Curing Mechanism of Phenolic Resin Binder for Oxide-Carbon Refractories

ISIJ International, Vol. 56 (2016), ISIJ No. International, 1 Vol. 56 (2016), No. 1, pp. 44 49 Curing Mechanism of Phenolic Resin Binder for Oxide-Carbon Refractories Jiu ZHANG, Guohui MEI,* Zhi XIE and Shumao ZHAO College of Information Science and Engineering, Northeastern University, China, Mailbox 321#, Northeastern University, Shenyang, Liaoning, 110819 China. (Received on May 6, 2015; accepted on September 29, 2015) Phenolic resin has been widely used as the binder of oxide-carbon refractories, and its curing process has a great effect on the bonding strength. In order to optimize the curing process, the curing mechanism, which involved the heat release reaction, weight change and chemical structure evolution, has been investigated in this study. The dehydration-condensation reactions of the curing started at 363 K. Then, the formation and volatilization of water began and accelerated the rate of weight loss of the binder. Meanwhile, these reactions led to the increase of the molecular size of the binder, which might cause the improvement of the binder s bonding strength. As the temperature increased up to 403 K, the exothermic reaction by the dehydration-condensations of the binder increased and the endothermic reaction by the volatilization of free molecules decreased. In the range from 403 K to 523 K, main dehydration-condensation reactions occurred. These reactions let the molecular size of the binder increase obviously and the bonding strength might be improved significantly. Therefore, a slow heating rate was needed. Above 523 K, the curing gradually approached the end and a higher heating rate was needed. Moreover, when the temperature reached 543 K, the curing was completed because the binder almost had the stable chemical structure, constant weight and very small exothermic reaction. KEY WORDS: curing; mechanism; phenolic resin; binder; oxide-carbon refractories. 1. Introduction Oxide-carbon refractories have excellent resistance to erosion, good thermal shock resistance and low cost, which have been widely applied in metallurgical industry. 1,2) Generally, they are composed of the oxide aggregates, flake graphite, anti-oxidation additives and a binder of phenolic resin. Quantitative investigations have been performed on the compositions, which are as follows: (1) anti-oxidation additives, such as Al, Mg, Si, B 4 C, SiC and MgB 2, 3 5) have attracted the most attention; (2) many oxides, such as Al 2 O 3, MgO and ZrO 2, have been investigated to improve the corrosion resistance of the refractory; 6 8) (3) an anti-oxidation coating of the flake graphite has been developed through the sol-gel method to improve the oxidation resistance. 9) The manufacture process of the fired carbon-containing refractories, such as the stopper rod, long nozzle, submerged nozzle and molten steel temperature sensor, is generally constituted of several processes which include mixing, drying, cold isostatically pressed (CIP) formation, curing and firing. 2,10) To be concrete, the raw materials are mixed homogeneously and coated by the liquid phenolic resin at first. 10) Secondly, the coated materials are dried for releasing volatiles. Thirdly, the materials are molded to produce shaped refractories through CIP formation (about 150 MPa). * Corresponding author: E-mail: meiguohui@ise.neu.edu.cn DOI: http://dx.doi.org/10.2355/isijinternational.isijint-2015-260 And then, the liquid phenolic resin is converted into a solid state of the macromolecular network through curing process (usually below 573 K). 11) Finally, the decomposition and volatilization of phenolic resin occurs and the rigid carbon skeleton is formed through firing process (usually above 1 173 K). 10) During the curing and firing, the binder converts from the liquid phenolic resin into the rigid carbon skeleton. This affects the bonding strength among the granules of the refractory. So the curing and firing are very important processes for the oxide-carbon refractory. The mechanism of the firing process has been investigated in the previous work. 2) In this work, the curing mechanism was studied. According to the investigation of curing mechanism reported by Shunsuke Irie et al. and H. Ishida et al., 10,12) the dehydration-condensation reactions were regarded as the main mechanism. And the methylene and ether linkage formed and built bridges with the increase of temperature, which led to the solidification of the phenolic resin. From this, it is predicated that the temperature affects the mechanism and the curing performance largely. If the curing temperature is too low, the physical and chemical reactions of curing will be uncompleted. Besides, if the heating rate is too high, the rapid volatilization of curing will result in more pores. A lot of defects will be generated in the macromolecular network of the binder. Therefore, the aim of this work was to clarify the curing mechanism of the phenolic resin binder, especially the con- 2016 ISIJ 44

trol of the heating rate and curing temperature. The curing mechanism involves the physical and chemical reactions of the binder which can be reflected by the heat release reaction, weight change and chemical structure evolution. They were evaluated by differential scanning calorimetry (DSC), 13) thermogravimetry-derivative thermogravimetry (TG-DTG), 14) and infrared (IR) spectroscopy, 15 17) respectively. 2. Experimental Procedure 2.1. Raw Materials The resol-type phenolic resin (Type: 5408, SQhepworth, Yingkou, China) was used as the binder of the oxide-carbon refractory. It was a brown-red liquid synthesized from the formaldehyde and phenol with an alkaline catalyst. Characteristic parameters of the phenolic resin are showed in Table 1. 2.2. Phenolic Resin Characterization It has been known that the curing of the phenolic resin involves complex physical and chemical reactions, which may affect on the heat release reaction, weight change and chemical structure evolution. They were evaluated by DSC- TG-DTG (NETZSCH STA 409C/CD) and IR spectroscopy (Spectrum One FTIR Spectrometer), respectively. In order to take the DSC-TG-DTG analysis, the sample was heated from room temperature to 573 K with a heating rate of 5 K/ min in Ar. And the sample was 15 mg liquid phenolic resin. The preparation of the IR spectroscopy analysis was as follows: (1) the 5 g liquid phenolic resin was heated up to the desired temperatures (temperature range: 403 573 K, heating rate: 20 K/min), and then it was held for 1 h; (2) the phenolic resin was ground into powders ( 200 mesh) after the heat-treatment; (3) the 2 mg powders were mixed with the potassium bromide (KBr, 80 mg), and then they were pressed into a wafer (thickness: 0.5 mm); (4) the wafer was mounted in the chamber for the IR spectroscopy measurement. 2.3. Kinetic Modeling On the basis of the TG data with the different heating rates, the apparent activation energy (E) can be calculated by the Flynn-Wall-Ozawa (FWO) method. 18) The basic assumption of the model-free method is that the reaction model is not dependent on temperature or heating rate. 19,20) The FWO method is known as an isoconversional integral method. This method is one of the model-free isoconversional methods, unlike the model fitting method, because this method can avoid the error caused by improper reaction mechanism functions. 19,21,22) This method will also obtain more reliable apparent activation energy values, E, without specifying a kinetic model. However, the model Phenolic resin Table 1. Characteristic parameters of the phenolic resin. Viscosity Carbon residue Solid content H 2O content Free phenol 630 mpa.s 40 wt.% 69 wt.% 2.8 wt.% 11.5 wt.% fitting method is not suitable for obtaining reliable apparent activity energy, because it brings highly uncertain kinetic triplets. 20) The major advantage of this method is that it does not require any assumptions concerning the form of the kinetic equation other than the Arrhenius type temperature dependence. 21) The relationship between the apparent activity energy, E(α), and conversion ratio, α, was investigated through the FWO method, which was given by Eq. (1): 23) AE( α) ( α) logβ = log.. RG( α) E 2 315 0 4567... (1) RT where A is apparent factor, E(α) is apparent activation energy, β is the heating rate, R is the gas constant (8.314 J/mol/K) and G(α) is the integral of conversion ratio (α). The conversion ratio can be expressed as Eq. (2), which is based on the TG data with different heating rates (5, 10 and 20 K/min), because the curing of the binder may bring the weight loss as shown in Eq. (3). 18) m0 mt α =... (2) m0 m where m 0 is the initial weight of the phenolic resin, m T and m refer to the weight at temperature T and at the end of the curing process, respectively. Initial phenolic resin... (3) Residual phenolic resin+ Volatiles When the α is the same, the G(α) is constant, the log1/β has the linear relationship with 1/T as shown in Eq. (4): d log β E( α) = 0. 4567 = K... (4) 1 R d T where K is the slope of the Eq. (1). E(α) can be calculated from the inclination of the regression line when a liner relationship can be assumed between logβ and 1/T as expressed in Eq. (5): KR E( α ) =... (5) 0. 4567 3. Results and Discussion 3.1. Heat Release Reaction and Weight Change The key point of curing process of the phenolic resin is to control the temperature on the basis of the corresponding physical and chemical reactions. Generally, the physical and chemical reactions lead to the endothermic and exothermic reactions, weight change and chemical structure evolution. Figures 1 and 2 depict the curves of the DSC and TG-DTG for the phenolic resin below 573 K. It was found that a small exothermic peak existed below 339 K. Meanwhile, the corresponding weight almost kept constant. This might be associated with the further synthesis reaction of the free phenol and aldehyde in the phenolic resin. And this had little influence on the curing performance. In the temperature range of 339 403 K, there was an endothermic peak. And the DTG curve had a minimum value and a maximum value at 347 K and 363 K, respectively. This phenomenon was presumed to be related with 45 2016 ISIJ

Fig. 1. DSC curve of phenolic resin in curing. Fig. 2. TG-DTG curve of phenolic resin in curing. the weight loss. The weight loss was the result from the reaction that the volatility of the free molecules such as water and phenol was occurred. 11) The volatilization was accelerated with the increase of the temperature from 339 K to 347 K, due to the increase of the activated volatile molecules. Meanwhile, this led to the monotonic decreasing of the DTG curve. In the range from 347 K to 363 K, the DTG curve increased monotonically, which was related to the decrease of the volatilization rate of the free molecules. This phenomenon was considered as the decreasing of the content of volatile molecule gradually. When the temperature was higher than 363 K, the volatile rate of the free molecules increased again. This was attributed to the forming of the new volatile substances, which was caused by the beginning of the dehydration-condensation reactions. 10) These reactions led to the increase of the molecular size of the binder, which might cause the improvement of the binder s bonding strength. However, the molecular size of the binder was small (molecular weight: about 150 500) before the dehydration-condensation reactions, 11) which resulted in the poor bonding strength of the binder. Furthermore, since the viscosity of the binder decreased with the increase of the temperature, the bonding strength of the binder would be reduced. Therefore, that might be the one reason why the deformation of the oxide-carbon refractory was easy to occur in this temperature range. As the temperature increased up to 403 K, the exothermic reaction by the dehydration-condensations of the binder increased and the endothermic reaction by the volatilization of free molecules decreased. In the range from 403 K to 473 K, the dehydration-condensation reactions were enhanced significantly. There was an obvious exothermic reaction, which was corresponded to 0.21 W/g at 443 K. The weight decreased from 91.2% to 83.9%, which was equal to 40.5% in the total reduction. Therefore, it was necessary to heat up the phenolic resin with a low heating rate so as to satisfy the demand of the intense dehydration-condensation reactions. To be concrete, the process of holding temperature should be needed at 443 K. In the range from 473 K to 543 K, the exothermic reaction became slower and gradually trended to end. The heat release decreased from 0.12 to 0.03 W/g. And the weight loss was 7.8% in the total reduction during 473 523 K. That meant that the curing reaction might be approaching the end. It was thought that the curing reaction at 523 543 K could be regarded as a heat release behavior without an obvious weight change. From these results, it was also thought that the heating rate in this temperature range could be increased suitably. Above 543 K, the weight loss was little (1% in total reduction) and the exothermic reaction was very small (the amount of heat released: 0.01 W/g), respectively. This implied that the bonding of the binder could be almost stable state and the curing reactions were completed at this temperature. 3.2. The Result of Kinetic Analysis for Curing In order to discuss the kinetics of curing of the phenolic resin, the relationship between the apparent activation energy (E) and conversion ratio (α) was investigated in this section. The conversion ratio was calculated by the Eq. (2) on the basis of the TG data with the different heating rates of 5, 10 and 20 K/min. The values of m 0 and m are the weight of the phenolic resin at 333 K and 573 K, respectively. From Figs. 1 and 2, the weight loss began at 333 K and the curing was completed before 573 K. Figure 3 shows the TG curves with different heating rates. It was found that the weight decreased more quickly when the heating rate was higher. Under the condition of the slower heating rate, the phenolic resin would have more weight loss at the same temperature. Because it had relatively longer heating time and absorbed more heat in this condition, which led to a higher curing degree. The relationship between the conversion ratio (α) and temperature is illustrated in Fig. 4. When the heating rate was higher, the phenolic resin had relatively shorter heating time and absorbed less heat at the same temperature, which might result in slower curing reaction. As a result, the lower value of α was obtained compared to that with lower heating rate. It was thought that a higher temperature which was corresponded to the heat for conversion reaction should be needed under this condition. According to the Eq. (1), the log1/β and 1/T should have a linear relationship as shown in Fig. 5. It was found that the different special slopes, which were corresponded to the different conversion ratio, were got from the assumption of liner relationship. On the basis of these, the apparent activation energy can be calculated by the Eq. (5). Figure 6 represents the relationship between the conversion ratio (α) 2016 ISIJ 46

Fig. 3. TG curves of phenolic resin with different heating rates. Fig. 5. Relationship between logβ and 1/T with different conversion ratios (α). Fig. 4. Relationship between the conversion ratio (α) and temperature. Fig. 6. Relationship between apparent activation energy (E) and conversion ratio (α). and apparent activation energy (E). It was found that the apparent activation energy was approximately proportional to the value of α which was less than 0.6 (i.e. E=44 kj/mol at α = 0.1, E=72 kj/mol at α = 0.6). This might be due to the fact that the increase of E with α was mainly dependent on the increase of volatilization of free molecules which was regarded as the main curing mechanism. And this could also be because the dehydration-condensation reactions of curing would be small in this conversion ratio range. This apparent activation energy, which was corresponded to the volatilization of free molecules, might increase in proportional to α. When α reached 0.6, the main curing mechanism would change from the volatilization of free molecules to dehydration-condensation reactions. The dehydration-condensation reactions would be corresponded to higher activation energy because of the cutting and formation of the chemical bonds. As the value of α increased up to 0.8, E increased rapidly from 72 to 122 kj/mol. This result was in agreement with the activation energy (62 135 kj/mol) of similar phenolic resin reported by A. Tejado et al., M. V. Alonso et al. and Byung-Dae Park et al. 24 26) Besides, this increase of E implied that the dehydration-condensation reactions increased significantly in this conversion ratio range. This might indicate that the methylene and ether linkage formed rapidly in the phenolic resin and built bridges with the increase of α. From above discussion, it was found that the change of the apparent activation energy was well consistent with the heat release reaction of the phenolic resin. 3.3. Chemical Structure Evolution From the result of the DSC analysis, the main exothermic reaction of the curing occurred above 403 K. In order to identify the chemical structure evolution directly in this temperature range, the IR spectroscopy was used. After the heat-treatment at different temperatures, the IR spectra of the treated phenolic resin samples were measured. The result is summarized in Fig. 7. According to the correlation described in the literatures, 15 17) the major corresponding absorption peaks of the IR spectra for the treated phenolic resin samples were identified as listed in Table 2. It was found that the absorption peaks at 1 475 and 1 200 cm 1, which were corresponded to the bending vibration of the -CH 2 - bridges and the stretching vibration of the C-O of a phenol group, respectively, were observed remarkably. The methylene (-CH 2 -) and ether linkage (C-O) of phenol group were synthesized by the following dehydration-con- 47 2016 ISIJ

Fig. 7. Infrared spectroscopy spectra of the phenolic resin after being heat treated at desired temperatures. Table 2. IR Absorption bands of the treated phenolic resin. Wavenumber (cm 1 ) Functional group 3 505 O-H stretching of phenol group 3 330 O-H stretching of phenol group 3 009 Ar-H sretching 2 900 -CH 2- stretching 1 595 C-C aromatic ring stretching 1 475 -CH 2- bending 1 440 -CH 2- bending 1 200 C-O stretching of phenol group 882, 820, 755 Ar-H out-of-plane bending densation reactions as shown in Eqs. (6) and (7). 17) It has been known that these reactions are the main reactions of the curing, and they bring the chemical structure evolution of the phenolic resin.... (6)... (7) In addition to the absorption peaks of the methylene and ether linkage of phenol group, other absorption peaks were also observed. These absorption peaks were identified as follows: (1) the peaks at 3 505 and 3 330 cm 1, which were attributed to the stretching vibration of the phenol O-H groups, were well consistent with the absorption peaks of near 3 500 and 3 330 cm 1 reported by K. A. Trick and G. Carotenuto et al., 16,17) respectively; (2) the bands near 3 009 and 2 900 cm 1 caused by the stretching of the aromatic hydrogen (Ar-H) and methylene (-CH 2 -), respectively, were in agreement with the absorption peak of near 3 000 and 2 900 cm 1 reported by K. Ouchi; 15) (3) the peak at 1 595 cm 1 resulted from the stretching of C-C bonds of benzene rings was well consistent with the absorption peak of near 1 598 cm 1 reported by G. Carotenuto et al.; 17) (4) the absorption peak of 1 440 cm 1, which was partially covered by the strongest band of 1 475 cm 1, was corresponded to the -CH 2 - adjacent to oxygen as in dibenzyl ether linkages (-CH 2 -O-) 17) that might be formed as seen in Eq. (7); (5) a group peaks at 882, 820 and 755 cm 1, which were derived from the deformation vibrations of C-H bonds in benzene rings, were in agreement with the absorption peaks of near 880, 828 and 756 cm 1 reported by K. Ouchi and K. A. Trick et al. 15,16) At 403 K, some absorption peaks were observed, in spite of not so strong ones. That indicated the beginning temperature of the dehydration-condensation reactions could be below 403 K. These reactions accompanied the formation and volatilization of water and caused the weight loss of the binder. Based on the analysis of the weight change, as seen in Fig. 2, the relatively maximum value of the DTG curve was observed at 363 K. It was found that the dehydrationcondensation reactions started at this temperature, because the beginning of the formation and volatilization of the water let the rate of the weight loss accelerate. In the range from 403 K to 523 K, the intensities of the absorption peaks significantly increased with the increase of temperature, which was corresponded to the rapid exothermic reaction. This implied that the main dehydrationcondensation reactions occurred. These reactions might cause the increase of the molecular size of the phenolic resin and result in the significant increase of bonding strength of the binder. Besides, a lot of water was generated and volatized. This was corresponded to the obvious weight loss, which might cause higher porosity of the binder and subsequently a high porosity of the refractory. Therefore, a slow heating rate should be needed to improve the bonding strength and to reduce the porosity of the oxide-carbon refractory from above discussion. Above 523 K, the absorption peaks almost kept constant. This indicated that the chemical structure of phenolic resin was nearly stable, which implied that the curing reactions almost were completed. However, the chemical structure approached the constant state at 523 K before the heat release reaction (543 K). It was thought that this was derived from the following reasons: (1) the phenolic resin samples, which were used to identify the chemical structure evolution and to measure the heat release reaction, might have different curing degrees due to their different heat-treatment processes; (2) the difference of the IR spectra was very small at 523 543 K. From above discussions of the DSC-TG and IR spectroscopy analyses, if the temperature increased up to 543 K, the phenolic resin binder in the oxide-carbon refractory might have good bonding strength. Therefore, it was found that a higher heating rate should be needed in the range of 523 543 K, and the suitable curing temperature should be 543 K. 4. Conclusion Phenolic resin has been widely used as the binder of oxide-carbon refractories, whose curing process has a great effect on the bonding strength. The curing mechanism, which involved to the heat release reaction, weight change and chemical structure evolution, was investigated to opti- 2016 ISIJ 48

mize the curing process. Results obtained in this study are summarized as follows: (1) At 363 K, the dehydration-condensation reactions of the curing started. The formation and volatilization of water began and accelerated the rate of weight loss of the binder. Meanwhile, these reactions led to the increase of the molecular size of the binder, which might cause the improvement of the binder s bonding strength. As the temperature increased up to 403 K, the exothermic reaction by the dehydration-condensations of the binder increased and the endothermic reaction by the volatilization of free molecules decreased. (2) According to the result of IR spectra, it was predicted that main dehydration-condensation reactions would occur in the range from 403 K to 523 K. These reactions let the molecular size of the binder increase obviously and the bonding strength might be improved significantly. Therefore, a slow heating rate was needed in this temperature range. (3) When the conversion ratio was in the range from 0.1 to 0.6, it was thought that the main curing mechanism changed from the volatilization of free molecules to dehydration-condensation reactions. On the other hand, when the conversion ratio increased from 0.6 to 0.8, the apparent activation energy increased rapidly from 72 to 122 kj/mol. The degree increase was much higher than that in the range from 0.1 to 0.6. (4) Above 523 K, the curing approached the end and a higher heating rate was needed in this temperature range. Moreover, when the temperature reached 543 K, the curing was completed, which was due to the fact that the binder almost had the stable chemical structure, constant weight and very small exothermic reaction. Acknowledgements The authors gratefully acknowledge the financial support received from the National Natural Science Foundation of China (No. 61333006 and No. 61473075) and the research support from Taihe Metallurgical Measurement and Control Technologies Co., Ltd, China. REFERENCES 1) W. E. Lee and S. Zhang: Int. Mater. Rev., 44 (1999), 77. 2) J. Zhang, G. Mei, S. Zhao and Z. Xie: ISIJ Int., 54 (2014), 553. 3) A. S. Gokce, C. Gurcam, S. Ozgen and S. Aydin: Ceram. Int., 34 (2008), 323. 4) S. Uchida, K. Ichikawa and K. Niihara: J. Am. Ceram. Soc., 81 (1998), 2910. 5) M. N. Khezrabadi, J. Javadpour, H. R. Rezaie and R. Naghizadeh: J. Mater. Sci., 41 (2006), 3027. 6) S.-m. Zhao, G.-h. Mei and Z. Xie: J. Iron Steel Res. Int., 8 (2011), 12. 7) A.-H. Bui, S.-C. Park, I.-S. Chung and H.-G. Lee: Met. Mater. Int., 12 (2005), 435. 8) S. Zhang and W. E. Lee: J. Eur. Ceram. Soc., 21 (2001), 2393. 9) Sk. A. Ansa, S. Bhattacharya, S. Dutta, S. S. Ghosh and S. Mukhopadhyay: Ceram. Int., 36 (2010), 1837. 10) L. Pilato: Phenolic Resins: A Century of Progress, Springer, Berlin, (2010), 503. 11) N. Li, H. Gu and H. Zhao: Refractory Materials, Metallurgical Industry Press, Beijing, (2010), 234. 12) H. Ishida and Y. Rodriguez: Polymer, 36 (1995), Issue 16, 3151. 13) G. Rivero, V. Pettarin, A. Vázquez and L. B. Manfredi: Thermochim. Acta, 516 (2011), 79. 14) Y. Hua, Z. Zhang and Q. Qin: J. Appl. Polym. Sci., 88 (2003), 1410. 15) K. Ouchi: Carbon, 4 (1966), 59. 16) K. A. Trick and T. E. Saliba: Carbon, 33 (1995), 1509. 17) G. Carotenuto and L. Nicolais: J. Appl. Polym. Sci., 74 (1999), 2703. 18) K. Açıkalın: J. Therm. Anal. Calorim., 109 (2012), 227. 19) C. Zhang, W. K. Binienda, L. Zeng, X. Ye and S. Chen: Thermochim. Acta, 523 (2011), 63. 20) S. Vyazovkin and C. A. Wight: Thermochim. Acta, 340 341 (1999), 53. 21) P. Ravi, A. A. Vargeese and S. P. Tewari: Thermochim. Acta, 550 (2012), 83. 22) Y. Zhao, R. Bie, J. Lu and T. Xiu: Chem. Eng. Commun., 197 (2010), 1033. 23) R. Hu and Q. Shi: Thermal Analysis Kinetics, Science Press, Beijing, (2001), 50. 24) A. Tejado, G. Kortaberria, J. Labidi, J. M. Echeverria and I. Mondragon: Thermochim. Acta, 471 (2008), 80. 25) M. V. Alonso, M. Oliet, J. García, F. Rodríguez and J. Echeverría: Chem. Eng. J., 122 (2006), 159. 26) B.-D. Park, B. Riedl, Y. S. Kim and W. T. So: J. Appl. Polym. Sci., 83 (2002), 1415. 49 2016 ISIJ