EFFECT OF THERMO-OXIDATION ON PERMEATION RESISTANCE OF BROMOBUTYL COMPOUNDS
ABSTRACT
The impact of thermo-oxidation on the permeation process in a series of BIIR or bromobutyl compounds has been investigated. Methyl salicylate (MS) was used as the simulating chemical agent, and the permeation rate was measured gravimetrically using vapometers. Heat aged samples at 120 °C increased in stiffness with an accompanying decline in ultimate properties and network chain density. A reduction in permeability was found, driven primarily by the reduced solubility of MS in the thermo-oxidized BIIR matrix. The formation of a secondary oxidized chain network has been proposed to explain the decline in MS solubility. The loss of plasticizer during the heat aging process is responsible for the reduction of the MS diffusibility. Diffusion-limited oxidation during thermo-oxidation is also a likely factor in affecting the MS transport process.
INTRODUCTION
IIR or butyl rubber is the material of choice with respect to barrier properties given its chemical resistance and its low permeability to gases and liquids.1 To have exceptional barrier characteristics, the polymer structure must interfere with the diffusion process and it must not be chemically similar to the penetrant molecule. The low permeability of IIR is caused by the steric hindrance of the two pendant methyl groups on every other main chain carbon atom in the long isobutylene sequences. The improvement in permeation resistance in butyl compounds by incorporating layered silicates such as Montmorillonite clay to impede small molecule penetration is a subject of current interest.2,3 With respect to military applications, articles such as respirators, gloves, boots, and overcoats are a standard part of personal protection equipment against the threat of chemical warfare agents (CWAs) and are manufactured from IIR or from its halogen-activated derivatives based on bromine or chlorine (i.e., bromobutyl or chlorobutyl).
The blistering agent bis(2-chloroethyl) sulfide or sulfur mustard (SM) is perhaps the most well-known CWA; consequently, it is often used as a benchmark for permeation investigations. Because of its high toxicity and obvious risk in handling, investigations into the use of nontoxic chemical-simulating agents have been undertaken.4–10 Methyl salicylate (MS) has been reported as a suitable chemical simulant for SM for operational training under most testing conditions.4 Many of its physical and chemical characteristics are similar to that of SM, including molecular weight and volume, melting and boiling points, flash point, vapor density, and pressure in case of vapor hazards as well as density, surface tension, spread factors, viscosity, and diffusion coefficient for liquid contact.
Permeability data determined through use of an absorption–desorption time cure and gravimetric techniques have been calculated in liquid contact with the rubber for a series of ester-based plasticizing agents including MS.11 Transport behavior was interpreted by the size, shape, and chemical nature of the penetrant molecules. The most important commercial elastomers were studied, except IIR was omitted. Bakken12 measured equilibrium solubilities of MS in chlorobutyl rubber at 318 and 333 K for an array of MS vapor pressures. The derived Flory interaction parameters, calculated through the enthalpy of mixing, were independent of the MS concentration level. Extensive permeability data for IIRs upon exposure to MS are still lacking in the literature.
As reviewed by van Amerongen13, rubber oxidation has been reported to have a decreasing effect on diffusivity and permeability. Oxidation may cause an increase in the number of cross-links and attach polar oxygen-containing groups to the rubber molecules, both of which cause a decrease in diffusibility. In particular, polar group addition to rubber molecules can create high intermolecular forces between polymer chains, resulting in an increase in the activation for diffusion and lower permeability and diffusion. Polar group insertion has the effect of raising the glass transition temperature. The formation of a hard skin on the material due to oxidation effects also enhances diffusion resistance. As reviewed by Gillen et al.,14 diffusion limited oxidation (DLO) conditions hinder the migration of oxygen into the membrane cross section, causing lower degradation rates at higher aging temperatures. We have been investigating the nature of IIR thermo-oxidation to better predict both shelf and service lives of butyl-based articles in use by the military.15–17 Its thermo-oxidative behavior has been broken down into two competing reactions: one reaction affecting the unsaturated regions at low temperatures and a second set of reactions occurring primarily beyond 70 °C in the isobutylene sequences. Thermo-oxidation is known to be detrimental to physical properties and affects cross-linking level. This study explored the effect of thermo-oxidation on the IIR barrier properties upon exposure to MS. BIIR or bromobutyl is used as the permeation barrier and the effect of removing plasticizers and fillers upon permeability data is also explored. Permeation rate data were collected by using vapometers.
THEORY
Besides the influence of pressure, temperature, and penetrant concentration, the transport or diffusive properties depend on the natures of the polymer membrane and the penetrant molecule. For rubber membranes, this depends on free volume and segmental mobility of the polymer chains that are affected by the presence of bulky and/or polar groups, inter- and intra-chain interactions, and the chain molecular weight (i.e., number of free chain ends). Cross-linking level affects transport properties, and crystallinity has an effect if the polymer can crystallize. In filled rubbers, the presence of carbon black reinforcement and inert fillers affects diffusive properties as does the addition of plasticizing agents. Besides its concentration, the shape and size of the penetrant molecule are important factors influencing permeation.
Hole theory is commonly used to explain the diffusion of the penetrant molecule through the barrier rubber.13,18,19 The micro-Brownian motion of the rubber chain segments in combination with thermal energy fluctuations causes the formation of a wide range of holes of differing volumes that are continuously created and destroyed according to a specific size distribution. Now the diffusion of the penetrant molecule progressing from hole to hole through the rubber takes place through Brownian motion, but the rate of its diffusion depends wholly on the concentration of holes that are large enough to accept them. Hole volume in the activated state is a combination of the average free volume in the normal state plus an increase above this average due to activation. The concentration of holes decreases exponentially with their size, resulting in the reduction of the diffusion rate for large diffusing molecules.
It will be assumed that the diffusion process obeys both Fick's first and second laws and that the system has reached a steady state of permeation, signifying that the diffusion flow is constant and that the diffusion coefficient (D) is independent of the concentration of the penetrant. The transport process involves sorption, diffusion, and evaporation and is driven by the penetrant concentration difference on either side of the membrane. The concentration of the penetrant is assumed to change linearly over the thickness (h) of the membrane, which is isotropic in its properties.
Henry's law is applied in the case of gas or vapor diffusion that relates the gas concentration in the rubber in equilibrium with the gas pressure outside the rubber. The ratio of these two quantities is known as the solubility (S). The permeation rate is measured by using the expression of (q/t)·(h/A), where q is the volume of gas at standard temperature and pressure (STP, 298.15K, 1 bar) diffusing through a membrane of area A during a time t. The units of the permeation rate are square meters per second, but they often are expressed as square centimeters per second. The permeability (Q) is proportional to the permeation rate, but inversely proportional to the pressure difference ΔP [P1 (high partial pressure) – P2 (low partial pressure)] across the membrane as per Eq. 1:
Upon exposure to liquid vapor, ΔP can be replaced by the vapor pressure Po that refers to the high pressure side of the membrane. Q is a characteristic quality depending on the type of rubber and gas used as well as temperature. It can be expressed as the product of D·S, where S is described as the volume of vapor in the membrane at STP divided by the volume of rubber in the wetted membrane corrected for 1 bar of gas pressure. A simple method of measuring the value of S is to measure the weight gain of the wetted membrane during steady-state conditions. Ideal gas behavior of the penetrant can be assumed for cases of low swelling in the rubber membrane (i.e., incompatible rubber–gas system and low vapor pressures). Solubility may be increased or decreased based on the polarity of the gas. The solubility of a rubber gas encompasses rubber/gas solubility, gas adsorption onto filler, and gas entrapment into voids around the particles. The independent measurements of both Q and S now allow for the calculation of D, and D and S obey Fick's and Henry's law, respectively. The diffusion coefficient, also called the diffusivity, describes the nature of the transport process and of the molecular interactions involved. Strong polymer–penetrant interactions cause D to depend on the penetrant concentration.
Arrhenius-type behavior can also be applied to the permeation process because temperature influences both the diffusibility and solubility of liquids and gases in elastomers as illustrated by Eqs. 2 and 3. The temperature dependence of the diffusion process is determined by the increase in free volume with temperature due to thermal volume expansion. ED represents the activation energy needed to produce an opening between polymer chains that is large enough to permit penetration of the diffusing molecule. The factor Do relates to the density and size of the free volume. A stronger temperature dependence of D results in a greater activation energy and slower expected diffusion. The value ΔHS is the heat of sorption comprising of both Henry's and Langmuir sorption modes involving endothermic and exothermic processes, respectively. In most polymer–solvent systems, the overall process is endothermic, giving rise to positive values of ΔHS:
Because Q = D·S, it follows that the permeability may also be expressed in terms of Arrhenius behavior via Eq. 4:
where EQ = ED + ΔHS.
By applying a simplified solution to Fick's first law of diffusion by considering the unidirectional diffusion of the low-molecular-weight penetrant molecules through a thin rubber membrane, the estimation of the diffusant retardation time (τ) may be calculated by way of Eq. 5:
This equation is often used to estimate a value for the diffusion coefficient by using so-called time-lag measurements.
EXPERIMENTAL
The formulations of three BIIR rubber compounds (Table I) were mixed according to ASTM Standard D 3182 by using a Plasti-Corder internal mixer (420 cm3 chamber volume; Brabender, Duisburg, Germany) equipped with cam blades. The reference gas mask formulation was taken from ref 20. Compound refinement took place on a Farrell two roll laboratory mill. A hydraulic compression press (Wabash MPI, Wabash, IN, USA) set at 160 °C was held at a pressure of 90 short tons for tensile slab preparation by using a 10 min cure time.
Hardness measurements of the rubber tensile sheets were carried out using a digital Shimana type A tester following ASTM D 2240. A universal tensile testing machine (Tech Pro, Cuyahoga Falls, OH) possessing a 500 N load cell was used to measure the mechanical properties of the rubber membranes before and after testing at ambient temperature. Testing was carried out following ASTM D 412 guidelines and a MET D4482 die was used to cut out the test samples.
Oxidative heat aging for 72 and 240 h was carried out according to ASTM D 573 by placing the samples in an air circulating oven (Binder, Bohemia, NY) set at 120 °C.
MS or 2-hydroxy-benzoic acid methyl ester (152.15 g/mol) was obtained from Fisher Scientific, Nepean ON. Its vapor pressure as a function of temperature is known.21
Vapometers (Thwing-Albert Instrument Company, West Berlin, NJ, USA) (ϕ = 63.5 mm, depth = 50.8 mm) were used to follow the weight change (Explorer balance, OHaus Corp, Parsippany, NJ) due to vapor transmission through the rubber membrane according to ASTMs E 96 and D 814. Permeation rate data were collected in triplicate at 60 °C, and final results were averaged. Membrane weights were recorded before and at the end of testing to calculate solubility.
The network chain density (n) was calculated through using the equilibrium solvent swelling technique and eq. 6. For butyl-based rubbers, cyclohexane was the preferred solvent and a value of 0.44 was used for the Flory–Huggins interaction parameter (χ).22 A phantom network model derived through the presence of cross-link point fluctuations was chosen and a chain functionality of 4 was assumed. Vs is the molar volume for cyclohexane and νr is the volume fraction of rubber.
The apparent volume fraction of rubber νrf was corrected depending on the filler volume fraction ϕ in the compound.23 Equations 7 and 8 were used considering the volume fraction of the inactive and active fillers respectively.
Here c is a constant depending on the type of reinforcing filler and its extent of interaction with the rubber chains. These interactions lead to a reduction of the swelling capacity of the rubber. An increase in the value of c signifies increased interaction and adhesion of the polymer chains at the filler surface.
RESULTS AND DISCUSSION
Initial and heat aged properties
The initial unaged properties of the reference starting mask formulation and its two variants are summarized in Table II. Its hardness and mechanical property characteristics are typical of general respirator facepiece rubber requirements.24 The removal of 5 parts per hundred of rubber (phr) of paraffinic plasticizer in compound 2 brings about an increase in both hardness and 50% modulus with an accompanying minimal effect on both tensile strength and elongational characteristics. Polymer chain flexibility has been reduced, leading to an increased overall stiffness. Removal of the carbon black and talc fillers in formulation 3 causes substantial drops in specific gravity, hardness, modulus, and tensile strength. The reduction of mechanical properties due to the removal of reinforcing particulate fillers is well known. Strong chain–filler interactions increase the stiffness of the network, contributing to mechanical property enhancement. The elastomer-rich sulfur cross-linked network provides for a highly elastic compound characterized by a high ultimate elongation. The density of the network chains using the phantom model calculation for the unfilled compound 3 was 1.02 × 10−4 mol/cm3. Values of the chain–carbon black interaction parameter c for compounds 1 and 2 were 2.01 and 2.16, respectively. Plasticization is observed to slightly lessen the extent of the favorable adhesion between the rubber chains and the carbon black surface. It will be assumed that interaction parameter c remains constant during the heat aging process for the network chain calculations.
The effect of heat aging on the hardness, mechanical properties, and network chain density on the three compounds is shown in Figure 1. Assuming a simple linear Arrhenius reaction rate for the oxidative reactions taking place, the 3 and 10 day aging at 120 °C corresponds approximately to 6.3 and 21 yr of shelf life at 25 °C. Given the high temperature and short duration of the heat aging, DLO effects can not be discounted. The property changes in all three compounds show similar trends in that there is stiffening taking place with a loss of ultimate property performance. The percentage of change in network chain density decreases with aging time. The increase in hardness and modulus is partly due to the loss of plasticizer in compound 1. Furthermore, the compound stiffening can be explained by the introduction of oxidized functional groups into the degraded BIIR matrix. Oxidized chains are able to interact through hydrogen bonding and dipole–dipole interactions, thereby providing a stiffness increase to the macroscopic properties. Simultaneously, a loss in network chain density is occurring due to the scission of polymer chains in both the unsaturated and saturated regions. The chain radicals react with oxygen to form an array of end products including end chain methyl ketones, carboxylic acids, aldehydes, and alcohols.15,17 In summary, all three compounds exhibited significant property changes upon heat aging due to the oxidation process, thus providing ideal candidate materials to test for their effect on the permeability characteristics.
Citation: Rubber Chemistry and Technology 90, 1; 10.5254/rct.16.83750
Temperature dependence of permeation
Compound 1 was used to test the MS permeation rate behavior as a function of temperature (Figure 2). Linear steady-state permeation conditions were attained after ∼20, 30, and 70 days at 60, 40, and 23 °C. Data collection was stopped after 60 days for 60 and 40 °C, whereas ∼100 days was required to obtain sufficient permeation data for linear extrapolation at 23 °C. The higher permeation rate as the temperature is raised can be attributed to the increased kinetic energy of MS as well as to the thermal expansion of the compound that brings about an increase in polymer free volume due to the facilitation of the thermal motion of the chain segments. Upon raising the temperature, the rate of the formation of holes large enough to allow passage of MS through the membrane increases.
Citation: Rubber Chemistry and Technology 90, 1; 10.5254/rct.16.83750
The permeation rate data collected from Figure 2 and the membrane weight change data (solubility) were used to deduce the diffusion coefficients at the three temperatures. The data were plotted according to the Arrhenius equations reported above (Figure 3). A diffusion coefficient value of 5.5 × 10−9 cm2/s was calculated from the 23 °C permeation data. This value is at the high end of the accepted range for SM permeation into filled butyl membranes.5,6 Polymer types and formulation differences may strongly influence measured diffusion results. Linear Arrhenius dependence with excellent data fit is achievable and can be used to extract the activation energies and heat of absorption values for BIIR. Calculated values of Ep and ED were 72.7 ± 3 and 67.3 ± 5 kJ/mol, respectively. An ED value of 90.5 kJ/mol has been reported for SM by using cast films of filled IIR rubber in a similar temperature range and breakthrough time testing.8 The BIIR diffusion activation energy is greater than elastomers containing polar functionalities such as NBR, CR, and polyurethane that possess favorable intra- and inter-chain interactions (dipole–dipole, hydrogen bonding) that enable lower penetrant permeabilities.11 The highly symmetric structure of the long isobutylene sequences along the polymer backbone permits an optimal macromolecular packing density, resulting in a higher diffusion resistance compared to the polar elastomers.
Citation: Rubber Chemistry and Technology 90, 1; 10.5254/rct.16.83750
A heat of absorption (ΔHS) of 5.4 ± 2 kJ/mol was calculated through the solubility of MS in the BIIR membrane. This value is good agreement with the enthalpy of mixing of 4.7 ± 3 kJ/mol for the chlorobutyl/MS system.12 MS solubility in BIIR was found to increase slightly with temperature. Positive endothermic values of ΔHS generally indicate little or no specific interactions are taking place between the polar MS and nonpolar butyl backbone. It is expected that the large majority of these interactions are weak due to van der Waal's forces, but some stronger interactions are expected at the main chain unsaturation. As in the case of other nonpolar elastomers such as EPDM and NR, this suggests a Henry's mode of sorption that requires the formation of a hole and the immediate dissolution of MS into that site. The heat of solution through temperature dependence of the solubility is largely dependent of the adsorbing properties of the filler.
Permeation parameters: unaged and heat aged biir
Figure 4 summarizes the averaged permeation rate for the unaged and aged samples at 60 °C. Steady-state permeation conditions were achieved from ∼20 days until the end of data collection. With respect to the unaged samples, it is observed that compound 3 possesses the steepest slope and hence fastest permeation rate, followed by compounds 1 and then 2. The removal of 5 phr of plasticizer had a significant effect of decreasing the rate of permeation (compound 1 compared to compound 2). Upon removal of both the carbon black and talc fillers (compound 3), the permeation rate increases and is in fact faster than that of the starting formulation. The reasoning behind the effect of both plasticizing agents and filler particles on penetrant diffusion is straightforward. Plasticizers enhance chain segment mobility and reduce cohesive forces between polymer chains, which aids in the migration of small diffusing molecules across the membrane's cross section. Fillers act as impermeable geometric obstructions in the path of the gas migrating through the rubber, causing a longer tortuous path and a decrease in the permeation rate. After heat aging of the three compounds, the steady-state permeation rate is lower in all cases. The magnitude of permeation rate loss seems larger for thermo-oxidized compounds 1 and 3. From the permeation rate data together with membrane weight change, the permeability (Q), solubility, diffusion coefficient, and steady-state retardation time were derived to better comprehend the exact nature of the permeability process. A summary of these data is illustrated in Table III.
Citation: Rubber Chemistry and Technology 90, 1; 10.5254/rct.16.83750
The unaged permeability Q values follow the order of the permeation rate as explained for Figure 4. Compound 2, which does not contain any plasticizer, provides the best permeability resistance. This is noted in the D value, which mirrors the same trend as the Q values. Plasticizer addition (compound 1) or filler removal (compound 3) causes the permeability to approximately double. The solubility of MS does not change upon plasticizer removal, whereas removing the fillers causes its value to increase by ∼17%. The elastomer-rich formulation 3 allows more MS absorption into the membrane due to the lack of the additional chain–filler interactions taking place in the other compounds. The steady-state retardation times depend essentially on the D values because the membrane swelling was minimal (<5%). The longest retardation time was found in the presence of the fillers and without the plasticizer (compound 2).
As noted in Figure 4, the heat aging of the BIIR compounds causes the magnitude of the permeability to decrease. In all cases, the solubility parameters are also seen to decrease upon heat aging and thermo-oxidation. The diffusibility behavior of the penetrant represented by the D value, however, only decreases for formulations 1 and 3 and remains unaffected for formulation 2. In this case, the permeability decrease is driven by the lower MS solubility in the thermo-oxidized BIIR membrane.
The increase in stiffness of the matrix as seen by the hardness and 50% modulus increases due to thermo-oxidation can be explained by the introduction of oxidative end groups consisting primarily of carboxylic acids, methyl ketone, aldehydes, alcohols, and β-lactones due to the oxidative degradation of the isobutylene25,26 and unsaturated regions of BIIR.17 These polar end groups are able to create high intermolecular forces that can limit permeability.11 In the context of degraded BIIR, it is hypothesized that strong forces come about through dipole–dipole and hydrogen bonding interactions, thereby forming a “secondary” chain network. In the case of equilibrium solvent swell measurements for network density measurements, the nonpolar cyclohexane molecule is able to diffuse and swell the thermo-oxidized BIIR due to the largely nonpolar nature of the matrix. Upon swelling, it is surmised that the oxidized chain network is destroyed, as a network density loss was measured. In the case of MS absorption into the thermo-oxidized matrix, it is assumed that some favorable interactions may occur between the polar ester group and the oxidized BIIR but that the strong dipole–dipole and hydrogen bonding are able to resist and hold together the secondary oxidized chain network. This may be a possible explanation for the reduced solubility of MS in the thermo-oxidized BIIR compounds.
Plasticizer loss is occurring during the heat aging process at 120 °C as seen by the increased stiffening effect for compound 1 (Figure 1a). This can account for the observed decrease of the D value during heat aging, as seen in compound 1. The presence of a thermo-oxidized network does not seem to be affecting the diffusive ability of MS, given the constant results of formulation 2. The reduction of the diffusion coefficient in the unfilled and unplasticized formulation 3 may be related to the influence of DLO, but it is not clear due to the constant D seen for MS in formulation 2. In another investigation, we have observed the onset of DLO in BIIR compounds for aging temperatures >85 °C.27 The extent of DLO effects in the 120 °C heat aged samples cannot be underestimated because the transport process may have been affected by the changing thermo-oxidized morphology from the surface through to the bulk phase, making the cross-sectional membrane anisotropic in properties. In addition to DLO effects, the choice of base polymer and formulation differences can lead to significant changes in the permeation resistance.
Finally, the maximum error that was estimated for the permeability data of Table III was ∼10%. The sources of error in conducting permeation measurements by using the vapometer method are not negligible. Leakage and edge effects along with membrane sagging can influence permeation data collection. The estimation of the linear steady-state region for the permeation rate data is not without error; it seemed that the permeation rate increases with a slight exponential curvature for longer times. The so-called transient behavior has not been addressed, so actual breakthrough times are likely shorter. Compound 3 was the most difficult to mix and mold due to the absence of fillers and therefore presented the largest variation in test permeation results, given sample inhomogeneity and thickness variation effects.
CONCLUSIONS
The diffusion coefficient of the MS/BIIR system was found to be comparable to values reported for SM. The energies of activation of the permeation process were higher compared to polar elastomers. MS solubility in BIIR was found to increase slightly with temperature following a Henry's mode of sorption. Filler and plasticizer concentration directly influence the permeation rate. The thermo-oxidation of BIIR compounds was found to enhance permeation properties. The heat aged samples possessed an increased stiffness with a reduction in both ultimate properties and network chain density. The reduction in permeation was driven by the lower solubility of MS in the thermo-oxidized versus the unaged matrix. It is hypothesized that the creation of a secondary oxidized chain network created through dipole–dipole and hydrogen bonding may explain the decline in MS solubility. The presence of the thermo-oxidized network does not seem to affect MS diffusion. The anisotropic cross-sectional property distribution due to DLO effects in the thermo-oxidized samples may also influence MS diffusion. In addition, the loss of plasticizer during the heat aging process decreases the chain segment mobility and subsequently was found to decrease the extent of MS diffusibility.
Percentage of change in the property data of the hardness, mechanical properties, and network chain density for compounds (a) 1, (b) 2, and (c) 3 upon heat aging for 72 and 240 h at 120 °C. The y axis scale is the same for all three figures to allow for comparison of magnitude changes between compounds.
Permeation rate data collected for compound 1. The heavy black lines are shown as an indication of the steady-state permeation region where linear extrapolation was undertaken.
Plot showing the Arrhenius behavior of Q, S, and D as a function of inverse temperature for compound 1. Linear fits of the data points and goodness of fit values are provided.
Graph depicting the averaged permeation rate data for the three BIIR compounds upon exposure to MS at 60 °C for 60 days. The dark lines represent unaged samples; the light line, 72 h/120 °C; and the dashed line, 240 h/120 °C.
Contributor Notes