Zeitschrift Kunststofftechnik Journal of Plastics Technology

Transcription

1 Zeitschrift Kunststofftechnik Journal of Plastics Technology eingereicht/handed in: angenommen/accepted: Javier C. Cruz, M.Sc., Prof. Prof. h.c. Dr. Tim A. Osswald, Department of Mechanical Engineering, University of Wisconsin-Madison Wissenschaftlicher Arbeitskreis der Universitäts- Professoren der Kunststofftechnik archivierte, peer-rezensierte Internetzeitschrift des Wissenschaftlichen Arbeitskreises Kunststofftechnik (WAK) archival, peer-reviewed online Journal of the Scientific Alliance of Polymer Technology Precise Curing Analysis of Highly Accelerated Unsaturated Polyester Reactions This paper presents a method for monitoring highly accelerated curing reactions. The method consists of using a reaction chamber where a thin sample is placed inside and its heating rate is controlled by a system that reacts to direct changes in sample temperature. Simultaneously, a Raman spectroscopy system is used to monitor the cure by shooting the laser through a lens over the reactor. The method is compared to differential scanning calorimetry. It was found that due to the high reaction rates of the pultrusion grade unsaturated polyester resin used, overheating effects were uncontrollable by DSC. It is shown that with an adequate sample thickness and with a temperature control system that reacts directly to changes in sample temperature, overheating effects can be minimized and eliminated leading to very consistent linear heating rates. Genaue Analyse von hochbeschleunigten Härtungsreaktionen ungesättigter Polyester Der Artikel veranschaulicht eine Methode, hochbeschleunigte Härtungsreaktionen zu verfolgen. Dabei wird eine Reaktionskammer mit einer dünnen Probe verwendet, deren Heizrate durch ein System gesteuert wird, das auf Änderungen in der Probentemperatur reagiert. Die Härtung wird durch ein Raman-Spektroskopie-Systeme verfolgt, indem ein Laser durch eine Linse oberhalb des Reaktionsgefäßes auftrifft. Die Methode ist der Differential Scanning Calorimetry vergleichbar. Aufgrund der hohen Reaktionsgeschwindigkeit des verwendeten Pultrusions-UP-Harzes ließen sich die Überhitzungseffekte mittels herkömmlicher DSC nicht kontrollieren. Mit geeigneter Probendicke und dem hier verwendeten Temperaturkontrollsystem können die Effekte eliminiert werden, so dass man lineare Aufheizraten erhält. Carl Hanser Verlag Zeitschrift Kunststofftechnik / Journal of Plastics Technology 4 (28) 3

2 PRECISE CURING ANALYSIS OF HIGHLY ACCELERATED UNSATURATED POLYESTER REACTIONS J. C. Cruz, T. A. Osswald 1 INTRODUCTION When processing most thermosetting polymers materials have to be exposed to some kind of heat treatment. Essentially the heat rise will control the activation of reactive sites which will initiate and propagate a reaction within the material and ultimately control the quality and final properties of the part. For this reason there is a desire to be able to accurately describe or model the reaction throughout the different stages of cure. The ability to understand reaction kinetics of thermosets assures a better understanding of the final material properties and aids in the determination of optimal processing conditions. The main concern when determining cure in time is to consider all the variables that may affect the curing behavior which include but are not limited to: reaction temperatures, initiator, catalysts, concentrations, reaction rates, exothermic heat of reaction, thickness, specific heat, and the thermal conductivity of the material. The solidification process of thermosets such as epoxies, unsaturated polyesters, phenolics, and polyurethanes is governed by an exothermic chemical reaction [1]. There is usually some kind of monomer or crosslinking agent which will bond with other polymeric chains through an induced molecular excitation. As molecules bond, a release of energy occurs, which for highly accelerated reactions may lead to high temperature rises. Unsaturated polyesters (UP) resins are generally produced through a condensation polymerization of dicerboxylic acids, glycols and anhydrides of dicarboxylic acids after which the product is diluted in an unsaturated monomer, usually styrene. Since both the monomer and the UP have C=C unsaturation, with the proper temperature range and initiators, free radicals may form and react with neighboring C=C bonds leading to a mostly saturated crosslinked structure. With excessive heat or highly reactive initiators or catalysts, reactions may be accelerated to a great extent. Due to the desire of reducing process time special material combinations have been produced which lead to very fast curing times and in many cases excessive uncontrollable overheating. Currently, many different studies have been performed in order to try to characterize the curing reaction of unsaturated polyester resins and their degree of cure. Among them, Takemoto [2] described how variations in the chemical structure, molecular weight, and initial mole fraction of styrene would affect the final conversion of UP resin. Koeing and Shin [3] studied and identified main changes in structure of unsaturated polyester to aid in monitoring cure by use of Raman spectroscopy. Journal of Plastics Technology 4 (28) 3 2

3 Several researchers [4-6] studied the final individual conversions of styrene and fumarate bonds in UP resin during low temperature isothermal reactions when certain changes in chemistry or concentration variations were imposed. Many researchers [7-1] use differential scanning calorimetry (DSC) to continuously monitor the curing kinetics of UP resins for low temperature reactions isothermally and at constant heating rates. Others like Huang and Chen [11-12] have described variations in curing at different comonomer compositions for medium and high temperature reactions with DSC and IR spectroscopy. Gorovaya and Korotkov [13] have analyzed quick cure of thermoset composites through a one-dimensional heat transfer model. For quick reactions that lead to high exotherms during the process overshoots of temperature in the sample may alter the isothermal temperatures or linear heating rates assumed during cure and create errors in final kinetic results. Heat conduction effects which are directly related to the temperature rise in the sample have been modeled by Barone and Caulk [14] for curing of fiber reinforced unsaturated polyester resin during compression molding. They describe how non-uniform temperature distributions may be seen in thick parts, how rapid cure can result in a reaction time comparable to the time scale of thermal diffusion and how progressing cure as it couples with heat conduction may lead to non-uniform curing in parts and excessive temperature rises. In order to characterize the curing of a material, kinetic models such as the Kamal and Sourour model [15] have been created in which dependant constants have to be derived for a particular material from experimental results. In order to achieve accurate results isothermal or constant heating rate assumptions must be held throughout the experiments. Janeschitz-Kriegl et al. [16-17] described how there is a thermal delay between the sample, pan and surface-to-thermocouple which has to be accounted for in order to obtain accurate and correct heating rates of samples. For fast exothermic reactions heat released from the sample will have to travel through these surfaces before reaching the thermocouple. Temperature rises within the sample may be significant enough to cause isothermal or constant heating rates of the sample to deviate from their true values. The following work is to show this effect and to develop an alternate method to monitor highly accelerated reactions where temperature overshoots may be directly monitored and controlled. For highly accelerated reactions commonly seen in unsaturated polyester pultrusion processes assuming possible overheating is a must even for small sized samples as will be shown. Journal of Plastics Technology 4 (28) 3 3

4 2 EXPERIMENTAL 2.1 Materials A pultrusion grade Unsaturated Polyester Resin was used for all the experiments. The pultrusion grade resin provided was made through a combination of Maleic Anhydride, poly(ethylene terephthalate), a mix of glycols and determined to have been diluted in 3% styrene by weight. Pultrusion processing conditions require fast reactions to achieve optimal properties as soon as the material exits the die. Hence, three different initiators were used to perform the reaction. The first initiator is di-(4-t-butylcyclohexy)peroxydicarbonate known as Perkadox 16, with initiation temperatures between 45 C-95 C. This low temperature initiator assures that the reaction will begin at a low enough temperature in order to reach the gel point fast enough to maintain better structural rigidity of the part during fiber impregnation. The second initiator was a mixture of 75% tert-amyl-peroxy-2-ethylhexanoate and 25% butyl benzyl phthalate known as Trigonox 121-BB75, with initiation temperatures between 85 C-125 C. The third initiator used was t-butylperoxy benzoate known as Trigonox C, a high temperature initiator covering the range of 135 C-17 C. By the combination of these three initiators the whole temperature range in which the UP Resin will cure during pultrusion process is covered and reactions will be highly accelerated. Samples were prepared by mixing the unsaturated polyester resin with styrene- 5 PPM Inhibitor to assure a 2:1 ratio of styrene to unsaturated polyester C=C bonds. By obtaining the average number molecular weight of UP and average number of C=C bonds per polyester molecule the necessary amount of styrene for a 2:1 ratio was determined knowing that styrene has a molecular weight of 14g/mol with one C=C bond active site. Approximately 1.46g of styrene per gram of UP was to be in solution for the correct ratio, therefore, knowing the initial weight percent of styrene in solution an accurate ratio could easily be obtained. In order to perform a uniform mixing of Perkadox 16 powder, initial dilution in the styrene was performed. Full dilution of such as small sample size could be achieved in less than 1 drops of styrene. Perkadox 16 initiator was mixed at.2wt%, followed by the addition of the liquid initiators Trigonox 121- BB75 and Trigonox C at.4wt% and.2wt% respectively. A total batch of 27.95g was made with initiators accounting for.8wt% of the total batch. Initiators were mixed for 1 minutes then batch was stored in closed container at 3 C to prevent any activation or diffusion of volatiles. Journal of Plastics Technology 4 (28) 3 4

5 2.2 Instrumentation and Procedures Differential Scanning Calorimetry (DSC) A NETZSCH DSC 2 PC developed by Netzsch-Gerätebau GmbH (Selb / Bayern, Germany) was used for all comparative analysis. The thermoset curing reaction can be monitored by use of the DSC due to the heat release of the molecules during crosslinking. This is one of the most common methods of monitoring curing reactions because it directly relates to the molecular level but does not require perfect understanding of the molecular behavior. Sample preparation consisted of using the stored sample batch, mixing for 5 minutes, extracting a small sample with a syringe, and by using a precisely drilled DSC lid, inserting a thin thermocouple directly into the sample inside the DSC dish. Thermocouple wire passed through the internal cavity of the DSC to a temperature data-logger which would accurately record the actual sample temperature. Scanning runs from 25 C-19 C were performed for the different samples at rates of 2.5K/min, 5K/min and 1K/min Raman Spectroscopy and Chamber Design A RamanRxn1 analyzer provided by Kaiser Optical Systems, Inc. (Ann Arbor, MI) was used to monitor the reactions. The system consisted of a TE-cooled CCD detector, a high-powered Invictus 785-nm NIR laser and Holoplex grating technology to provide fast, simultaneous collection of Raman data. The laser power used was 3mV, which does not induce any local heating effects. Spectra were recorded every.3 seconds. A burst mode option is used which provides the ability of recording many spectrums per second for our fast reactions. A fiber-optic sampling probe with an optical lens that has a focal distance of 2.5 inches was adapted to a microscope stand for focusing. A cylindrical shape reaction vessel, as depicted in Fig.1, was designed to allow for precise temperature and pressure control. The vessel consists of a.5 inch quartz lens over the top, entrance and exit N 2 pressure lines, heaters, insulation, a high-pressure 3-way thermocouple fitting, and a bottom port for sample insertion. Sample temperature is accurately controlled with a temperature control system that responds to the controlling thermocouple which is inserted directly onto the sample. Reactor pressure control and purging is done through two ports over the sample cavity. If an accurate excessive overheating control is required, for example, during fast heating rates, liquid nitrogen may be adapted to entrance port with a control valve that will respond to any overshoots in temperature. Journal of Plastics Technology 4 (28) 3 5

6 Pressure Port Quartz Lens Pressure Port Figure 1: Top view of reactor (left). Side view of reactor aligned to direction of thermocouple port (right) Samples are prepared as described for the DSC but placed in a thin aluminum shim. Thickness is controlled based on the sample mass and by use of a glass microscope cover slide which helps uniformly distribute the material. Sample thicknesses have been kept low to eliminate sample overheating and not require any nitrogen cooling. Thicknesses were varied between.2mm 1mm with heating rates of 5K/min, 1K/min and 2K/min. Simultaneously, Raman spectra was obtained at constant time intervals controlled by a PC software called Holograms which is then exported to Holoreact, GRAMS and other managing software for further peak change-in-time analysis. All spectra are normalized to a reference peak and peak height and area directly related to conversion. 3 RESULTS AND DISCUSSION Thermocouple Port Sample 3.1 Differential Scanning Calorimetry Reactions of heat activated thermosetting resins are accelerated with increased heating rates, therefore, heat generated and sample overheating will be greater with faster heating rates. Due to the low thermal conductivity of the sample it is expected that samples of greater thickness will overheat much more than thinner samples. With the fixed geometry of the DSC dish we can assume that thickness of the sample although hard to control will be related to the mass. Results of the temperature profile generated by DSC did not show any sample overheating and showed uniform linear heating rates for the highly accelerated pultrusion resins. When the temperature of the sample was monitored with a Journal of Plastics Technology 4 (28) 3 6 \

7 thermocouple directly imbedded into the sample, results were very different. Fig. 2 shows a graph of results for a heating rate of 2K/min. Tests for each heating rate were repeated five times. While small differences were perceived between the tests, mainly attributed to the dependence on the samples mass, the repeatability between the results was satisfactory. Average overshoots from the programmed temperature profile of above 6 C where obtained for 2K/min heating rates. Table 1 shows averages of overshoots obtained between the five samples at the different heating rates tested. One may see how the overshoots are higher for the larger heating rates. Examples of mass effect in overheating will be shown later. Heating Rate 2 K/min 1 K/min 2.5 K/min Average Weight 34 ± 4mg 35 ± 3mg 37 ± 4mg Ts-Tf at Overshoot 68 ± 22K 54 ± 17K 6 ± 1.5K Table 1: Effect of heating rate on overshoot of the sample temperature (Ts) averaged between five samples per rate. As shown in Fig. 2, the sample overheats at a similar time to when the exothermic peak generated by the DSC occurs. There can also be seen a delay between the response of the DSC relative to the actual temperature variations in the sample. DSC Output (mw/mg) DSC/(mW/mg) DSC Tf Temp Ts measured Temp time (min) Figure 2: DSC exotherm profile and Temperature output (Tf) as compared to Data-logger actual sample temperature (Ts) for a heating rate of 2K/min and a sample mass of 32.6mg Journal of Plastics Technology 4 (28) Temperature (C)

8 As the sample absorbs heat from the furnace of the DSC, heat has to be transferred from the furnace and wall unto the surface of the pan then into the low conductivity resin. For this reason we can see a continuous delay in time between the DSC temperature (Tf) and the actual sample temperature (Ts) of approximately 2.5 C. When the sample is reacting, heat is generated from the sample, which also, like the heat of the furnace but in opposite direction, has to travel out of the sample, through the pan and surface-to-thermocouple. Therefore, we can also note a delay between the DSC output of heat generation and the overheating occurring in the sample. For the case shown in fig. 2, there is a difference in time between the maxima of Ts and the DSC output peak of approximately 8 seconds. In the initial stages of overheating, where the peak growth begins, similar differences may be found. This means that the sample is overheating because of the excessive exothermic heat of reaction but due to the system and surroundings there is an 8s offset to when this heat release is detected and reported. The temperature increase of the sample was found to be very large and may not be neglected. If a sample is not undergoing a constant heating rate any temperature increase will accelerate the reaction and in turn affect the heat release leading to a deviation of results obtained if the sample were really maintained at a constant heating rate. Comparable overshoots and delays were observed for 1K/min samples. For the sample shown in fig. 3 there was a difference of approximately 12 seconds in the maxima of the temperature peak Ts relative to the DSC heat output peak and a 1s average delay between the sample and furnace temperatures during heating. Depending on the weight of the sample maximum temperature rise at overshoot peak ranged between 2 C for lower mass samples to 8 C for greater mass samples. Due to the hardships of inserting a thermocouple directly into the sample, mass of sample could not be kept too low in order to assure good thermocouple contact with resin. Fig. 4 presents results for a sample with a lower mass of only 17.7mg at a rate of 1K/min. Overheating was still observed for this sample with 16 C of overshoot and about 6 seconds difference between the maxima of Ts and the DSC output peak. Journal of Plastics Technology 4 (28) 3 8

9 DSC Output (mw/mg) Time vs DSC Output DSC Tf Temp Ts measured Temp time (min) Figure 3: DSC exotherm profile and Temperature output (Tf) as compared to actual sample temperature (Ts) for a heating rate of 1K/min and a sample mass of 28.6mg DSC Output (mw/mg) time (min) Time vs DSC Output DSC Tf Temp Ts measured Temp Figure 4: DSC exotherm profile and Temperature output (Tf) as compared to actual sample temperature (Ts) for a heating rate of 1K/min and a sample mass of 17.7mg Journal of Plastics Technology 4 (28) Temperature (C)

10 Similarly, fig. 5 shows how for heating rates of 2.5K/min delays and overshoots may also be noted on higher weight samples but to a lesser extent due to the slower heating rate. DSC Output (mw/mg) time (min) Time vs DSC Output DSC Tf Temp Ts measured Temp Figure 5: DSC exotherm profile and Temperature output (Tf) as compared to actual sample temperature (Ts) for a heating rate of 2.5K/min and a sample mass of 41.4mg Nevertheless, even at such slow heating rates which would in no way simulate real process conditions overshoots of temperature are still present. For 2.5 K/min samples, differences between Tf and Ts during heating time are less than half a degree. Maximum temperature difference at overshoot was 8 C and the temperature overshoot occurs exactly inside the DSC heat output curve indica ting that for these slow rates the response time is much accurate. Only a small misalignment is seen between the maxima of the peaks but do to the curvature of the Ts peak, it is difficult to provide an accurate comparison. Anyhow, as seen in the graph, the growth of the Ts peak occurs well within the time period of the DSC output peak. Table 2 provides examples of results where variations in overheating due to sample mass may be perceived. As observed in the table, for all other variables constant, if the sample mass decreases the sample overheating decreases. Similarly, if the heating rate decreases delays and differences between temperatures Ts and Tf will decreases because changes are slower and there is more time for equilibrium and heat transfer. It is important to mention that there was not a way to accurately determine the position of the thermocouple inside the sample. Results may vary depending on the location of the thermocouple therefore many runs were performed until accurate results were obtained. For samples where the thermocouple might have been slightly closer to the surface Journal of Plastics Technology 4 (28) Temperature (C)

11 of the material rather than imbedded in the center of the sample overheating detected may have been smaller. Heating Rate (K/min) Sample Weight (mg) Average Heating Temp. Difference Ts-Tf ( C) Max Temperature difference at overshoot Ts-Tf ( C) Relative Peak Time Difference (sec) 2 tr ± tr ± tr ± tr ± tr ± tr ± tr ± Table 2: Variations in results of heat absorption and release, detected by the DSC and a thermocouple directly imbedded into the sample for various heating rates and sample weights. 3.2 Raman Spectroscopy The analysis method of Raman spectroscopy for monitoring curing reactions of unsaturated polyester crosslinked with styrene is ideal and very precise due to the high Raman sensitivity of carbon-carbon double bond vibrations leading to large peak intensities. C=C bonds are consumed in time during the reaction, for both UP and styrene molecules, therefore monitoring consumption of these bonds can directly provide a value for reaction conversion. Fig. 6 shows a region of the unsaturated polyester spectrum denoting peaks that may be used for analysis of conversion. Peaks designated with an S or UP are particular to a styrene or unsaturated polyester vibrations respectively. Among the many possible peaks that may be used to monitor the reaction progress, three of the strongest peaks that provide very reliable conversion values are the 1661cm -1, 1632cm -1 and 1413cm -1. The 1661cm -1 peak represents a C=C stretching mode in UP, whereas the 1413cm -1 peak represents the CH 2 wag-deformation (δ ) present on the C=CH 2 bond particular to styrene before reacting [18]. The 1632cm -1 peak denotes the overall C=C trans-stretching vibrations of both the UP and the styrene C=C bonds. The 1632cm -1 peak is the largest in the mid-spectrum range and will be the main peak used to monitor the overall progress of the reaction. Journal of Plastics Technology 4 (28) 3 11

12 Figure 6: Central region of an unsaturated polyester resin spectrum during initial stages of cure Through the use of an accurate temperature control system which reacts directly to changes in sample temperature, very precise and controllable heating rates may be obtained. As mentioned previously, sample size is determinant in how complicated the control of overheating for a sample may be. Sample thickness of.5mm was found to be relatively good to ease calibration and optimization of temperature control system. Diameter of sample was maintained constant at 9.9mm to provide an adequate surface area for heat transfer and to improve thermocouple contact. Fig. 7 shows a graph of conversion in time for a UP reaction at 5K/min and a sample thickness of.5mm. Conversion 774S UP time (min) 1413S 1632S/UP 1661UP Conv 5K/min HT Rate 1732UP Figure 7: Conversion and temperature in time for a.5mm sample of UP heated at 5K/min from room temperature to 18 C. Heating rate curve is the actual temperature of the sample recorded in time. Journal of Plastics Technology 4 (28) Temperature (C)

13 Conversion time (min) Conv 1K/min HT Rate Figure 8: Conversion and temperature in time for a.5mm sample of UP heated at 1K/min from room temperature to 18 C. As shown, a very uniform heating rate with no overheating of the sample may be obtained. Smooth heating leads to accurate conversion results that may be directly related to the temperature changes in the sample. Due to the slow rate of heating, slower reaction rates take place as may be seen by the low slope of the conversion curve during the accelerated reaction stage. Fig. 8 shows results for a reaction occurring at a rate of 1K/min and a sample thickness of.5mm. Again, we may notice a very accurate and well controlled heating rate with no sample overheating. Reaction rates are higher for this case with an accelerated reaction period starting around 4.5min into the test and lasting close to 4min, as compared to a period of 7.5min for the 5K/min reaction. Even at 1K/min overheating is well controlled for these thin sample sizes with our accurate temperature control. It is worth mentioning how by use of Raman spectroscopy the actual conversion in time is reported. There is no need for normalization as required with the DSC; therefore, final conversion values do not show a 1% conversion which is unreal but the actual final conversion of the sample. Any molecules that have not reacted are noted by a residual peak in which its height and area may be related to the unreacted bonds. Results shown with the DSC demonstrated that big overshoots in temperature occurred for heating rates of 2K/min. The set of graphs in fig. 9 describe how there may also be extreme sample overheating for thicker samples with our Raman reactor system and how an accurate heating control with a TC directly imbedded into the sample will lower the overheating greatly. Fig. 9a presents results for a sample reacted at 2K/min and a thickness of 1mm. Journal of Plastics Technology 4 (28) Temperature (C)

14 Conversion Conversion Conversion Conversion 2K/min HT rate Time (min) Time (min) time (min) Conv 2K/min HT Rate Tdsc HT Rate Ts Figure 9: Conversion and temperature in time 1mm sample of UP at 2K/min controlled no LN 2 (a) 1mm sample of UP at 2K/min uncontrolled (b) UP sample of 32.2mg analyzed by DSC at 2K/min (c) Conversion 2K/min Journal of Plastics Technology 4 (28) 3 14 HT rate Temperature (C) Temperature (C) Temperature (C)

15 Due to the unavailability of rapidly cooling this thicker sample with LN 2 in order to maintain an accurate heating there is an overshoot in the temperature that could not be eliminated. If purposely an inaccurate temperature control is used which assumes no overheating of the sample a huge overshoot will occur as seen in fig. 9b. When comparing the previous result to a DSC sample temperature (T s ) result (fig. 9c) we may clearly see a similar temperature rise to the one encountered by the DSC at such high rates. Key differences are seen when comparing the conversion curves. Since the Raman spectroscopy equipment is monitoring the actual molecular reaction the overshoot of the temperature leads to an extreme increase in reaction rate that leads to a sharp rise up to maximum conversion. Fig. 9a shows a similar behavior but to a lower extent due to the lower overheating. On the contrary, the DSC does not report such a behavior but shows a smooth curved transition to full conversion as if the reaction had lowered its reaction rate after 8% conversion. This shows that for these highly accelerated reactions the heat release in time detected by the DSC is not directly related to the real conversion of the resin due to the heat transfer delays present and the overheating of the sample. In order to be able to control the heating rate at 2K/min without the need of LN 2 or other advanced cooling systems the thickness of the sample had to be reduced. Fig. 1 shows a result for a sample thickness of.5mm. Overheating of the sample was greatly controlled by the control system but not fully eliminated. A hump on the heating rate curve may be seen right over the steepest part of the conversion curve, demonstrating how the sample overheated and simultaneously the reaction rate increased. For this case due to the better control of sample overheating a decrease in the reaction rate is seen after 85% conversion. Conversion time (min) Conv 2K/min HT rate Figure 1: Conversion and temperature in time for a.5mm sample of UP heated at 2K/min from room temperature to 18 C. Journal of Plastics Technology 4 (28) Temperature (C)

16 By reducing the sample size again, this time to.2mm as shown in fig. 11 the sample overheating was eliminated. No need for advanced cooling systems was required but our temperature control proved efficient at this sample thickness. A lower overall reaction rate may be observed, seen by the lower slope of the conversion curve. Smoother changes were obtained during the initial and final stages of the reaction with a period of time where the reaction was accelerated of about 2.5min and a very uniform heating rate. Conversion Conv 2K/min time (min) Figure 11: Conversion and temperature in time for a.2mm sample of UP heated at 2K/min from room temperature to 18 C. 4 CONCLUSIONS The reacting chamber developed for monitoring highly accelerated unsaturated polyester curing reactions provided very consistent repeatable heating rates. By use of Raman spectroscopy curing reactions were accurately monitored and final conversion values and reaction rates determined. Raman spectroscopy provided real-time information of the actual conversion values which may be directly related process conditions. The high sensitivity temperature control system reported a real sample temperature and produced very smooth heating rates. It was shown how there is a delay between the heating rate reported by the differential scanning calorimetry system and the real sample temperature due to heat transfer between surfaces and the low conductivity resin. The exothermic peak generated by the DSC was not coincident in time with the temperature rise of the sample during overheating but was reported at a later time. The faster the heating rate the greater the delay. When the reaction commenced and high reaction rates occurred, overheating which was not shown in the DSC heating rate output was observed in the actual sample temperature. Journal of Plastics Technology 4 (28) 3 16 HT rate

17 Rates were varied from 2.5K/min to 2K/min and sample mass from 17mg- 4mg. Sample overheating of as much as 8 C-1 C for larger mass higher rate samples and 3 C-16 C for slower rate smaller samples were detected. Conversion curves of overheated samples also deviated from what was shown by Raman spectroscopy analysis. The Raman analysis showed that due to sample overheating reaction rates increased greatly and were maintained until a full conversion value was reached. Maximum conversion values never reached 1% conversion because some molecules end trapped between each other and never reach a neighboring molecule to react. It was determined that for our particular resin, optimal thickness values that would allow monitoring reactions at constant heating rates were under.5mm for heating rates of 5-1K/min and under.2mm for rates of 2K/min. Modeling of thermosets is a subject of much interest and several generalized equations have been developed to try to describe the curing process. The main difficulty is that many of these equations are based on empirical data which in turn depends on reaction conditions such as heating rates. For modeling purposes or for determination of constants, this Raman spectroscopy method may be used with confidence assuring consistent results with constant heating rates and real-time conversion values. 5 ACKNOWLEDGEMENTS The authors wish to thank Mark Kemper from Kaiser Optical Systems for providing the Raman Spectroscopy equipment and Pete Johnson from Teel Plastics for providing all necessary materials. 6 REFERENCES [1] T.A. Osswald; G. Menges [2] K. Sakaguchi; H. Takemoto Material Science of Polymers for Engineers. 2 nd edition, Hanser Gardner Publications, Inc., 334, 23. Conversion of Fumarate Double Bonds in Unsaturated Polyester Resins in Copolymerization with Styrene. J. Macromol. Sci., Part A 1(6):p , Journal of Plastics Technology 4 (28) 3 17

18 [3] J.L. Koeing; T.K. Shin Structure of Unsaturated Polyester Resins Crosslinked with Styrene as Studied by Raman Spectroscopy. Journal of Polymer Science, 1:p , [4] K. Sakaguchi Investigation of Conversion of Styrene and Fumarate Double bonds in Unsaturated Polyester Resins. J. Macromol. Sci., Part A 8(3):p , [5] T. Imai Infrared Study of the Cure of Unsaturated Polyester. Journal of Applied Polymer Science, 11(7):p , [6] W.D Cook; O. Delatycki [7] C.D. Han and K.W. Lem [8] P.B. Zetterlund; A.F. Johnson [9] Y.S. Yang; L.J. Lee Re-examination of the Crosslinking Process in Styrene-Unsaturated Polyester Systems. J. Macromol. Sci., Part A 12(5):p , Chemorheology of Thermosetting Resins I. The Chemorheology and Curing Kinetics of Unsaturated Polyester Resin. Journal of Applied Polymer Science, 28(1):p , Free volume-based modeling of free radical crosslinking polymerization of unsaturated polyesters. Polymer, 43:p , 22 Comparison of Thermal and Infrared Spectroscopic analyses in the formation of polyurethane, unsaturated polyester, and their blends. Journal of Applied Polymer Science, 36(6):p , Journal of Plastics Technology 4 (28) 3 18

19 [1] P. Penczek, E.; Rudnik, B. Arczewska, R. Ostrysz [11] Y.J. Huang; C.J. Chen [12] Y.J. Huang; C.J. Chen [13] T.A. Gorovaya; V.N. Korotkov [14] M.R. Barone; D.A. Caulk [15] M.R. Kamal and S. Sourour [16] H. Janeschitz- Kriegel; H. Wippel; Ch. Paulik; G. Eder [17] C.H. Wu; G. Eder; H. Janeschitz- Kriegel Curing of unsaturated polyester resins with low exotherm peak. Die Angewandte Makromolekulare Chemie, 229:p.15-27, 1995 Curing of Unsaturated Polyester Resins Effects of Comonomer Composition. III. Medium- Temperature Reactions. Journal of Applied Polymer Science, 48:p , 1993 Curing of Unsaturated Polyester Resins: Effects of Comonomer Composition. II. High-Temperature Reactions. Journal of Applied Polymer Science, 47:p , 1993 Quick cure of thermosetting composites Composites Part A, 27A:p , 1996 The Effect of Deformation and Thermoset Cure on Heat Conduction in a Chopped-Fiber Reinforced Polyester During Compression Molding Int. J. Heat Mass Transfer, 22:p , 1979 Kinetics and Thermal Characterization of Thermoset Cure Polymer Engineering and Science, 13(1):p.59-64, 1973 Polymer Crystalization Dynamics as Reflected by Differential Scanning Calorimetry Colloid. Polym. Sci., 271:p , 1993 Polymer Crystalization Dynamics as Reflected by Differential Scanning Calorimetry, Colloid. Polym. Sci., 271:p , 1993 Journal of Plastics Technology 4 (28) 3 19

20 [18] J.L. Koeing; T.K. Shin Structure of Unsaturated Polyester Resins Crosslinked with Styrene as Studied by Raman Spectroscopy Journal of Polymer Science, 1:p , 1972 Keywords: Curing Reactions, Overheating, Unsaturated Polyester, Raman Spectroscopy, Differential Scanning Calorimetry Schlüsselworte: Härtungsreaktionen, Überhitzung, Ungesättigte Polyester, Raman-Spektroskopie, Differential Scanning Calorimetry Journal of Plastics Technology 4 (28) 3 2

21 Author/Autor: Javier C. Cruz, M.Sc. (author) Prof. Prof. h.c. Dr. Tim A. Osswald (Professor) Department of Mechanical Engineering University of Wisconsin-Madison 1513 University Avenue Madison, WI 5376 USA Editor/Herausgeber: Europe/Europa Prof. Dr.-Ing. Dr. h.c. G. W. Ehrenstein, verantwortlich Lehrstuhl für Kunststofftechnik Universität Erlangen-Nürnberg Am Weichselgarten Erlangen Deutschland Phone: +49/()9131/ Fax.: +49/()9131/ ehrenstein@lkt.uni-erlangen.de Publisher/Verlag: Carl-Hanser-Verlag Jürgen Harth Ltg. Online-Services & E-Commerce, Fachbuchanzeigen und Elektronische Lizenzen Kolbergerstrasse Muenchen Phone.: 89/ Fax: 89/ harth@hanser.de osswald@engr.wisc.edu Website: Phone.: +1/ Fax.: +1/ The Americas/Amerikas Prof. Dr. Tim A. Osswald, responsible Polymer Engineering Center, Director University of Wisconsin-Madison 1513 University Avenue Madison, WI 5376 USA Phone: +1/ Fax.: +1/ osswald@engr.wisc.edu Editorial Board/Beirat: Professoren des Wissenschaftlichen Arbeitskreises Kunststofftechnik/ Professors of the Scientific Alliance of Polymer Technology Journal of Plastics Technology 4 (28) 3 21