**2. Results**

## *2.1. Structural Characterization of DESG*

The preparation process of DESG consisted of two steps. In the first step, the P–H bond of DOPO and the double bond of EA underwent an addition reaction to prepare phosphatized EA (DEA). In the second step, the hydroxyl group of SG reacted with the anhydride group of DEA to obtain DESG. This process is shown in Figure 1.

The structures of SG, DEA and DESG were characterized using FTIR spectroscopy and 1H-NMR. Figure 2a,b show the 1H-NMR and FTIR spectra of SG, DEA and DESG, respectively. Compared with those of DOPO [31] and EA [32], in the 1H-NMR spectrum of DEA the P–H signal (at δ 8.82 ppm in DOPO) and the peak (at approximately δ 6.5 ppm) attributed to the double bond of EA disappeared. However, new absorption peaks located at δ 3.6 and 2.1~2.8 ppm appeared, which demonstrated the occurrence of an addition reaction between DOPO and EA. For DESG, the signal that appeared in the range of δ 3.9~5.2 ppm was from SG; peaks at approximately δ 7~8 ppm could be assigned to the biphenyl group in DOPO, indicating that DEA was successfully grafted onto SG. All of this can also be confirmed in the FTIR spectra of DEA and DESG. For DEA, the stretching vibration band of double bond at 1628 cm<sup>−</sup><sup>1</sup> and the characteristic peak of P–H at 2436 cm<sup>−</sup><sup>1</sup> vanished. In the FTIR spectrum of DESG, the absorption peak of the anhydride group at 1864 cm<sup>−</sup><sup>1</sup> disappeared, but the absorption peaks of P–O–Ph (1159, 912, 758 cm<sup>−</sup>1), P=O (1232, 1208 cm<sup>−</sup>1) and P–Ph (1596, 1478 cm<sup>−</sup>1) [33,34] were still present.

**Figure 2.** SG and DESG curves: (**a**) 1HNMR, (**b**) FTIR, (**c**) TGA and (**d**) DTG.

The TGA and DTG curves of SG and DESG are shown in Figure 2c. Thermal stabilities were determined from the temperature at 5% weight loss, which was defined as the onset degradation temperature. From Figure 2c, we can see that the weight loss of SG occurred in two steps, including dehydration and thermolysis processes. The onset degradation temperature of SG was approximately 237.6 ◦C. DESG underwent only one stage of thermal degradation due to the thermolysis of the polysaccharide component. The onset degradation temperature shifted to a higher value (282.5 ◦C) compared with that of SG, which indicated that the introduction of the phosphorus element improved the early thermal stability of DESG. However, it can be found from the DTG curve that the maximum degradation rate temperature of DESG dropped approximately 28 ◦C. This was because the

high temperature broke down the phospholipid bonds and the resulting acidic substances promoted the decomposition of SG, leading to a decrease in the maximum decomposition temperature. The final char residue at 800 ◦C was 33.5%, which was much higher than that of SG (11.0%); therefore, the high char residue and speed of carbon formation make DESG a potentially excellent carbon source for an IFR.

#### *2.2. Flame Retardant Properties of DESG/APP on Polylactic Acid*

DESG was combined with APP and then added to PLA, resulting in PLA-based composites with flame retardancy. Here, DESG was used as a carbon source and APP acted as both an acid source and a gas source. The flame retardancy of PLA composites was measured using UL-94 and LOI. The results are given in Table 1. The neat PLA sample showed a LOI value of 19.5%. For binary composites with 20 wt% SG or DESG added to PLA alone, the LOI did not change significantly. Similarly, SG or DESG alone were not helpful in upgrading the UL rating for PLA substrate. The incorporation of APP alone improved the LOI value of the A20 binary composite to 26.2%. However, the molten droplets produced ignited the degreasing cotton; the composite achieved only a UL V-2 rating. For a ternary composite, A12S8, containing 12 wt% APP and 8 wt% SG, the LOI value increased to 29.6%, but the UL rating was still V-2. This indicated that the synergistic flame retardancy of nonphosphatized SG and APP on PLA was weak. While 12 wt% APP and 8 wt% DESG were introduced into PLA, the A12D8 composite achieved a LOI value of 32.2% and a V-0 rating. The improvement in flame-retardant properties was due to the synergy effect between APP and DESG. As a result of this synergy, adding only 10 wt% flame-retardant, like A7D3 or A5D5, made the composite reach the V-0 rating. The effect of the addition ratio of APP and DESG on the flame-retardant properties of PLA composites was also investigated. The results from Table 1 show that the LOI value of the composites increased gradually and then decreased with an increase in APP content.


**Table 1.** Formulations and flammability of PLA and flame-retardant PLA composites.

NR = no rating.

a

In order to further investigate the synergistic effect of DESG and APP on PLA, a cone calorimeter (CONE) test, which is used to simulate the burning behavior of materials in a real fire, was conducted. The heat release rate (HRR) and total heat release (THR) curves of PLA and PLA composites are shown in Figure 3. In Figure 3a, one can see that neat PLA presented the largest peak heat release rate (pHRR) of 523.5 kW/m2. The pHRR value of D20 and A20 were 501.6 and 482.5 kW/m2, respectively. This indicated that the addition of DESG or APP alone cannot effectively reduce the HRR value. However, the pHRR value of the A12D8 sample decreased to 341.6 kW/m2, exhibiting a 34.7% decrease compared to that of neat PLA. The HRR curve of A12D8 showed a double peak, while only a single

peak was observed for other samples. This bimodal phenomenon was characteristic of the intumescent flame-retardants [29,35]. As shown in the THR curves of PLA and PLA composites (Figure 3b), the THR values of neat PLA, D20 and A20 were 78.1, 72.4 and 65.8 MJ/m2, respectively. Similarly, although adding DESG or APP alone reduced the THR value compared with PLA, it was not as effective as adding both. The A12D8 sample showed the lowest THR value, 52.7 MJ/m2, which was a 32.5% reduction compared to that of neat PLA.

**Figure 3.** PLA, D20, A20 and A12D8 curves: (**a**) HRR and (**b**) THR.

Compared with the results reported for polysaccharide-added PLA flame-retardant composites [26,36], A12D8 composites achieved a higher LOI value as well as lower pHRR and THR values. For A12D8 composites, a V-0 rating can be achieved by adding only 10 wt% flame-retardants; however, 15–25 wt% was added in some studies [28,35,37].

## *2.3. Thermal Stabilities of Flame-Retardant PLA Composites*

TGA was employed to evaluate the thermal stabilities of flame-retardant PLA composites. TGA and DTA thermograms of PLA and flame-retardant PLA composites in a nitrogen atmosphere (from 100 to 650 ◦C) are illustrated in Figure 4a,b. The initial decomposition temperature (Ti), defined as the 5% weight loss temperature, the temperature of maximal weight loss rate (Tmax) and char yields at 400 and 600 ◦C are summarized in Table 2. For PLA, Ti and Tmax were 309.4 ◦C and 338.7 ◦C, respectively. At approximately 400 ◦C, its decomposition was essentially complete; the char yield after 400 ◦C under N2 was only 0.3%. As a result of SG's low decomposition temperature, the Ti of D20 decreased to 282.5 ◦C, with a char yield of 4.1% at 600 ◦C. The Ti of A20 was 296.5 ◦C and its char yield significantly increased at 400 ◦C. This occurred because the APP was decomposed in advance to generate phosphoric acid or polyphosphoric acid, which catalyzed the char formation of PLA itself. However, the char layer was unstable at high temperatures and underwent secondary decomposition at approximately 500 ◦C. For the A12D8 sample, thermal stability was higher after 300 ◦C than it was when adding APP or DESG alone; the presence of APP improved the thermal stability of DESG at a low temperature. Moreover, the weight loss of DESG at a low temperature was reduced compared with other samples. The char yield of A12D8 was as high as 32.1% and 16.8% at 400 ◦C and 600 ◦C, respectively. These results demonstrated the existence of the synergistic effect of APP and DESG combined in promoting char formation.


**Table 2.** TGA and DTG data for PLA, D20, A20 and A12D8.

**Figure 4.** PLA, D20, A20 and A12D8 thermograms: (**a**) TGA and (**b**) DTG.

#### *2.4. Dispersion of DESG/APP Flame-Retardant in PLA Matrix*

To investigate the dispersion of DESG/APP flame-retardants in the PLA matrix, FESEM images of the fracture surfaces of D20, A20 and A12D8 samples are shown in Figure 5. It can be seen in Figure 5a,b that the fracture surfaces of the S20 and A20 samples were rough; there was an obvious phase interface between the SG or APP particles and the PLA matrix, indicating poor compatibility between the dispersed phase and the matrix. On one hand, because APP and SG are hydrophilic, there was a polarity difference between them and the hydrophobic PLA matrix. On the other hand, both APP and SG have strong intermolecular forces and are prone to agglomeration, which resulted in their poor dispersibility in PLA. After SG was phosphorylated by DEA, the hydrophobicity of SG improved significantly. Additionally, the carboxyl groups still existing on DESG could undergo esterification with the terminal hydroxyl groups of PLA during processing, which further improved the compatibility of DESG and PLA [38,39]. Therefore, homogeneous dispersion was found in the FESEM micrograph of the A12D8 sample (Figure 5c). In addition, the introduction of DESG also enhanced the compatibility of APP with the PLA matrix. This may be due to the reaction between the hydroxyl groups on DESG and the carboxyl groups on APP, which promoted better dispersion of the flame-retardant in the PLA matrix and further improved the synergistic flame-retardant efficiency of the flame-retardants.

**Figure 5.** FESEM micrographs of PLA composites: (**a**) S20, (**b**) A20 and (**c**) A12D8.

It can be seen from the tensile properties of PLA and its composites (as shown in Table 3), that adding 20% DESG (sample D20) or APP (sample A20) alone significantly reduced the tensile strength and Young's modulus of the composites, which ranged from 67.6 to 48.1 MPa and from 1.75 to 1.08 GPa, respectively. This indicated that the interfacial compatibility between DESG or APP and PLA was very poor, resulting in a serious decline in the mechanical properties of composites. For A20, the elongation at break increased slightly, which was attributed to the dilution effect of APP on the PLA molecular chain [19,40]. After adding both DESG and APP (sample A12D8), the decrease in the strength and modulus of the composites was effectively alleviated; it was speculated that the incorporation of DESG could improve compatibility between the dispersed phase and the PLA matrix and reduce the influence on the mechanical properties of the composites.


**Table 3.** Tensile properties of PLA and its composites.

## *2.5. Flame-Retardant Mechanism*

To understand the flame-retardant mechanism in the condensed phase, the Raman spectra of char residues of D20, A20 and A12D8 samples after cone calorimeter tests were investigated; the results are shown in Figure 6. The vibrations of disordered carbon and graphitic carbon were detected near 1340 cm<sup>−</sup><sup>1</sup> and 1585 cm<sup>−</sup>1, which correspond to the D and G bands, respectively. The degree of graphitization was assessed using the value of the area ratio of D to G bands (AD/AG); a lower value suggested a higher graphitization degree of char layer, better shielding effect and higher thermal stability [41,42]. The AD/AG values of D20, A20 and A12D8 samples were 3.1, 2.9 and 2.3, respectively. The A12D8 sample had a lower AD/AG value compared to the others, indicating that the addition of both DESG and APP facilitated forming more graphitized char residue with a shielding effect for PLA.

**Figure 6.** Raman spectra of (**a**) D20, (**b**) A20 and (**c**) A12D8 after cone calorimeter test.

Figure 7 presents a possible synergistic mechanism of the DESG/APP flame-retardant. When the composite material was heated and burned, APP, which was an acid source and a gas source, first decomposed to generate ammonia and polyphosphoric acid (or polymetaphosphoric acid). Ammonia, a non-flammable gas, could have a diluting effect in the gas phase. An esterification reaction between hydroxyl groups in the DESG molecules and polyphosphoric acid might occur. Meanwhile, the polyphosphoric acid, with strong dehydration, can accelerate the carbonization of the ester to form a phosphorus–carbon char layer containing cross-linked structures of P–O–C and P–O–P. With a continuous increase in the external temperature, the cross-linked structures then underwent various chemical reactions such as dehydrogenation, carbonization, breakage of the chemical bond, etc., to form char residues rich in phosphorus and carbon. These char residues appeared macroscopically as black carbonaceous skeletons. The water vapor generated by the reaction and some incombustible gases (such as NH3) swelled the char layer, forming an intumescent protective layer. This intumescent layer, with high strength, good compactness and excellent thermal stability, covered the surface of the PLA matrix to produce a fine shielding effect, which further delayed the degradation rate of the material through effective thermal insulation and oxygen insulation and improved the flame-retardant performance. It is noteworthy that the esterification of DESG and polyphosphoric acid caused the initial weight loss of the composite, but at the same time a more stable carbon layer formed. This also explains why A12D8 had a 5% weight loss at a lower temperature of 273.0 ◦C, but less weight loss and a higher char residue at a higher temperature.

**Figure 7.** A possible synergistic mechanism of DESG/APP flame-retardant.

## **3. Materials and Methods**
