**4. Hydrolysis near the critical point**

In order to extend the results obtained by Tagaya et al. (1999), several experiments have been performed by adopting liquid water in the more convenient subcritical conditions. The results obtained in decomposing PC and producing BPA, will be reported in the following.

## **4.1 Experimental**

#### **4.1.1 Materials**

The reagent grade Poly(bisphenol A carbonate) (average MW 64000) [CAS 25037-45-0], Bisphenol A (>99%) [CAS 80-05-7], Phenol (>99%) [CAS108-95-2] and other chemical used were all purchased from Sigma-Aldrich. Water was twice distilled. Commercial Recordable Compact Discs crashed for de-polymerization tests were Verbatim Datalife CD-R.

#### **4.1.2 Apparatus and methods**

De-polymerization tests were carried out in a 316 stainless steel tubular batch reactor (internal diameter 7.8 mm, length 150 mm) having an internal volume of 7.1 mL. The tests were performed by first weighting the empty reactor and then charging it with 1 g of BAPC and 1 g of water. The reactor, exactly weighted after charge, was put into a laboratory fan assisted furnace (Heraeus M110 electronic), preheated at the temperature set point, over a support disposed along the symmetry axis of the oven. A thermocouple fixed on the external wall of the reactor measured temperature level. Both horizontal and vertical disposition were studied. At the end of the experiment the reactor was recovered from the oven, rapidly cooled and newly weighted at ambient temperature. Not more than ±0.2 mg of weight difference from initial and final total weight was accepted as a probe of no spill and good capping for the test, otherwise the test was repeated. The reactor was unhead and the evolved gas collected and GC analyzed. The degassed reactor was finally weighted and the difference was assumed as CO2 produced. The yield was evaluated from this latter information. A Mettler Toledo analytical precision balance (model B154-S) was employed. By cooling the reactor, the resulting content (at room temperature) is constituted only by condensed phases. The liquid and solid internal material was discharged, identified and quantified by FTIR and GC.

### **4.1.3 Some analytical aspects**

122 Material Recycling – Trends and Perspectives

Piñero et al. (2005) used methanol-water mixtures and developed (Piñero et al., 2006) a shrinking particle model to describe the reactive dissolution of the BPA-PC particles in semicontinuous de-polymerization of polycarbonate. Jie et al. (2006) studied decomposition in ethanol that has a critical point practically at same temperature and a lower pressure than that of methanol, allowing lower operating temperatures and pressures. Comparing the use of methanol and ethanol as solvents, it is reported that PC completely decomposes in supercritical methanol at 290°C and 9.96 MPa, in supercritical water above 374°C and pressure higher than 22.1 MPa, while in ethanol this is possible above 243 °C and 6.38 MPa producing BPA and diethylcarbonate (DEC). The mechanism consists in the random reaction along the polymer chain of the ester linkage with the solvent , that produces two smaller polymer chains, which can still react by ester exchange reaction until the polymer is

In all the reported examples of alcoholysis yields of 90% of BPA (in terms of weight of obtained BPA divided the weight of initial PC) were obtained. The activation energy

In order to extend the results obtained by Tagaya et al. (1999), several experiments have been performed by adopting liquid water in the more convenient subcritical conditions. The results obtained in decomposing PC and producing BPA, will be reported in the following.

The reagent grade Poly(bisphenol A carbonate) (average MW 64000) [CAS 25037-45-0], Bisphenol A (>99%) [CAS 80-05-7], Phenol (>99%) [CAS108-95-2] and other chemical used were all purchased from Sigma-Aldrich. Water was twice distilled. Commercial Recordable

De-polymerization tests were carried out in a 316 stainless steel tubular batch reactor (internal diameter 7.8 mm, length 150 mm) having an internal volume of 7.1 mL. The tests were performed by first weighting the empty reactor and then charging it with 1 g of BAPC and 1 g of water. The reactor, exactly weighted after charge, was put into a laboratory fan assisted furnace (Heraeus M110 electronic), preheated at the temperature set point, over a support disposed along the symmetry axis of the oven. A thermocouple fixed on the external wall of the reactor measured temperature level. Both horizontal and vertical disposition were studied. At the end of the experiment the reactor was recovered from the oven, rapidly cooled and newly weighted at ambient temperature. Not more than ±0.2 mg of weight difference from initial and final total weight was accepted as a probe of no spill and good capping for the test, otherwise the test was repeated. The reactor was unhead and the evolved gas collected and GC analyzed. The degassed reactor was finally weighted and the difference was assumed as CO2 produced. The yield was evaluated from this latter information. A Mettler Toledo analytical precision balance (model B154-S) was employed.

Compact Discs crashed for de-polymerization tests were Verbatim Datalife CD-R.

completely converted to BPA and another product depending on the solvent.

deduced from experimental data are resumed in table 5 (paragraph 6).

**4. Hydrolysis near the critical point** 

**4.1 Experimental 4.1.1 Materials** 

**4.1.2 Apparatus and methods** 

The solid sample obtained from the tests was characterized by Shimadzu FTIR: IRAffinity-1 in the 500 - 4000 cm-1 range (KBr disc). Gas chromatographic analyses of solid products were carried out (using methanol as solvent) on a GC-FID HP 6890plus with SLB-5ms capillary SUPELCO column (30 m length, 1 μl injection volume, split ratio 1:10, Helium carrier 1.4 mL/min, constant flow). The temperature was held at 65°C for the first 2 minutes, then increased at 255°C at a heating rate of 10°C/min and kept at this temperature for 15 minutes. The main products were identified and quantified by comparing the retention time with standard compounds.

Figure 3 shows a typical gas chromatographic analysis of the condensed phase discharged at the end of a test at high conversion. As it can be observed BPA is the main product. Gas chromatographic analyses of evolved gas were performed with a column 0.53 mm ID, Molecular Sieve 5A as stationary phase, GC-T as detector, isothermal at T=25°C. Permanent gases and CO2 were detected.

Fig. 3. Typical gas chromatographic analysis of solid product obtained by the tests.

Figure 4 shows FT transmittance plot showing the FTIR analyses of commercial poly(bisphenol A carbonate): (A) starting material obtained from a CD crash, (B) low conversion depolymerization solid discharged material, (C) medium conversion material, (D) solid material discharged at complete depolymerization. The FTIR spectra of reagent grade BPA, p-isopropylphenol and phenol are also reported. The figure 4 D shows the same shape as pure BPA.

Poly(bisphenol A carbonate) Recycling: High Pressure Hydrolysis Can Be a Convenient Way 125

This was also confirmed by some preliminary tests that have been performed. Different residence times have been adopted. Because of the competition existing among the concerted path hydrolysis reactions and the radical path (non selective) reactions, the selected temperature range has to be considered optimal. In facts the apparent activation energy of the pyrolysis is quite higher (40-80 kJ/mol more). The hydrolysis yields have been evaluated by means of both the amount of CO2 released and the weight loss of the sample (of course the last one is comprehensive of all the gaseous matter produced). In the preliminary tests it has been verified that the horizontal position of the reactor inside the furnace is largely preferable. In fact the reaction takes place starting from the contact surface between the molten and swollen polymer and the water, producing fractures and pores that favor the further penetration of water (as also reported by Dongpil et al., 2009 in the case of methanolysis). It is important to point out that the horizontal configuration allows enhancing the surface contact between the molten immiscible polymer and the water and,

Figure 5 reports the conversion versus time at different temperatures. One test has been performed for a time as long as 4 h. There is an apparent induction time for the process that can be essentially attributed to the time required for the heating of reagents inside the reactor. In order to understand whether some autocatalytic effect can be related to the products, it has been decided to perform also some tests by adding small amounts of BPA: practically no significant effect has been observed. The result obtained by Kitahara et al.

The obtained data have been used for deducing the reaction rate and the apparent activation energy of the reaction. Figure 6 reports the experimental reaction rate constant as a function of 1/T. This diagram incidentally can be considered preliminary to deduce in a simple way the activation energy of the reaction that results to be about 80 kJ/mol (in agreement with

The gases produced by the hydrolysis reaction was evaluated not only in terms of their total amount but, periodically, they were also analyzed by GC in terms of their composition. First of all the analyses have clearly shown that N2/O2 ratio, in the effluent gases, were practically the same of the original air inclusion. This result simply means that no oxidative degradation was taking place during hydrolysis. Indirectly, the last considerations are also showing that practically no extra components are entering during the periodical filling and emptying of the reactor. In the experiments with the highest de-polymerization degree the products obtained have been analyzed after mixing with methanol. The results show always a high purity in BPA in comparison with byproducts like phenol, p- propyl and propylidene phenols. This confirms that the polymer decomposition through breakage of the bond C aromatic-C isopropylic (i.e. pyrolysis) is negligible in practice and that this parasitic radical

The FTIR analysis on the reaction residue after evaporation of the water shows a spectrum that is coherent with a progressive hydrolysis of the carbonate bond, without appreciable evidences of de-alkylation followed by formation of terminal propylidene and phenol. The IR spectra at higher conversion are practically coincident with those of standard BPA.

On the basis of these evidences, it can be affirmed that the de-polymerization reaction proceeds in the condensed phase and, under the preferred conditions, it regenerates the

mechanism doesn't occur significantly in the selected operating conditions.

therefore, the apparent global reaction rate.

other experiences).

(2009) in subcritical conditions are also reported in the figure.

#### **4.2 Results and discussion**

The experimental tests on hydrolysis have been performed at temperatures ranging from 240 to 290 °C (corresponding to a pressure range of 3.5-8 MPa). The latter temperature has been selected as a maximum, because it is known that BPA, obtained by hydrolysis of PC, starting from 300°C decomposes giving place to phenol (Tagaya et al., 1999).

Fig. 4. FT transmittance plot from starting material until 100% conversion compared with BPA, p-isopropylphenol and phenol FTIR spectra.

The experimental tests on hydrolysis have been performed at temperatures ranging from 240 to 290 °C (corresponding to a pressure range of 3.5-8 MPa). The latter temperature has been selected as a maximum, because it is known that BPA, obtained by hydrolysis of PC,

Fig. 4. FT transmittance plot from starting material until 100% conversion compared with

BPA, p-isopropylphenol and phenol FTIR spectra.

starting from 300°C decomposes giving place to phenol (Tagaya et al., 1999).

**4.2 Results and discussion** 

This was also confirmed by some preliminary tests that have been performed. Different residence times have been adopted. Because of the competition existing among the concerted path hydrolysis reactions and the radical path (non selective) reactions, the selected temperature range has to be considered optimal. In facts the apparent activation energy of the pyrolysis is quite higher (40-80 kJ/mol more). The hydrolysis yields have been evaluated by means of both the amount of CO2 released and the weight loss of the sample (of course the last one is comprehensive of all the gaseous matter produced). In the preliminary tests it has been verified that the horizontal position of the reactor inside the furnace is largely preferable. In fact the reaction takes place starting from the contact surface between the molten and swollen polymer and the water, producing fractures and pores that favor the further penetration of water (as also reported by Dongpil et al., 2009 in the case of methanolysis). It is important to point out that the horizontal configuration allows enhancing the surface contact between the molten immiscible polymer and the water and, therefore, the apparent global reaction rate.

Figure 5 reports the conversion versus time at different temperatures. One test has been performed for a time as long as 4 h. There is an apparent induction time for the process that can be essentially attributed to the time required for the heating of reagents inside the reactor. In order to understand whether some autocatalytic effect can be related to the products, it has been decided to perform also some tests by adding small amounts of BPA: practically no significant effect has been observed. The result obtained by Kitahara et al. (2009) in subcritical conditions are also reported in the figure.

The obtained data have been used for deducing the reaction rate and the apparent activation energy of the reaction. Figure 6 reports the experimental reaction rate constant as a function of 1/T. This diagram incidentally can be considered preliminary to deduce in a simple way the activation energy of the reaction that results to be about 80 kJ/mol (in agreement with other experiences).

The gases produced by the hydrolysis reaction was evaluated not only in terms of their total amount but, periodically, they were also analyzed by GC in terms of their composition. First of all the analyses have clearly shown that N2/O2 ratio, in the effluent gases, were practically the same of the original air inclusion. This result simply means that no oxidative degradation was taking place during hydrolysis. Indirectly, the last considerations are also showing that practically no extra components are entering during the periodical filling and emptying of the reactor. In the experiments with the highest de-polymerization degree the products obtained have been analyzed after mixing with methanol. The results show always a high purity in BPA in comparison with byproducts like phenol, p- propyl and propylidene phenols. This confirms that the polymer decomposition through breakage of the bond C aromatic-C isopropylic (i.e. pyrolysis) is negligible in practice and that this parasitic radical mechanism doesn't occur significantly in the selected operating conditions.

The FTIR analysis on the reaction residue after evaporation of the water shows a spectrum that is coherent with a progressive hydrolysis of the carbonate bond, without appreciable evidences of de-alkylation followed by formation of terminal propylidene and phenol. The IR spectra at higher conversion are practically coincident with those of standard BPA.

On the basis of these evidences, it can be affirmed that the de-polymerization reaction proceeds in the condensed phase and, under the preferred conditions, it regenerates the

Poly(bisphenol A carbonate) Recycling: High Pressure Hydrolysis Can Be a Convenient Way 127

These results seem to be in partial contradiction with some of those obtained by Tagaya et al. (1999) indicating low BPA yields (around 35%) after 2 h reaction at 300°C . In that paper no indication was given on modalities of contact between molten polymer and water. A hypothesis on the difference of the obtained results can be related to a non adequate surface contact in their experiments. On the contrary other data supports our results (Kitahara et al., 2009, see figure 5). In facts, after hydrolysis of PC with water for 30 minutes at 270°C, they

Regarding the experiments performed with pure PC or with crashed CDs, essentially no

It is interesting to observe that PC hydrolysis is mainly dominated by a six center concerted path reaction mechanism. By analogy with other reactions following the same mechanism, involving hydrolysis of esters, the kinetic constant suggested for every elementary de-

Of course the radical reactions path becomes more important at higher temperatures (and of course in practical absence of water). In our conditions water becomes a powerful reactant. It has also to be pointed out that water approaching critical conditions has an increased solubility. Moreover, in this range of temperatures (>230-250 °C), the swelling of the molten polymer offers a large increase of diffusion coefficient into the polymer phase. Formally the

homogeneous reaction can be schematized as shown in figure 7.

= − l/mol/s (1)

<sup>9</sup> <sup>84000</sup> *<sup>k</sup>* 10 exp *RT*

This figure is coherent also with other activation energies as represented in Table 5.

obtained a BPA yield of 16% that is substantially in line with our results.

difference was observed.

polymerization act is:

Fig. 7. Hydrolysis mechanism.

**5. Hydrolysis kinetic mechanism** 

monomer. It is important to point out that after the tests at the highest conversion level (i.e. BPA yield >90% wt), the melting point of the product resulted over 145-150 °: this is another excellent indication of the substantial purity of the obtained raw monomer.

Fig. 5. Yield of BPA versus time at different temperatures.

Fig. 6. Experimental kinetic constant (s-1) as a function of -1/T [K-1].

These results seem to be in partial contradiction with some of those obtained by Tagaya et al. (1999) indicating low BPA yields (around 35%) after 2 h reaction at 300°C . In that paper no indication was given on modalities of contact between molten polymer and water. A hypothesis on the difference of the obtained results can be related to a non adequate surface contact in their experiments. On the contrary other data supports our results (Kitahara et al., 2009, see figure 5). In facts, after hydrolysis of PC with water for 30 minutes at 270°C, they obtained a BPA yield of 16% that is substantially in line with our results.

Regarding the experiments performed with pure PC or with crashed CDs, essentially no difference was observed.
