Printer Friendly

Moisture diffusion behavior in bismaleimide resin subjected to hygrothermal cycling.

Hung-Jue Sue (*)

Absorption-desorption-reabsorption hygrothermal cycling was performed in bismaleimide resin. Dynamic mechanical analysis and swelling experiments were conducted to investigate the effect of hygrothermal cycling on the moisture diffusion behavior in the resin. Non-Fickian diffusion behavior was observed in both the absorption and the reabsorption processes, while the desorption process is found to be Fickian. The molecular network structure of BMI appears to have changed during the hygrothermal cycling. This, in turn, altered the subsequent reabsorption process.


Bismaleimide (BMI), currently used as a matrix for high-temperature aerospace structural applications, experiences extreme environmental conditions during the expected life of service. The effects of hygrothermal aging on BMI resins and their composites have been studied by many researchers (1-5). Changes in material properties, such as glass transition temperature ([T.sub.g]) (2-4) and density (4) due to hygrothermal conditioning, have been reported. Both Fickian (2) and non-Fickian diffusion (1. 4. 5) behaviors have been observed in BMI resins.

It has been reported that the non-Fickian diffusion behavior observed in BMI resin is caused by hydrogen bonding between water and polymer (6), but some studies (7-9) have indicated that the non-Fickian diffusion behavior in polymers may be due to other factors. Chemical modification, such as the advancement of crosslinking networks, oxidation, or hydrolysis reactions during the conditioning, is considered a possible cause for non-Fickian moisture diffusion in polymers (7). Sahlin and Peppas (8) found that as the sorption temperature is increased, the non-Fickian behavior observed in epoxy resins disappears. Whitney and Browning (9) observed a two-stage diffusion process in neat epoxy resin at 71[degrees]C. They suggest that time-dependent matrix cracking is the mechanism associated with the two-stage diffusion process.

Hygrothermal cycling behavior in polymeric materials has been studied by many researchers (10-13). Hahn et al. (10) found that the initial absorption process in virgin specimens of AS/3501-5 graphite/epoxy composites facilitates the subsequent diffusion. Residual stresses appear to be the reason why the absorption process is slower than the desorption process. during the early stage of moisture exposure. The study by Cataldis et al. (11) shows that the diffusion coefficients in the absorption tests in epoxy are lower than those of the corresponding desorption tests. The diffusion coefficients in the first cycle were higher than those of the second cycle. They proposed that the decrease in diffusion coefficients is associated with the physical microdamage induced by the sorption of the first cycle at high relative humidity. Weitsman (12) reported that the diffusion process is reproducible in the unstressed epoxy. but not in the stressed specimen. For a given final humidity and temperature, it is found that the apparent equilibrium moisture content depends on the previous history of exposure of the specimen (13) even when no temperature change is involved. Specifically, at a given final humidity, a polymeric material appears to have a greater capacity for holding water if it is saturated initially at a higher humidity than if it is initially dry or saturated at a lower humidity.

Detailed studies on how the crosslinking density and hygrothermal conditioning temperatures affect the moisture diffusion behavior in BMI have been reported earlier (6). The objective of the current study is to further investigate the nature of non-Fickian diffusion behavior via absorption-desorption-reabsorption hygrothermal cycling. The cycling is expected to accelerate the hygrothermal aging process in the resin. It also allows us to examine the possible factors that lead to non-Fickian diffusion behavior in the resin, such as further curing, micro-damage and change of network structure. It is hoped that this study will contribute to a better understanding of the nature of non-Fickian diffusion behavior in BMI resin.


2.1. Material

The Cytec 5250-4-RTM BMI resin (14) used in this study is a three-component EMI, consisting of 4,4-bismaleimidiphenylmethane (BMPM), O,O'-diallyl bisphenol A (DABPA), and BMI-1,3-tolyl. Plaques were prepared using an infusion molding process. Resin was cured at 190[degrees]C for 6 hrs and then postcured at 230[degrees]C for another 6 hrs. This curing schedule is recommended by the resin supplier. No observable microcracks were found in the bulk resin upon cooling even after careful optical microscopy examinations.

2.2. Hygrothermal Conditioning

Specimens with dimensions of 25 x 25 x 2 (length x width x thickness) [mm.sup.3] were cut from the laminates using a diamond saw. Specimens were dried in a vacuum oven at 80[degrees]C under a vacuum of 762 mm-Hg until a constant weight was achieved. They were then placed in a humidity chamber and exposed to humid air with 100% relative humidity (RH) and at 70[degrees]C.

The desorption experiment was conducted on the saturated specimens from the absorption specimens above; Specimens were dried at 70[degrees]C and 762 mm-Hg vacuum until a constant weight was achieved. After desorption, specimens were returned to the humidity chamber for the second absorption experiment, i.e., reabsorption.

During the absorption-desorption-reabsorption cycling, specimens were removed from the chamber and weighed periodically using a balance with a precision of [+ or -]0.0001 g. Measuring time was subtracted from the calculated exposure time. Weight gain or weight loss of the specimen, M, is calculated using the following M = W - [W.sub.d]/[W.sub.d] x 100%, where [W.sub.d] is the weight of the dried specimen obtained at 80[degrees]C and 762 mm-Hg vacuum according to ASTM D5229, and W is the weight of the specimen at a given conditioning time. M vs. the square root of exposure time over thickness was plotted to generate diffusion curves. At least three specimens were used for the diffusion experiments. The average values were reported.

2.3. Dynamic Mechanical Analysis (DMA)

Dynamic mechanical analysis (Rheometrics RMS-800) was performed on BMI resin in a torsional mode, with 5[degrees]C per step. A 0.1% sinusoidal strain and a frequency of 1 Hz were used. The samples were tested at temperatures ranging from 25 to 350[degrees]C. The temperature at which the maximum Tan [delta] value is located is defined as [T.sub.g].

2.4. Swelling Experiment

Dimensional changes of the specimens due to hygrothermal conditioning were quantified using [DELTA]V = V - [V.sub.o]/[V.sub.o] x 100%, where [V.sub.o] is the volume of the dry sample before the absorption test, and V is the apparent volume of polymer and absorbed moisture during the hygrothermal conditioning. V is calculated using V = M/[rho], where M is mass, and p is the density of the specimen. Density was determined according to the ASTM standard D792, using [rho] = [w.sub.a]/[w.sub.a] - [w.sub.s] x [[rho].sub.s], where [W.sub.a] is the weight of sample in the air, [w.sub.s] is the weight of sample in the solvent, [[rho].sub.s] is the density of solvent (isopropanol, [[rho].sub.s] = 785 kg/[m.sup.3]). At least three specimens were used for each measurement.


3.1. Moisture Diffusion in BMI

Figure 1 shows the absorption-desorption-reabsorption behavior in BMI at 70[degrees]C. The moisture absorption and reabsorption behaviors in BMI are found to be similar. They both have a short-term Fickian and a long-term non-Fickian behavior. On the other hand, the Fickian curve can fit the desorption data well. As discussed in the previous paper (6), the non-Fickian diffusion behavior is due to the hydrogen bonding between water and the resin. The Langmuir model (15), which accounts for the hydrogen bonding interaction, is employed to describe the absorption behavior in BMI.

The Langmuir model assumes that diffusivity, [D.sub.[gamma]], is independent of both concentration and stress. Two types of water, bound and unbound, exist during the diffusion process with certain probabilities at [gamma] and [beta], respectively. An equilibrium moisture uptake. [M.sub.[infinity]], is obtained when the number of mobile molecules per unit volume, n, and the number of bound molecules per unit volume, N, approach values such that [gamma]n = [beta]N. A useful approximation for the total moisture uptake is (15):

[M.sub.t] [congruent to] [M.sub.[infinity]] {[beta]/[gamma] + [beta] [e.sup.-[gamma]t][1 - 8/[[pi].sup.2] [summation over ([infinity](odd)/l=1)] [e.sup.-[kl.sup.2]t]/[l.sub.2]] + [beta]/[gamma] + [beta] ([e.sup.-[beta]t] - [e.sup.-[gamma]t]) + (l - [e.sup.-[beta]t])}; 2[gamma], 2[beta] [less than or equal to] k (1)

where [M.sub.[infinity]] is the equilibrium moisture uptake, [M.sub.t] is the moisture uptake percentage at time t, l is the specimen thickness, and k = [[pi].sup.2] [D.sub.[gamma]]/[l.sup.2]. The BMI resin used in this study contains polar groups such as hydroxyl groups. The hydroxyl group can form hydrogen bonds with water during the diffusion process. Therefore, the Langmuir model is used to describe the absorption process in BMI. As seen in Fig. 1, the Langmuir curves fit all the experimental data well for both the absorption and reabsorption processes.

After desorption, a small amount of water still remains at 70[degree]C (Table 1). Diffusivities in desorption and reabsorption processes are the same, and they are higher than that in the absorption process. Although the moisture uptake at the Fickian plateau ([M.sub.f]) of the reabsorption process is higher than that of the absorption process, the equilibrium moisture uptakes ([M.sub.x] in Table 2) determined from the Langmuir model for the two processes are about the same. As mentioned in the previous paper (6). it is more reasonable to compare [M.sub.x] instead of [M.sub.f]. The probability for unbound water, [beta] is increased while the probability for bound water [gamma] remains the same. The above diffusion results suggest that hygrothermal history has an effect on the diffusion behavior in BMI.

3.2. DMA Spectra of BMI

Network structures of BMI after absorption-desorption-reabsorption hygrothermal conditionings were characterized by dynamic mechanical analysis (Fig. 2). Before the DMA test, the moisture saturation contents for specimens after the absorption and reabsorption processes were 4.5 wt% and 4.7 wt%. Respectively. The desorption specimen was dried to a constant weight from the saturated specimen. Although moisture content is expected to be reduced during the DMA testing, the effect of absorbed moisture on DMA spectra of BMI conditioned by absorption, desorption, and reabsorption is still significant. Figure 2 suggests that absorbed moisture acts as plasticizer in the resin. Shear storage moduli, G'. of the saturated specimens from absorption and reabsorption conditionings, start to decrease at around 120[degrees]C, and the rubber plateau shear storage moduli are also lower than those of the desorption specimen. The plasticization effect is more apparent for the reabsorption conditioned specimen than for the absorption conditioned specimen. i.e., the shear storage moduli were further decreased after reabsorption conditioning. The damping peak is broader in the saturated BMI. The [T.sub.g], if defined as the temperature at which the peak value of the Tan [delta] curve is located, of all three samples are the same. This s uggests that no further curing reaction has taken place in BMI resin during the hygrothermal cycling. Shear storage moduli after desorption are increased. The [beta] transition, which appears as a shoulder of the [alpha] transition is moved to a higher temperature after the absorption-desorption cycling. The above findings suggest that the network structure of BMI can be changed during different hygrothermal conditionings, and the absorbed moisture acts as a strong plasticizer, which lowers the moduli of the resin.

3.3. Swelling Experiment

Figures 3 and 4 are the swelling profiles for the samples conditioned during absorption, desorption and reabsorption, respectively. The percentage volume change ([DELTA]V/[V.sub.0], %) in the resin is plotted against the hypothetical volume change of absorbed water ([V.sub.water]/[V.sub.0], %). The swelling efficiency with the slope equal to 1 represents swelling that would be expected if the volumes of the dry resin and the absorbed water were additive. Swelling behavior during reabsorption is found to be similar to that seen in the absorption process. The slopes, obtained from linear regression for these two processes, are close to each other in regions I and II (Table 3). Region I is from zero volume change to about 2.2% volume change, which corresponds to the initial stage of absorption. The swelling of the resin in Region I is far less than the swelling of the resin in Region II, where the swelling efficiency is increased to close to 1 at volume change above 2.2%.

In the desorption process, regions I a nd II are assigned reversely since the desorption process is opposite to the absorption process (Fig. 4). The shrinkage efficiency of the desorbed sample becomes 1 in region II. Swelling coefficient, [lambda] is defined by the following equation:

[lambda] = [(V - [V.sub.0]).sub./[V.sub.0,%]]/Moisture uptake, wt% (2)

It is found that [lambda] for both absorption and reabsorption are about the same (Table 4).


Absorption, desorption and reabsorption cycling has an effect on the hygrothermal diffusion behavior in BMI. Both absorption and reabsorption are found to be non-Fickian. Diffusivities of the desorption and reabsorption processes are the same, and both are higher than that of the absorption process (Table 1). Since the Fourier transform infrared spectroscopy (FTIR) results show that no new peak appears (e.g., ether group appears due to the hydrolysis reaction) and no peak disappears during each conditioning process, the non-Fickian diffusion in the reabsorption process is not because of chemical reaction in BMI. No observable microscopic or macroscopic damages are found in the matrix even after exhaustive examination of the samples using optical microscopy. This further confirms that the non-Fickian diffusion behavior is caused by the interaction between water and BMI matrix.

The network structure of BMI is believed to have changed during this hygrothermal cycling. It is also believed that the network structure of BMI is relaxed after hygrothermal conditioning (6). The above findings are consistent with the present experimental results, indicating that the final volume of the desorbed specimen is slightly smaller than the original volume of the unconditioned specimen. At the same time, a small amount of residual water is found to remain in the system. These findings suggest that the network structure is altered, resulting in a slightly packed volume with tightly bound water inside the BMI resin. A modified diffusion path may be created after the absorption-desorption process since the diffusivity in desorption is the same as in the subsequent reabsorption process. The modified diffusion path makes moisture easier to diffuse in the beginning of the reabsorption process, i.e., the diffusivity of reabsorption is higher than that of absorption, and the [beta] value is increased in th e reabsorption process. Since the amount of free volume and the number of polar groups in the system remain about the same, [M.sub.x] and the swelling behavior are not much affected by the hygrothermal cycling.

Residual water after desorption is observed in the system. This phenomenon has been observed by many researchers (16-20). It has been reported that some of the absorbed water could not be removed unless a higher desorption temperature is used. Two types of bound water were postulated by using nuclear magnetic resonance (NMR) technique and desorption experiments (20). Type I involves water molecules forming a single hydrogen bond and/or dispersion bonding with epoxy resin network. This type of water is more easily desorbed from the resin. Type I bound water acts as a plasticizer, facilitating the movement of the network. Type II bonding results from water forming multiple hydrogen bonds with the resin network, possessing a higher activation energy and the water is harder to be removed from the resin. Type II bound water does not act as plasticizer but forms bridges between structural segments in BMI. The increases in ([beta]-transition temperature observed in the desorbed sample (Fig. 2) may be related to Type II bound water in the system. It is clear that the rate of hydrogen bonding process in BMI is rather slow (6). It may take years for material to become saturated with water if at all possible. Material may degrade before it is saturated with water.

A large amount of water can condense and form water dusters in polymers, leading to the formation of blister and other damage. Molecular chains are rearranged during hygrothermal conditioning. Although it is not the case for BMI resins, chemical reactions such as oxidation, hydrolysis, or depolymerization are all possible. One should take into account the environmental stability of the material when studying long-term hygrothermal diffusion behavior of polymers.


Non-Fickian diffusion behavior is observed during the absorption and reabsorption processes in BMI. The non-Fickian behavior is caused by the formation of hydrogen bonds between water and BMI. Network structure is believed to be altered during hygrothermal cycling. Residual water after desorption may be caused by the presence of tightly bound water, which needs a higher temperature to remove. No signs of microcrack formation in BMI are found during hygrothermal cycling.


The authors would like to thank Dow-UT for assisting in the fabrication of composite panels. The funding by the Air Force Office of Scientific Research (Grant No. F49620-98-1-0149) is greatly appreciated.

[Figure 1 omitted]

[Figure 2 omitted]

[Figure 3 omitted]

[Figure 4 omitted]
Table 1

Summary of Hygrothermal Cycling Data at 70[degrees]C.

 Absorption Desorption Reabsorption

Diffusivity, [mm.sup.2]/ 2.80 3.80 3.80
 sec, x [10.sup.-6]
(*) Mf, % 4.20 0.03 (**) 4.50

(*)[M.sub.f]: moisture uptake at Fickian plateau.

(**)Residual moisture.
Table 2.

Fitting Parameters for the Absorption and Reabsorption Behavior Based
on the Langmuir Model.

 [beta] x [10.sup.-4] [gamma] x [10.sup.-4]
Conditioning 1/hr 1/hr

Absorption 7.0 2.0
Reabsorption 10.0 2.0

Conditioning %

Absorption 5.3
Reabsorption 5.2
Table 3.

Linear Regression Slopes of the Swelling Experiments.

Conditioning Region I Region II

Absorption 0.48 0.77
Desorption 0.44 1.00
Reabsorption 0.49 0.80
Table 4.

Swelling Coefficients of BMI Under Absorption and Reabsorption

Conditioning (%[delta]V/% moisture)

Absorption 0.62
Reabsorption 0.63

(*.) Corresponding author. E-mail:


(1.) Y. Li, X Tang, J. Miranda, H.-J. Sue, J. Whitcomb and W. Bradley, 57th Annu. Tech. Conf.-Soc. Plast. Eng., 3423 (1999).

(2.) J. Cinquin and P. Abjean, 38th International SAMPE Symp. Exhib., 1539 (May 1993).

(3.) P.O. Biney, Y. Thong, and J. Zhou. Int. SAMPE Symp. Exhib., 43, 120 (1998).

(4.) M. Wilenski. The Improvement of the Hygrothermal and Mechanical Properties of Bismaleimide and K3B/1M7 Carbon Fiber Composites Through a Systematic Study of the Interphase, Ph.D. Thesis, Michigan State University (1997).

(5.) L. Bao and A. F. Yee, Moisture Transport and Damage Initiation in Advanced Composites, AFOSR Report, Sept 1999.

(6.) Y. Li. J. Miranda, and H.-J. Sue, "Hygrothermal Diffusion Behavior in Bismaleimide Resin," Polymer, in press (2000).

(7.) T. C. Wong and L. Broutman, Polym. Eng. Sci., 25, 529 (1985).

(8.) J. J. Sahlin and N. A. Peppas, Ind. Eng. Chem. Res., 30, 211 (1991).

(9.) J. M. Whitney and C. E. Browning, in Advanced Composite Materials-Environmental Effects, ASTM STP 658, p. 43, J. R Vinson, ed., American Society for Testing and Materials, Philadelphia (1978).

(10.) H. T. Hahn and R. Y. Kim, in Advanced Composite Materials--Environmental Effects, ASTM STP 658, p. 98, J. R. Vinson. ed., American Society for Testing and Materials. Philadelphia (1978).

(11.) C. Cataldis, R. Tessieri. A. Apicella, and C. Carfagna, in Interrrelations Between Processing Structure and Properties of Polymeric Materials. p. 605, J. C. Seferis and P. S. Theocaris, eds., Elsevier Science Publishers B. V., Amsterdam (1984).

(12.) Y. Weitsman, J. Mech. Phys. Solids, 35(1), 73 (1987).

(13.) C. D. Shirrell. Composite Materials: Testing and Design (Fourth Conference), ASTM STP 617, p. 514, American Society for Testing and Materials (1977).

(14.) G. Wei and H.-J. Sue, J. Appl. Polym. Sci, 74. 2539 (1999).

(15.) H. G. Cater and K. G. J. Kibler, J. Composite Materials, 12, 118(1978).

(16.) G. Z. Xiao and M. E. R. Shanahan, Polymer. 39(14), 3253 (1998).

(17.) P. Moy and F. E. Karasz, Polym. Eng. Sci., 20, 315 (1980).

(18.) L. L. Marsh. R. Lasky, D. P. Seraphim, and G. S. Springer. In Environmental Effects on Composite Materials, Vol. 3, p. 51, G. S. Springer, ed., Technomic Publishing Co., Westport. Conn. (1988).

(19.) Z. D. Xiang and F. R. Jones. Proc. of the International Conference on Composite Materials ICCM-9, Vol. 5, 601, Spain (1993).

(20.) J. Zhou and J. P. Lucas, Polymer, 40, 5505 11998).
COPYRIGHT 2002 Society of Plastics Engineers, Inc.
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 2002 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Author:Li, Yanmei; Miranda, John; Sue, Hung-Jue
Publication:Polymer Engineering and Science
Date:Feb 1, 2002
Previous Article:A step towards understanding the heating phase of laser transmission welding in polymers.
Next Article:Investigation of appropriate cooling times for electrofusion jointing of PE pipes for gas distribution.

Related Articles
Effect of moisture absorption on the tensile properties of short glass fiber reinforced poly(butylene terephthalate).
Water sorption and diffusion in starch/polyolefin blends.
Newspaper reinforced plastic composite laminates: mechanical and water uptake characteristics.
Tick, tick, tick.
Plexiglas Resins Illuminate Lighting Applications.
Moisture diffusion through vinyl ester nanocomposites made with montmorillonite clay.
Methodology and application of simulated wind-driven rain infiltration in building envelope experimental testing.
Japan Vilene to market fuel cell components.
Evaluating stucco-clad building wall systems through hygrothermal modeling.

Terms of use | Copyright © 2017 Farlex, Inc. | Feedback | For webmasters