Vacuum injection is a manufacturing technique suitable for large load-carrying composite and sandwich structures, such as marine vessels, cooling trailers, etc. As these structures are becoming more optimized, in order to reduce weight and material cost, the requirements on mechanical properties are increasing in terms of performance and consistency. One aspect, which may have a strong influence on mechanical properties, is void content. During injection, voids may form due to local variations in reinforcement permeability at the resin flow front, leakage, boiling of volatile components in the resin or gas dissolved in the resin coming out of solution. After injection, voids can also be formed as a result of shrinkage of the resin.

This study focuses on the formation of voids due to gaseous components, which are dissolved in the resin, and which can come out of solution during the vacuum injection process. The conventional approach of dealing with dissolved gasses in resins is degassing the resin prior to injection simply by applying vacuum. In this paper, more effective degassing mechanism are proposed and tested. The results from the experiments show a much lower void content in laminates made with resin degassed by the new degassing methods than made with resin degassed with the conventional degassing method.

The quality of fiber composites is to a large extent governed by the process parameters and materials used in the manufacturing. One of the more important quality aspects is the void content in the finished part. This is due to the detrimental effect which voids have on many properties, such as mechanical properties, dielectric properties, surface finish, etc (reference 3 and 7). In other words, it is beneficial to keep the void content to a minimum.

Two aspects of voids in composites have attracted most attention, namely, void formation and influence of voids on mechanical properties. Concerning void formation, the main causes are generally due to variations in permeability on a filament and filament bundle scale, outgassing of dissolved gas in the resin, evaporation of volatile components in the resin, shrinkage of the resin and leakage in connections and mold.

One way of reducing the void content is to use degassed resin. There are basically two advantages of using degassed resin:

Ad 1. The gas concentration at equilibrium generally increases linearly with respect to the absolute pressure (Henry's law), where the gas concentration at absolute vacuum is zero. Thus, for manufacturing methods where the mold is evacuated (vacuum injection, Resin Transfer Molding (RTM), etc.), it is likely that the resin will become over-saturated with gas when exposed to a lower absolute pressure. For an epoxy resin, Wood and Bader (reference 9) measured the equilibrium gas concentration of nitrogen at atmospheric pressure to approx. 1.7 % by volume. These 1.7 % of dissolved gas may not seem like a considerable amount of gas, but at a vacuum pressure of 20 mbar, which is commonly used in vacuum injection, this would expand 50 times if brought out of solution.

Ad 2. If bubbles are formed during flow due to ill impregnation, and the resin is not yet saturated, the gas from the bubble can dissolve in the resin. The amount of gas, that can dissolve, depends on the saturation level of the resin, the diffusion speed of the gas molecules in solution in relation to the geltime of the resin, the absolute pressure and of the characteristics of the gas-resin system.

The standard procedure to degas resin is simply to expose the resin to partial vacuum. The idea is to make use of the fact that the gas solubility decreases as the pressure is reduced (Henry's law), where at absolute vacuum the gas solubility is zero. So, if the pressure is decreased, at a certain moment, the resin will become over-saturated and gas should come out of solution. But the dissolved gas is dispersed as molecules and not as bubbles. Therefore, gas will only come out of solution if bubbles or bubble nuclei are already present in the resin. What in fact will happen when the pressure is reduced, is that the bubbles, which have been whipped in during mixing of the resin, will increase in size (Gas Law). With increasing size, the rising speed of the bubbles also increases (Archimedes' Law). This will result in a foaming resin, suggesting that the resin is being degassed. In fact, the resin is mainly "de-bubbled"! Of course, some of the dissolved gas will indeed diffuse into these rising bubbles which resulted from mixing or pouring the resin in a different container thereby entrapping air in scratches or imperfections in the container. If no bubbles or bubble nucleation sites have been added, the standard degassing procedure will not cause any outgassing at all.

So, standard degassing is highly questionable if performed by simply reducing the pressure. If no bubble nucleation sites or bubbles are present there will not be any outgassing. Lundström (reference 8) degassed resin by using high vacuum in combination with stirring, which reduced the amount of dissolved gas in the (polyester, vinylester) resin. This may be explained by the stirring causing evaporation of the styrene, which in turn brought gas out of solution (cavitation).

The use of degassed resin would increase the solubility of entrapped gas in the fiber reinforcement. To limit the scope of this investigation, only polyester resin and its components will be considered. The gases considered are nitrogen and oxygen. Nitrogen because it is involved in the production of polyester resin but also because it is the main component (78 %) in air. The second largest component in air is oxygen (21 %), for which reason this is also considered. In the present paper a theoretical estimation of the amount of gas which may dissolve in polyester resin is presented. Two methods to degas resin are presented and investigated.

A liquid consists of molecules which are held together by attractive van der Waals forces which are due to electromagnetic interaction between molecules, where the molecules are said to be dipoles. The degree of polarity is directly related to the magnitude of the attractive forces. For a solution to occur, the solvent molecules must overcome the intermolecular attractive forces of the solute in order to dissolve. This is best accomplished when the attraction between the molecules in the solute and the solvent is similar.

To estimate the amount of gas, which may be dissolved in liquids and polymer solutions the Hildebrand solubility theory (reference 5) is generally used. For single component liquids, the molar fraction of gas dissolved in the liquid is given by (reference 4):

(1)

where is the average partial molar volume of the gas in a range of liquids, and are the Hildebrand solubility parameters for the liquid and gas, R the gas constant (8.13441 J K-1 mol-1) and T the temperature. The "ideal" molar fraction of gas , the molar fraction of a gas, which does not interact energetically with the liquid at temperature T, is defined as

where is the enthalpy of vaporization of the gas at its normal boiling point Tb. For multi component liquids is replaced in Equation (1) by the so called effective Hildebrand parameter defined by

where is the volume fraction of component i in the liquid and . If the molar volume of the gas differs considerably from that of the liquid, the molar fraction of gas predicted by Equation (1) will deviate from the actual molar fraction. To compensate for these deviations a Flory-Huggins type correction term may be added to Equation (1), resulting in (reference 2 and 6)

(2)

where is the average molar volume of the multi component liquid defined by

(3)

where is the molar fraction and the molar volume of component i in the liquid. Equation (2) is not only applicable to multi component liquids with molar volumes considerably different from that of the gas, but also to polymer solutions which have the same characteristics.

Estimation of Gas Solubility

The amount of gas, which may dissolve in polyester resin, may now be estimated by utilizing Equation (2). The estimations will be limited to a model resin consisting of only one polyester polymer (diethylene glycol isophthalate) mixed with styrene. The parameters for gas solubility calculations, with Flory-Huggins correction, are compiled in Table 1. The material properties for the polyester resin and its components are compiled in Table 2.1, where the properties for the resin are calculated using Equation (1) and (3).

Table 2.1: Parameters for gas solubility calculations, with Flory-Huggins correction, at 25°C (reference 6)

 

Table 2.2: Hildebrand parameters for monomer and polymer components in polyester resins at 25°C (reference 1)

The molar fraction of gas, which may dissolve in styrene, polyester and polyester resin is calculated by applying Equation (2) and solving for , using the material properties in Table 2.1 and 2.2. The results are compiled in Table 2.3. The molar fractions of air are calculated by assuming the air to consist of only nitrogen and oxygen, in proportions of 78 % and 21 %, respectively.

Table 2.3: Molar fractions of gas, which may dissolve in styrene, polyester and polyester resin, at atmospheric pressure and 25°C

Transforming the molar fractions of gas, found in Table 2.3, into volume fractions of gas (in gaseous state) gives insight in the potential problem of outgassing, see Table 2.4. The transformation calculation is performed by utilizing that one mole of gas corresponds to 24.1 dm3, at normal atmospheric pressure and temperature.

Table 2.4: Volume fractions of gas, in gaseous state, which may dissolve in styrene, polyester and polyester resin, at normal atmospheric pressure and temperature

It can be seen from table 2.4 that in a normal polyester resin about 2.6 to 3.4 % by volume of air can dissolve. These volumes are calculated for normal atmospheric pressures and temperatures. At reduced pressure, these volumes are of course increased. In table 2.4, only volumes are considered. An estimation of the speed of solubility (dissolving) is not yet executed.

Vacuum injection experiments with different fiber reinforcement materials have shown quite a different void content and void distribution. Research has shown (reference 8) that the main cause of voids with these experiments was outgassing of the resin. Apparently, some reinforcement materials (like Unifiloä) exhibit better bubble nucleation properties than other materials, and therefore will result in laminates with a higher void content.

If the resin would be brought into contact with such a nucleation material at reduced pressure prior to the actual injection, this could lead to a much more effective degassing procedure. This would then lead to a better laminate quality during and after injection.

Consequently, the reinforcement material with the best nucleation properties (and thus the worst laminate quality) is suited best for pre-injection degassing of the resin. Also other materials than reinforcement materials have been tested for nucleation properties, such as pumice stone, chalk, etc. Scotch Briteä came out as the most promising material.

In figure 3.1, a schematic illustration is given, which might explain the working principle of the nucleation process. An over-saturated solution comes into contact with the bubble nucleation material. This material contains bubble nuclei as entrapped air in cavities (figure 3.1a). The difference in gas concentration between the entrapped air and the dissolved gas causes gas molecules to diffuse from the resin into the entrapped air. The entrapped air bubble grows and some of the air forms a bubble. This bubble rises to the surface and a new bubble starts to grow. When the bubble rises through resin, the size of the bubble increases due to two mechanisms:

As long as nucleation sites (microscopic bubbles) remain present in the nucleation material, bubbles will be formed until the resin is no longer over-saturated. This will more likely be the case when the nucleation material exhibits a low surface tension with regard to the dissolved gas (mixture).

Another method would be to completely skip the bubble nucleation stage and add bubbles to the resin at reduced pressure, the so-called sparging. This can be done by the setup sketched in figure 3. 2.

A container is filled with resin. The pressure in this container is reduced to a pressure below the injection pressure to be used during the vacuum injection process. At the bottom, air is fed into this container. The air is forced through a very fine filter, thus creating many small bubbles. These bubbles rise through the resin. At the reduced pressure, the resin will be over-saturated with air (or components of air). The difference in gas concentration between the air bubble and the dissolved gas causes gas molecules to diffuse from the resin into the bubble. This process continues until a new equilibrium situation is reached, e.g. the resin is saturated (but no longer over-saturated) with air.

If degassed resin (either by adding a nucleation material during degassing or by sparging) is used during the vacuum injection process, and the injection pressure is higher than the degassing pressure, there is no risk of outgassing. There will even be the possibility of dissolving some bubbles, which have been formed during the flow of resin, entrapping air in fiber bundles.

To test the effectiveness of a degassing procedure, the amount of gas dissolved in the resin prior and after degassing has to be determined. Since there is no simple, reliable procedure to directly measure the amount of gas dissolved in a resin, an indirect method is adopted by measuring the void content in the resulting laminates.

Four experiments are conducted. From previous experiments, it is known that injections with Unifilo give a high void content, mainly due to outgassing of gaseous components in the resin. Therefore, injections on this material should give a good indication if the degassing procedure is effective. Four procedures are tested.

The laminates consisted all of 3 layers of Unifilo 450gr/m2. The degassing pressure was 10 mbar and the injection pressure was 50 mbar. After injection, first the resin inlet hose was closed, and the outlet pressure remained 50 mbar. After curing, the laminates were tested visually of void content and distribution.

The results are given in table 5.1

With vacuum injection, the laminate quality with regard to void content can be improved considerately by degassing the resin with the use of a bubble nucleation material. Scotch Briteä was one of the most promising materials found so far.

Placing a piece of Scotch Briteä in a container and reducing the pressure works very well for small amounts of resin (up to 5 liters).

The sparging degassing method is even more promising. Since it seems to be a faster method and more easily useable for larger quantities.

It would be desirable to develop a continuous degassing method. Research on this subject is still going on. If possible, the resin manufacturer should supply the resin in a degassed condition. Some gas can then still be whipped in the resin during mixing, but if the mixing of all ingredients is done carefully, this represents only a very small amount of gas. And these bubbles can be removed easily by a normal degassing procedure. Most of these bubbles will rise to the surface anyway, certainly with the low-viscous resin used with vacuum injection. Therefore, it is always a good idea to wait several minutes after the mixing of the resin.

A simple method should be developed to directly measure the amount and type of gas dissolved in a resin. Some research has been done on this subject, but the results have not been verified yet.

The work presented in this paper was supported by the Dutch Institute for Maritime Research (NIM) and the Dutch Ministry of Economical Affairs. The work was carried out in co-operation with Jan Söderlund, Department of Aeronautics, Division of Lightweight Structures, Royal Institute of Technology, Stockholm, Sweden.