Effect of Exogenous Carboxyl and Hydroxyl Groups on Pyrolysis Reaction of High Molecular Weight Poly(L-Lactide) under the Catalysis of Tin

The effect of exogenous hydroxyl, carboxyl groups and/or Sn2+ on pyrolysis reactions of poly(L-lactide) (PLLA) was investigated by thermogravimetric analysis (TGA). The activation energy (Ea) of pyrolysis reactions was estimated by the Kissinger-Akahira-Sunose method. The kinetic models were also explored by the Malek method, and the random degradation behavior was determined by comparing the plots of In{-In[1 - (1 - w)0.5]} versus 1/T for experimental data from TGA with model reactions. The pyrolysis reaction rate of PLLA was affected slightly by exogenous hydroxyl and carboxyl groups at lower levels of Sn with 65–70 mg·kg-1 but increased appreciably in the presence of extraneous Sn2+, -COOH/Sn2+, or -OH/Sn2+. The Ea values for the pyrolysis reactions of the PLLAs that provided lactide were different under the catalysis of Sn2+ in different chemical environments because Sn2+ can form the new Sn-carboxylate and Sn-alkoxide with exogenous carboxyl and hydroxyl groups, which were different in steric hindrance for the formation of activated complex between Sn2+ and PLLA. Under the catalysis of Sn2+, a lactide molecule can be directly eliminated selectively at a random position of PLLA molecular chains, and the molecular chain of PLLA cannot change two PLLA fragments at the elimination site of lactide. However, it was regenerated into a new PLLA molecule with the molecular weight reduced by 144 g·mol-1.


INTRODUCTION
Poly(lactic acid) (polylactide, PLA), as a well-known biodegradable polymer, has wide applications in the fields of medical, pharmaceutical and environmentally-friendly polymeric materials. High molecular weight PLA is usually prepared by the ring-opening polymerization of L-, D-, and meso-lactides. The ring-opening polymerization is a reversible reaction, so the content of lactide monomer is temperature-dependent. [1] Therefore, the thermal depolymerization of PLA can also regenerate lactides. The actual thermal degradation of PLA involves a complex reaction process in relation to the pyrolysis conditions and can be enhanced in the presence of active groups, active chain end-groups, residual monomers, residual catalysts, inorganic/organic fillers, plasticizers, H 2 O and other impurities. Furthermore, in addition to lactide as the main product, the thermal degradation of PLA can generate linear and cyclic oligomers, diastereoisomers of the oligomers, CO, CO 2 , methylketene and acetaldehyde. [2−5] The thermal depolymerization or stability of PLA is an extensive concern because of its direct relevance to production, processing, application, thermal recycling and complete life cycle. [6−8] Several techniques have been used to improve the thermal stability of PLA. [9] For example, residual monomers are strictly removed after polymerization; [10] active -OH end-groups are capped with anhydrides or isocyanates; [11,12] some stabilizers such as phosphites (i.e., tris(nonylphenyl) phosphite (TNPP) and triphenyl phosphite (TPP)), epoxides (i.e., JONCRYL®ADR-4370-S, BASF), and carbodiimide are used; [9,13−15] PLA resin is dried strictly to remove moisture since PLA, as a type of aliphatic polyester, is sensitive to moisture. [16] The thermal degradation of PLA has been investigated by many technologies, such as thermogravimetric analysis (TGA), gel permeation chromatography (GPC), [6,8,17] TGA/Fourier transform infrared spectroscopy (TGA-FTIR), [18,19] and pyrolysis gas chromatography/mass spectrometry (Py-GC/MS). [7,19,20] The pyrolysis process of PLA has been proved as random main chain scission and unzipping depolymerization reac-tions. The random degradation reaction involves hydrolysis, oxidative degradation, cis-elimination, and intra-/inter-molecular transesterification. [21] The unzipping depolymerization reaction mainly involves a selective elimination reaction for lactide under the catalysis, such as Sn 2+ . [3,22,23] The degradation behavior is different between pure and Sn-containing PLAs, [3] and is affected by other compounds of metals, such as Sn, Zn, Al and Fe. [24] The TGA curves of poly(L-lactide) (PLLA)-Sn samples with an increase in Sn content shift to a lower degradation temperature range, and a decrease in activation energy is observed. [23] The pyrolysis reaction of PLA results in molecular weight reduction by random scissions of molecular chains [25,26] and in weight-loss by release of volatile products.
In the present work, we investigated the effect of exogenous hydroxyl, carboxyl groups, and/or Sn 2+ on the pyrolysis reaction of PLLA. Exogenous hydroxyl, carboxyl groups, and/or Sn 2+ were introduced into high-molecular-weight PLLA matrix by solution blending. These PLLAs were depolymerized by TGA to obtain the kinetic parameters and mechanism of the pyrolysis reaction, that is, the high-molecularweight PLLA provided lactide in the presence of -COOH, -OH, and/or Sn 2+ .

EXPERIMENTAL Instruments and Materials
TGA was conducted on a TA instrument TGA-Q50. A Ni standard reference was used for the temperature correction of TGA. The nitrogen flow rate was 40.0 mL·min -1 through the balance and 60.0 mL·min -1 over the sample. The mass of the loaded sample was 4−6 mg in a platinum pan. Inductively Coupled Plasma-Mass Spectrometer (ICP-MS) (XSERIES 2, Thermo Fisher Scientific, US) was used to quantify the Sn content in the samples. The sample was decomposed with HNO 3 (65 wt%) in a sealed microwave digestion vessel at 120 °C for 6 h and then used to determine the Sn content. The high-molecular-weight PLLA (Revode 190, Optical purity 99.5%) was obtained from Zhejiang Hisun Biomaterials Co., Ltd., Taizhou, Zhejiang, China. The number-average, weight-average molecular weights (M n , M w ), and molecular weight polydispersity index (PDI) of the polymer are M n =84.1 kg·mol −1 , M w =155.9 kg·mol −1 , and PDI=1.85, respectively. Stannous octanoate (Sn(Oct) 2 , 95%) and stearic acid (CH 3 (CH 2 ) 16 COOH) (STA, 98%) were purchased from Shanghai Aladdin Biochemical Technology Co., Ltd. (Shanghai, China). Poly(ethylene glycol) (PEG-600) (CP) with di-hydroxyl end-groups was produced from Xilong Scientific Co., Ltd. (Guangdong, China). Dichloromethane (AR) was made in China and used directly without further purification.

Kinetic parameters
The kinetic parameters, including activation energy (E a ) and pre-exponential factor (lgA), of the pyrolysis reactions of PLLAs were calculated by the Kissinger-Akahira-Sunose (KAS) method, [22,27,28] as shown in Eq.
(1). The relationship between and 1/T is a straight line at a certain conversion (α) for a set of TG curves at different heating rates (β), and the values of ln ( β where β is the heating rate (K·min −1 ), A is the pre-exponential factor, E a is the activation energy, R is the gas constant, and T is the absolute temperature (K).

Mechanism model
The Malek method [22,29−34] was used to explore the most probable mechanism model of pyrolysis reactions, as shown in Eqs. (2) and (3). The most probable kinetic model functions and can be inferred by comparing the shapes of and .
dα/dT f(α) and G(α) T 0.5 (dα/dT) 0.5 f(0.5) , and G(0. 5) where and are the standard model function and the experimental curve; α is the fractional conversion; T and are the absolute temperature (K) and the temperature differential; are the differential and integral forms for the most probable mechanism functions; , , are the values of the absolute temperature, the temperature differential, the differential, and integral forms of the most probable mechanism functions at α=0.5 as the reference point.

Random scission kinetic analysis
Random scission behavior was determined by the plots of versus 1/T for model reactions and experimental data from TGA for polymers in the initial stage of the pyrolysis reactions (w=0.99−0.85) according to the reference. [22] In the equation, w and T are the residual weight fraction and the absolute temperature (K), respectively.

Reactions of Sn(Oct) 2 with Hydroxyl and Carboxyl Groups
The volatility of 2-ethylhexanoic acid (Oct) is far higher than that of STA, since the boiling point of Oct is far lower than that of STA, which are 228 and 383 °C at 101.325 kPa, respectively. PEG is a non-volatile polymer. In the heating process, the double decomposition reactions of Sn(Oct) 2 with the carboxyl and hydroxyl groups of STA and PEG-600 can occur and thus form Sn-carboxylate and Sn-alkoxide, as well as Oct released in the form of gas at high temperatures, as shown in Fig. 1. The reactions were further confirmed using TGA. The TG thermograms of Sn(Oct) 2 , STA, PEG-600, Sn(Oct) 2 /STA (3/7) and Sn(Oct) 2 /PEG-600 (3/7) are shown in Fig. 2. In the main stage of thermal weight-loss, the weight-loss temperature for STA was higher than that for Sn(Oct) 2 and lower than that for PEG-600 at an identical percentage of weight-loss. In the heating process, the weight-loss was mainly due to the volatilization and decomposition for Sn(Oct) 2 , the volatilization for STA, and the decomposition for PEG-600. However, the shape of the TG curves for the blends of Sn(Oct) 2 with STA or PEG-600 can testify the reactions between the blending components. The weightloss temperatures of the Sn(Oct) 2 /PEG-600 (3/7) blend or the Sn(Oct) 2 /STA (3/7) blend were between Sn(Oct) 2 and PEG-600 or STA in the initial period of the TG curves but were higher than those for PEG-600 or STA in the middle and later period of the TG curves. The weight-loss temperatures of TG curves for the binary blends were higher than those for any component, indicating that the two components can react to form new products, which were more difficult to volatilize or decompose at higher temperatures. Therefore, the Sn-carboxylate (Oct-Sn-STA and Sn(STA) 2 ) and the Sn-alkoxide (Oct-Sn-PEG, SnPEG, PEG-Sn-PEG and (SnPEG) m ) can be produced through the reactions of Sn(Oct) 2 with STA and PEG at high temperatures ( Fig. 1). These Sn salts were more difficult to volatilize or decompose than Sn(Oct) 2 .
The thermal weight-loss properties of Sn(Oct) 2 , STA, PEG-600, the Sn(Oct) 2 /STA (3/7) blend and the Sn(Oct) 2 /PEG-600 (3/7) blend are shown in Table 1. The Sn-alkoxide can be produced through the reaction of PEG with Sn(Oct) 2 , and the mo- blend respectively, at 425 °C. These data indicated that the thermal chemical reaction of Sn(Oct) 2 and PEG-600 was complicated and that the coking and carbonization of the Sn(Oct) 2 /PEG-600 (3/7) blend were relatively significant, thus the residual percentage reached 20 wt% for the Sn(Oct) 2 /PEG-600 (3/7) blend at 425 °C. Regardless of the products obtained from the reaction of Sn(Oct) 2 and PEG-600, one or several stable Sn salts were formed in the heating process of the Sn(Oct) 2 /PEG-600 blend, thereby effectively inhibiting the volatilization loss of Sn 2+ at high temperatures.

Pyrolysis Reaction of PLLA
This work aimed to study the pyrolysis process and mechanism of selective lactide elimination for high-molecular-weight PLLA under the catalysis of Sn 2+ , and to study the effect of extraneous -OH and -COOH on the pyrolysis reaction. These PLLAs contained Sn in the range of 65−380 mg·kg −1 , and thus they showed a preceding selective depolymerization step, which produced lactide exclusively according to previous research work. [3,22,35] The thermogravimetric and derivative thermogravimetric (TG/DTG) curves are shown in Fig. 3, and the thermal decomposition properties are shown for PLLAs with extraneous -OH, -COOH, and/or Sn 2+ in Table 2. Compared with raw PLLA, the initial thermal decomposition temperature reduced slightly for PLLA(-COOH) and PLLA(-OH) with merely extraneous -COOH and -OH. However, at the middle and later stages (α≥0.20), thermal decomposition properties were consistent for raw PLLA, PLLA(-COOH) and PLLA(-OH), because their TG/DTG curves coincided roughly. Nevertheless, the thermal decomposition temperature decreased considerably for the PLLAs with higher Sn contents of 349−380 mg·kg −1 , including PLLA(Sn 2+ ), PLLA(-COOH, Sn 2+ ) and PLLA(-OH, Sn 2+ ), compared with the corresponding PLLAs without extraneous Sn 2+ , including raw PLLA, PLLA(-COOH), and PLLA(-OH). In the starting pyrolysis reaction stage, for the PLLAs with extraneous Sn 2+ , the temperature for 5% weight-loss (T 5% ) decreased in the order PLLA(Sn 2+ )> PLLA(-OH, Sn 2+ )>PLLA(-COOH, Sn 2+ ). However, in the main pyrolysis reaction stage, the temperature for 50% weight-loss (T 50% ) decreased in the order PLLA(Sn 2+ )>PLLA(-COOH, Sn 2+ )>PLLA(-OH, Sn 2+ ). These phenomena indicated that at  lower Sn levels of 65-70 mg·kg −1 , exogenous hydroxyl and carboxyl groups can slightly affect the reaction rate of PLLA thermal decomposition. However, the pyrolysis reaction rate of PLLA increased appreciably in the presence of extraneous Sn 2+ , -COOH/Sn 2+ , or -OH/Sn 2+ . According to the DTG curves, PLLA(Sn 2+ ) and PLLA(-COOH, Sn 2+ ) contained at least two steps of decomposition, given that the double or multiple peaks of DTG curves appeared evidently, but the decomposition process was relatively simple for raw PLLA, PLLA(-COOH), PLLA(-OH) and PLLA(-OH, Sn 2+ ) due to the single peak of DTG curves.
The thermal analysis kinetics for PLLAs was evaluated by the KAS method. The TG thermograms of the PLLAs with exogenous -OH, -COOH and/or Sn 2+ at different heating rates are shown in Fig. S1 (in the electronic supplementary information, ESI). The relationships of E a and lg A versus α of the pyrolysis reactions for these PLLAs are shown in Fig. 4. In our previous work, [22] we found that the low-molecularweight PLAs (L-PLA) containing merely carboxyl or hydroxyl end-groups (R-L-PLA-COOH with M w =13.6 kg·mol −1 or R-L-PLA-OH with M w =18.8 kg·mol −1 ) had an approximately identical E a value of pyrolysis reactions at a certain conver-sion for α=0.1−0.40 under the catalysis of Sn 2+ (497 and 461 mg·kg −1 ). R-L-PLA-COOH and R-L-PLA-OH can form Sncarboxylate and Sn-alkoxide chain-ends. Thus, the pyrolysis reaction of the L-PLAs can selectively produce lactide through the backbiting reaction caused by the Sn-carboxylate and Snalkoxide chain-ends. The E a values were equal for the pyrolysis reactions that PLA selectively produce lactide through backbiting reactions from the carboxyl and hydroxyl endgroups. In the present work, the exogenous -OH, -COOH, and/or Sn 2+ from PEG-600, STA, and/or Sn(Oct) 2 were introduced into the PLLA matrix, and the formed Sn-carboxylate and Sn-alkoxide were located outside the molecular chains of PLLAs. Thus, they can produce different effects on the pyrolysis reaction of PLLA, compared with those located at the chain-ends of PLLAs.
For raw PLLA, PLLA(-COOH), and PLLA(-OH) with lower Sn levels of 65−70 mg·kg −1 , the E a values of pyrolysis reactions decreased in the order of PLLA(-COOH)>PLLA>PLLA(-OH) at the initial stage (α=0.05−0.10) because the Sn 2+ was located at the end of the PLLA molecular chains (PLLA-O-CO-CH(CH 3 )-O-Sn-X), and the depolymerization of PLLA primarily occurred through the backbiting reaction. The -COOH introduced from STA could snatch Sn 2+ located at the end of the PLLA molecular chains and generate a small amount of STA-Sn-X. Then, STA-Sn-X formed an activated complex transition state with PLLA. Finally, the depolymerization of PLLA occurred, slightly increasing the pyrolysis activation energy. The -OH introduced from PEG can also snatch Sn 2+ located at the end of the PLLA molecular chains and generate PEG-Sn-X, but the snatching Sn 2+ competitiveness for -OH was weaker than that for -COOH. Thus, the generated trace PEG-Sn-X had no significant effect on the activation energy of thermal degradation. The reduced activation energy of PLLA(-OH) with PEG content of 100 mmol·kg −1 (3.0 wt%) was primarily due to the pyrolysis effect of PEG in PLLA. T 5% s of PEG and PLLA are 266.7 and 326.8 °C, respectively (Tables 1 and 2), and thus PEG could decompose at lower temperature than PLLA. The activation energy of PEG should be further lower than that of PLLA, thus might reducing the activation energy of pyrolysis reaction of PLLA(-OH).
For PLLA(Sn 2+ ), PLLA(-COOH, Sn 2+ ) and PLLA(-OH, Sn 2+ ), the chemical environment of Sn 2+ was located as follows: Sn(Oct) 2 , Sn(Oct) 2 /Oct-Sn-STA/Sn(STA) 2 and Sn(Oct) 2 /Oct-Sn-PEG/SnPEG/PEG-Sn-PEG/(SnPEG) m (Fig. 1). In the early stage of pyrolysis, the produced Oct-Sn-PEG, SnPEG, PEG-Sn-PEG, and (SnPEG) m were very little for PLLA(-OH, Sn 2+ ), which had no evident effect on the activation energy. In addition, T 5% of PLLA(-OH, Sn 2+ ) was 251.2 °C, less than that of PEG (266.7 °C), so the pyrolysis of PEG (3 wt%) has little contribution to the E a values of the blend. Therefore, PLLA(Sn 2+ ) and PLLA(-OH, Sn 2+ ) had approximately identical E a values at the initial stage of pyrolysis. The initial activation energy for PLLA(-COOH, Sn 2+ ) was relatively lower than that of PLLA(Sn 2+ ), because Oct in Sn(Oct) 2 contained the α-substituted branched structure, which could produce a steric hindrance to the formation of activated complex between Sn 2+ and PLLA. However, in the presence of STA, Sn(Oct) 2 could be partially converted into Oct-Sn-STA and Sn(STA) 2 . STA is a linear molecule, and the generated Oct-Sn-STA and Sn(STA) 2 could reduce the steric hindrance to the formation of activated complex between Sn 2+ and PLLA. Therefore, the activation energy of pyrolysis reaction was much lower for PLLA(-COOH, Sn 2+ ) than that for PLLA(Sn 2+ ). In the later stage of pyrolysis reaction, the loss of Sn 2+ in PLLA(Sn 2+ ) and PLLA(-COOH, Sn 2+ ) increased with increasing temperature. Thus, the activation energy increased in the later stage of the reaction. However, the loss of Sn 2+ in PLLA(-OH, Sn 2+ ) was the lowest because the formed Oct-Sn-PEG, SnPEG, PEG-Sn-PEG and (SnPEG) m were more difficult to volatilize and decompose. Moreover, PEG has no branched structure, and the steric hindrance to the formation of activated complex between Sn 2+ and PLLA was lower for Oct-Sn-PEG, SnPEG, PEG-Sn-PEG, and (SnPEG) m than that for Sn(Oct) 2 ; thus, the activation energy decreased.

Pyrolysis Kinetic Model
PLA can selectively produce lactide under the catalysis of Sn 2+ , and the most probable kinetic model can be identified by the Malek method. The plots of the standard model functions for common kinetic models, , and the experimental curves, , versus conversion (α) for PLLAs are shown in Fig. 5. The differential and integral forms [ and ] of the common mechanism models for pyrolysis processes are shown in the supplementary data of the reference. [22] When a pyrolysis process can be described by a single kinetic model, should be overlapped with . However, when a pyrolysis process is controlled by two or multiple mechanism functions, the experimental curve must deviate from any standard model function.
for raw PLLA, PLLA(-COOH), and PLLA(-OH) overlapped roughly with of Nos. 10−20 at α<0.95. However, for PLLA(Sn 2+ ), PLLA(-COOH, Sn 2+ ), and PLLA(-OH, Sn 2+ ) was away from all the standard model functions. These phenomena indicated that the pyrolysis processes of PLLA, PLLA(-COOH) and PLLA(-OH) with the Sn content of 65−70 mg·kg −1 tended to be described by a single kinetic model. However, the pyrolysis processes of PLLA(Sn 2+ ), PLLA(-COOH, Sn 2+ ) and PLLA(-OH, Sn 2+ ) with the Sn content of 349−380 mg·kg −1 must be controlled by not less than two processes. For PLLA(Sn 2+ ), PLLA(-COOH, Sn 2+ ) and PLLA(-OH, Sn 2+ ), the initial Sn contents were very high, and in the pyrolysis processes, the Sn contents evidently decreased with temperature due to the volatilization of Sn(Oct) 2 (boiling point 228 °C at 101.325 kPa). Moreover, the Sn 2+ in the PLLAs had different losses due to the different volatility of Sn 2+ under different chemical environments. According to the previous section, the reduction in Sn content was maximum for PLLA(Sn 2+ ) and minimum for PLLA(-OH, Sn 2+ ). Thus, the pyrolysis mechanism and processes were more complicated in the later stage of the decomposed reactions for PLLA(Sn 2+ ), PLLA(-COOH, Sn 2+ ) and PLLA(-OH, Sn 2+ ). However, the pyrolysis processes were unaffected in the initial stage of the reactions, so the pyrolysis mechanism could be investigated for the polymers in the presence of Sn 2+ , -COOH/Sn 2+ , and -OH/Sn 2+ .

Evaluation of Random Scission
The cleavage of bonds in random scission processes and the n thorder model reactions (half-, zero-, first-, and second-order reactions) of PLLA were evaluated using the plots of versus 1/T for the experimental data from TG with model reactions, as shown in Fig. 6. For random degradation, the least number of repeating units of the residual polymer (L) should be variable for the model simulations. [39] Compared with the model reactions, the degradation behavior of these PLLAs did not present an n th order (n=0.5, 0, 1, and 2) weight-loss process and a random scission process (L=2−10) in the initial stage (w=0.99−0.85). Therefore, the random mainchain scission for raw PLLA, PLLA(-COOH), PLLA(-OH), PLLA(Sn 2+ ), PLLA(-OH, Sn 2+ ) and PLLA(-COOH, Sn 2+ ) did not occur in the initial pyrolysis stage. Therefore, under the catalysis of Sn 2+ , an activated complex center could be formed between PLLA and Sn 2+ , a lactide molecule was eliminated selectively at a random position of PLLA molecular chains, and the PLLA molecular chain was not broken but regenerated into forming a new PLLA with the molecular weight reduced by 144 g·mol −1 (molecular weight of lactide).

Mechanism of PLLA Pyrolysis
Under the catalysis of Sn 2+ , PLLA can be decomposed for producing lactide. According to previous research results, PLLA can form Sn-carboxylate and Sn-alkoxide from the PLLA molecular chain end-groups. Thus, it can continuously generate lactide through the backbiting reaction or directly eliminate a lactide molecule at a random position of PLLA molecular chains. However, to this day, when a lactide molecule is eliminated directly at a random position of PLLA molecular chains, whether the molecular chains of PLLA were broken remains unknown. We found that after the selective lactide elimination at a random position of PLLA molecular chains, the molecular chain of PLLA could neither break nor form two PLLA fragments at the elimination site of lactide but could be regenerated into a new PLLA molecule with the molecular weight reduced by 144 g·mol −1 (molecular weight of lactide), as shown in Fig. 7.

CONCLUSIONS
Exogenous hydroxyl, carboxyl groups, and/or Sn 2+ were introduced into high-molecular-weight PLLA by solution blending with PEG-600, STA, and/or Sn(Oct) 2 . Thus, the obtained PLLA films, including raw PLLA, PLLA(-COOH), PLLA(-OH), PLLA(Sn 2+ ), PLLA(-COOH, Sn 2+ ) and PLLA(-OH, Sn 2+ ), were depolymerized by TGA to obtain the kinetic parameters and mechanism of the pyrolysis reaction in which high-molecularweight PLLA provided lactide in the presence of exogenous -COOH, -OH, and/or Sn 2+ . The pyrolysis reaction rate of PLLA was affected slightly by exogenous hydroxyl and carboxyl groups at lower Sn levels of 65−70 mg·kg −1 , but increased appreciably in the presence of extraneous Sn 2+ , -COOH/Sn 2+ , or -OH/Sn 2+ (Sn contents of 349−380 mg·kg −1 ). The E a values for the pyrolysis reactions of the PLLAs with higher Sn 2+ contents decreased appreciably, compared with those with lower Sn 2+ contents. The E a values for the pyrolysis reactions of the PLLAs with higher Sn 2+ contents differed with each other because the newly formed Sn-carboxylate (Oct-Sn-STA and Sn(STA) 2 ) and Sn-alkoxide (Oct-Sn-PEG, SnPEG, PEG-Sn-PEG and (SnPEG) m ) with the exogenous carboxyl (STA) and hydroxyl (PEG) groups were different in steric hindrance for the formation of activated complex between Sn 2+ and PLLA. In the early stage of pyrolysis, PLLA(Sn 2+ ) and PLLA(-OH, Sn 2+ ) had approximately identical E a values, given that the produced Oct-Sn-PEG, SnPEG, PEG-Sn-PEG, and (SnPEG) m were very little for PLLA(-OH, Sn 2+ ) and had little effect on the activation energy. However, the E a value was much lower for PLLA(-COOH, Sn 2+ ) than that for PLLA(Sn 2+ ) because Oct in Sn(Oct) 2 contained the α-substituted branched structure, which could produce the steric hindrance to the formation of activated complex between Sn 2+ and PLLA. However, in the presence of STA, Sn(Oct) 2 could be partially converted into Oct-Sn-STA and Sn(STA) 2 , thereby reducing the steric hindrance to the formation of activated complex between Sn 2+ and PLLA. The pyrolysis processes of PLLA, PLLA(-COOH), and PLLA(-OH) with lower Sn levels of 65−70 mg·kg −1 tended to be described by a single kinetic model. However, the pyrolysis mechanism and processes of PLLA(Sn 2+ ), PLLA(-COOH, Sn 2+ ), and PLLA(-OH, Sn 2+ ) with higher Sn contents of 349−380 mg·kg −1 were complicated in the later stage of the decomposed reactions because Sn 2+ had significant losses when heated. Under the catalysis of Sn 2+ , a lactide molecule can be directly eliminated selectively at a random position of PLLA molecular chains. In addition, the molecular chain of PLLA cannot form two PLLA fragments at the elimination site for lactide but was regenerated into a new PLLA molecule with the molecular weight reduced by 144 g·mol −1 .

Electronic Supplementary Information
Electronic supplementary information (ESI) is available free of charge in the online version of this article at http://doi.org/ 10.1007/s10118-021-2557-4.  Fig. 7 Possible pyrolysis mechanism of high-molecular-weight polylactide for producing lactide under the catalysis of Sn 2+ .