- Email: [email protected]
Measurement of glass transition temperature of crosslinked EVA encapsulant by thermal analysis for photovoltaic application
Renewable Energy 43 (2012) 218e223
Contents lists available at SciVerse ScienceDirect
Renewable Energy journal homepage: www.elsevier.com/locate/renene
Measurement of glass transition temperature of crosslinked EVA encapsulant by thermal analysis for photovoltaic application K. Agroui a, *, G. Collins b, J. Farenc c a
Silicon Technology Development Unit (UDTS) 2, Bd. Dr. Frantz Fanon, BP 140 Alger 7 Merveilles, Algiers, Algeria Department of Biomedical Engineering, New Jersey Institute of Technology, Newark, NJ 07102, USA c Plasma Laboratory and Energy Conversion (LAPLACE) UMR 5213, Paul Sabatier University, 118, route de Narbonne 31062 Toulouse cedex 9, France b
a r t i c l e i n f o
a b s t r a c t
Article history: Received 13 May 2011 Accepted 9 November 2011 Available online 16 December 2011
The purpose of the experiment was to better understand the changes due to thermal transitions and the molecular organizations of EVA encapsulant material after cross linking process by thermal analysis methods as DSC, TSC and DMTA. DSC experiments on EVA show a glass transition at about 33.1 C, which is characteristic of crystalline phase and an endothermic peak at temperature of 55 C characteristic of amorphous phase. The basic results by TSC technique is that there are two relaxations that are reproducibly observed in crosslinked EVA encapsulant material. At temperature polarization 60 C, a low temperature relaxation occurs at temperature 24.4 C and a high temperature relaxation occurs at temperature þ30.4 C. DMTA results exhibit two tand peaks located at 14.9 C and þ66.6 C. In addition, our results reveal that the glass transition temperature determined by TSC experiments in depolarization mode is more relevant than DSC and DMTA methods. TSC was chosen due to its low equivalent frequency consideration, useful to study encapsulant material exhibiting multiple relaxations. Ó 2011 Elsevier Ltd. All rights reserved.
Keywords: EVA Crosslinking PV module Thermal analysis Glass transition temperature
1. Introduction The qualification of photovoltaic (PV) modules described in the IEC tests, did not appear to provide the structure changes of the encapsulant based on polymeric material during PV module encapsulation process or during outdoor exposure. In general, polymers are designed for a certain range of operating temperatures where they possess desired properties; it is therefore suitable that the polymer’s glass transition region is outside the range of operating temperatures. Generally, the region between 0 C and 40 C is called the glass transition region and the selection of encapsulant materials is based on very low glass transition temperature (Tg). Low temperature phase transition behaviour in polymers is important, because this transition has a strong impact on materials mechanical properties and PV module encapsulation reliability. For this reason, the Tg of encapsulant must be lower or equal to 40 C as criteria requirement for polymers in PV [1]. Thermal analysis techniques are used to characterize the transitions and relaxations of polymers, where the challenge is actually to detect very weak Tg and even sub-Tg relaxations. The Tg can be
* Corresponding author. Tel/fax. þ213 21 43 35 11. E-mail address: [email protected] (K. Agroui). 0960-1481/$ e see front matter Ó 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.renene.2011.11.015
determined readily only by observing the temperature range in which a significant change takes place in some specific electrical, mechanical, thermal, or other physical property. It seems obvious that to standardize Tg determination and to obtain Tg values compatible with each other it is necessary to select the most popular way of processing curves and to optimize experimental condition parameters for all experiments [2]. Thermally Stimulated Current (TSC) is a driven technique that provides a better sensitivity to Tg and sub-Tg glass relaxations [3] and also has been frequently employed to investigate the molecular motions in polymeric materials [4]. The high interest of TSC technique is due to the very low equivalent frequency and its capability to resolve complex dielectric transitions. TSC complements perfectly other thermal analysis methods like as Differential Scanning Calorimetry (DSC), Dynamic Mechanical Thermal Analysis (DMTA) to determine fundamental properties of EVA such as phase transition, rheology and molecular mobility characteristics. The aim of this work is to analyze the changes in the relevant thermal behaviour of EVA encapsulant material after crosslinking process over a broad temperature around Tg by DSC, TSC, and DMTA thermal analysis. Special interest will be focussed on the specific TSC relaxation parameters like activation energy and relaxation frequency determination by using the initial rise method.
K. Agroui et al. / Renewable Energy 43 (2012) 218e223
219
2. Material and experimental analysis techniques 2.1. Sample preparation EVA (ethylene vinyl acetate) sample test is based on EVA A9918 (33% of vinyl acetate) standard cure from STR manufacturer is referred to as original EVA [5,6]. The original EVA film was inserted between two pieces of Teflon and placed on the low chamber of the SPIRE-LAMINATOR 240. The laminationecure cycle is executed under vacuum of 0.1 mm Hg in order to produce EVA without bubbles. During the first 10 min the encapsulant temperature is raised to 160 C where it is maintained for 15 min to the end of the cycle [7]. Original EVA film submitted to the cure step cycle is referred to as crosslinked EVA. The heating and cooling rate during the process is fixed at 20 C/min and 40 C/ min respectively. 2.2. Thermal analysis description Fig. 1. Thermal program for the TSDC experiment at 0 C polarization temperature.
2.2.1. DSC DSC is a calorimetric method that measures heat flow as a function of temperature associated with thermally active transitions such as crystallization, melting and glass transitions. All DSC experiments were conducted in a nitrogen atmosphere by using a TA Q100 instrument. The DSC program used to evaluate the behaviour of the previously cured EVA sample was a heat-cool-heat cycle. The first heating was done at 7 C/min; the cooling was done at 40 C/min; the final heating was done at 7 C/min. Furthermore the high value of cooling rate was chosen in order to avoid crystallization of the polymer sample. The heating program was a conventional temperature ramp from 50 C to 85 C,as the expected thermal events lie within this range. Whereas DSC measures heat flow as a function of a constant rate of change in temperature, modulated DSC (MDSC) superimpose a sinusoidal temperature modulation on this rate. The sinusoidal change in temperature permits the measurement of heat-capacity effects simultaneously with the kinetic effect. MDSC is an enhancement to conventional DSC whereby the total heat flow is separated into reversing and non-reversing components. The reversing signal contains heat-capacity events such as the glass transition and melting. The non-reversing signal contains kinetic events such as crystallization, crystal perfection and reorganization, cure, and decomposition [8]. MDSC experiments were conducted using the TA Q100 in the MDSC mode. The MDSC experiments conditions were a temperature oscillation of 2 C amplitude with a period of 60 s and a heating rate of 1 C/min. 2.2.2. TSC For crystalline material, the glass transition could not be detected easily using DSC. Therefore, thermally stimulated current (TSC) as a more sensitive thermal analysis technique was used to determine the Tg. TSC can be simply described as a non-calorimetric method for detecting the transitions that depend on changes in the mobility of molecular scale dipolar structural units. In this work, TSC analysis is performed in Thermally Stimulated Depolarization Current (TSDC) mode. The instrument used for the study the TSDC spectra of EVA is the TSC/RMA Model 9000 from Thermold, LP in the temperature range of 80 C to 60 C because the glass transition of EVA lies within this range. In TSDC, a voltage of 200 VDC is applied across the sample in order to align the dipoles of relaxation processes in the internal structure at polarization temperature Tp, the electric field is about 0.44 106 V/m. After this, the material is brought rapidly by lowering the temperature to 130 C in a nitrogen
atmosphere, which the electric field still applied as described in Fig. 1. Once the polar groups are frozen in the alignment, the electric field is turned off and the material is heated at a specific heating rate of 7 C/min (the temperature program is identical to that used in DSC) until temperature Tf. As the material is heated, the oriented polar groups again become mobile but this time without the electric field. This leads to the polar groups to reorganize in the more stable unaligned configuration. As the dipoles reorganize, the motion of the polar groups generates a current. This current is thermally stimulated because in this technique the current that is generated, as response to the electric field, is analyzed as function of temperature [9,10]. 2.2.3. DMTA The complex dynamic modulus E expression is given by the relation (1):
E ¼ E0 þjE00 E0
(1) E00
Where: is the storage modulus, is the loss modulus. The loss factor tgd expression is given by the relation (2):
tg d ¼
E00 E0
(2)
Thermo mechanical properties of the crosslinked EVA were measured using a Rheometric DMTA IV in a nitrogen atmosphere. Crosslinked EVA specimens were cut from the samples and mounted in the rectangular tension fixture. The typical sample dimensions were 10 mm long, 3.3 mm wide and 0.4 mm thick. During the dynamic temperature ramp experiment, the heating rate was 7 C/min and the frequency was 1 Hz. The static force was maintained at 20% higher than the dynamic force. 3. Results and discussions 3.1. Gel content measurements First, EVA samples are subjected to gel content measurements, by extracting method, which represents the amount of crosslinking occurring during the cure. The minimum acceptable gel content recommended for PV module encapsulation is 65%. The gel content measurements produced in EVA before and after cure cycle is 85% according to STR test procedure by using toluene as a solvent [11]. It is noted that a residual amount of the peroxide curing agent will remain in the EVA material after curing.
220
K. Agroui et al. / Renewable Energy 43 (2012) 218e223
3.2. Melting and transition temperatures Fig. 2 shows the heat-cool-heat data profile on the DSC instrument for crosslinked EVA sample. During the rapid cooling a large exothermic peak appears at about 36 C due to exothermic crystallization. This behaviour is better seen in Fig. 3, where the heating cycles are expanded.On first heating there is one endothermic process clearly visible with peak at 47.5 C with a shoulder at 55 C [12]. The enthalpy change for a given phase transition may be found by integrating over the area in which the transition is seen to occur on the DSC plot according to ISO 11357e3 [13]. The magnitude of the integrated temperature peak for EVA is 31.9 J/g. And the endothermic temperature is described as the result of melting of larger, more regularly formed crystallites [14]. After cooling, the second heating is designed to remove the possible thermal history of the sample and also it ensures good contact with the DSC pan to enable good data to be taken. On the second heating, the endothermic peak is no longer observed due to the fact that the structure of the material has changed as a result of curing at high temperature; instead, there is a very broad endothermic with peak at 55 C and the magnitude of the integrated temperature peak for EVA is 14.4 J/g. The broad endothermic on the second heating is suggestive of a broad distribution of crystallite sizes. Apparently during the rapid cooling cycle, the crystallization process is not fast enough to develop well-formed crystal structures. In order to examine the effect of encapsulation process on EVA crystallinity, the degree of crystallinity cc was calculated via the total enthalpy method [15] according to the Equation (3):
cc ð%Þ ¼
DHm 100% DH100
(3)
Where: DHm is the specific enthalpy of melting of the sample studied and DH100 is the specific enthalpy of melting for 100% crystalline Polyethylene (288 J/g). Since the previous thermal history of a polymer affects the calculated degree of crystallinity, EVA crystallinity was evaluated after being subjected to a second heating designed to remove the effect of prior thermal history [16,17]. Based on EVA formulation used in our experiments, the degree of crystallinity of EVA is not affected after crosslinking reaction and is calculated to be 5%. Based on this result the crosslinked EVA is a semi crystalline material. In the DSC technique, the Tg appears as a step transition and not a peak such as might be seen with a melting transition. Analysis of the Tg using the total heat flow signal with conventional DSC is
Fig. 3. DSC curves of the crosslinked EVA in heat-cool-heat mode with expanded heating cycles.
difficult due to the overlapping enthalpic relaxation. For this reason, MDSC is well suited for this purpose. The MDSC data in Fig. 4 shows the Tg and overlapping enthalpic relaxation events clearly separated into the reversing and non-reversing heat flow signals respectively. During the MDSC experiment EVA exhibits a broad sigmoidal decrease in reversible heat flow in the range of 50 C to about 10 C, and there is also a distinct sigmoidal decrease over the range 10 C to about 40 C. As such, a Tg should be observable as a sigmoidal change in heat flow and is determined from the midpoint of the reversible heat flow transition on MDSC curve and is found to be 33.1 C.
Fig. 4. MDSC curves of the crosslinked EVA.
3.3. Effect of heating rate on endothermic peak Fig. 5 shows the DSC curves of EVA at different heating rates between 2 and 10 C/min. It can be seen that the peak temperature of melting increased with the increasing heating rate. The second approach is the evaluation of the activation energy of endothermic peak by using the Kissinger method [18]. The apparent activation energy is deduced from the Equation (4):
lnðqÞ ¼ C þ Fig. 2. DSC curves of the crosslinked EVA in heat-cool-heat mode.
W RTm
(4)
K. Agroui et al. / Renewable Energy 43 (2012) 218e223
Fig. 5. DSC scans of the crosslinked EVA material at various heating rates.
Where: C is a constant, W is the activation energy of the melting, R is the perfect gas constant (8314 J/mol K), Tm is the peak temperature of melting and q is the heating rate. The activation energy W can be obtained from the plot of ln(q) vs. 1000/Tm. as seen in Fig. 6. The activation energy of the main endothermic peak is calculated to be 45.6 kJ/mol, which is in good agreement with literature results [19].
Fig. 7. TSDC curves of the crosslinked EVA at 0 C and 60 C polarization temperatures.
population of crystalline perfection [21,22]. The literature discussion explains this. The Polymer literature indicates that polyethylene can have a Tg in the region of 30 C, and poly(vinyl acetate) can have a Tg in the range of 25e39 C. The complex behaviour of these peaks suggests that there are two processes occurring in this temperature range due to the structural relaxation mechanism in EVA material; first there is the relaxation characteristic of the Tg, and second, there is the crystal melting as indicated in the DSC results. It is noted that the TSDC clearly identifies the high glass transition temperature, however the DSC clearly also identifies the melting process occurring near the hotter glass transition. The complexity of the high temperature TSDC peak makes it difficult to separately characterize the Tg and the crystal melting. The width of the DSC peak temperature may overlap the position of the high temperature TSDC peak, the DSC peak temperature does not correspond to anything that was observed in the TSDC experiments. Fig. 8 provides direct comparison results of the TSDC at 60 C polarization temperature and the MDSC by measuring reversible heat flow. The two sigmoidal changes in heat flow are consistent with the two Tg relaxations that have been described in TSDC experiments.
Fig. 6. Plot of ln(q) vs. 1000/Tm of the crosslinked EVA for the main endothermic peak.
3.4. Thermal structural relaxation 3.4.1. Glass transition temperature in TSDC mode In order to examine the thermal relaxations dipoles of EVA material, we run TSDC analysis in the same sequence by using polarization at 0 C followed by polarization at 60 C. The TSDC spectra of crosslinked EVA sample at 60 C polarization temperature appears to have a low and a high temperature relaxation process as seen in Fig. 7. Crosslinked EVA has a somewhat broad low temperature peak at 36.7 C, and a high temperature peak at 30.7 C related respectively to b and a relaxations [20]. There are examples of this endothermic behaviour that have been published in the technical literature as the consequence of
221
Fig. 8. TSDC and MDSC comparative curves of the crosslinked EVA.
222
K. Agroui et al. / Renewable Energy 43 (2012) 218e223
3.4.2. Relaxation parameters For the determination of the characteristic parameters of the process i.e. the activation energy and the relaxation frequency the Initial rise method is used [4,23]. This method is based on the fact that for a single relaxation process and for T<
W JðTÞ ¼ J0 exp kT
(5)
Where: J0 is the constant, W is the activation energy, T is the temperature, and k is the Boltzmann’s constant. The activation energy can be determined by the slope of the plot ln(J(T)) versus 1/T. The relaxation frequency f(Tmax) at the peak temperature Tmax can be extracted from the elementary spectrum by using the relation (6):
Tmax
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi qW ¼ kfðTmax Þ
(6)
Where q is the heating rate. The Arrhenius plots for the low and high TSDC relaxation temperature are illustrated in Figs. 9 and 10 respectively. The relaxation parameters are summarized in Table 1.
Fig. 9. Plot of ln(J) vs. 1000/T of crosslinked EVA at low temperature relaxation peak.
Fig. 10. Plot of ln(J) vs. 1000/T of crosslinked EVA at high temperature relaxation peak.
Table 1 Low and high TSDC relaxation parameters at 60 C polarization temperature.
Activation energy (kJ/mol) Relaxation frequency (103 Hz)
Low temperature peak
High temperature peak
27.0 6.8
74.2 0.1
3.5. Dynamic thermo mechanical behaviour Fig. 11 shows the DMTA curves of the crosslinked EVA, where the storage modulus (E’), the loss modulus (E’’) and the loss tangent (tand) curves were plotted at frequency of 1Hz in the range of 60 C to 100 C. EVA modulus shows a strong dependency on the temperature and there are at least two distinct drops in storage modulus. At temperature 37 C, crosslinked EVA exhibits a bending point and high elastic modulus value around 541 MPa, which decrease significantly to 16.7 MPa at temperature 10 C corresponding to the first drop modulus. This strong decrease can be attributed to the b relaxation of EVA [24] as manifested by tand peak at about 15 C in the tand vs. temperature graph. This low temperature tand peak can be assigned as the Tg in accordance to IEC 61006 [25], which is much higher than Tg of 33.1 C and 36.4 C measured by DSC and TSC methods respectively. Because PV encapsulant materials provide mechanical support to the cells, one would prefer the Tg measured by DMTA for mechanical considerations [26]. As the temperature increases a second modulus drop attributed to the a transition is observed and the storage modulus is 1.2 MPa at temperature 60 C characterized by tand peak temperature at 66 C. It appears clearly that in the temperature range located between the a and b relaxations, i.e., the broad range that includes those temperatures at which polymers are practically applied, the thermal properties of EVA undergo pronounced changes around glass transition temperature [27,28]. The low temperature mechanical loss process is taken to correspond to the low temperature relaxation peak observed in the TSDC experiments. It is also taken to be responsible for the broad sigmoidal decrease in reversible heat flow seen at low temperature in the MDSC experiments. The high temperature drop in modulus is fairly complex for the EVA sample and may reflex the complexity of the loss processes associated with an overlapping glass transition and crystal melting as discussed above. EVA films showed a decrease in elastic modulus values by a factor of 830 over the whole temperature range from 37 C to 85 C. These transitions, however, cause EVA to be embitter at low temperatures (w15 C)
Fig. 11. Variation of DMTA curves of the crosslinked EVA vs. temperature.
K. Agroui et al. / Renewable Energy 43 (2012) 218e223
and to be very soft at high temperatures (>40 C). This large change in mechanical properties under application relevant temperatures and also impact events at cold temperature could cause some severe problems during the service life time of a PV module, starting with delamination and a reduced ability to withstand mismatches in thermal expansion and eventually cracking of the cell or the wiring [29]. 4. Conclusion The determination of the glass transition temperature for EVA material by DSC, TSC and DMTA methods is found to be not similar but approximately compatible with each other under optimal test conditions. The observed temperature can vary significantly depending on the property chosen for observation and on details of the experimental technique (e.g., cooling and heating rate, test frequency, polarization temperature). The analysis of EVA sample by TSC has indicated that the position of the low temperature peak has been found to be near the glass transition temperature region. For the EVA under investigation the glass transition region temperature is not outside the range of operating temperatures between 40 and þ85 C based on measurement procedures for encapsulant materials used in photovoltaic modules according to IEC standard. Also, the temperatures in desert deployed modules can certainly reach 100 C, which is greater than 85 C used in qualification tests. TSC is a very promising technique to study mobility and glass transition in polymeric materials, DSC is more sensitive to melting and crystallization, while DMTA reflects these two behaviours. The results presented in this paper, provide an important basis for the further development of TSC thermal analysis and particularly the Tg determination, as a quality control test, in polymer encapsulant provided from different manufactures and based on different raw materials. References [1] Lewis KJ. Encapsulant material requirements for photovoltaic modules. In: Geblein CG, Williams DJ, Deanin RD, editors. Polymers in solar energy utilization. Washington, DC: ACS; 1983. p. 367e85. [2] Mazurin OV. Problems of compatibility of the values of glass transition temperatures published in the world literature. Glass Phys Chem 2007;33(1): 22e33. [3] Saffell JR, Matthiesen A, McIntyre R, Ibar JP. Comparing thermal stimulated current (TSC) with other thermal analytical methods to characterize the amorphous phase of polymers. Thermochim Acta 1991;192:243e64. [4] Moura Ramos JJ, Correia NT, Diogo HP. TSDC as a tool to study slow molecular mobility in condensed complex systems. J Non-Cryst Solids 2006;352(42e49): 4753e7. [5] Yamaki SB, Prado EA, Atvars TDZ. Phase transitions and relaxation processes in Ethylene Vinyl Acetate copolymers probed by fluorescence spectroscopy. Eur J Polym 2002;38:1811e26.
223
[6] Cuddihy EF, Coulbert CD, Liang RH, Gupta A, Willis P, Baum B. Applications of Ethylene Vinyl Acetate as an encapsulation material for terrestrial photovoltaic. Jet Propulsion Laboratory (JPL), JPL Publication; April 15, 1983. pp. 83e35. [7] Burger DR, Cuddiky EF. Vacuum lamination of photovoltaic modules. In: Geblein CG, Williams DJ, Deanin RD, editors. Polymers in solar energy utilization. Washington, DC: ACS; 1983. p. 408e19. [8] Folmer JCW, Franzen S. Study of polymer glasses by modulated differential scanning calorimetry. Chem 2003;80(7):813e8. [9] Recber AC. Thermal analysis of poly(Desmino Arylates). Master thesis, New Jersey University; 2004. [10] Collins G, Yoo SU, Recber A, Jaffe M. Thermal analysis of complex relaxation processes in Poly(Desaminotyrosyl-Tyrosine Arylates). Polymer 2007;48(4): 975e88. [11] Determining the gel content of ethylene vinyl acetate, Specialized Technology Resources, Inc., Available on http://www.PhotoCap.net. [12] Stark W, Jaunich M. Investigation of Ethylene Vinyl Acetate copolymer (EVA) by thermal analysis DSC and DMA. Polym Test 2011;30(2):236e42. [13] ISO 11357-3. Determination of temperature and enthalpy of melting and crystallization. In: Plastics-Differential scanning calorimetry (DSC). 1st ed.; 1999. [14] Androsch R. Melting and crystallization of Poly(Ethylene-co-Octene) measured by modulated DSC and temperature resolved X-ray diffraction. Polymer 1999;40:2805e12. [15] Reyes-Labarta JA, Olaya MM, Marcilla A. DSC and TGA study of the transitions involved in the thermal treatment of binary mixtures of PE and EVA copolymer with a crosslinking agent. Polymer 2006;47:8194e202. [16] Brogly M, Nardin M, Schultz J. Effect of Vinyl Acetate content on crystallinity and second-order transitions in Ethylene Vinyl acetate copolymers. J Polym Sci 1997;64(10):1903e12. [17] McEvoy RL, Krause S, Wut P. Surface characterization of ethylene vinyl acetate (EVA) and ethylene-Acrylic acid (EAA) co-polymers using XPS and AFM. Polymer 1998;39(2l):5223e39. [18] Kissinger HE. Reaction kinetics in differential thermal analysis. Anal Chem 1957;29:1702e6. [19] Bregulla M, Köhl M, Lampe B, Oreski G, Philipp D, Wallner G, Andersnd Weiß K. Degradation mechanisms of Ethylene Vinyl Acetate copolymer: new Studies including ultra fast cure foils, In: Proceedings of the 22nd European photovoltaic solar energy conference, Milan (Italy), 3e7 September 2007; pp. 2704e2707. [20] Agroui K, Collins G. Characterization of EVA encapsulant material by thermally stimulated current technique. Sol Energy Mater Sol Cells 2003;80(2):33e45. [21] Bistec S, Kunemann P, Schultz J. Crystalline modifications of Ethylene-Vinyl Acetate copolymers induced by a tensile drawing: effect of the molecular weight. Polymer 1998;39(20):4875e81. [22] Tsocheva D, Tsanov T, Terlemezyan L. Structure of composite films containing polyaniline studied by DSC. J Therm Anal Cal 2001;66:415e22. [23] Mandowski A, Swiatek J. Generalization of initial rise methods for TSC and TL. In: Proceedings of the8th international symposium on electrets, (ISE 8), Paris (France), 7e9 September 1994; pp. 455e460. [24] Woo L, Ling MTK, Westphal SP. Dynamic mechanical studies on secondary relaxations of Ethylene copolymers. Thermochimica Acta 1994;243:147e54. [25] IEC 61006. Methods of test for the determination of the glass transition temperature of the electrical insulating materials; 1991. [26] Kempe M, Jorgensen GJ, Terwilliger KM, McMahon TJ, Kennedy CE, Borek TT. Acetic acid production and glass transition concerns with Ethylene Vinyl Acetate used in photovoltaic devices. Sol Energy Mater Sol Cells 2007;91(4): 315e29. [27] Nielsen L. Transitions in Ethylene polymers. J Polym Sci 1960;XLII:357e66. [28] Popli R, Glotin M, Mandelkern L, Bens RS. Dynamic mechanical studies of alpha and beta relaxations of Polyethylene. J Polym Sci Polym Phys Ed 1984; 22(3):407e48. [29] Sadok M, Mehdaoui A. Outdoor testing of photovoltaic arrays in the Saharan region. Renewable Energy 2008;33(12):2516e24.
Contents lists available at SciVerse ScienceDirect
Renewable Energy journal homepage: www.elsevier.com/locate/renene
Measurement of glass transition temperature of crosslinked EVA encapsulant by thermal analysis for photovoltaic application K. Agroui a, *, G. Collins b, J. Farenc c a
Silicon Technology Development Unit (UDTS) 2, Bd. Dr. Frantz Fanon, BP 140 Alger 7 Merveilles, Algiers, Algeria Department of Biomedical Engineering, New Jersey Institute of Technology, Newark, NJ 07102, USA c Plasma Laboratory and Energy Conversion (LAPLACE) UMR 5213, Paul Sabatier University, 118, route de Narbonne 31062 Toulouse cedex 9, France b
a r t i c l e i n f o
a b s t r a c t
Article history: Received 13 May 2011 Accepted 9 November 2011 Available online 16 December 2011
The purpose of the experiment was to better understand the changes due to thermal transitions and the molecular organizations of EVA encapsulant material after cross linking process by thermal analysis methods as DSC, TSC and DMTA. DSC experiments on EVA show a glass transition at about 33.1 C, which is characteristic of crystalline phase and an endothermic peak at temperature of 55 C characteristic of amorphous phase. The basic results by TSC technique is that there are two relaxations that are reproducibly observed in crosslinked EVA encapsulant material. At temperature polarization 60 C, a low temperature relaxation occurs at temperature 24.4 C and a high temperature relaxation occurs at temperature þ30.4 C. DMTA results exhibit two tand peaks located at 14.9 C and þ66.6 C. In addition, our results reveal that the glass transition temperature determined by TSC experiments in depolarization mode is more relevant than DSC and DMTA methods. TSC was chosen due to its low equivalent frequency consideration, useful to study encapsulant material exhibiting multiple relaxations. Ó 2011 Elsevier Ltd. All rights reserved.
Keywords: EVA Crosslinking PV module Thermal analysis Glass transition temperature
1. Introduction The qualification of photovoltaic (PV) modules described in the IEC tests, did not appear to provide the structure changes of the encapsulant based on polymeric material during PV module encapsulation process or during outdoor exposure. In general, polymers are designed for a certain range of operating temperatures where they possess desired properties; it is therefore suitable that the polymer’s glass transition region is outside the range of operating temperatures. Generally, the region between 0 C and 40 C is called the glass transition region and the selection of encapsulant materials is based on very low glass transition temperature (Tg). Low temperature phase transition behaviour in polymers is important, because this transition has a strong impact on materials mechanical properties and PV module encapsulation reliability. For this reason, the Tg of encapsulant must be lower or equal to 40 C as criteria requirement for polymers in PV [1]. Thermal analysis techniques are used to characterize the transitions and relaxations of polymers, where the challenge is actually to detect very weak Tg and even sub-Tg relaxations. The Tg can be
* Corresponding author. Tel/fax. þ213 21 43 35 11. E-mail address: [email protected] (K. Agroui). 0960-1481/$ e see front matter Ó 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.renene.2011.11.015
determined readily only by observing the temperature range in which a significant change takes place in some specific electrical, mechanical, thermal, or other physical property. It seems obvious that to standardize Tg determination and to obtain Tg values compatible with each other it is necessary to select the most popular way of processing curves and to optimize experimental condition parameters for all experiments [2]. Thermally Stimulated Current (TSC) is a driven technique that provides a better sensitivity to Tg and sub-Tg glass relaxations [3] and also has been frequently employed to investigate the molecular motions in polymeric materials [4]. The high interest of TSC technique is due to the very low equivalent frequency and its capability to resolve complex dielectric transitions. TSC complements perfectly other thermal analysis methods like as Differential Scanning Calorimetry (DSC), Dynamic Mechanical Thermal Analysis (DMTA) to determine fundamental properties of EVA such as phase transition, rheology and molecular mobility characteristics. The aim of this work is to analyze the changes in the relevant thermal behaviour of EVA encapsulant material after crosslinking process over a broad temperature around Tg by DSC, TSC, and DMTA thermal analysis. Special interest will be focussed on the specific TSC relaxation parameters like activation energy and relaxation frequency determination by using the initial rise method.
K. Agroui et al. / Renewable Energy 43 (2012) 218e223
219
2. Material and experimental analysis techniques 2.1. Sample preparation EVA (ethylene vinyl acetate) sample test is based on EVA A9918 (33% of vinyl acetate) standard cure from STR manufacturer is referred to as original EVA [5,6]. The original EVA film was inserted between two pieces of Teflon and placed on the low chamber of the SPIRE-LAMINATOR 240. The laminationecure cycle is executed under vacuum of 0.1 mm Hg in order to produce EVA without bubbles. During the first 10 min the encapsulant temperature is raised to 160 C where it is maintained for 15 min to the end of the cycle [7]. Original EVA film submitted to the cure step cycle is referred to as crosslinked EVA. The heating and cooling rate during the process is fixed at 20 C/min and 40 C/ min respectively. 2.2. Thermal analysis description Fig. 1. Thermal program for the TSDC experiment at 0 C polarization temperature.
2.2.1. DSC DSC is a calorimetric method that measures heat flow as a function of temperature associated with thermally active transitions such as crystallization, melting and glass transitions. All DSC experiments were conducted in a nitrogen atmosphere by using a TA Q100 instrument. The DSC program used to evaluate the behaviour of the previously cured EVA sample was a heat-cool-heat cycle. The first heating was done at 7 C/min; the cooling was done at 40 C/min; the final heating was done at 7 C/min. Furthermore the high value of cooling rate was chosen in order to avoid crystallization of the polymer sample. The heating program was a conventional temperature ramp from 50 C to 85 C,as the expected thermal events lie within this range. Whereas DSC measures heat flow as a function of a constant rate of change in temperature, modulated DSC (MDSC) superimpose a sinusoidal temperature modulation on this rate. The sinusoidal change in temperature permits the measurement of heat-capacity effects simultaneously with the kinetic effect. MDSC is an enhancement to conventional DSC whereby the total heat flow is separated into reversing and non-reversing components. The reversing signal contains heat-capacity events such as the glass transition and melting. The non-reversing signal contains kinetic events such as crystallization, crystal perfection and reorganization, cure, and decomposition [8]. MDSC experiments were conducted using the TA Q100 in the MDSC mode. The MDSC experiments conditions were a temperature oscillation of 2 C amplitude with a period of 60 s and a heating rate of 1 C/min. 2.2.2. TSC For crystalline material, the glass transition could not be detected easily using DSC. Therefore, thermally stimulated current (TSC) as a more sensitive thermal analysis technique was used to determine the Tg. TSC can be simply described as a non-calorimetric method for detecting the transitions that depend on changes in the mobility of molecular scale dipolar structural units. In this work, TSC analysis is performed in Thermally Stimulated Depolarization Current (TSDC) mode. The instrument used for the study the TSDC spectra of EVA is the TSC/RMA Model 9000 from Thermold, LP in the temperature range of 80 C to 60 C because the glass transition of EVA lies within this range. In TSDC, a voltage of 200 VDC is applied across the sample in order to align the dipoles of relaxation processes in the internal structure at polarization temperature Tp, the electric field is about 0.44 106 V/m. After this, the material is brought rapidly by lowering the temperature to 130 C in a nitrogen
atmosphere, which the electric field still applied as described in Fig. 1. Once the polar groups are frozen in the alignment, the electric field is turned off and the material is heated at a specific heating rate of 7 C/min (the temperature program is identical to that used in DSC) until temperature Tf. As the material is heated, the oriented polar groups again become mobile but this time without the electric field. This leads to the polar groups to reorganize in the more stable unaligned configuration. As the dipoles reorganize, the motion of the polar groups generates a current. This current is thermally stimulated because in this technique the current that is generated, as response to the electric field, is analyzed as function of temperature [9,10]. 2.2.3. DMTA The complex dynamic modulus E expression is given by the relation (1):
E ¼ E0 þjE00 E0
(1) E00
Where: is the storage modulus, is the loss modulus. The loss factor tgd expression is given by the relation (2):
tg d ¼
E00 E0
(2)
Thermo mechanical properties of the crosslinked EVA were measured using a Rheometric DMTA IV in a nitrogen atmosphere. Crosslinked EVA specimens were cut from the samples and mounted in the rectangular tension fixture. The typical sample dimensions were 10 mm long, 3.3 mm wide and 0.4 mm thick. During the dynamic temperature ramp experiment, the heating rate was 7 C/min and the frequency was 1 Hz. The static force was maintained at 20% higher than the dynamic force. 3. Results and discussions 3.1. Gel content measurements First, EVA samples are subjected to gel content measurements, by extracting method, which represents the amount of crosslinking occurring during the cure. The minimum acceptable gel content recommended for PV module encapsulation is 65%. The gel content measurements produced in EVA before and after cure cycle is 85% according to STR test procedure by using toluene as a solvent [11]. It is noted that a residual amount of the peroxide curing agent will remain in the EVA material after curing.
220
K. Agroui et al. / Renewable Energy 43 (2012) 218e223
3.2. Melting and transition temperatures Fig. 2 shows the heat-cool-heat data profile on the DSC instrument for crosslinked EVA sample. During the rapid cooling a large exothermic peak appears at about 36 C due to exothermic crystallization. This behaviour is better seen in Fig. 3, where the heating cycles are expanded.On first heating there is one endothermic process clearly visible with peak at 47.5 C with a shoulder at 55 C [12]. The enthalpy change for a given phase transition may be found by integrating over the area in which the transition is seen to occur on the DSC plot according to ISO 11357e3 [13]. The magnitude of the integrated temperature peak for EVA is 31.9 J/g. And the endothermic temperature is described as the result of melting of larger, more regularly formed crystallites [14]. After cooling, the second heating is designed to remove the possible thermal history of the sample and also it ensures good contact with the DSC pan to enable good data to be taken. On the second heating, the endothermic peak is no longer observed due to the fact that the structure of the material has changed as a result of curing at high temperature; instead, there is a very broad endothermic with peak at 55 C and the magnitude of the integrated temperature peak for EVA is 14.4 J/g. The broad endothermic on the second heating is suggestive of a broad distribution of crystallite sizes. Apparently during the rapid cooling cycle, the crystallization process is not fast enough to develop well-formed crystal structures. In order to examine the effect of encapsulation process on EVA crystallinity, the degree of crystallinity cc was calculated via the total enthalpy method [15] according to the Equation (3):
cc ð%Þ ¼
DHm 100% DH100
(3)
Where: DHm is the specific enthalpy of melting of the sample studied and DH100 is the specific enthalpy of melting for 100% crystalline Polyethylene (288 J/g). Since the previous thermal history of a polymer affects the calculated degree of crystallinity, EVA crystallinity was evaluated after being subjected to a second heating designed to remove the effect of prior thermal history [16,17]. Based on EVA formulation used in our experiments, the degree of crystallinity of EVA is not affected after crosslinking reaction and is calculated to be 5%. Based on this result the crosslinked EVA is a semi crystalline material. In the DSC technique, the Tg appears as a step transition and not a peak such as might be seen with a melting transition. Analysis of the Tg using the total heat flow signal with conventional DSC is
Fig. 3. DSC curves of the crosslinked EVA in heat-cool-heat mode with expanded heating cycles.
difficult due to the overlapping enthalpic relaxation. For this reason, MDSC is well suited for this purpose. The MDSC data in Fig. 4 shows the Tg and overlapping enthalpic relaxation events clearly separated into the reversing and non-reversing heat flow signals respectively. During the MDSC experiment EVA exhibits a broad sigmoidal decrease in reversible heat flow in the range of 50 C to about 10 C, and there is also a distinct sigmoidal decrease over the range 10 C to about 40 C. As such, a Tg should be observable as a sigmoidal change in heat flow and is determined from the midpoint of the reversible heat flow transition on MDSC curve and is found to be 33.1 C.
Fig. 4. MDSC curves of the crosslinked EVA.
3.3. Effect of heating rate on endothermic peak Fig. 5 shows the DSC curves of EVA at different heating rates between 2 and 10 C/min. It can be seen that the peak temperature of melting increased with the increasing heating rate. The second approach is the evaluation of the activation energy of endothermic peak by using the Kissinger method [18]. The apparent activation energy is deduced from the Equation (4):
lnðqÞ ¼ C þ Fig. 2. DSC curves of the crosslinked EVA in heat-cool-heat mode.
W RTm
(4)
K. Agroui et al. / Renewable Energy 43 (2012) 218e223
Fig. 5. DSC scans of the crosslinked EVA material at various heating rates.
Where: C is a constant, W is the activation energy of the melting, R is the perfect gas constant (8314 J/mol K), Tm is the peak temperature of melting and q is the heating rate. The activation energy W can be obtained from the plot of ln(q) vs. 1000/Tm. as seen in Fig. 6. The activation energy of the main endothermic peak is calculated to be 45.6 kJ/mol, which is in good agreement with literature results [19].
Fig. 7. TSDC curves of the crosslinked EVA at 0 C and 60 C polarization temperatures.
population of crystalline perfection [21,22]. The literature discussion explains this. The Polymer literature indicates that polyethylene can have a Tg in the region of 30 C, and poly(vinyl acetate) can have a Tg in the range of 25e39 C. The complex behaviour of these peaks suggests that there are two processes occurring in this temperature range due to the structural relaxation mechanism in EVA material; first there is the relaxation characteristic of the Tg, and second, there is the crystal melting as indicated in the DSC results. It is noted that the TSDC clearly identifies the high glass transition temperature, however the DSC clearly also identifies the melting process occurring near the hotter glass transition. The complexity of the high temperature TSDC peak makes it difficult to separately characterize the Tg and the crystal melting. The width of the DSC peak temperature may overlap the position of the high temperature TSDC peak, the DSC peak temperature does not correspond to anything that was observed in the TSDC experiments. Fig. 8 provides direct comparison results of the TSDC at 60 C polarization temperature and the MDSC by measuring reversible heat flow. The two sigmoidal changes in heat flow are consistent with the two Tg relaxations that have been described in TSDC experiments.
Fig. 6. Plot of ln(q) vs. 1000/Tm of the crosslinked EVA for the main endothermic peak.
3.4. Thermal structural relaxation 3.4.1. Glass transition temperature in TSDC mode In order to examine the thermal relaxations dipoles of EVA material, we run TSDC analysis in the same sequence by using polarization at 0 C followed by polarization at 60 C. The TSDC spectra of crosslinked EVA sample at 60 C polarization temperature appears to have a low and a high temperature relaxation process as seen in Fig. 7. Crosslinked EVA has a somewhat broad low temperature peak at 36.7 C, and a high temperature peak at 30.7 C related respectively to b and a relaxations [20]. There are examples of this endothermic behaviour that have been published in the technical literature as the consequence of
221
Fig. 8. TSDC and MDSC comparative curves of the crosslinked EVA.
222
K. Agroui et al. / Renewable Energy 43 (2012) 218e223
3.4.2. Relaxation parameters For the determination of the characteristic parameters of the process i.e. the activation energy and the relaxation frequency the Initial rise method is used [4,23]. This method is based on the fact that for a single relaxation process and for T<
W JðTÞ ¼ J0 exp kT
(5)
Where: J0 is the constant, W is the activation energy, T is the temperature, and k is the Boltzmann’s constant. The activation energy can be determined by the slope of the plot ln(J(T)) versus 1/T. The relaxation frequency f(Tmax) at the peak temperature Tmax can be extracted from the elementary spectrum by using the relation (6):
Tmax
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi qW ¼ kfðTmax Þ
(6)
Where q is the heating rate. The Arrhenius plots for the low and high TSDC relaxation temperature are illustrated in Figs. 9 and 10 respectively. The relaxation parameters are summarized in Table 1.
Fig. 9. Plot of ln(J) vs. 1000/T of crosslinked EVA at low temperature relaxation peak.
Fig. 10. Plot of ln(J) vs. 1000/T of crosslinked EVA at high temperature relaxation peak.
Table 1 Low and high TSDC relaxation parameters at 60 C polarization temperature.
Activation energy (kJ/mol) Relaxation frequency (103 Hz)
Low temperature peak
High temperature peak
27.0 6.8
74.2 0.1
3.5. Dynamic thermo mechanical behaviour Fig. 11 shows the DMTA curves of the crosslinked EVA, where the storage modulus (E’), the loss modulus (E’’) and the loss tangent (tand) curves were plotted at frequency of 1Hz in the range of 60 C to 100 C. EVA modulus shows a strong dependency on the temperature and there are at least two distinct drops in storage modulus. At temperature 37 C, crosslinked EVA exhibits a bending point and high elastic modulus value around 541 MPa, which decrease significantly to 16.7 MPa at temperature 10 C corresponding to the first drop modulus. This strong decrease can be attributed to the b relaxation of EVA [24] as manifested by tand peak at about 15 C in the tand vs. temperature graph. This low temperature tand peak can be assigned as the Tg in accordance to IEC 61006 [25], which is much higher than Tg of 33.1 C and 36.4 C measured by DSC and TSC methods respectively. Because PV encapsulant materials provide mechanical support to the cells, one would prefer the Tg measured by DMTA for mechanical considerations [26]. As the temperature increases a second modulus drop attributed to the a transition is observed and the storage modulus is 1.2 MPa at temperature 60 C characterized by tand peak temperature at 66 C. It appears clearly that in the temperature range located between the a and b relaxations, i.e., the broad range that includes those temperatures at which polymers are practically applied, the thermal properties of EVA undergo pronounced changes around glass transition temperature [27,28]. The low temperature mechanical loss process is taken to correspond to the low temperature relaxation peak observed in the TSDC experiments. It is also taken to be responsible for the broad sigmoidal decrease in reversible heat flow seen at low temperature in the MDSC experiments. The high temperature drop in modulus is fairly complex for the EVA sample and may reflex the complexity of the loss processes associated with an overlapping glass transition and crystal melting as discussed above. EVA films showed a decrease in elastic modulus values by a factor of 830 over the whole temperature range from 37 C to 85 C. These transitions, however, cause EVA to be embitter at low temperatures (w15 C)
Fig. 11. Variation of DMTA curves of the crosslinked EVA vs. temperature.
K. Agroui et al. / Renewable Energy 43 (2012) 218e223
and to be very soft at high temperatures (>40 C). This large change in mechanical properties under application relevant temperatures and also impact events at cold temperature could cause some severe problems during the service life time of a PV module, starting with delamination and a reduced ability to withstand mismatches in thermal expansion and eventually cracking of the cell or the wiring [29]. 4. Conclusion The determination of the glass transition temperature for EVA material by DSC, TSC and DMTA methods is found to be not similar but approximately compatible with each other under optimal test conditions. The observed temperature can vary significantly depending on the property chosen for observation and on details of the experimental technique (e.g., cooling and heating rate, test frequency, polarization temperature). The analysis of EVA sample by TSC has indicated that the position of the low temperature peak has been found to be near the glass transition temperature region. For the EVA under investigation the glass transition region temperature is not outside the range of operating temperatures between 40 and þ85 C based on measurement procedures for encapsulant materials used in photovoltaic modules according to IEC standard. Also, the temperatures in desert deployed modules can certainly reach 100 C, which is greater than 85 C used in qualification tests. TSC is a very promising technique to study mobility and glass transition in polymeric materials, DSC is more sensitive to melting and crystallization, while DMTA reflects these two behaviours. The results presented in this paper, provide an important basis for the further development of TSC thermal analysis and particularly the Tg determination, as a quality control test, in polymer encapsulant provided from different manufactures and based on different raw materials. References [1] Lewis KJ. Encapsulant material requirements for photovoltaic modules. In: Geblein CG, Williams DJ, Deanin RD, editors. Polymers in solar energy utilization. Washington, DC: ACS; 1983. p. 367e85. [2] Mazurin OV. Problems of compatibility of the values of glass transition temperatures published in the world literature. Glass Phys Chem 2007;33(1): 22e33. [3] Saffell JR, Matthiesen A, McIntyre R, Ibar JP. Comparing thermal stimulated current (TSC) with other thermal analytical methods to characterize the amorphous phase of polymers. Thermochim Acta 1991;192:243e64. [4] Moura Ramos JJ, Correia NT, Diogo HP. TSDC as a tool to study slow molecular mobility in condensed complex systems. J Non-Cryst Solids 2006;352(42e49): 4753e7. [5] Yamaki SB, Prado EA, Atvars TDZ. Phase transitions and relaxation processes in Ethylene Vinyl Acetate copolymers probed by fluorescence spectroscopy. Eur J Polym 2002;38:1811e26.
223
[6] Cuddihy EF, Coulbert CD, Liang RH, Gupta A, Willis P, Baum B. Applications of Ethylene Vinyl Acetate as an encapsulation material for terrestrial photovoltaic. Jet Propulsion Laboratory (JPL), JPL Publication; April 15, 1983. pp. 83e35. [7] Burger DR, Cuddiky EF. Vacuum lamination of photovoltaic modules. In: Geblein CG, Williams DJ, Deanin RD, editors. Polymers in solar energy utilization. Washington, DC: ACS; 1983. p. 408e19. [8] Folmer JCW, Franzen S. Study of polymer glasses by modulated differential scanning calorimetry. Chem 2003;80(7):813e8. [9] Recber AC. Thermal analysis of poly(Desmino Arylates). Master thesis, New Jersey University; 2004. [10] Collins G, Yoo SU, Recber A, Jaffe M. Thermal analysis of complex relaxation processes in Poly(Desaminotyrosyl-Tyrosine Arylates). Polymer 2007;48(4): 975e88. [11] Determining the gel content of ethylene vinyl acetate, Specialized Technology Resources, Inc., Available on http://www.PhotoCap.net. [12] Stark W, Jaunich M. Investigation of Ethylene Vinyl Acetate copolymer (EVA) by thermal analysis DSC and DMA. Polym Test 2011;30(2):236e42. [13] ISO 11357-3. Determination of temperature and enthalpy of melting and crystallization. In: Plastics-Differential scanning calorimetry (DSC). 1st ed.; 1999. [14] Androsch R. Melting and crystallization of Poly(Ethylene-co-Octene) measured by modulated DSC and temperature resolved X-ray diffraction. Polymer 1999;40:2805e12. [15] Reyes-Labarta JA, Olaya MM, Marcilla A. DSC and TGA study of the transitions involved in the thermal treatment of binary mixtures of PE and EVA copolymer with a crosslinking agent. Polymer 2006;47:8194e202. [16] Brogly M, Nardin M, Schultz J. Effect of Vinyl Acetate content on crystallinity and second-order transitions in Ethylene Vinyl acetate copolymers. J Polym Sci 1997;64(10):1903e12. [17] McEvoy RL, Krause S, Wut P. Surface characterization of ethylene vinyl acetate (EVA) and ethylene-Acrylic acid (EAA) co-polymers using XPS and AFM. Polymer 1998;39(2l):5223e39. [18] Kissinger HE. Reaction kinetics in differential thermal analysis. Anal Chem 1957;29:1702e6. [19] Bregulla M, Köhl M, Lampe B, Oreski G, Philipp D, Wallner G, Andersnd Weiß K. Degradation mechanisms of Ethylene Vinyl Acetate copolymer: new Studies including ultra fast cure foils, In: Proceedings of the 22nd European photovoltaic solar energy conference, Milan (Italy), 3e7 September 2007; pp. 2704e2707. [20] Agroui K, Collins G. Characterization of EVA encapsulant material by thermally stimulated current technique. Sol Energy Mater Sol Cells 2003;80(2):33e45. [21] Bistec S, Kunemann P, Schultz J. Crystalline modifications of Ethylene-Vinyl Acetate copolymers induced by a tensile drawing: effect of the molecular weight. Polymer 1998;39(20):4875e81. [22] Tsocheva D, Tsanov T, Terlemezyan L. Structure of composite films containing polyaniline studied by DSC. J Therm Anal Cal 2001;66:415e22. [23] Mandowski A, Swiatek J. Generalization of initial rise methods for TSC and TL. In: Proceedings of the8th international symposium on electrets, (ISE 8), Paris (France), 7e9 September 1994; pp. 455e460. [24] Woo L, Ling MTK, Westphal SP. Dynamic mechanical studies on secondary relaxations of Ethylene copolymers. Thermochimica Acta 1994;243:147e54. [25] IEC 61006. Methods of test for the determination of the glass transition temperature of the electrical insulating materials; 1991. [26] Kempe M, Jorgensen GJ, Terwilliger KM, McMahon TJ, Kennedy CE, Borek TT. Acetic acid production and glass transition concerns with Ethylene Vinyl Acetate used in photovoltaic devices. Sol Energy Mater Sol Cells 2007;91(4): 315e29. [27] Nielsen L. Transitions in Ethylene polymers. J Polym Sci 1960;XLII:357e66. [28] Popli R, Glotin M, Mandelkern L, Bens RS. Dynamic mechanical studies of alpha and beta relaxations of Polyethylene. J Polym Sci Polym Phys Ed 1984; 22(3):407e48. [29] Sadok M, Mehdaoui A. Outdoor testing of photovoltaic arrays in the Saharan region. Renewable Energy 2008;33(12):2516e24.