Impact of Hard Domains on Mechanical Properties of Hybrid Networks, Notas de estudo de Engenharia de Produção

Baixe Impact of Hard Domains on Mechanical Properties of Hybrid Networks e outras Notas de estudo em PDF para Engenharia de Produção, somente na Docsity! Polyurethane and unsaturated polyester hybrid networks: 2. Influence of hard domains on mechanical properties Ludovic Valette, Chih-Pin Hsu* Cook Composites and Polymers, Corporate Research and Development, 820 E. 14th Avenue, North Kansas City, MO 64116, USA Received 9 June 1997; revised 17 April 1998; accepted 28 May 1998 Abstract The influence of hard domains on the mechanical and thermomechanical properties of polyurethane and unsaturated polyester hybrid networks has been investigated. The hybrid networks consist of a polyurethane linkage formed by reacting unsaturated polyester polyol with polymeric 4,49-diphenylmethane diisocyanate (MDI) and free-radical crosslinking through styrene monomer and vinylene groups in the unsaturated polyester. Hard segments were formed by condensing two different types of chain extender, ethylene glycol (EG) and 1,6- hexanediol (HD), with MDI. Incorporation of chain extenders in the hybrid networks varied from 0% to 12% by weight based on the weight of unsaturated polyester polyol. The thermomechanical properties of the polyurethane and unsaturated polyester hybrid networks were characterized by heat distortion analysis and by dynamic mechanical analysis. Flexural three-point bend tests and unnotched Izod impact analysis were used to investigate mechanical properties at ambient temperature. Hybrid networks with hard segments formed by MDI and EG showed an increase in the glass transition temperature. A second glass transition was found with incorporation of more than 6 wt% EG due to the formation of phase-separated hard domains. The rubber plateau of the hybrid networks decreased owing to the lower crosslinking density when chain extenders were incorporated. Phase-separated hard domains enhanced the rubber plateau by acting as physical crosslinks in the hybrid network until the glass transition was reached. The hybrid network had improved mechanical properties when more hard segments were added into it without creating the phase-separated hard domains. A dramatic drop in mechanical properties was observed for the sample with a two-phase structure. q 1999 Elsevier Science Ltd. All rights reserved. Keywords: Unsaturated polyester; Polyurethane; Hybrid networks 1. Introduction Unsaturated polyester resins are one of the most widely used thermosetting materials in the composites industry. They offer reasonably good mechanical properties and are relatively inexpensive. The properties of the cured resins can be enhanced by adding various additives to the unsatu- rated polyester. The fracture properties of cured resins can be improved by blending these materials with reactive liquid rubbers [1,2]. The shrinkage that occurs during the crosslinking reaction of the unsaturated polyester resins with styrene can be eliminated by the incorporation of low-shrinkage or low-profile agents [3,4]. Improvements in the thermal properties and elastic modulus have been investigated recently by creating a polymer network of unsaturated polyester and bismaleimide resins [5]. The mechanical properties of the unsaturated polyester resin can be greatly improved by incorporating the polyurethane linkage into the polymer network. The mechanical proper- ties of polyurethane and unsaturated polyester hybrid net- works also can be altered with the techniques used in segmented polyurethanes. Segmented polyester polyur- ethanes are well-known materials that have been studied extensively by Cooper and co-workers [6–11] as well as other researchers [12–14]. The polyurethane and unsatu- rated polyester hybrid networks consist of soft segments from the crosslinked unsaturated polyester and hard seg- ments from the extension of a diisocyanate with a low-mole- cular-weight diol. The structures of hybrid networks are different from those of segmented polyurethanes. The poly- urethane and unsaturated polyester hybrid networks are thermosetting materials, while segmented polyester polyur- ethanes are usually thermoplastics. The unique structure of the polyurethane and unsaturated polyester hybrid networks mentioned in this study was dis- cussed in a previous paper [15]. The soft segment is com- posed of a styrene/unsaturated polyester crosslinking network and forms the continuous phase for most of the 0032-3861/99/$ - see front matter q 1999 Elsevier Science Ltd. All rights reserved. PII: S0032-3861(98)00428-5 * Corresponding author. Tel.: +1-816-391-6000; Fax: +1-816-391-6254 Polymer 40 (1999) 2059–2070 samples. The hard segment comprises a polymeric 4,49- diphenylmethane diisocyanate (MDI) condensed either with ethylene glycol (EG) or 1,6-hexanediol (HD). The hard segment can be either the overall polyurethane seg- ment or simply the chain-extended polyurethane [15]. If two or more hard segments stay next to each other through hydrogen bonding, they could form a region rich in hard segments called a hard domain. The hard domains may be dispersed in the soft domain, forming a phase-mixed struc- ture, or they may coagulate and create phase-separated hard domains. The phase-separated hard domains may also con- tain soft segments because they are linked to the hard seg- ments through urethane linkages in the hybrid network. The previous study investigated the influence of hard domains on phase structure and its influence on thermal properties [15]. This paper aims to determine the influence of hard domains on the mechanical behaviour of polyurethane and unsaturated polyester hybrid networks. Flexural tests and impact analysis were run at ambient temperature. Thermomechanical properties were investi- gated by heat deflection analysis and dynamic mechanical analysis. The elastic behaviour of polymer networks can be described by either the affine or the phantom network model. For the affine network model [16], the shear mod- ulus is given by: Gaff ¼ nRT (1) where n is the molar number of elastic chains per unit volume of network, R is the gas constant and T is absolute temperature. The phantom network [17] also considers the effect of elastically active junctions. The shear modulus of the phantom network is lower than that of an affine network and is given by: Gphant ¼ (n ¹ m)RT (2) where m is the molar number of elastically active junctions per unit volume of the network. The phantom network usually describes the elasticity of perfect networks. However, most of the networks have less than perfect elasticity in real networks. Non-idealities such as pendant chains will decrease n and entrapped entanglements will increase n [18]. The molecular chains will also interact with each other and reduce the junction fluctuations. In the case of strong interactions, the junctions do not fluctuate at all and are displaced affinely with macroscopic strain [19]. The shear modulus is equal to Gaff. Because strong interaction among molecular chains exists in the hybrid network, the shear modulus G of the hybrid network will be correlated by the affine network model as: G ¼ rRT Mc (3) where r is the network density at the given temperature T and 〈Mc〉 is the number-average molecular weight between crosslinks. 2. Experimental 2.1. Materials Polyurethane and unsaturated polyester hybrid networks were made with the same materials as described in a pre- vious paper [15]. The unsaturated polyester polyol resin was made by reacting an excess of diethylene glycol with 50 mol% of isophthalic acid and 50 mol% of maleic anhy- dride. It was dissolved in styrene monomer with a 76% solids content. The modified 4,49-diphenylmethane diiso- cyanate (MDI) was provided as a 90% solids content solu- tion in styrene. The isocyanate content of the MDI solution was 20.5%. Chain extenders were two a,q-aliphatic diols: ethylene glycol (EG) and 1,6-hexanediol (HD). Various amounts of chain extender were blended into the unsatu- rated polyester polyol in styrene solution before preparing the hybrid network. All materials were used as-received without further purification. The polyurethane and unsaturated polyester hybrid net- works were prepared by reacting the unsaturated polyester polyol solution with MDI solution and about 1.5 wt% of benzoyl peroxide at ambient temperature. The reaction between MDI and hydroxyl groups took place first, forming the polyurethane linkage. The crosslinking reaction even- tually occurred through the unsaturated polyester and styr- ene monomer. Two sets of samples were prepared with various EG or HD contents and their compositions are listed in Table 1. Clear castings about 3 mm in thickness were prepared by moulding the samples between two glass plates. The clear castings were cured at ambient temperature for 24 h to convert the isocyanate groups fully and then post- cured at 1208C for 1 day to complete the unsaturated polye- ster/styrene crosslinking reaction. The clear castings were cooled slowly for 2 h to ambient temperature after postcure. Quenched samples were prepared by heating postcured sam- ples to 2308C and then quickly cooling them in cold water. Table 1 Composition of the samples used in this study Chain extender Wt%a Sample name Molar ratiob None 0 REF (1:0.81:0) EG 1.5 EG1.5 (1:1.02:0.27) 3 EG3 (1:1.25:0.55) 4.5 EG4.5 (1:1.48:0.84) 6 EG6 (1:1.72:1.14) 8 EG8 (1:2.04:1.54) 10 EG10 (1:2.38:1.96) 12 EG12 (1:2.74:2.42) HD 3 HD3 (1:1.05:0.28) 6 HD6 (1:1.29:0.6) 10 HD10 (1:1.63:1.04) aWt% calculated with respect to the unsaturated polyester polyol solution. bMolar ratio based on pure products (without styrene): (x:y:z) ¼ unsaturated polyester resin:modified MDI:chain extender. 2060 L. Valette,C.-P. Hsu/Polymer 40 (1999) 2059–2070 plateau was a constant for all HD and EG samples at 1.5 GPa. This value agrees with the storage shear modulus of segmented polyurethanes found in the literature [26,27]. The G9 rubber plateau varied with the content of hard seg- ments. The more chain extender incorporated, the lower was the ultimate rubber plateau due to the change in crosslinking density. The weight fraction of unsaturated polyester polyol in the resin mixture decreased when more chain extender was added to the system. Since crosslinks come from the reaction between unsaturated polyester and styrene, a lower crosslinking density is expected with higher incorporation of chain extender. To correlate the rubber plateau modulus with the hybrid resin composition, the number-average molecular weight between crosslinks, 〈Mc〉, was chosen for the calculation. 〈Mc〉 was calculated from: Mc ¼ 422:4 wUPR (4) where wUPR is the weight fraction of pure unsaturated polye- ster polyol resin. The value of 422.4 is the number-average molecular weight between unsaturations as calculated from the polymer composition. Eq. (4) does not consider the effect of pendant chains which exist in the hybrid network. Table 3 lists the 〈Mc〉 values of all samples calculated from Eq. (4). The relationship between G at the rubber plateau and 〈Mc〉 as given by Eq. (3) was examined at 120 and 1808C. The entire hybrid network was in the rubbery state at 1808C, while only the soft domain of the phase-separated hybrid network reached the rubbery state at 1208C. The hybrid network density was calculated from the specific thermal expansion of 5 3 10¹4 cm3 g¹1 K¹1 [28] and a network density of 1.20 g cm¹3 at 258C. The calculated hybrid net- work density was equal to 1.14 g cm¹3 at 1208C and 1.10 g cm¹3 at 1808C. The rubber storage shear modulus, G0, was used in the comparison since the phase angle was small at both temperatures. The comparisons of measured and calculated rubber storage shear modulus, G9, as a func- tion of 1/〈Mc〉 are shown in Fig. 2. A very good correlation between the measured and calculated results was observed for the phase-mixed system. However, the predicted results were slightly higher than the experimental results. The dif- ference is due to pendant chains which were not included in the 〈Mc〉 calculation. The pendant chains should decrease n and result in a lower G0 as described by Patel et al. [18]. The intermediate storage shear modulus measured at 1208C for phase-separated samples was much higher than the value predicted by Eq. (3). This is due to the existence of phase- separated hard domains. The phase-separated hard domains act like physical crosslinks within the rubber soft domain when the temperature is lower than their Tg. Fedors described the phenomenon as ‘virtual crosslinking’, although no chemical crosslinks are present [29]. Therefore, the number-average molecular weight between crosslinks was smaller than the value calculated by Eq. (4). The phase structure of the hybrid network can be changed by quenching the sample from the rubbery state to the glassy Table 2 Transition temperatures determined by h.d.a., d.m.a. and modulated d.s.c. (in 8C) HDT D.m.a. D.s.c. Peak b of tan d Peak a of tan d Peak b of G0 Peak a of G0 Peak height of b trans. in tan d Tg REF 54 70.8 — 55 — 0.89 50.2 EG1.5 53.5 72.5 — 57 — 0.92 52.4 EG3 59 75.9 — 65 — 1.04 56.5 EG4.5 55 77.6 — 62 — 1.07 60.2 EG6 63 81.1 125 65 125 1.03 62.5 EG8 61.5 82.5 130 67 130 0.87 62.5 130.0a EG10 71 86.4 135 72 132 0.62 65.0 136.6a EG12 69 87.6 138 72 133 0.56 67.2 134.6a HD3 52 71.1 — 55 — 0.9 51.6 HD6 55 71 — 60 — 1.01 52.7 HD10 58 71.4 — 60 — 1.04 52.2 aSecond Tg. Table 3 Average-number molecular weight between crosslinks, 〈M c〉 (in g) Incorporation of chain extender (wt%) EG HD 0a 745 745 1.5 808 3 873 823 4.5 940 6 1009 905 8 1104 10 1203 1024 12 1307 Number-average of molecular weight between unsaturations for pure unsaturated polyester polyol ¼ 422 g/CyC. aREF. 2063L. Valette,C.-P. Hsu/Polymer 40 (1999) 2059–2070 state as described in the previous paper [15]. Fig. 3 shows G9 and tan d vs. temperature for EG12 before and after quenching. The hybrid network changed from a phase- separated system into a phase-mixed system after quenching. The high-temperature a transition was not present on G9 nor tan d curves after quenching. G9 yield as well as tan d peak were much broader for the quenched samples. The half-height width of the tan d peak changed from 408C to 608C for EG12 after quenching. The height of the tan d peak also increased after quenching. These results indicate that the volume fraction of the soft phase was higher since the hard domains were dissolved in the soft domain instead of forming phase-separated hard domains. The dynamic mechanical properties of these samples were also investigated by creating master curves of shear storage modulus and tan d vs. frequency following the time–temperature superposition (TTS) principle [30] using 1068C as the reference temperature. The shift factor, aT, is defined by the Williams–Landel–Ferry (WLF) Fig. 2. Rubber storage shear modulus, G9, vs. 1/〈M c〉: (W) EG at 1208C; (A) HD at 1208C; (X) EG at 1808C; (B) HD at 1808C; (- - -) extrapolation of G9 by Eq. (3) at 1208C (rT ¼ 448 g K cm¹3); (——) extrapolation of G9 by Eq. (3) at 1808CðrT ¼ 496 g K cm¹3). Fig. 3. Storage shear modulus G9 and tan d of EG12 before and after quenching. 2064 L. Valette,C.-P. Hsu/Polymer 40 (1999) 2059–2070 equation as shown below: log aT ¼ ¹ c01 T ¹ T0 ÿ
c02 þ T ¹ T0 ÿ
(5) where c01 and c 0 2 are the shift coefficients and T0 the shift reference temperature. c01 and c 0 2 appeared to be constant for a given chain extender within experimental error. c01 and c 0 2 were 14.3 and 168.48C respectively for hybrid networks with EG incorporation, and 12.9 and 165.68C for hybrid networks with HD as chain extender. They are of the same order of magnitude as those found in the literature [30,31]. The activation energy, Ea, of the shift factor was calculated by the Rheometrics software using an Arrhenius equation of the form: aT ¼ Aa exp ¹ Ea RT   (6) where aT is the shift factor and Aa is a constant. All samples showed good correlation, with a correlation factor r2 between 0.85 and 0.97. The average activation energy for polyurethane and unsaturated polyester hybrid networks, 〈Ea〉EG, was 282 kJ mol¹1 with EG incorporation. The aver- age activation energy for hybrid networks with HD as chain extender, 〈Ea〉HD, was 242 kJ mol¹1. This is in the same range of activation energy found for polyurethane systems by Senich and MacKnight [26] and by Hartmann and co- workers [31,32]. Changes in the dominant relaxation frequency of the mas- ter curves, f0, have been shown to be directly related to changes in Tg determined by the maximum of tan d [33]. For a phase-mixed polymer system the relationship between log(f0) and 1/Tg is linear. Samples with a higher f0 have a lower Tg. Experimental results are given in Fig. 4 showing log(f0) plotted against 1/Tg. The b transition of tan d in d.m.a. tests was used as the Tg to be consistent with the method. All HD samples were almost superimposed because they had about the same Tg. Polyurethane and unsa- turated polyester hybrid networks with low contents of EG as chain extender, i.e., before phase separation, followed a linear relationship. When the system was phase-separated, experimental results did not agree with the linear correla- tion. f0 increased as the Tg of each sample increased. The deviation is due to the inhomogenity of the phase-separated hybrid network. To clarify this point, the same tests were conducted for quenched samples with higher EG contents. In this case, a higher Tg led to a lower f0 and agreed with the theoretical expectation. A linear relationship was found among all phase-mixed samples. The results of hybrid net- works with HD as the chain extender, with EG incorporation lower than 6 wt%, or quenched samples with higher EG incorporation, could be extrapolated by a single straight line. Fig. 5 shows storage shear modulus at the rubber plateau, G0, vs. the weight fraction of hard segments in the hybrid network. Data obtained from time–temperature superposi- tion experiments gave good confirmation of the results of d.m.a. tests. G0 decreased as the hard segment fraction increased. However, we can note that after phase separation, G0 determined from shifted data was much larger. This means that G0 was not the ultimate rubber plateau but the intermediate rubber plateau. The TTS principle can be used only with a phase-mixed system, which was not the case for the hybrid network with high content of EG. When the tests were conducted with quenched samples, the experimental results agreed with expectation. All values can be described by a decreasing line as the weight fraction of hard segments increases. No significant difference in G0 was found between EG and HD hard segments. Fig. 4. Log(f 0/Hz) vs. 1000/Tg: (X) EG; (W) EG quenched; (B) HD. 2065L. Valette,C.-P. Hsu/Polymer 40 (1999) 2059–2070 The flexural modulus of hybrid networks vs. weight frac- tion of total hard segments is given in Fig. 8. The modulus of samples with HD as the chain extender decreased slightly when the HD content increased, while the modulus of sam- ples with EG as the chain extender increased with increasing EG content. As previously discussed, this is due to the chain length of the chain extender. A significant increase in flexural modulus was observed for the samples with phase-separated hard domains. In this case, hard domains form a co-continuous phase with soft domains as described by Ophir and Wilkes [14] and give a higher flexural mod- ulus. The quenched samples had a lower flexural modulus and the modulus decreased as the incorporation of chain extender increased for both sets of samples. Because hard segments do not form a phase-separated hard domain in quenched samples, the apparent flexural modulus of hybrid networks of quenched samples was controlled by the soft domain. Fig. 9(a), Fig. 9(b) show the stress at yield and strain at yield as a function of the weight fraction of hard segments. Both the flexural stress and strain at yield increased slightly for samples with low incorporation of chain extender. For higher amounts of chain extender, both flexural stress and strain at yield decreased, especially after phase separation. Mekhilef and Verhoogt explained the fact that the co-con- tinuous structure of polymer blends usually possesses weak mechanical properties by virtue of the weak interfacial interaction between the two polymers, although both com- ponents are continuous and thus could fully contribute to the properties of the blend [36]. However, quenched samples showed a decrease in yield stress and an increase in yield strength as the content of chain extender increased. Shifts in the flexural stress and strain at yield were also observed before and after quenching. Quenched samples had a lower yield stress and a higher strain than unquenched sam- ples. Flat round cracks were found in quenched samples that had previously been phase-separated. The cracks decreased the flexural stress and stain at yield dramatically. 4. Conclusion The thermomechanical properties of polyurethane and unsaturated polyester hybrid networks have been investi- gated by heat distortion analysis and dynamic mechanical analysis. Transition temperatures agreed with the results measured by modulated d.s.c. as reported in a previous study. Multiple transitions shown by d.m.a. for samples with EG contents above 6 wt% were related to the phase- separated structure. The a transition was related to the phase-separated hard domain and the b transition was related to the soft domain. The rubber plateau of the storage shear modulus was correlated to the crosslinking density. However, theoretical predictions at temperatures lower than the glass transition temperature of the hard domains did not agree with experimental results for the phase-separated hybrid networks. The phase-separated hard domains acted as ‘virtual crosslinks’ and reduced the number-average molecular weight between crosslinks. Mechanical properties at room temperature were generally improved by the incorporation of a chain extender. HD increased the flexibility of polymer chains, resulting in higher defor- mation and impact resistance of the hybrid networks. Hybrid networks with EG as the chain extender were stif- fer by virtue of the rigid hard domains. Hybrid networks with an EG content greater than 6 wt% showed a high Fig. 8. Flexural modulus vs. wt% of hard segments: (– • –) EG; (· · · W · · ·) EG quenched; (- - - B - - -) HD; (— – —A— – —) HD quenched. 2068 L. Valette,C.-P. Hsu/Polymer 40 (1999) 2059–2070 flexural modulus, but poor ultimate mechanical properties because of the formation of phase-separated hard domains. References [1] Malinconico M., Martuscelli E., Volpe MG. Int J Polym Mater 1987;11:295. [2] Martuscelli E, Musto P, Ragosta G, Scarinzi G, Bertotti E. J Polym Sci—Phys 1993;31:619. [3] Bucknall CB, Davies P, Partridge IK. Polymer 1985;26:109. [4] Crosbie GA, Phillips MG. J Mater Sci 1985;20:182. [5] Martuscelli E, Musto P, Ragosta G, Scarinzi G. Polymer 1996;37 (18):4025. [6] Cooper SL, Tobolsky AV. J Appl Polym Sci 1966;10:1837. [7] Seymour RW, Estes GM, Cooper SL. Macromolecules 1970;3 (5):579. [8] Seymour RW, Cooper SL. Macromolecules 1973;6 (1):48. [9] Cooper SL, West JC, Seymour RW. Encycl Polym Sci Technol 1976;1:521. [10] Srichatrapimuk VW, Cooper SL. J Macromol Sci—Phys 1978;B15 (2):267. [11] Van Bogart JWC, Lilaonitkul A, Cooper SL. Adv Chem Ser 1979;176:3. [12] Paik Sung CS, Hu CB, Wu CS. Macromolecules 1980;13:111. Fig. 9. (a) Stress at yield vs. weight fraction of hard segments and (b) strain at yield vs. weight fraction of hard segments: (– • –) EG; (· · · W · · ·) EG quenched; (- - - B - - -) HD; (— – — A — – —) HD quenched. 2069L. Valette,C.-P. Hsu/Polymer 40 (1999) 2059–2070 [13] Ophir ZH, Wilkes GL. Polym Prepr, Am Chem Soc, Div Polym Chem 1978;19 (1):26. [14] Ophir ZH, Wilkes GL. In: Cooper SL, Estes GM, editors. Multiphase polymers, Adv Chem Ser, vol 176. 1976:53. [15] Valette L, Hsu CP. Polymer (in press). [16] Flory PJ. Principles of polymer chemistry. Ithaca (NY): Cornell University Press, 1953. [17] James HM, Guth E. J Chem Phys 1947;15:669. [18] Patel SK, Malone S, Cohen C, Gillmor JR, Colby RH. Macromolecules 1992;25:5241. [19] Painter PC, Shenoy SL. J Chem Phys 1993;99:1409. [20] ASTM D 256-88. Philadelphia (PA): American Society for Testing and Materials, May 1988. [21] Nielsen LE. Mechanical properties of polymers. New York: Reinhold Publishing Corporation, 1962:Ch. 6. [22] ASTM D 790-86. Philadelphia (PA): American Society for Testing and Materials, September 1986. [23] Seefried CG Jr, Koleske JV, Critchfield FE. J Appl Polym Sci 1975;19:2503. [24] Melot D, Escaig B, Lefebvre JM, Eustache RP, Laupêtre F. J Polym Sci—Phys 1994;32:249. [25] Hesketh TR, Van Bogart JCW, Cooper SL. Polym Eng Sci 1980;20:190. [26] Senich GA, MacKnight WJ. Adv Chem Ser 1979;176:97. [27] Pandya MV, Deshpande DD, Hundiwale DG. Br Polym J 1987;19:1. [28] Van Krevelen DW. Properties of polymers. 2nd ed. Amsterdam: Elsevier Scientific, 1976. [29] Fedors RF. J Polym Sci 1969;C26:189. [30] Ferry JD. Viscoelastic properties of polymers. 3rd ed. New York: John Wiley, 1980. [31] Hartmann B, Lee GF. J Non-Cryst Solids 1991;131133:887. [32] Hartmann B, Lee GF, Lee JD. In: Raju PK, editor. Vibro-acoustic characterization of materials and structures, Am Soc Mech Eng, NCA vol 14. 1992:21. [33] Fedderly JJ, Lee GF, Ferragut DJ, Hartmann B. Polym Eng Sci 1996;36 (8):1107. [34] Grillo DJ, Housel TL. In: Physical properties of polyurethanes from polyesters and other polyols, Proc. SPI 34th Polyurethane Technical/ Marketing Conference. New York: Society of the Plastics Industry, Inc., 1992:552. [35] Miller JA, Lin SB, Hwang KKS, Wu KS, Gibson PE, Cooper SL. Macromolecules 1985;18:32. [36] Mekhilef N, Verhoogt H. Polymer 1996;37 (18):4069. 2070 L. Valette,C.-P. Hsu/Polymer 40 (1999) 2059–2070