for hydrocarbons, CO 2 , and carbonyl compounds are shown in Figure 10b-d, respectively. As can be seen, the absorption intensity for hydrocarbons, CO 2 , and carbonyl compounds increased with the addition of APP and decreased with the addition of ATH in WPCs. A decrease in the generation rate of organic volatiles implies that there is less "fuel" available to feed flaming combustion at the peak of mass lose rate, thus, the PHRR might be reduced [61]. TGA-IR data clearly show the smoke suppression effect of ATH during combustion. The reduction of hydrocarbons can in fact suppress soot generation [63]. These data agree with the measured low TSR value and SDR value for the ATH-containing WPCs in the cone calorimetry and the smoke density tests, respectively.

Evolved Gas Analysis
The volatiles produced during the anaerobic pyrolysis of WPCs were investigated by TGA-IR. Figure 10a displays the FTIR spectra at the maximum of mass loss rate. The FTIR characteristic peaks of the volatilized gases in the spectra agree with previously reported data and are assigned to the generation of H2O (3800 cm −1 to 3200 cm −1 ), hydrocarbons (2800 cm −1 to 3000 cm −1 ), CO2 (2340 cm −1 ), and different carbonyl compounds (1750 cm −1 to 650 cm −1 ) [61]. The intensities of the characteristic absorption for hydrocarbons, CO2, and carbonyl compounds are shown in Figure  10b-d, respectively. As can be seen, the absorption intensity for hydrocarbons, CO2, and carbonyl compounds increased with the addition of APP and decreased with the addition of ATH in WPCs. A decrease in the generation rate of organic volatiles implies that there is less "fuel" available to feed flaming combustion at the peak of mass lose rate, thus, the PHRR might be reduced [61]. TGA-IR data clearly show the smoke suppression effect of ATH during combustion. The reduction of hydrocarbons can in fact suppress soot generation [63]. These data agree with the measured low TSR value and SDR value for the ATH-containing WPCs in the cone calorimetry and the smoke density tests, respectively.

Conclusions
In this study, different APP to ATH mass ratios were incorporated into wood-flour/polypropylene composites (WPCs). Water soaking tests indicated that the replacement of APP with ATH reduced the water absorption of WPCs, likely due to the decreased hygroscopicity. The flexural properties were not significantly affected by the APP to ATH mass ratio. The addition of 30 wt % APP and/or

Conclusions
In this study, different APP to ATH mass ratios were incorporated into wood-flour/polypropylene composites (WPCs). Water soaking tests indicated that the replacement of APP with ATH reduced the water absorption of WPCs, likely due to the decreased hygroscopicity. The flexural properties were not significantly affected by the APP to ATH mass ratio. The addition of 30 wt % APP and/or ATH decreased the flexural strength of the neat WPCs (no flame retardants) by about 15% and increased the flexural modulus by about 25%. The impact strength appeared to increase with the APP to ATH mass ratio decrease. No reduction in impact strength was observed in the WPCs containing 30 wt % ATH as compared to the neat WPCs. Cone calorimetry data showed that the partial replacement of APP with ATH decreased the heat release rate and smoke generation during combustion. Among all of the WPCs investigated here, WPC/APP-20/ATH-10 showed the lowest PHRR and THR values with a reduction of about 50% and 32%, respectively, as compared to the neat WPCs. WPC/APP-20/ATH-10 was also the sample with the best char (the highest residue with the most compact carbonaceous structure and the highest degree of graphitization) that could act as a physical barrier to protect the underlying matrix. Therefore, the formation of an effective physical barrier was likely playing a key role in terms of fire behavior. In conclusion, the data showed that the combination of APP and ATH at a proper mass ratio had a beneficial effect on smoke and heat generation of WPCs during forced combustion.