Contactless optoelectronic technique for monitoring epoxy cure Andrea Cusano, Vincenzo Buonocore, Giovanni Breglio, Antonio Calabro`, Michele Giordano, Antonello Cutolo, and Luigi Nicolais
We describe a novel noninvasive optical technique to monitor the refractive-index variation in an epoxy-based resin that is due to the polymerization process. This kind of resin is widely used in polymer matrix composites. It is well known that the process of fabricating a thermoset-based composite involves mass and heat transfer coupled with irreversible chemical reactions that induce physical changes. To improve the quality and the reliability of these materials, monitoring the cure and optimization of the manufacturing process are of key importance. We discuss the basic operating principles of an optical system based on angle deflection measurements and present typical cure-monitoring results obtained from optical characterization. The method provides a flexible, high-sensitivity, material-independent, low-cost, noninvasive tool for monitoring real-time refractive-index variation. © 2000 Optical Society of America OCIS codes: 120.4290, 120.5710, 160.5470, 220.4840.
1. Introduction
Thermoset polymer matrix composites are finding an increasing number of applications. They are light, and their mechanical properties are comparable with those of steel, titanium, and aluminum. However, the properties of polymer and polymer–matrix composites are depend strongly on the processing parameters: temperature, time, and pressure sequence. The chemorheological behavior of the resin is a function of the operational environment in which processing occurs. Depending on the nonhomogeneity and variability of the initial materials, the exothermic nature of the reaction, the low thermal diffusivity of the materials, the characteristics of the curing oven, When this research was performed, A. Cusano, G. Breglio 共email address,
[email protected]兲, and A. Cutolo were with the Department of Electronic Engineering, University of Naples “Federico II,” Via Claudio, 80125 Naples, Italy. A. Cutolo 共
[email protected]兲 is now with the Faculty of Engineering, University of Sannio, Corso Garibaldi, 107 82100, Benevento, Italy. V. Buonocore, M. Giordano 共
[email protected]兲, and L. Nicolais are with Institute of Composite Materials Technology, Piazza le Tecchio, 80, 80125 Naples, Italy. A. Calabro` 共e-mail address,
[email protected]兲 is with Italian Aerospace Research Center, Via Maiorise, 41043 Capua, Italy. Received 6 August 1999; revised manuscript received 22 November 1999. 0003-6935兾00兾071130-06$15.00兾0 © 2000 Optical Society of America 1130
APPLIED OPTICS 兾 Vol. 39, No. 7 兾 1 March 2000
and the geometry of the mold, the curing process may vary from place to place in a large molded structure. The curing process of composites is therefore the most critical and costly stage in the manufacturing of composite structures.1,2 Generally, the cure of low-molecular-weight prepolymer involves the transformation of a fluid resin into a rubber and into a solid glass; this is the result of the exothermic chemical reactions of the reactive groups that are present in the system, which develop a progressively denser polymeric network. Growth and branching of the polymeric chains are due to inframolecular reactions that initially occur in the liquid state until a critical degree of branching is reached and an infinite network and an insoluble material are formed. After gelification, successive cross-linking reactions increase the cross-link density, and the stiffness of the polymer is steadily increased, leading, at the end of the process, to the glassy structure of the fully cured thermoset.3 Considerable research and development effort has been expended to improve the quality and reliability of these types of material and to transform the curing process from the fixed, standard cycle offered by the resin maker to a scientific operation resulting in lower cost, higher yield, and higher-quality composite parts. To control the processing operation with feedback capability requires real-time in situ information about the condition of the material being processed. Moreover, off-line techniques may be used
for optimization of the control system. This requires the use of a sensing device. The need for suitable cure-monitoring sensors has been well recognized, and in recent years considerable efforts have been made to develop fiber optic cure sensors. Fiber optic sensors are able to measure temperature, pressure, strain, resin viscosity, and the chemical state of the reactants. Their miniature size and freedom from electromagnetic interference have made it possible to embed the fiber optic sensors into composite structures. Moreover, curing determines chemical and physical modifications in a large number of properties of a reacting resin. Among others, these optical properties 共i.e., absorbance and refractive index兲 are strongly correlated to the structural features of the developing polymeric network through the variation of density. Various approaches proposed and developed for cure monitoring are based primarily on evanescent wave interaction, Rayleigh and Brillouin backscattering, refractive-index measurements, transmission spectrum analysis, polymer fluorescence, and Fourier-transform Raman spectroscopy.4 The techniques based on spectroscopy require tunable wavelength sources that cover specific wavelength ranges. However, the refractive-index-based sensors can be used with a single-wavelength source. Therefore they are much cheaper.5 In particular, monitoring the variations of the refractive index is a suitable method for analyzing the evolution of thermoset resin polymerization. The refractive index of the resin is a nonlinear function of the temperature and of the degree of cure 共DOC, conversion兲. Under isothermal cure conditions the effect of the temperature on the refractive index can be eliminated. In this situation the optical response of the sensor depends only on the cure kinetics and the state of the cross linking in the material. From initial mixing to final cure, the indices of refraction of various epoxy– resin systems increase in the range of 0.01– 0.04.6 –10 Based on this line of argument, in this paper we discuss the basic principles of operation and present typical cure-monitoring results obtained from optical characterization. In this research to investigate the optical properties of epoxies, we used a typical epoxy– amine system. 2. Theoretical Background A.
Principle of Operation
Figure 1 shows the basic configuration used for refractive-index measurements. The laser beam impinges perpendicularly upon the resin–air interface. The transmitted beam is reflected by a sloped mirror. Then the reflected signal is transmitted in air with an output angle u, depending on the refractive index of the epoxy resin. The output laser spot is viewed onto a linear charge-induction device 共CID兲 camera located at a distance do from emission point 共a兲. Indicating by np the refractive index of the epoxy sample and by p the incidence angle of the reflected ray at the resin–air interface, we can express the
Fig. 1. Basic configuration for refractive-index measurements.
dependence of the epoxy’s refractive index on the output angle as sin共u兲 ⫽ npsin共p兲.
(1)
The geometric configuration of the sensor ensures the independence of the epoxy’s refractive index from incidence angle p at the resin–air interface, which is given by p ⫽ 2d,
(2)
where d is the known plane inclination. The resin’s refractive-index variation during the cure process causes an angle deflection that changes u of light beam that reaches the camera. By monitoring laser spot displacements on the receiving camera as a function of the epoxy’s refractive index np, we obtain both physical and chemical variations during a cure process in real time. The sensor’s sensitivity is an increasing function of d and do. To avoid internal total reflection at the resin–air interface, we can express the maximum useful value for plane inclination by dmax ⫽ arcsin共1兾npmax兲兾2.
(3)
The upper limit for do is determined to the required measurement accuracy; too great a value for do will cause excessive laser spot enlargement with a consequent diminution in accuracy. For an epoxy resin the refractive index will change from approximately 1.54 to 1.59 during cure reaction. This implies a plane inclination dmax equal to 19.5°. Another important feature of our approach is due to the independence of the laser intensity fluctuations, but, in light of the dispersive features of these kinds of material, it requires a laser probe with a constant operation wavelength. B.
Factors that Affect the Measurements
The previous analysis is based on the hypothesis that the height of the epoxy–air interface does not vary during the test. Usually this assumption is not verified because of thermal expansion of either the apparatus or the epoxy and volume shrinkage 1 March 2000 兾 Vol. 39, No. 7 兾 APPLIED OPTICS
1131
冨
冊冨
␥appLtot ⫺ ␣Lepo
do ⬎⬎ kg
冉
␣˙ k T ⫹ ␣
␣˙
,
(8)
where v is the heating rate and ␣˙ represents the conversion rate. During isothermal scans the sensor response depends only on the degree of cure, and inequality 共8兲 becomes Fig. 2. Shift observed in the camera in the case of a height change during refractive-index monitoring.
during the resin cure. Hence, with reference to Fig. 2, the shift detected on the camera, ⌬Y, is the result of two factors: One, ⌬Yn, relies on the change in index and is a consequence of the output angle, and the other, ⌬Y␦, represents the unwanted shift that is due to the change in height ␦. Then we can write ⌬Y ⫽ ⌬Yn ⫹ ⌬Y␦ ⫽ do⌬ ⫹ ␦kg.
(4)
In particular, kg takes account into the relative orientation of the detector and the output angle; it is less than 1. It is worthwhile to note that ⌬Y␦ is independent of the receiver– epoxy distance. Thus, by increasing the epoxy–receiver distance until ⌬Yn ⬎⬎ ⌬Y␦, we can neglect the second term of the Eq. 共4兲; in this case the sensor’s response is only refractive-index sensitive. Based on this line of argument, the total change in height can be expressed as the result of the thermal expansion of the metallic apparatus, the thermal expansion of the resin, and the shrinkage that is due to the cure reactions as follows: ␦ ⫽ ␦epo ⫹ ␦app ⫺ ␦␣ ⫽ 共␥epoLepo ⫹ ␥appLapp兲⌬T ⫺ ␣Lepo⌬␣,
(5)
where ␥ is the linear thermal expansion factor, L is the height,  is the linear shrinkage coefficient, and ␣ is the degree of cure 共conversion兲. Typically, ␥app ⬎ ␥epo, which implies that ␦ ⬍ ␥appLtot⌬T ⫺ ␣Lepo⌬␣.
(6)
The refractive-index variation that is due to temperature and index changes is given by
冉
冊
n n ⌬Yn ⫽ do ⌬T ⫹ ⌬␣ ⫽ do k共T⌬T ⫹ ␣⌬␣兲. n T ␣ (7) In light of these considerations, the condition ⌬Yn ⬎⬎ ⌬Y␦ is ensured when 1132
APPLIED OPTICS 兾 Vol. 39, No. 7 兾 1 March 2000
do kg␣ ⬎⬎ . Lepo k␣
(9)
In the case of temperature changes of an unreacting resin, we can obtain the same condition by setting ␣˙ ⫽ 0 in inequality 共8兲: kg␥app do ⬎⬎ . Ltot kT
(10)
Furthermore, this analysis shows how by appropriately increasing the epoxy–receiver distance it is possible to obtain a robust system that is able to reject the effects of small vibrations. As is explained clearly in Section 4 below, we have chosen a value of do that ensures that ⌬Yn ⬎⬎ ⌬Y␦ in all cases that we investigated. 3. Apparatus Setup A.
Sensor Configuration
A 25-mW cw helium–neon laser at 632 nm was used as the light source. To measure the change in angle deflection d we monitored the laser spot displacements over the sensing plate of a solid-state camera. The camera sensor was equipped with a 512 ⫻ 512 array 共7.5 mm ⫻ 7.68 mm兲 of photodiodes 共pixels兲. Each pixel size was 15 m ⫻ 15 m, and the wavelength range extended from 0.190 to 1.1 m. The camera was connected to a PC computer, which performed all data acquisition, display, and storage functions. The sample holder was prepared with aluminum to improve the heat exchange between the sample and the heater. We positioned small holes in which to place thermocouples to record the temperature profiles during each isothermal cure experiment. To maximize the sensor sensitivity consistently with total internal reflection we chose the slope of the holder equal to 18°. The thermocouples were connected to a controller to stabilize the temperature profile during the cure reaction. B.
Signal Elaboration
In Fig. 3 we show the laser spot viewed by CID camera in the presence 关Fig. 3共a兲兴 and in the absence 关Fig. 3共b兲兴 of the epoxy sample. As one can see, the presence of the epoxy diffuses the laser spot. Furthermore, the maximum value of the received signal is difficult to detect. This observation suggested our
Fig. 4. Conversion versus time during an isothermal scan at 40 °C.
we observed the amplitude of the background noise and of the spurious reflections in a large number of tests. 共A Matlab code will perform this task automatically兲. 4. Experiment
The epoxy system used in this study was the epoxy resin EPON 828 共a product of the Shell Chemical Company兲 mixed with the stoichiometric quantity of TETA amine curing agent. The refractive index of the curing epoxy was determined as a function of time. A portion of the same freshly mixed epoxy sample was placed in a differential scanning calorimeter for calorimetric analysis. A.
Fig. 3. Laser spot viewed with the CID camera 共a兲 in the absence of the resin sample and 共b兲 in the presence of the resin.
using a center-of-gravity motion analysis to monitor the laser spot displacements. Nevertheless, the observation of a transmission that is essentially constant and of a diffusion that is substantially isotropic justifies the use of a center-of-gravity analysis. The center-of-gravity analysis uses a threshold principle with the aim of rejecting the noise effect and, as a consequence, improving the signal-to-noise ratio. Furthermore, two polarizing filters were used to attenuate the main spurious reflections that reached the receiver. To find the optimum threshold value
Calorimetric Analysis
The reaction rate is assumed to be proportional to the exothermic heat evolved per minute during cross linking per gram of epoxy sample. Differential scanning calorimetry is a standard technique for measuring the reaction rate as a function of time at a given temperature. We can find the extent of cure as a function of time by integrating the heat flow during the reaction and comparing that integrated heat flow with the total heat that evolves when the sample undergoes complete cure. We used a Mettler DSC 30 calorimeter to measure the degree of cure of the EPON 828; it was programmed to reproduce the same temperature profile as experienced by the epoxy sample used for optical characterization. A sample mass of 10 ⫾ 1 mg was measured into standard aluminum differential scanning calorimetry pans that were then covered with lids. Figure 4 shows the DOC of the epoxy of interest as a function of cure time during an isothermal cure experiment at 40 °C. As one can see, the epoxy samples did not achieve a complete cure. As the cure temperature increased, the DOC increased and the cure time for completion of the cure reaction decreased. 1 March 2000 兾 Vol. 39, No. 7 兾 APPLIED OPTICS
1133
Fig. 5. Epoxy’s refractive index during an isothermal scan at 40 °C.
B.
Camera Positioning
The distance of the receiver do from the apparatus and the dimensions of the apparatus were chosen in accordance with the theoretical analysis described above. As the linear shrinkage coefficient is 10⫺3 m⫺1,8 the refractive-index sensitivities to temperature and conversion are, respectively ⬇4.5 ⫻ 10⫺4 °C⫺1 共zero-cure reacting resin兲 and ⬇2 ⫻ 10⫺2,9 we chose the distance of the receiver do from the apparatus to be 20 cm, the maximal epoxy height to be less than approximately 1 cm, and the total height not to exceed 4 cm. With this setup the epoxy shrinkage during isothermal scans and the thermal expansion during both temperature ramp-up and testing of unreacting resin affect less than 1% of the sensor’s response. C.
Isothermal Test
Figure 5 shows the profile of the epoxy’s refractive index measured with the optical technique during the isothermal cure experiment at the same temperature used during thermocalorimetric characterization. The sensor’s response increases monotonically as a function of cure time, which indicates that the mismatch between the epoxy resin and air increases as the DOC increases. Because the experiments were carried out under isothermal conditions, the effects of temperature on refractive index were eliminated. Thus the variation of the optical response is determined only by the cross-linking kinetic rate and the
Fig. 7. Neat resin’s refractive index versus temperature.
amount of cross linking in the material, or the DOC. The profiles of the optical response contain the information on the cure rate and on the DOC of the polymerization. D.
Dynamic Scan
This particular cure cycle consisted of a temperature ramp-up from room temperature to 40 °C and a 40min 40 °C cure. In Fig. 6 the refractive-index profile is shown as a function of cure time. During the ramp-up the epoxy’s refractive index decreases as a consequence of the temperature increase. Then the refractive index starts to increase as the resin system cures. At the initiation of the cure process, the viscosity of the resin decreases as the temperature increases, because high temperature leads directly to high mobility of the molecules and to large spaces among the molecules. When the material becomes a more optically loose medium, the refractive index decreases. In the initial heating stage, the variation of the properties of the mixture resin– hardener was dominated by a physical phenomenon; i.e., as the temperature rose, the viscosity and refractive index decreased. However, as the cure reaction took place, the change in the optical properties of the sample was dominated by the chemical cross-linking reaction. The cure process tries to reduce the space among the molecules and to increase the density of material. Thus the refractive index increases with cure. E.
Temperature Effects
Figure 7 shows the refractive-index variations of the neat resin 共without the curing agent兲 during a temperature scan in the range 20 –70 °C. In absence of a curing agent, the refractive index decreases monotonically as the temperature increases because no cross-linking formation occurs. As one can see, the refractive-index profile is found to be linear with temperature with an angle coefficient of 2 ⫻ 10⫺4兾°C. 5. Conclusions
Fig. 6. Epoxy’s refractive index during a dynamic cure cycle. 1134
APPLIED OPTICS 兾 Vol. 39, No. 7 兾 1 March 2000
A novel contactless optoelectronic technique with which one is able to measure the refractive-index variations in reacting materials has been developed and tested with a commercial epoxy resin. Preliminary results indicate the capability of the sensor to
monitor the refractive-index variations that are due to both temperature and conversion. The proposed approach seems to have the potential for contactless high-sensitivity material-independent monitoring of refractive-index variation. Future research will address the use of the technique described here to explain clearly the relationship between the extent of curing and the refractive-index changes induced by polymerization. Improvements in the proposed technique will take into consideration the alignment and planarity of the air–material interface by design of a full transmission sensor based on a similar configuration. In addition, data collected by the technique will be useful in the design and the development of a fiber optic refractive-indexbased sensor for in situ cure monitoring. References 1. A. Maffezolli, J. M. Kenny, A. Trevisano, L. Torre, and L. Nicolais, “Process modeling of thermoset based composites,” Adv. Mater. 16B, 747–753 共1993兲. 2. J. M. Kenny, A. Apicella, and L. Nicolais, “A model for the thermal and chemoreological behavior of thermosets. 1. Processing of epoxy based composites,” Polym. Eng. Sci. 29, 973–983 共1989兲.
3. L. A. Berglund and J. M. Kenny, “Processing science for high performance thermoset composites,” SMPTE J. 27, 271–280 共1991兲. 4. Y. M. Liu, C. Ganesh, J. P. H. Steele, and J. E. Jones, “Fiber optic sensor development for real-time in-situ epoxy cure monitoring,” J. Compos. Mater. 31, 87–102 共1997兲. 5. P. A. Crosby, G. R. Powell, G. F. Fernando, D. N. Waters, C. M. France, and R. C. Spooncer, “A comparative study of optical fiber cure monitoring methods,” in Smart Structures and Materials 1997: Smart Sensing, Processing, Instrumentation, O. R. Claus, ed., Proc. SPIE 3042, 157–172 共1997兲. 6. M. A. Afromowitz and K. Y. Lam, “The optical properties of curing epoxies and application to the fiber-optic cure sensor,” Sens. Actuators A 21–23, 1107–1110 共1990兲. 7. M. A. Afromowitz, “Fiber optic polymer cure sensor,” J. Lightwave Technol. 6, 1591–1594 共1988兲. 8. M. A. Afromowitz and K. Y. Lam, “Fiber optic cure sensor for thermoset composites,” in Fiber Optic Smart Structures and Skins, E. Udd, ed., Proc. SPIE 986, 135–138 共1988兲. 9. M. A. Afromowitz and K. Y. Lam, “Fiber-optic epoxi composite cure sensor. I. Dependence of refractive index of an autocatalytic reaction epoxy system at 850 nm on temperature and extent of cure,” Appl. Opt. 34, 5635–5638 共1995兲. 10. M. A. Afromowitz and K. Y. Lam, “Fiber-optic epoxy composite cure sensor. II. Performance characteristics,” Appl. Opt. 34, 5639 –5643 共1995兲.
1 March 2000 兾 Vol. 39, No. 7 兾 APPLIED OPTICS
1135
