Stereo-chemical contributions to the glass transition and liquid–liquid phase separation in high molecular weight poly(N-vinyl carbazole)

Abstract

The effects of thermal history and purification on poly(N-vinyl carbazole) (PVK) was investigated focusing on thermal, dielectric and structural analyses. A single glass transition temperature, T_g at 225 °C was observed in the case of the material as received, while two distinct T_gs were identified in the purified samples. Thermal annealing at a temperature above 275 °C also introduced a transition identical to that observed on purification. The nature of the two transitions was confirmed to be that of α-transitions having activation enthalpies of 400 ± 50 kJ mol⁻¹ as determined by dielectric relaxation spectroscopy. No change in molecular weight was observed and ¹H-NMR spectroscopy showed that there was little or no change in tacticity as a result of these treatments. It is suggested that liquid–liquid phase separation occurs by separation of isotactic rich segments from a matrix which is predominantly atactic, as indicated by the T_g temperature. It is considered that the liquid–liquid phase separation was driven by the differences in solubility of two stereo-isomers. This mechanism is preferred to that of phase separation by molecular weight driven by the higher molecular weight phase forming a liquid crystalline structure since the observed T_gs at 196, 215 and 225 °C have differences too great to be accounted for by changes in molecular weight alone but consistent with the known T_g values of the stereoregular polymers.

---

1 Introduction

Poly(N-vinyl carbazole), PVK, is a speciality rather than a commodity plastic; the combination of many excellent properties makes it a focus of attention of numerous research studies for application as a conducting and photo-conducting material. These properties include its unusual electrical properties as well as its ease of polymerisation by different mechanisms, its interesting morphological features and solution properties.^1–4 Moreover, the chemical modification of PVK has been successfully reported by sulfonation which occurs on the carbazole moiety. The degree of sulfonation was controlled by changing the molar ratios of the sulfonating agent to the carbazole unit and the yield of the sulfonation products was estimated to be 41–47%. This process makes the polymer chemically compatible with other components in a hybrid system.⁵

Purification is of prime importance in particular when the polymer is required to have excellent conductivity. The various electro-photographic characteristics of PVK, such as photoconductivity and photosensitivity, have been reported to increase greatly on purification.⁶ The purification of PVK is normally carried out by extraction or precipitation from solution.^7,8 Extraction has advantages over precipitation including shorter time and less quantity of the used solvent, but it requires special apparatus.⁸ In the precipitation process, the polymer is dissolved in a suitable solvent, such as N,N-dimethyl formamide, tetrahydrofuran (THF), benzene, toluene, or methylene chloride, and is precipitated by adding methanol.⁹ Generally, the chemicals used in purification do not change the chemical structure of the material.

We have observed that a sample of PVK with a single glass transition temperature, T_g, progressively develops two on purification or on heat treatment. This has been attributed to the development of incompatibility or liquid–liquid phase separation in the system. Bai et al.¹⁰ have pointed out that high molecular weight PVK samples exhibit two transitions on repeated heating and cooling cycles but they were unable to explain or assign the second transition which occurred at 203 °C.

In this paper, PVK was purified using different solvent/non-solvent systems and the effects of the purification process were studied focusing on dielectric and thermal behaviour. The effects of thermal annealing on the properties of PVK were also investigated using DSC. The change to molecular conformation introduced by purification or annealing was analysed by ¹H NMR spectroscopy.

2 Experimental

2.1 Materials and reagents

PVK powder was purchased from Sigma-Aldrich Co. It was reported as having a density of 1200 kg m⁻³, melting point of 300 °C, a glass transition of 220 °C and an average weight molecular weight of 1100 kg mol⁻¹.

Tetrahydrofuran (THF), chloroform and 1,1,2,2-tetrachloroethane were used as casting solvents and methanol, acetone and diethyl ether as non-solvents for the purification process. These were reagent grade from Sigma-Aldrich Co. and used without further purification.

2.2 Sample preparation

PVK was dried in a vacuum oven at 110 °C for several hours before characterisation. It was purified by repeating precipitation from different solvent/non-solvent systems under vigorous stirring, and dried in vacuum at 110 °C. Films were cast from 50 mg of the appropriate sample in 100 cm³ of chloroform. After the polymer had dissolved, approximately 25 cm³ were deposited in Petri dishes and left under fume-cupboard to evaporate overnight at room temperature. These were subsequently dried for several days in a vacuum oven at 110 °C. Finally, the films were peeled by tweezers and they were solvent free, good optical quality films with a thickness of 25 μm (measured by a digital micrometer).

2.3 Measurements

Thermal properties were measured using a differential scanning calorimeter (Perkin-Elmer model DSC-2) interfaced to a PC computer. Measurements were carried out in nitrogen at a sensitivity of 20 mW. Samples of mass 15 ± 3 mg were sealed in DSC aluminium pans and heated from 120 to 270 °C at various heating rates. Temperature and enthalpy calibrations were made with ultra-pure metal standards: indium (melting point, mp: 156.63 °C, ΔH = 29.2 J g⁻¹), tin (mp: 231.91 °C) and lead (mp: 327.50 °C). A baseline was established for the instrument using two empty aluminium DSC pans and subtracted from the heat flow-temperature response.

The change in dielectric constant, ε′, dielectric loss, ε′′, and tan δ with applied frequency and temperature was followed using a dielectric thermal analyser, DETA, manufactured by Polymer Laboratories Ltd. The measuring cell was equipped with circular parallel plate electrodes of 20 mm in diameter and placed in a thermo-stated furnace with a temperature range from −150 to 300 °C. The DETA measurements were carried out using PVK films with thicknesses between 15 to 35 μm. A fixed voltage of 1.0 V was used with a set of frequencies in the range of 10 to 10⁵ Hz and in the temperature range from room temperature up to 285 °C.

FTIR spectra were measured on Nicolet spectrophotometers, models 1860 and 8700 with a DTGS-KBR detector using the KBr method. All spectra were recorded at a resolution of 4 cm⁻¹ and total of 64 scans were accumulated for each spectrum along with the background.

High resolution ¹H-NMR spectra were measured using a Bruker AC 300 spectrometer. Solutions of the polymer were prepared in deuterated chloroform with a small quantity of tetramethylsilane (TMS), as reference.

3 Results and discussions

3.1 The effects of purification

Generally, commercial polymer can be contaminated with up to 6% monomer, NVK, up to 500 ppm of carbazole, anthracene, and sulfur compounds in the ppm range.⁷ Since, the solubility of polymer in non-solvent decreases with increasing molecular weight, monomer as well as low molecular weight PVK are dissolved in the solvent and on precipitation the impurities and monomer content is estimated to be reduced by a factor of 10 on each cycle.^6,11 In this work, PVK was purified by precipitation using different solvent/non-solvent systems as quoted in Table 1. The process was repeated 3–7 times and the recovery yields were higher than 90%.

Table 1 PVK samples preparation, purification and characterisation

Sample ID	Purification	Characterizations
	Solvent/non-solvent	DSC observation^a Tg ± 1.0 °C		GPC analysis			^1H NMR IH/IL
				Mn	Mw	PD	±0.05
				kg mol^−1
a Extrapolated to zero heating rate.
PVK-0	As received	A single Tg	225.0	78.8	378	4.8	0.32
PVK-1	Chloroform/methanol	Two Tgs	196.0	106.5	374	3.5	0.41
			208.1
PVK-2	Chloroform/diethyl ether	Two Tgs	194.5	—	—	—	0.35
			209.8
PVK-3	Chloroform/acetone	Two Tgs	193.5	—	—	—	0.38
			214.1
PVK-4	Tetrachloroethane/methanol	A single Tg	225.0	—	—	—	0.31
PVK-5	Thermally annealed at 300 °C for 40 minutes and scanning at 40 °C min^−1	Two Tgs	205.9	74.6	380	5.1	0.33
			233.2

---

Gel permeation chromatography (GPC) was carried out on PVK samples in tetrahydrofuran solutions before and after purification. The molecular weight averages are quoted in Table 1 as polystyrene equivalents. The number average, M_n, increased on purification while the polydispersity, M_w/M_n, decreased due to loss of low molecular weight material. Similar values have been reported for commercial PVK but the weight averages were higher, i.e. M_w = 1.5 × 10⁶, PD ≈ 5.5.^4,12

The samples were characterized by DSC to evaluate their glass transition temperatures, T_g, see Fig. 1 and one T_g was observed for the original un-purified material, at 225 °C consistent with the polymer being atactic.^4,13 However, two T_gs were observed at 193–196 and 208–214 °C with the purified samples. Tetrachloroethane/methanol alone reproduced the original material with a single T_g. The presence of two transitions was taken to indicate the presence of two glass phases which separated on precipitation of these samples from chloroform. Similar results were found using different heating rates, i.e. 10 and 20 °C min⁻¹.

Fig. 1 DSC thermal response of PVK samples before and after purification (heating rate is 40 °C min⁻¹).

Sample PVK-1 (75 mg) was dissolved in chloroform (20 cm³) and fractionated by drop wise addition of methanol as a non-solvent. The appearance of a milky precipitate was taken to be the end point of each fraction and the resulted solid was filtered, weighed and analyzed by DSC. Two fractions were isolated after the addition of 3.1 and 4.0 cm³ of methanol. No solid could be isolated even after the addition of 8.0 cm³ of methanol as it was too small to isolate. In the DSC experiment of the isolated fractions (15 mg and 30 mg), samples were heated to 270 °C and held for 2 minutes then cooled down. The T_g was recorded on reheating at 40 °C min⁻¹. This procedure eliminated any residual solvent and imposed a standard cooling rate on cooling through the transition. As shown in Fig. 2, two T_gs were observed and were identical to the observed transition with the purified samples. This imply that the observed splitting of T_g is not related to molecular weight.

Fig. 2 DSC thermal response of PVK-1 and the two fractions obtained after addition of 3.1 and 4.0 cm³ of methanol (heating rate is 40 °C min⁻¹).

To confirm that the two transitions were second order, the samples were physically aged for various periods at 5–10 °C below each T_g and the extent of enthalpic relaxation measured from the endotherm produced on measuring the glass transition by DSC,¹⁴ see Fig. 3(a) and (b). Annealing at 200 °C produced considerable amounts of ageing in the lower transition, as shown by the large increase in the endothermic peak while ageing at 220 °C only enable the higher transition to age. Clearly while the two transitions relax towards the mobile liquid on physical ageing they appear to behave independently of one another and implied that two separate and distinct glasses were present where previously there was one.

Fig. 3 Enthalpic relaxation in sample PVK-1; (a) below the lower transition and (b) between the two transitions above the lower and below the second transition. Heating rate is 40 °C min⁻¹.

3.2 The effects of thermal history

PVK, as received, was not crystalline but since thermal annealing has been reported to increase the polymer's crystallinity,^4,15 the samples were heated, at 300, 340 and 360 °C to induce them to crystallize. Further heating to higher temperatures showed no evidence of melting and sample degradation alone was observed.

Samples of PVK were also annealed in nitrogen in the temperature range from 280 to 330 °C and for extended periods of time, see Fig. 4(a) and (b) as examples. A lower temperature transition was again observed at 205 ± 1.0 °C, consistent with that observed on precipitating the samples in different solvents/non-solvents and in agreement with the literature.¹⁰ Moreover, the lower transitions developed progressively with increasing annealing period.

Fig. 4 DSC thermal response of sample PVK-0 after annealing at 300 °C (a) and at 325 °C (b). Heating rate is 40 °C min⁻¹.

GPC was carried out on a sample of PVK-0 after annealing at 300 °C in flowing N₂ to investigate the presence of degradation or cross-linking. There were only slight changes in the MW averages in comparison with PVK-0 before annealing, see Table 1. It was concluded that no cross-linking or chain scission had occurred on annealing and annealing had little or no effect on the measured molecular weights.^12,16

To study further the effect of repeated heating on PVK, samples PVK-0 and PVK-1 were cast as films using chloroform and then dried at 110 °C in air for several days. The dielectric thermal responses of these samples were recorded using different frequencies in the temperature range from 140 to 285 °C.

The dielectric relaxation spectrum of PVK has been reported in the temperature range from −180 to +240 °C and four relaxation processes were observed.¹⁷ In the temperature range used in our study, only the α-relaxation, which is associated with the glass transition of the polymer, was considered. Two relaxation processes were observed at 260 and 240 °C. On cooling and re-heating, these shifted to lower temperatures, 240 and 210 °C respectively confirming the previous DSC observations; see Fig. 5(a). A close examination of Fig. 5(a) and (b) revealed that the dielectric loss maximum is reduced in intensity on repeated scans of the sample, a second maximum developed after the third scan and increased in intensity on further scans; both maxima in the dielectric loss shifted to higher temperature with increasing frequency.

Fig. 5 Dielectric loss as a function of temperature for sample PVK-0; (a) at 100 Hz, (b) different frequencies and (c) Arrhenius plots of the two transitions.

In order to establish the nature of the dielectric transition the frequency dependences of the temperature corresponding to the maxima in the dielectric loss, both the 1^st and 2^nd, were analysed as Arrhenius plots, according to the following equation:

f = A exp(−ΔH/RT) (1)

where ΔH is the activation enthalpy (J mol⁻¹), R is the gas constant (J mol⁻¹ K⁻¹), A is a constant, T is the temperature of relaxation (K) and f is the frequency (Hz).¹⁸

As shown in Fig. 4(c), two distinct dependences were observed with similar slopes corresponding to activation enthalpy of 400 ± 50 kJ mol⁻¹. This agrees with a similar study in the literature¹⁷ and obviously both transitions are second order.

3.3 FTIR spectroscopic analysis

The FT-IR spectrum of PVK has been reported and discussed in the literature.^19–21 A typical spectrum in the fingerprint region of PVK is shown in Fig. 6. The absorption bands are assigned according to the groups in the repeating unit below,

Fig. 6 FTIR spectra of PVK samples.

Important spectral features of PVK include deformation of the substituents on the aromatic ring at 722 cm⁻¹, CH₂ rocking vibration due to a tail-to-tail addition at 744 cm⁻¹, C–H in-plane deformation of the aromatic ring and the C–N stretching vibration of ethylene carbazole group at 1220–1230 cm⁻¹, C–H in-plane deformation of the vinylidene group at 1320 cm⁻¹ and of the aromatic ring at 1130–1160 cm⁻¹ and ring vibration of the carbazole moiety at 1460–1480 cm⁻¹, the CH₂ deformation of the vinylidene group at 1410 cm⁻¹, stretching of ethylene group at 1600–1630 cm⁻¹, and the C–C and C=C stretching of the benzene ring at 1595 and 1630 cm⁻¹.^19–21 No sulfonic acid group was present as a trace impurity.

FT-IR spectroscopy was carried out on samples after purification and on annealing. No differences in the spectra were observed compared to that of the original sample PVK-0, see Fig. 6. There were no carbonyl or peroxide peaks present which could be attributed to oxidation and no chemical change on precipitation from solution or on annealing which could account for the changes in the observed T_g.

3.4 ¹H-NMR spectroscopic analysis

¹H-NMR spectra of the PVK samples are shown in Fig. 7, and the peaks were consistent with those reported in the literature.^4,22,23 Repeated dissolution and precipitation, three to seven times, was sufficient to remove the lowest molecular weight impurities which appeared as a sharp peak at 2.2 ppm. The bands at 3.70 and 3.85 ppm are attributed to methine protons, and 1.1 to 1.60 ppm to methylene ones. The others are due to aromatic C–H bonds. The ratio of the areas of methine peaks, I_H/I_L was taken to be a measure of isotactic to syndiotactic dyads from which the isotactic fraction was determined.²² The results are quoted in Table 1, which indicated that there were little changes on tacticity with repeated purification cycles, but it also showed that the two phase polymers did not change after thermal annealing.

Fig. 7 ¹H-NMR spectra of different PVK samples shown in Table 1.

As shown in Fig. 7, there was an interesting different behavior of the methylene protons (1–2 ppm) on purification and thermal treatment; the peaks sharpen and increased in intensity after thermal treatment, but reduced with increasing the number of purification cycles. It has been reported that the ¹H NMR spectra of PVK is temperature dependent and the shape and multiplicity of the peak changes with increasing temperature.^4,22,23 As we have done our work at room temperatures, this difference could be attributed to improved backing efficiency or due to orientation of the sample on heating treatments.

4 Discussion

Annealing PVK at a temperature as high as 275 °C split the glass transition into two separate transitions which have the characteristics of second order transitions; there was no change in enthalpy but only a step change in specific heat;²⁴ it underwent physical ageing on ageing below the transition temperatures; DETA analysis showed that both transitions have large but similar activation enthalpies. Both glass transitions were lower than the original and so cannot be explained in terms of the sample crystallizing or separating into high and low molecular weights. Similar effects were observed on precipitating the polymer from solution. The observed change in molecular weight averages could not account for the lowering of the glass transition. Apart from loss of low molecular weight additives, there was no crosslinking, chain session, chemical decomposition or chemical reaction i.e. complex formation in both purified sample and the sample held at high temperature. This was confirmed using different techniques including NMR, GPC and FTIR.

Three separate values for the T_g of PVK have been reported for samples prepared with a cationic catalyst, and have been explained as the results of stereo-blocks present in the polymer chains. These three T_g values are 126, 227 and 276 °C corresponding to isotactic, atactic and syndiotactic polymer respectively.^4,13 For ideal solution behaviour changing the tacticity by increasing the isotactic fraction should lower the T_g following the ideal behaviour line shown in Fig. 8.

Fig. 8 The variation of glass transition with tacticity.

The observed dependence of T_g on isotactic fraction for the solution precipitated polymers deviates below the ideal line but this may reflect the assumption of ideal behaviour and random sequences of d and l isomers. However, there is a decrease in T_g with increasing isotactic fraction, see Fig. 8. Okamoto et al.²² have polymerized N-vinyl carbazole with various initiators which altered the isotactic/syndiotactic ratio. Their results are also presented in Fig. 8. Only the low isotactic fraction polymers exhibited the same dependence as that observed by the ideal behaviour line. This could of course arise from non-random statistics of the co-ordinated ionic initiator used to prepare these higher isotactic polymers. This also indicates that ¹H NMR does not measure the individual composition of the two phases but an average over the two phases. For the present results, if the T_g values reflect compositional differences then the lower T_g is richer in isotactic isomer and would be at 0.5 mole fraction.

To throw more light on this interesting phenomenon, PVK samples were thermally annealed for various time periods at temperatures in the range 280 to 325 °C and over periods up to 30 min, as shown in Table 2. It was observed that phase separation, as indicated by the two T_gs, was time dependent which is characteristic of the coexistence curve and not the spinodal. Each annealed sample exhibited an upper and lower glass transition; the lower is decreasing with decreasing annealing temperature.

Table 2 Effect of annealing temperature on the glass transition of PVK. Heating rate is 40 °C min⁻¹

Annealing temperature (°C)	Annealing time (min)	Glass transition temperature Tg ± 1.0 (°C)
		Tg(1)	Tg(2)
282	0	—	235.4
	10	—	235.4
	30	205.0	234.8
292	0	—	235.2
	16	205.2	235.2
302	0	—	235.4
	15	206.2	234.2
	40	207.7	235.5
325	0	—	234.2
	2	205.3	233.7
	5	205.4	232.4
	20	205.8	231.9

---

We can utilize the big difference in the measured T_g values of the two components to determine the tacticity of liquid phases which separate and construct a phase diagram for PVK system.^25,26 This is shown in Fig. 9 where the isotactic content was determined for each component from the measured T_g values assuming the applicability of the following equation:^4,13

(T_g − 399)X_iso + (T_g − 549)X_syn = 0 (2)

Fig. 9 Phase diagram for PVK showing one and two component liquid phases and the composition of the two phases x₁ and x₂ which separate on heat treatment. Initial isotactic composition 0.46 defines the Lower Critical Solution Temperature (LCST) for PVK.

The data obtained from purification and ¹H NMR analysis were also inserted in Fig. 9.

We conclude that phase separation of different stereo-chemical compositions take place in PVK on heat treatment and on precipitation from chloroform resulting in isotactic poor and rich glasses and liquid/liquid phase separation occurs. These phases exhibited an upper and lower T_g. The presence of two well separated transitions is consistent with liquid phase separation corresponding to the phase with the lower temperature T_g being richer in isotactic dyads and the other with an increased T_g richer in syndiotactic dyads.

We also conclude that PVK has a lower critical solution temperature and the polymer melt will undergo liquid–liquid phase separation by the increasing insolubility of the syndiotactic dyads in the isotactic dyads with increasing temperature and accordingly the phases which separate have a different tacticity.

5 Conclusion

Poly(N-vinyl carbazole) (PVK) was purified using different solvent/non-solvent reagents and also subjected to thermal annealing above the glass transition temperature. Two distinct glass transitions were observed in both cases as revealed from DSC and DETA experiments. This analysis indicated that liquid/liquid phase separation had occurred producing two incompatible amorphous phases which differed either in molecular weight or in stereo-regular composition. FT-IR and ¹H NMR spectroscopies further showed that there was no overall change in tacticity in the phase separated PVK although the individual liquid phases differed and this accounted for the presence of two well separated glass transitions.

This study is important as it details the effects of two important processes; purification and annealing; that are usually used in the preparation of any industrial product. For example, it has been shown that the acceptance potential, contrast potential, photosensitivity of PVK are greatly increased with increasing the number of purification cycle and it reached a maxima after 12 purification cycles.⁶

It also throws lights on the effects of solvents and thermal treatment on the interesting morphological features of PVK which has an effect on its liquid crystalline structure which is associated with the higher molecular weight phase only.¹⁰

Acknowledgements

The authors are grateful to the Atomic Energy Commission of Syria for their financial support during the tenure of the work. Thanks are also due to Mr F. Biddlestone for technical support.

References

1. T. Ye, J. Chen and D. Ma, Phys. Chem. Chem. Phys., 2010, 12, 15410–15413.
2. S. Lee, S.-M. Yoon, H.-J. Shin, W.-J. Joo, D. K. Yi, J.-Y. Choi, C. A. Amarnath and U. Paik, Nanotechnology, 2008, 19, 5606.
3. B. M. Bonon and T. D. Atvars, Photochem. Photobiol., 2012, 88, 801–809.
4. R. C. Penwell, B. N. Ganguly and T. W. Smith, J. Polym. Sci., Part D: Macromol. Rev., 1978, 13, 63–160.
5. S. Wang, Z. Zeng, S. Yang, L.-T. Weng, P. C. L. Wong and K. Ho, Macromolecules, 2000, 33, 3232–3236.
6. S. K. Agarwal, S. S. Hemmadi and N. L. Pathak, Polymer, 1979, 20, 867–871.
7. R. Aldag and P. Neumann, US Pat., 4 987 220, 1991.
8. C. R. Costin, T. W. Hazell, G. W. Ceska and J. P. Horgan, US Pat., 5 175 248, 1992.
9. J. K. Fink, in High Performance Polymers, A Volume in Plastics Design Library, Elsevier Inc., USA, 2014, pp. 1–42.
10. Y.-W. Bai, X.-F. Chen, X.-H. Wan and Q.-F. Zhou, Chin. J. Polym. Sci., 2002, 20, 155–159.
11. W. Klöpffer, J. Chem. Phys., 1969, 50, 2337–2343.
12. R. C. Penwell and W. M. P. Jr, Polymer, 1978, 19, 537–541.
13. J. M. Pearson, Vinylcarbazole Polymers in Encyclopedia of Polymer Science and Engineering, Wiley, New York, 1989, p. 257.
14. J. N. Hay, Pure Appl. Chem., 1995, 67, 1855–1858.
15. A. Kimura, S. Yoshimoto, Y. Akana, H. Hirata, S. Kusabayashi, H. Mikawa and N. Kasai, J. Polym. Sci., Part B: Polym. Phys., 1970, 8, 643–648.
16. J. C. Chu and M. Stolka, J. Polym. Sci., Polym. Chem. Ed., 1975, 13, 2867–2870.
17. J. M. Pochan, D. F. Hinman and R. Nash, J. Appl. Phys., 1975, 46, 4115–4119.
18. A. G. Al Lafi, Polym. Bull., 2012, 68, 2269–2283.
19. M. Baibarac, M. Lira-Cantú, J. O. Sol, I. Baltog, N. Casañ-Pastor and P. Gomez-Romero, Compos. Sci. Technol., 2007, 67, 2556–2563.
20. M. Baibarac, I. Baltog, S. Lefrant and P. Gomez-Romero, Polymer, 2007, 48, 5279–5288.
21. P. Bertoncello, A. Notargiacomo and C. Nicolini, Polymer, 2004, 45, 1659–1664.
22. K.-I. Okamoto, M. Yamada, A. Itaya, T. Kimura and S. Kusabayashi, Macromolecules, 1976, 9, 645–649.
23. A. Karali, P. Dais, E. Mikros and F. Heatley, Macromolecules, 2001, 34, 5547–5554.
24. S. Z. D. Cheng, Phase Transitions in Polymers: The Role of Metastable States, Elsivier, 2008.
25. Y. Xu, W. Yu and C. Zhou, RSC Adv., 2014, 4, 55435–55444.
26. P. Atkins and J. D. Paula, Physical Chemistry, W. H. Freeman and Company, New York, USA, 9th edn, 2010, ch. 5, pp. 156–198.

---