# Preparation of Poly(ethylene terephthalate) Foams Using Supercritical CO as a Blowing Agent

Authored by: Ling Zhao , Tian Xia , Zhenhao Xi , Tao Liu

# Polymeric Foams

Print publication date:  October  2016
Online publication date:  November  2016

Print ISBN: 9781498738873
eBook ISBN: 9781315369365

10.1201/9781315369365-4

3.1 Introduction

### Contents

 3.1 Introduction 49 3.2 Controllable Sandwich Structure of PET Microcellular Foams Based on the Coupling of CO2 Diffusion and Its Induced Crystallization 51 3.3 Melt Foamability of Reactive Extrusion-Modified PET 57 3.4 Melt-Foaming Behavior of In-Situ–Modified PET in Batch and Continuous Processes 65 3.5 Integrated Process of Supercritical CO2-Assisted Modification and Foaming of PET 73 3.6 Summary and Outlook 78 Acknowledgments 81 References 81

#### 3.1  Introduction

Poly(ethylene terephthalate) (PET) foams show good mechanical properties, high temperature dimensional stability, and recyclability and have recently attracted extensive attention in academic and industrial circles. The wide applications of PET foams depend on cell morphology and matrix properties. Microcellular PETs with cell sizes less than 5 μm were developed by Furukawa Electric (Japan), and possessed the excellent property of over 99% reflectivity and 96% light diffusion rate, which increased lamps’ illuminance by 40%–60% without an extra light source as the reflector. Low-density PET foams with a cell size of hundreds of microns have been widely used as packaging materials and structure materials in wind energy, marine, and transportation applications. Supercritical CO2 has been used as a blowing agent for PET because it is inexpensive, nontoxic, and environmentally benign. CO2 dissolved into a PET matrix causes many changes in the physical properties of PET in both the melt and solid states due to the strong plasticization effect, which enhances the free volume and chain mobility. It depresses the glass-transition temperature and the crystallization temperature, and changes the crystallization kinetics of semi-crystalline PET, which may provide many opportunities for the manipulation of foaming processes.

In a solid-state foaming process, the diffusivity of CO2 in a PET matrix was very low owing to the low temperature, and CO2 saturation in PET required several days or even longer. The homogeneous PET/CO2 solution could not be obtained due to CO2 insolubility in the crystalline region. The crystals supplied a large number of stable cell nuclei when acting as heterogeneous nucleation sites and at the same time inhibited the cell growth by increasing the stiffness of matrix [1,2]. The solid-state foaming process was usually conducted in batches. Cell nucleation was initiated by heating the saturated PET near or above its glass-transition temperature [3,4]. Kumar et al. [5] reported a semicontinuous process for the solid-state foaming of PET in which a CO2-saturated PET roll was passed through a hot bath continuously and foamed at temperatures ranging from 50°C to 90°C. The melt-state foaming of PET could be conducted in a continuous process, such as extrusion foaming. The CO2 diffusivity was facilitated by a high foaming temperature, and PET could be saturated by CO2 in a short period. However, the semicrystalline PET usually had low melt strength, and elasticity resulted from the low molecular weight and well-structured backbone, which could not resist the intensive extensional deformations during cell expansion. The cells would merge into larger ones or rupture [68]. It is known that high molecular weight, broad molecular weight distribution, and a long-chain branch attached to the PET backbone were required to improve the polymer melt strength through enhancing the possibility of entanglements in the polymeric melt. Multifunctional chain extenders with electrophilic functional groups such as cyclic anhydride [7,9] and epoxide [10,11] were developed to react with the nucleophilic end groups of PET, which could be performed by means of in-situ polymerization and reactive extrusion.

Some typical PET foaming processes using supercritical CO2 as a blowing agent, from a solid state to a melt state, will be discussed in Sections 3.2 through 3.5, including the following:

• Controllable sandwich structure of PET microcellular foams prepared by solid-state foaming process
• Melt foamability of reactive extrusion-modified PET
• Melt-foaming behavior of in-situ–modified PET in batch and continuous processes
• Integrated process of supercritical CO2-assisted modification and foaming of PET

The PET foams involved in this chapter cover the range from microcellular foam that is characterized by closed cells with cell sizes that are lower than 10 μm and cell densities that are higher than 109 cells/cm3 to macrocellular foam (the typical cell size of hundreds microns). The polymer matrix included PET with linear molecular chain, PET/clay nanocomposites, PET/SiO2 nano-composites, and branched PET that is prepared by both reactive extrusion and in-situ modification.

#### 3.2  Controllable Sandwich Structure of PET Microcellular Foams Based on the Coupling of CO2 Diffusion and Its Induced Crystallization

The sandwich structure of PET microcellular foams was fabricated by the solid-state foaming process, as illustrated in Figure 3.1. The gradient of CO2 concentration developed gradually in amorphous PET sheets during the saturation stage. The CO2-induced crystallization took place when the CO2 concentration was high enough. Therefore, the layers near the sheet surfaces crystallized earlier, but PET near the central areas remained amorphous. The foaming temperature was chosen to be 235°C, just 20°C lower than the PET melting point, and was very different from that of reported patents [5] because higher foaming temperatures would create larger thermodynamic instability, which was favorable for cell nucleation, and the difference between the foaming morphologies of the crystalline layer PET and the amorphous layer PET could be more obvious at higher foaming temperatures. The sandwich structure of PET foams was produced with larger cells in the amorphous layer inside and small cells in the crystalline layers outside, as displayed in Figure 3.2 [12].

Figure 3.1   Schematic of the preparation process of sandwich-structure microcellular PET foams.

Figure 3.2   Typical boundary between the crystalline layer and the amorphous layer of PET foam.

(From D. Li et al., Aiche Journal, 58, 2512–2523, 2012.)

The CO2-induced isothermal crystallization of PET was determined by in-situ high-pressure Fourier Transform Infrared Spectroscopy (FTIR). No crystallinity increase could be detected under 4.5 MPa CO2 even after 2000 min, as shown in Figure 3.3, whereas a very long crystallization period of more than 1500 min was observed under 5.0 MPa CO2, indicating the lowest pressure, corresponding to the lowest critical CO2 concentration in PET. Ccritical for the CO2-induced crystallization of PET was 5.0 MPa at 25°C. The well-known Avrami equation could be used to evaluate the CO2-induced isothermal crystallization of PET films.

Figure 3.3   Changes in crystallinity of PET films during isothermal crystallization at 25°C and different CO2 pressures.

(From D. Li et al., Aiche Journal, 58, 2512–2523, 2012.)

The solubility and diffusivity of CO2 in an amorphous PET sheet at 25°C and 6 MPa were measured by the Magnetic Suspension Balance (MSB) method. The distribution of CO2 concentration was found to vary with saturation time. The CO2-induced crystallization took place at the surface layers, which would ×250 100 µm subsequently reject CO2 and could be adopted to explain the knee-like sorption profile, as demonstrated in Figure 3.4. The diffusion coefficient was 6.7 × 10−13 m2/s assuming that the diffusion could be expressed by Fick’s second law. The solubility of CO2 in amorphous regions of PET C0 at 25°C and 6 MPa was 11.0 wt%, and Ccritical, the equilibrium concentration of CO2 in amorphous PET under 5 MPa CO2, was estimated to be 9.17 wt% assuming that the sorption of CO2 would follow Henry’s law at relatively low CO2 pressure.

Figure   Sorption profiles for CO2 in PET at 25˚C and 6 MPa. (a) Mass of dissolved CO2; (b) Relative amount of dissolved CO2.

(From D. Li et al., Aiche Journal, 58, 2512–2523, 2012.)

After a certain saturation time, the CO2 concentration and crystallinity distributions in the PET sheet were controlled by CO2 diffusion and induced crystallization simultaneously. A model was proposed to investigate the coupled processes with assumptions that the PET sheet consisted of N well-contacted layers in the thickness direction, and the thickness of each layer was equal and so small that the CO2 concentration distribution was uniform in each individual layer, as illustrated in Figure 3.5. In such a case, the concentration of CO2 in each layer could be determined by the following equation:

3.1()$C ( t , i ) C 0 = 1 − 4 π ∑ n = 0 ∞ ( − 1 ) n 2 n + 1 e x p [ − D ( 2 n + 1 ) 2 π 2 t 4 l 2 ] c o s ( 2 n + 1 ) π x ( i ) 2 l$

Figure 3.5   Schematic diagram of the proposed model.

(From D. Li et al., Aiche Journal, 58, 2512–2523, 2012.)

where C (t,i) is the concentration of CO2 at saturation time t in layer i, x(i) is the location of layer i in the thickness dimension of PET sheet, and D is the diffusion coefficient. The crystallization rate, Rc, in each individual layer could be determined as follows:

3.2()$R c = d X t d t = K t n − 1 e x p ( − K t n )$

where n and K are the Avrami exponent and crystallization rate constant, respectively.

As shown in Figure 3.6, most of PET sheet was still unsaturated with oneday saturation, leading to small cell sizes in the areas near the surface and large cells in the central areas. After a five-day saturation, the distribution of CO2 concentration had been relatively uniform in the PET sheet, which led to uniform cell morphology in most areas of the sample. Crystallinity played a major role in the microcellular foaming process on cell nucleation mechanisms, resulting in larger cell densities due to heterogeneous nucleation at the amorphous/crystalline boundaries, and cell growth mechanisms, leading to smaller cell size due to the increased stiffness of semicrystalline matrix. The thickness of crystalline layers calculated using the proposed model together with those measured from PET foams at different saturation times were shown in Table 3.1. Basically, the model could well predict the evolution of crystalline layer against a different saturation time.

### Table 3.1   Comparison of the Evolution of Crystalline Layer Thickness Calculated Using the Proposed Model and in PET Foams

Saturation Time, Day

1

2

3

5

7

Thickness of crystalline layer, μm

In PET foams

0a

80

130

250

900

Modeling

20

50

100

300

complete

Source: D. Li et al., Aiche Journal, 58, 2512–2523, 2012.

Notes:

a  The crystalline layer was not detected in the PET foams due to the overlap of crystalline and the unfoamed skin layer of the PET specimen.

Figure 3.6   The overall cell morphology of PET foams prepared with different saturation times.

(From D. Li et al., Aiche Journal, 58, 2512–2523, 2012.)

The evolution of cell size and cell density of both crystalline and amorphous layers and the expansion ratio of PET foams against saturation time are shown in Figure 3.7. The characterization of cell morphology is summarized in Table 3.3. With the increase of saturation time, the CO2 concentration in the amorphous layer of PET specimen increased, and more CO2 was available to support the cell growth, which led to an increase in both cell size and expansion ratio within the saturation time of five days. When the saturation time was longer than seven days, the crystal structure of crystalline became more perfect, and a small amount of crystals could have formed in the amorphous layer, which restricted the cell growth and increased the cell density via changing the nucleation mechanism from homogeneous to heterogeneous nucleation. Especially at the saturation time of 15 days, ultra-microcellular PET foams with an average cell diameter of 193 nm and a cell density of 3.37 × 1013 were obtained.

Figure 3.7   Characterization of PET foams obtained at different saturation times: (a) cell size, (b) cell density, and (c) expansion ratio.

(From D. Li et al., Aiche Journal, 58, 2512–2523, 2012.)

### Table 3.3   Examples of PET Foams Obtained by Different Foaming Processes with Supercritical CO2

Parameters/Properties

Li et al. [12]

Xia et al. [13]

Zhong et al. [18]

Fan et al. [19]

Zhong et al. [22]

Foaming method Modifying method

Solid state/temperature rising N/M

Melt state/pressure quench Reactive extrusion

Melt state/pressure quench in-situ polymerization

Melt state/extrusion foaming in-situ polymerization +SSP

Integrated process of CO2-assisted melt polycondensation and foaming

Chain extender

N/M

PMDA

PMDA/PENTA

PENTA

PENTA

Intrinsic viscosity (dL/g)

1.0

0.88–1.36

0.865/0.860

1.5

0.860

Inorganic filler

N/M

N/M

N/M

Nano-SiO2

N/M

Foaming temperature (°C)

235

210–280

265–280

232–240

280

Cell size (μm)

3.40–6.24(a)/0.19–2.57(c)

15–37

38–57/35–49

265–512

32–62

Cell density (cells/cm3)

2.2 × 109–1.3 × 1010(a)/4.2 × 1010–3.4 × 1013(c)

6.2 × 108–1.6 × 109

5.8 × 106–2.8 × 107/7.0 × 106–3.5 × 107

0.19 × 105–4.6 × 105

1.0 × 107–4.0 × 107

Expansion ratio

0.2–9.5

10–50

20–40

8.0–10.2

4.0–26*

Note: (a) amorphous layer, (c) crystalline layer.

Notes:

*  Assuming that the density of unfoamed PET was 1.3 g/cm3.

#### 3.3  Melt Foamability of Reactive Extrusion-Modified PET

High melt elasticity and strength were essential in the melt foaming process of PET to avoid cell collapse and coalescence, which were acquired via the chain extension/branching reactions between the end groups of PET and the multifunctional chain extenders. And the long-chain branch was introduced to the PET backbone. In the melt foaming process of PET, cell nucleation was initiated by rapid depressurization. Cell growth was hindered, and the cell morphology was solidified by the increasing stiffness of the polymer matrix due to the nonisothermal crystallization from melts during the cooling stage, which would be influenced by the topological structure of the PET chain and CO2 atmosphere. Pyromellitic dianhydride (PMDA) was often selected to react with the hydroxyl end groups of PET in extruders because the melting point of PMDA, close to the PET processing temperature, together with tetrafunctionality, ensured that the modifying reactions were fast [13].

Dynamic shear rheological properties were sensitive to the topological structure of polymers. As shown in Figure 3.8a, the Newtonian zone of linear virgin polyethyleneterephthalate (VPET) was broad, and no obvious shear thinning was observed. The transition from a Newtonian plateau to a shear-thinning regime shifted to lower frequency with the increase of PMDA content, which was ascribed to the long-time-relaxation mechanism, such as entanglement couplings between high-molecular-weight fractions and those that are associated with the long-chain branch. In addition, a broad relaxation time distribution was also used to explain the shear thinning. Cole–Cole plots, in which the axes were represented as the logarithms of storage modulus G′ and loss modulus G″, were independent of temperature and molecular weight but strongly dependent on the long-chain branch and molecular weight distribution. With the introduction of a long-chain branch or broader molecular weight distribution, G′ increased at a given G″. As demonstrated in Figure 3.8b, modified samples got a broader molecular weight distribution and a long-chain branch due to the reactive extrusion process. Figure 3.8b also included the equal-modulus line, and the position of plots with respect to this line indicated the degree of melt elasticity. It could be easily concluded that the melt elasticity of the modified sample increased with PMDA content.

Figure 3.8   Dynamic rheological properties of VPET and modified PETs: (a) η* versus ω, (b) Cole–Cole plots, and (c) G′ and G″ versus ω. (MPET2, 5, 8 represented PET modified with 0.2, 0.5, 0.8 wt% PMDA.)

(From T. Xia et al., Polymer Engineering & Science, 55, 1528–1535, 2015.)

The cross point of G′ and G″ should shift to a lower-frequency region when the molecular weight increased and to a lower-modulus region when the molecular weight distribution became broader. It was inferred from Figure 3.8c that MPET8 had the broadest molecular weight distribution and the highest molecular weight.

The Avrami analysis was extended for the nonisothermal crystallization of a modified PET, as described by Xia et al. [13]. With the introduction of the long-chain branch, the nonisothermal crystallization rate constant Z C became higher in general, whereas the half-crystallization time t1/2 decreased, as exhibited in Table 3.2, indicating that the crystallization rate increased within the experimental range of branching degree. The introduction of a long-chain branch brought two opposite effects on crystallization. The long-chain branch, as a kind of defected of polymer chain when crystallizing, resulted in a longer nucleation time than is required to exclude the branching point from crystal nucleates. On the other hand, branching provided a larger crystal-growing space due to the mutual repulsion of chains, which was in favor of the crystals’ growth. For the crystallization of a modified PET, crystal growth was the rate-controlling step. Higher crystallization rates were obtained with CO2 dissolution since Z C enhanced whereas t1/2 reduced with the increase of CO2 pressure, as presented in Table 3.2. CO2 dissolved into a PET matrix increased the free volume and weakened the interchain interaction, which facilitated the retraction and fold of molecular chains and meanwhile decreased the crystallization enthalpies. And during the cooling stage in the melt foaming process, the residual CO2 in cells and polymer matrix could accelerate the crystallization process further to gain an ideal cell morphology. It was noteworthy that the values of the Avrami exponent n of all PET samples were between 3 and 4, which deduced that all PET samples were crystallized according to the three-dimensional growth of crystals with either a homogeneous or heterogeneous nucleation.

### Table 3.2   Parameters of Nonisothermal Crystallization under 0.1 MPa N2 and 5 MPa CO2for VPET and Modified PETs

Gas Atmosphere

PET

R (°C/min)

ZC (min−1)

t1/2 (min)

n

ΔHc (J/g)

0.1 MPa N2

VPET

2

0.024

8.34

3.33

−40.96

3

0.084

6.31

3.25

−39.86

4

0.177

5.65

3.78

−38.91

5

0.349

4.21

3.40

−38.22

2

0.039

8.42

3.07

−39.46

3

0.163

5.12

3.11

−38.17

MPET2

4

0.216

4.38

3.90

−37.44

5

0.375

4.20

3.16

−36.46

2

0.044

6.45

3.16

−34.05

3

0.181

4.58

3.13

−33.84

MPET5

4

0.285

4.15

3.27

−33.07

5

0.384

3.54

3.50

−32.09

2

0.044

6.17

3.23

−36.05

3

0.123

5.14

3.62

−35.31

MPET8

4

0.241

4.31

3.64

−33.96

5

0.408

3.39

3.37

−33.39

2

0.069

5.22

3.01

−32.45

MPET5

3

0.270

3.48

2.86

−32.29

4

0.450

2.83

2.72

−32.01

5 MPa CO2

5

0.603

2.08

2.27

−30.73

2

0.103

4.46

2.80

−34.00

MPET8

3

0.294

3.26

2.80

−29.23

4

0.546

2.40

2.94

−25.06

5

0.679

2.02

2.23

−20.56

Source: T. Xia et al., Polymer Engineering & Sciencece, 55, 1528–1535, 2015.

The highest foaming temperature (Tf-H) of a modified PET was determined by the melt strength, which reduced with the foaming temperature. The formed cells would collapse at a foaming temperature that is higher than Tf-H, and then no foams could be gained. The lowest foaming temperature (Tf-L) of melt foaming process was the onset temperature of nonisothermal crystallization from melts under compressed CO2, below which the original densities of PET were measured as a result of extensive crystallization, and no PET foam could be gained. As shown in Figure 3.9, nearly no declined trend of crystallization onset temperature was observed with increased of CO2 pressure for a modified PET, and the crystallization onset temperature of MPET5 was nearly equal to that of MPET8. The foaming temperature windows of MPET5 and MPET8 were 210°C–250°C and 210°C–280°C, respectively, as displayed in Figure 3.10, and no foaming temperature windows were explored for VPET and MPET2 due to lower melt strength and the crystallization rate.

Figure 3.9   Differential scanning calorimetry cooling traces of modified PET melts at a cooling rate of 5°C/min under different gas atmospheres: (a) MPET5 and (b) MPET8.

(From T. Xia et al., Polymer Engineering & Science, 55, 1528–1535, 2015.)

Figure 3.10   Scanning electron microscope (SEM) micrographs of modified PET foams obtained at different foaming temperatures.

(From T. Xia et al., Polymer Engineering & Science, 55, 1528–1535, 2015.)

With the increase of foaming temperature, the amount of CO2 available for cell nucleation decreased, and the surface tension between the CO2 and the melt, reducing with CO2 dissolution, increased as a result of lower CO2 solubility. Therefore, the nucleation rate decreased according to the classic nucleation theory, and a lower cell density was obtained, as shown in Figure 3.11. On the other hand, a higher foaming temperature decreased the viscosity of the polymer matrix, increased the diffusivity of CO2 within the matrix, and extended the cell growth period. As a consequence, the average cell diameter and the expansion ratio increased. The characterization of modified PET foams is summarized in Table 3.3. More cells of MPET5 foams collapsed or coalesced due to the lower melt strength, as presented in Figure 3.10. Thereby, apparent lower cell density, especially at a higher foaming temperature, and a larger cell size along with higher expansion ratio were obtained, compared to those of MPET8 foams. The cell size distribution was controlled by the cell growth period. A longer time was required to reach the crystallization temperature and to solidify the cell morphology with the increase of foaming temperature, and hence, the cell size distribution became broader, as demonstrated in Figure 3.12. In addition, the cell size distributions of MPET8 foams were more uniform than those of MPET5 foams at the same foaming temperature, since the cell growth was inhibited more intensively due to high melt strength.

Figure 3.11   Characterization of MPET foams obtained at different foaming temperatures: (a) cell density, (b) average cell diameter, and (c) expansion ratio.

(From T. Xia et al., Polymer Engineering & Science, 55, 1528–1535, 2015.)

Figure 3.12   Cell size distribution of MPET foams obtained at different foaming temperatures: (a) MPET5 and (b) MPET8.

(From T. Xia et al., Polymer Engineering & Science, 55, 1528–1535, 2015.)

In recent years, nanoclays, such as montmorillonite, have been widely used to improve the melt foamability of polymers because of their high aspect ratio, plate morphology, and natural abundance [14,15]. The polymer/clay nanocomposites, in which the clays were dispersed in nano-size by the intercalation of polymer chains into clay galleries, presented attractive properties such as reduced gas permeability, flame retardance, and enhanced mechanical properties. The nanocomposite foams were expected to possess the combined advantages of polymer foams and nanocomposites. The large interfacial areas between clay platelets and the polymer matrix supplied many more heterogeneous nucleation sites, which facilitated the cell nucleation process by reducing the energy barrier of nucleation, resulting in a higher nucleation rate and ultimately a larger number of cell nuclei. The inorganic platelets were nonpermeable to CO2 and thus enhanced the tortuosity and path length of CO2 diffusion in the matrix, which reduced the rate of cell growth. Meanwhile, the amount of CO2 available for cell growth decreased while more cells were nucleated, leading to a reduction of cell size. Furthermore, reorientation of clay platelets along the cell walls was observed by Okamoto et al. [16,17], which acted as a second cell wall in favor of protecting the cells from being destroyed by stretching force. The melt foamability of PET with relatively lower intrinsic viscosity (IV) and melt elasticity could be significantly improved by the well-dispersed clay, and the foaming temperature windows of PET/clay nanocomposites prepared by the melt blending method could be obviously broadened. It is worthwhile to note that the long-chain branch structure of PET molecular was still necessary to obtain a foamable PET, as demonstrated in Figure 3.13, although layered clays were added.

Figure 3.13   Melt foamability of PET/clay nanocomposites.

In the reactive extrusion process of recycled PET, the addition of heat stabilizer and antioxidant was necessary to avoid serious thermal and oxidation degradations in the extruder. It was discovered that triglycidyl isocyanurate (TGIC) was more effective than PMDA to increase the molecular weight and melt strength of recycled PET. And the TGIC-modified bottle-recycled PET could also be foamed in melt state, as displayed in Figure 3.14.

Figure 3.14   SEM micrographs of TGIC-modified bottle-recycled PET foams obtained at (a) 260°C and (b) 270°C under 16 MPa CO2.

#### 3.4  Melt-Foaming Behavior of In-Situ Modified PET in Batch and Continuous Processes

In in-situ polymerization–modification process, the third monomer, polyacid or polyalcohol with more than two functional groups, was added into the direct esterification–polycondensation process together with terephthalate acid and ethylene glycol (EG). The synthesis process and process conditions were similar to the traditional PET manufacture. PMDA could also be introduced into polymerization reactions to obtain a long-chain-branched PET [18]. The IV of in-situ modified PET increased with PMDA content, and the IV also reached a maximum value of 0.865 dL/g with a PMDA content of 0.8 wt%. With the continuous increase of PMDA content, gel formation was observed. Pentaerythritol (PENTA) was more effective in enhancing the IV and melt strength of PET as a chain extender, and the IV of PENTA-modified PET got up to 0.860 dL/g, as the PENTA content was only 0.35 wt%, because PENTA could react with the carboxyl group of terephthalate acid monomer, as well as the PET oligomer, and induce the formation of a branched structure in the esterification stage during the PET synthesis process. Gel also formed as the PENTA content exceeded 0.35 wt%. PENTA and PMDA in-situ modified PET were foamed at different temperatures, as shown in Figures 3.15 and 3.16, both of which obtained uniform closed-cell structures. It was noted that PMDA-modified PET foams possessed larger cell sizes and lower cell densities than those of PENTA-modified PET foams, which could be explained by the lower carboxyl content of PENTA-modified PET, due to the consumption of a carboxyl-reactive chain extender PENTA, along with higher thermal stability. The characterization of in-situ modified PET foams is summarized in Table 3.3. Furthermore, via the prereaction between the monomer of EG and the phosphorus-based fire retardant of 2-carboxyethyl(phenyl)phosphinic acid (CEPPA), fire-resistant PET was prepared by in-situ polymerization and remained foamable due to the existence of polyalcohol as a multifunctional chain extender.

Figure 3.15   SEM micrographs of PENTA-modified PET foamed at a saturation pressure of 14 MPa and different temperatures of (a) 265°C, (b) 270°C, (c) 275°C, and (d) 280°C.

(From H. Zhong et al., Chinese Journal of Chemical Engineering, 21, 1410–1418, 2013.)

Figure 3.16   SEM micrographs of PMDA-modified PET foamed at a saturation pressure of 14 MPa and different temperatures of (a) 265°C, (b) 270°C, (c) 275°C, and (d) 280°C.

(From H. Zhong et al., Chinese Journal of Chemical Engineering, 21, 1410–1418, 2013.)

The melt-state foaming process of in-situ modified PET could also be implemented continuously, such as extrusion foaming. However, compared with the batch foaming process, a higher molecular weight and melt strength were required, which would deteriorate in the extruder owing to degradations. Therefore, PENTA-modified PET had to be polymerized in a solid state for another 24 h, and the IV reached 1.5 dL/g [19]. The rheological properties, including the shear and extension rheological behaviors of the polymer melt, played key roles in the extrusion foaming process. Relaxation time spectrum revealed the most basic function relation between viscoelasticity and time or frequency. All sorts of material functions obtained by rheology measurements were based on the same relaxation time spectrum, and rheological properties could be expressed by the relaxation time spectrum of different movement patterns and contributions. Four different PETs were investigated by Fan et al. [19]. PET1 was a regular spinning-grade PET with a linear molecular structure, and the IV was 0.65 dL/g. PET2 was obtained by solid-state polymerization (SSP) for 24 h at 220°C under vacuum with PET1 as a raw material, and the IV reached 0.90 dL/g. PENTA-modified PET was also polymerized in a solid state for 15 h to get PET3 with an IV up to 1.30 dL/g. Another 9-h SSP was carried out to obtain PET4 with an IV of 1.50 dL/g. Tschoegl equations were adopted to estimate the viscoelastic behavior of PET:

3.3()$H ( τ ) = d G ' d ln ω − d G ' 2 d ln ω | 1 ω = τ 2$
3.4()$H ( τ ) = 2 π [ G ″ − 4 3 ( d G ″ d ln ω ) + 1 3 ( d G ″ d ln ω ) ] | 1 ω = τ 5$

where H (τ) is the relaxation spectra, and ω is the experimental frequency. For a better revelation of viscoelasticity in a long time, the relaxation time spectrum was calculated based on G′ results. As shown in Figure 3.17, the linear chain (PET1 and PET2) was easier to slip to overcome macromolecular entanglement so that a shorter relaxation time was needed. When the relaxation time reached 10 s, the relaxation spectra were lower than 10 Pa, reflecting the liquid-like relaxation behavior. PET3 and PET4 had higher relaxation spectra due to the long-chain branch, especially at a long relaxation time. Macromolecular entanglements were more complicated with the existence of the long-chain branch, which made the relaxation process longer and the relax modulus higher. Modified PETs also exhibited a rubber-like behavior.

Figure 3.17   Relaxation spectra of PETs.

(From C. Fan et al. Journal of Cellular Plastics, 52, 277–298, 2016.)

PET1, PET2, and PET3 all failed in the extensional rheology characterization with Sentmanat extensional rheometer (SER) unit at the temperature between 250°C and 280°C because the testing samples sagged during the heating, which as attributed to the lower melt elasticity. The extensional viscosity curves of PET4 are shown in Figure 3.18, and obvious strain-hardening behaviors were observed, which played an important role in stabilizing the cell structure and preventing cells from merging and collapsing. Moreover, the strain hardening was more obvious with the decrease of temperature, which made the strain-hardening behavior occur earlier and more significantly.

Figure 3.18   Extensional viscosities of PET4 (a) at 270°C under different deformation rates and (b) at 0.3 s−1 and different temperatures.

(From C. Fan et al., Journal of Cellular Plastics, 52, 277–298, 2016.)

A single-screw extruder (screw diameter = 50 mm, L/D = 45) equipped with a rod die (2 mm in diameter) was used for the extrusion foaming process by Fan et al. [19] A static mixer, with Sulzer mixture units aimed to facilitate gas dispersion and dissolution in a polymer matrix, was installed at the end of screw, as shown in Figure 3.19. Under proper operating conditions, the diameter of extrudates increased gradually with operating time, and PET foams were obtained, as demonstrated in Figure 3.20.

Figure 3.19   Schematic of the experimental apparatus for extrusion foaming.

(From C. Fan et al., Journal of Cellular Plastics, 52, 277–298, 2016.)

Figure 3.20   Schematic of extrusion foaming process: the extrudates expanded gradually from (a) the start of CO2 injection to (f) the steady state of PET foaming.

Die temperature was the key parameter in the extrusion foaming process. With the decrease of foaming temperature, the melt strength increased, and furthermore, higher CO2 solubility was obtained. Therefore, the expansion ratio of PET foam got higher, and the cell morphology became better. And the highest expansion ratio of 9.37 was acquired at a die temperature of 240°C. As shown in Figure 3.21, idea cell morphologies were obtained with a uniform cell size distribution at foaming temperatures of 236°C and 240°C. With the continuous decrease of foaming temperature, the foam expansion would be hindered by the increasing stiffness of melt.

Figure 3.21   SEM images of the center (left) and border (right) regions of PET foams at different die temperatures: (a, b) 232°C, (c, d) 236°C, and (e, f) 240°C.

(From C. Fan et al., Journal of Cellular Plastics, 52, 277–298, 2016.)

The extrusion pressure increased with the screw speed and reached a maximum value of 11.6 MPa, and the expansion ratio of extruded PET foams got up to 9.0. Thermodynamic instability was triggered by the rapid pressure drop when the melt passed through the die. CO2 solubility in the PET melt and the supersaturation of CO2 increased with extrusion pressure, which directly facilitated the cell nucleation process. As a result, the cell size decreased from 512 to 449 μm, and the cell density, as well as the expansion ratio, got higher with the increase of screw speed from 37 to 46 rpm.

With the increase of gas input from 0.1 to 5 mL/min, the expansion ratio increased almost linearly and then leveled off as the gas input was up to 10.0 mL/min. More CO2 dissolved in the PET melt with the increasing gas input, which could be used to explain the increase of cell density and the reduction of cell size. It was the noteworthy that the undissolved gas would cause the fluctuation of extrusion pressure, resulting in the instability and even the failure of extrusion foaming process, as the gas input exceeded the solubility.

The addition of nano-SiO2 offered a large number of heterogeneous nucleation sites that improved the nucleation process via decreasing the nucleation energy barrier and had an obvious impact on the cell morphology. With the addition of only 0.1 wt% nano-SiO2, the cell density increased by four times, whereas the cell size decreased from 499 to 265 μm, as displayed in Figure 3.22. Moreover, the cell size distribution was an important parameter to characterize the cell morphology in polymer foam, which also had a significant effect on the physical and mechanical properties of foams along with the cell size and cell density. Shortening the nucleation time interval was an effective method to obtain a narrow cell distribution, which could be reduced via accelerating the nucleation process. As shown in Figure 3.23, the fitted curves of the cell size distribution of PET foams with 0.1 wt% nanoSiO2 demonstrated higher peak height and narrower peak width, meaning narrower cell size distribution.

Figure 3.22   SEM images of PET foams: (a) without nano-SiO2 and (b) with 0.2 wt% nano-SiO2.

(From C. Fan et al., Journal of Cellular Plastics, 52, 277–298, 2016.)

Figure 3.23   The effect of nano-SiO2 on cell size distribution of PET foams. (Lines were fitted curves.)

(From C. Fan et al., Journal of Cellular Plastics, 52, 277–298, 2016.)

#### 3.5  Integrated Process of Supercritical CO2-Assisted Modification and Foaming of PET

It was well known that the reversible reaction was usually controlled by the removal of volatile condensates in the final polycondensation stages, which was facilitated conventionally by high vacuum. However, there always existed mass-transfer limitations of volatiles in the polymer matrix. Supercritical CO2, a kind of environmental alternative to traditional solvents, had been successfully applied to facilitate the devolatilization of the polymer [20,21]. The free volume of polymer swollen by CO2 increased, which promoted the diffusion of volatiles in polymer matrix and the mobility of polymer chains along with their end groups, resulting in a higher polycondensation rate. Furthermore, many volatile condensates of poly-condensation displayed significant solubility in liquid or supercritical CO2, which may serve as a more effective sweeping fluid to bring the volatiles away and drive the polycondensation forward. EG, a volatile condensate generated in PET polycondensation, was soluble in CO2 up to 2–3 wt%, which was much higher than the solubility of EG in the PET melt. The melt polycondensation–modification process was carried out with high-pressure CO2 sweeping using PENTA-modified PET that is prepared via the in-situ polymerization method as a raw material, based on which an integrated process of CO2-assisted PET further melt polycondensation–modification and foaming was carried out, as illustrated in Figure 3.24a [22].

Figure 3.24   Flow chart of different PET foaming processes in melt state: (a) Integrated process, (b) one-step process, and (c) separated process.

The foam morphologies obtained under different CO2 flow rates and pressures are displayed in Figure 3.25. The cell size and the foam density reduced, whereas the cell density increased, with CO2 flow rates. Further, polycondensation promoted by CO2 sweeping increased long-chain branches, which could improve the melt strength and was beneficial for the foaming process, and thus uniform cell a morphology could be expected. Although higher CO2 flow rate was more effective in removing the condensates, it was found that the difference between foam densities obtained with CO2 flow rates of 4 L/min and 5 L/min at each saturation pressure was undetectable, as shown in Figure 3.26. For each CO2 flow rate, the foam density first decreased with saturation pressure (before 12 MPa) and then increased at 14 MPa. A higher saturation pressure had dual effects on foam morphology. The cell density increased, whereas the foam density decreased, with CO2 pressure at first due to the increase of CO2 solubility and depressurization rate. At the same time, a stronger plasticization effect caused by the continuous increase of CO2 pressure would weaken the entanglements of the molecular chain and finally lead to more cell coalescence. Therefore, PET foams with relative lower expansion ratio and higher density were obtained.

Figure 3.25   SEM micrographs of obtained PET foams from the integrated process under different saturation pressures and CO2 flow rates at 280°C, 30 min.

(From H. Zhong et al., The Journal of Supercritical Fluids, 74, 70–79, 2013.)

Figure 3.26   Characterization of PET foams obtained by the integrated process under different saturation pressures and CO2 flow rates at 280°C, 30 min.

(From H. Zhong et al., The Journal of Supercritical Fluids, 74, 70–79, 2013.)

The cell structures of PET foams obtained at different saturation pressures and treating times are exhibited in Figure 3.27. The foam density increased whereas the cell density decreased with treating time, as shown in Figure 3.28, indicating that the thermal degradation may dominate. The melt strength and foamability of PET deteriorated as a result of thermal degradation. Differential scanning calorimetry (DSC) and melt flow index (MFI) were measured for modified PET foams that are obtained under different treating times. The melt point of PET foams decreased slightly, whereas the MFI increased with the treating time, which could be attributed to the branched structures and thermal degradations.

Figure 3.27   SEM micrographs of obtained PET foams from the integrated process under different saturation pressures and treating times at 280°C, 4 L/min.

(From H. Zhong et al., The Journal of Supercritical Fluids, 74, 70–79, 2013.)

Figure 3.28   Characterization of PET foams obtained by the integrated process under different saturation pressures and treating times at 280°C, 4 L/min.

(From H. Zhong et al., The Journal of Supercritical Fluids, 74, 70–79, 2013.)

For comparison, another two foaming processes in the melt state were also carried out. One was the one-step process, as illustrated in Figure 3.24b, in which CO2 was just introduced at the beginning to saturate polymer, i.e., no CO2 sweeping, followed by rapid depressurization. A preliminary modified PET with an IV of 0.72 dL/g was foamed via the integrated process and the one-step process, as shown in Figure 3.29. It was obvious that no foam could be obtained with the one-step process due to the low melt strength of the preliminary PET. In regard to the integrated process, it was notable that cellular morphology could be easily distinguished, implying that the continuous sweeping of CO2 led to further polycondensation and improved melt strength. Hence, it was deduced that the integrated process reduced the demand of the molecular weight of PET matrix in the melt foaming process.

Figure 3.29   SEM micrographs of low-melt-strength PET foamed through (a) one-step process with saturation time 30 min and saturation pressure 12 MPa and (b) integrated process with sweeping time 30 min, saturation pressure 12 MPa, and CO2 flow rate 4 L/min at 280°C.

(From H. Zhong et al., The Journal of Supercritical Fluids, 74, 70–79, 2013.)

The CO2-assisted polycondensation–modification process of preliminary modified PET and the CO2 foaming process were implemented separately in the so-called separated process, as illustrated in Figure 3.24c. It is clearly seen from Figure 3.30 that foams obtained by separated process were characterized by larger cell sizes, lower cell density, and thin cell walls. Cell collapse and cell-wall fractures could also be easily discerned. The cell morphologies were worse than those obtained by the integrated process. It was believed that degradation was the main reason, and DSC and MFI tests were carried out to investigate the decrease of the melting point (Tm) and the melt viscosity. There existed a 12°C–15°C reduction of Tm and a significant increase of MFI before and after foaming. It could be concluded that the degradation did occur seriously in the remelt process.

Figure 3.30   SEM micrographs of PET foams obtained by separated process with foaming conditions: (a) 280°C, 12 MPa, 30 min; (b) 275°C, 12 MPa, 30 min; and (c) 275°C, 10 MPa, 30 min.

(From H. Zhong et al., The Journal of Supercritical Fluids, 74, 70–79, 2013.)

#### 3.6  Summary and Outlook

Some industrially promising processes aided by CO2 have been explored to tailor the structure of PET foams or simplify the manufacturing process of PET foams for high efficiency. For the melt-state foaming process of PET, a polymer matrix with high melt elasticity and strength was demanded, which could not be satisfied by commercial PET with linear molecular chain and relatively low molecular weight. Different modifying methods aiming to introduce the long-chain branch to the PET backbone were described in this chapter, including reactive extrusion, in-situ polymerization, and the so-called integrated process. It should be noted that a much-higher melt strength was essential in the extrusion foaming process because severe degradations were inevitable in the extruder, and the corresponding modifying process costs much more time and energy. Layered organoclay could also be employed to compound with the PET matrix to improve the melt foam-ability of PET with low melt elasticity. The cell collapse and coalescence were unavoidable in the melt foaming process, although the melt strength was improved a lot. PET foams prepared by the melt foaming process were always with a cell size of hundreds of microns and a cell density that is lower than 109 cells/cm3. In other words, no microcellular PET foams could be acquired. Microcellular PET foams (cell size < 10 μm, cell density > 109 cells/cm3) with excellent optical and mechanical properties were often produced by a solid-state foaming process. Crystallization, which could be facilitated due to the plasticization effect of CO2, was the most important factor in the solid-state foaming process rather than the viscoelastic properties of polymer matrix, and thus the PET with a linear molecular chain could be used as a raw material without further modifying. The solid-state foaming process described in this chapter prepared a controllable sandwich structure of PET foams with two microcellular or even ultramicrocellular foamed crystalline layers outside and a microcellular foamed amorphous layer inside. The thickness of foamed crystalline layer could be regulated by the CO2 saturation time and calculated by the proposed model coupling CO2 diffusion and CO2-induced crystallization of PET. In addition, the blowing agent supercritical CO2 could also assist the regulation of the foaming process due to the strong plasticization effect on the polymer matrix, such as inducing the crystallization and facilitating the melt modification.

PET foams with better cell morphology and foam property should be prepared through a more simplified process in order to develop more applications. The long-chain branch structure of PET destroyed the ordered molecular structure and decreased the crystallinity of PET, as shown in Table 3.2, which reduced the mechanical and other properties of PET. The crystallinity enhancement of melt foaming products was necessary, which may be carried out via the addition of nucleation agents and/or increasing the melt foamability of PET with a linear molecular chain. Furthermore, the intensive scission of a PET chain in the extrusion foaming process decreased the melt strength again and thus weakened the foamability of PET. Hence, some new continuous foaming processes should be developed to avoid the degradation in the CO2-dissolving stage. The integrated process involved in this chapter supplied an outstanding solution, which should be implemented in a continuous process, for example, in a falling film reactor with larger mass transfer interface that is operated under a high-pressure CO2 atmosphere.

#### Acknowledgments

The authors are grateful to the National Natural Science Foundation of China (Grant No. 21176070 and 21306043), the National Programs for High Technology Research and Development of China (863 Project, 2012AA040211), the Research Fund for the Doctoral Program of Higher Education of China (Grant No. 20120074120019), the Fundamental Research Funds for the Central Universities, and the 111 Project (B08021).

#### References

V. Kumar, R.P. Juntunen, C. Barlow. 2000. Impact strength of high relative density solid state carbon dioxide blown crystallizable poly(ethylene terephthalate) microcellular foams. Cellular Polymers, 19 25–37.
S. Doroudiani, C.B. Park, M.T. Kortschot. 1996. Effect of the crystallinity and morphology on the microcellular foam structure of semicrystalline polymers. Polymer Engineering & Science, 36 2645–2662.
V. Kumar, P.J. Stolarczuk. 1999. Microcellular PET foams produced by the solid state process. Imaging and Image Analysis Applications for Plastics, 241–247.
D.F. Baldwin, M. Shimbo, N.P. Suh. 1995. The role of gas dissolution and induced crystallization during microcellular polymer processing: A study of poly(ethylene terephthalate) and carbon dioxide systems. Journal of Engineering Materials and Technology, 117 62.
V. Kumar, H.G. Schirmer. 1997. Semi-continuous production of solid state polymeric foams. In US Patent 5684055.
L. Di Maio, I. Coccorullo, S. Montesano, L. Incarnato. 2005. Chain extension and foaming of recycled PET in extrusion equipment. Macromolecular Symposia, 228 185–200.
M. Xanthos, C. Wan, R. Dhavalikar, G.P. Karayannidis, D.N. Bikiaris. 2004. Identification of rheological and structural characteristics of foamable poly(ethylene terephthalate) by reactive extrusion. Polymer International, 53 1161–1168.
R. Dhavalikar, M. Yamaguchi, M. Xanthos. 2003. Molecular and structural analysis of a triepoxide-modified poly(ethylene terephthalate) from rheological data. Journal of Polymer Science Part A: Polymer Chemistry, 41 958–969.
M.Y. Wang, Z.Q. Guo, B.Y. Lei, N.Q. Zhou. 2011. Rheological and thermal behavior of recycled PET modified by PMDA. Advanced Materials Research, 391–392 688–691.
L. Xiao, H. Wang, Q. Qian, X. Jiang, X. Liu, B. Huang, Q. Chen. 2012. Molecular and structural analysis of epoxide-modified recycled poly(ethylene terephthalate) from rheological data. Polymer Engineering & Science, 52 2127–2133.
S. Japon, Y. Leterrier, J.-A.E. Manson. 2000. Recycling of poly(ethylene terephthalate) into closed-cell foams. Polymer Engineering & Science, 40 1942–1952.
D. Li, T. Liu, L. Zhao, W. Yuan. 2012. Controlling sandwich-structure of PET microcellular foams using coupling of CO2 diffusion and induced crystallization. Aiche Journal, 58 2512–2523.
T. Xia, Z. Xi, T. Liu, X. Pan, C. Fan, L. Zhao. 2015. Melt foamability of reactive extrusion-modified poly(ethylene terephthalate) with pyromellitic dianhydride using supercritical carbon dioxide as blowing agent. Polymer Engineering & Science, 55 1528–1535.
B. Zhu, W. Zha, J. Yang, C. Zhang, L.J. Lee. 2010. Layered-silicate based polystyrene nanocomposite microcellular foam using supercritical carbon dioxide as blowing agent. Polymer, 51 2177–2184.
W. Zhai, T. Kuboki, L. Wang, C.B. Park, E.K. Lee, H.E. Naguib. 2010. Cell structure evolution and the crystallization behavior of polypropylene/clay nano-composites foams blown in continuous extrusion. Industrial & Engineering Chemistry Research, 49 9834–9845.
P.H. Nam, P. Maiti, M. Okamoto, T. Kotaka, T. Nakayama, M. Takada, M. Ohshima, A. Usuki, N. Hasegawa, H. Okamoto. 2002. Foam processing and cellular structure of polypropylene/clay nanocomposites. Polymer Engineering & Science, 42 1907–1918.
M. Okamoto, P.H. Nam, P. Maiti, T. Kotaka, T. Nakayama, M. Takada, M. Ohshima, A. Usuki, N. Hasegawa, H. Okamoto. 2001. Biaxial flow-induced alignment of silicate layers in polypropylene/clay nanocomposite foam. Nano Letters, 1 503–505.
H. Zhong, Z. Xi, T. Liu, L. Zhao. 2013. In-situ polymerization-modification process and foaming of poly(ethylene terephthalate). Chinese Journal of Chemical Engineering, 21 1410–1418.
C. Fan, C. Wan, F. Gao, C. Huang, Z. Xi, Z. Xu, L. Zhao, T. Liu. 2016. Extrusion foaming of poly(ethylene terephthalate) with carbon dioxide based on rheology analysis. Journal of Cellular Plastics, 52 277–298.
S. Ye, K. Jiang, C. Jiang, Q. Pan. 2005. Dynamic supercritical fluid devolatilization of polymers. Chinese Journal of Chemical Engineering, 13 732–735.
S. Alsoy, J.L. Duda. 1998. Supercritical devolatilization of polymers. Aiche Journal, 44 582–590.
H. Zhong, Z. Xi, T. Liu, Z. Xu, L. Zhao. 2013. Integrated process of supercritical CO2-assisted melt polycondensation modification and foaming of poly(ethylene terephthalate). The Journal of Supercritical Fluids, 74 70–79.