On the Modeling and Optimization of UV-Curable Fluoro-Siloxane Formulations and Their Solvent Absorption and Adhesion Properties.
Aerospace coatings can require durability to fuel, solvent, and aerothermal exposure while exhibiting strong adhesion. Fluoropolymers and fluoro-siloxanes exhibit superior durability to swelling, weight gain, and mechanical degradation compared to hydrocarbon polymers. Partial explanation of the resistance to solvent penetration in polymers containing fluorine is provided by the work of Schmidt et al.  and Hillmeyer and Lodge . Schmidt  showed that if there is enough mobility, perfluorinated (-[C.sub.8][F.sub.17]) pendant chains shows a tendency to orient to the surface. The presence of fluorine at the surface promotes a decrease in the penetration of solvents into the polymer (the so-called, surface solubility ). Doeff and Lindner  showed using contact angle measurements that strongly suggest that polysiloxane coatings are oriented on the surface of thin films so that perfluoroalkyl chains are approximately parallel to one another. Similarly, Sun et al.  showed low surface energy perfluoro side chains and migrate to the surface of thin films anchor to gold substrates. Sun et al. observed that these films oriented away from anchoring hydrocarbon chains, yielding a distinctly structured, self-organized polymer monolayer with a fluoroalkyl-enriched surface. O'Rourke Muisener et al.  also used angle dependent X-ray photoelectron spectroscopy (ADXPS) to show the surface migration of polymer blends of fluoro-silane terminated polystyrene. O'Rourke Muisener's work analyzed polystyrenes that were terminated with -Si[(C[H.sub.3]).sup.2]-[(CH).sub.2]-[(C[F.sub.2]).sub.5]--(C[F.sub.3])--(C[F.sub.3]) and also showed migration of fluorine to the surface. They found that the surface concentration of fluorine was proportional to the concentration of the functional polymer. Segregation of the fluorine increased with increasing molecular weight up to 127 kDa and then decreased for high molecular weights. For a review of the surface effects of polymers containing fluorine, see Ref. .
Fluoropolymers, and in particular fluoro-silicones, is becoming widely known for their ability to withstand harsh chemical environmental conditions, increased thermal stability, and increased hydrophobicity, compared to silicones alone. Fluorine and siloxane functionality work synergistically to increase hydrophobicity and reduce organic solvent absorption and dimensional swelling. The addition of chemical functionality such as hydroxyl groups increases the hydrophilic character of the polymer. In general, increased hydrophilic functionality is important for adhesion to primed surfaces, and for post-cure processing of the polymer like painting or bonding, but can produce polymers that also absorb solvents and increase dimensional swelling.
With the cost of doing research and development, scientists and engineers are forced to be more efficient in the way the scientific method is exercised. Efficient cost-effective experimental methods are being adopted to focus research, limit cost, increase reliability, and reduce risk to a research effort especially when attempting to optimizing polymer processing conditions [8-10]. Companies are realizing that early recognition of system requirements and efficient experimentation can significantly reduce costs of development. The use of combinatorial research and development has been shown to significantly reduce time to market for high-risk technologies . Combinatorial research and development facilitates the controlled fabrication of material compositions that normally would not be made, which creates opportunities of serendipity discoveries. Liu et al.  discusse how N1ST is advancing materials processing with the integration of hardware and software-based infrastructure that can rapidly develop libraries of response surfaces. The libraries contain rational chemical synthesis and process information, which help to reduce the number of samples and experiments while reducing risk of false positive/negatives.
This article discusses work on the modeling and optimization of a series UV curable fluoro-silicones polymer coatings. Requirements for the desired polymer coating had strongly opposing requirements regarding solvent absorption and adhesion. An efficient mixture test matrix was developed that sampled the input variables and produced estimated coefficients that model the response surface. The empirically derived coefficients were then used to determine the optimized formulation. This study will show the sensitivity of level of fluorination on several physical properties of a family of UV-curable fluoro-silicone coatings, including adhesion, solvent absorption, stiffness, and thermal oxidative stability. While not explicitly used in this analysis, dynamic mechanic testing is also performed to provide mechanical property context for the individual formulations.
Cationic Curing System
Cycloaliphatic epoxy-terminated siloxane curing resins are attractive materials for use as coating for several reasons. The ring-opening reaction means that cured polymer will have low shrinkage. The cationic curing mechanism results in the formation of ether linkages, which are chemically stable in harsh environments. Cationic curing epoxies also create low amounts of free hydroxyl groups which results in low moisture absorption and swelling in damp and hydrophilic environments. Cationic curing epoxies cure through the release of protons from a source like a photo-initiator, which facilitates continued curing even after UV exposure. The continued curing mechanism should facilitate through thickness curing of thick films.
Currently, there are no sources for commercially available fluorinated epoxy resins. There are commercially available sources for epoxy-terminated siloxanes. Polyset (Mechanicsville, NY) is a supplier of unique low molecular weight cycloaliphatic epoxy siloxanes. PC 1035 has a molecular weight of 648 g/mol and epoxy equivalent weight of 3.9 x [10.sup.-3] g/eq. PC1035 was chosen over a lower molecular weight PC 1000 because of the desire to create an elastomeric UV curable fluorosilicone coating. When cured alone, these materials are extremely brittle and have glass transition temperatures above 150[degrees]C.
To create high elongation elastomeric polymer and take advantage of the hydrophobicity of the siloxane backbone, a mono-functional fluorinated alcohol would be required. Monofunctional alcohols will combine with a difunctional epoxy to create a linear polymer. Alcohol functionality greater than two combined with PC 1035 will crosslink the polymer. PC 1035 was combined with a combination of three alcohols, neo-pentyl glycol (R2490), 4,4'-(hexafluoroisopropylidene) diphenol (HFPD), and 2,2,3,3,4,4,5,5-octafluoropentanol (OFP). OFP has 65.5% by weight fluorine, a molecular weight of 232.1 g/mol and equivalent weight of 4.31 X [10.sup.-3]eq/g. The structures of these compounds are shown in Fig. 1. OFP creates a pendent fluorinated chain on the polymer backbone. The pendent chain will likely improve the hydrophobicity of the polymer but will likely not contribute to enhancing the thermal oxidative durability. HFPD is a difunctional alcohol with 34% wt/wt fluorine, a molecular weight of 336.23 g/mol. HFPD will incorporate fluorine into the backbone of the polymer, so it should increase the thermal oxidative stability of the polymer because it increases the thermal oxidative durability of the polymer backbone. R2490 is a non-fluorinated difunctional alcohol that will increase the flexibility of the polymer and dilute the fluorine concentration for determining the effect solvent absorption, thermal degradation, and adhesion as a function of weight percent fluorine content. R2409 has a molecular weight and equivalent weight of 220 g/mol and 8.73 x [10.sup.-3] eq/g, respectively.
Dynamic mechanical tests were performed on the neat polymers to test for glass-transition temperature and modulus. The goal for all formulations is for all to be rubbers at room temperature. Thermogravimetric analysis (TGA) was performed to establish the thermal decomposition temperature in air.
Because of the requirement that a polyurethane paint must be applied and remain adhered to the cured formulations, the stoichiometry was offset by 10% in favor of the alcohols to promote paint adhesion. It will be shown later that this had a negative consequence in percent weight loss at 232[degrees]C.
Development of Test Matrix Formulations
To efficiently test a range of fluorinations levels a few of the environmental challenges were selected and a statistical design of experiments was formulated based on the equivalent weights. The most challenging tests were the solvent absorption and adhesion for reason discussed earlier. The experimental design method used XVERT  to establish the sampling points in the constrained mixture space.
McClean and Anderson  in 1966 proposed the construction of experimental designs by forming all possible combinations in two level factorial design of upper and lower limits of proportions [b.sub.i] and [a.sub.i] of the (q-1) component in the mixture then computing the level of the qth component so that for each combination such that the proportion of the gth component satisfies Eq. 1.
[summation over (1[less than or equal to]i[less than or equal to]q)] [X.sub.i] = 1 0 [less than or equal to] [a.sub.i] [less than or equal to] [X.sub.i] [less than or equal to] [b.sub.i] [less than or equal to] 1 for 1 [less than or equal to] i [less than or equal to] q (1)
Lebaal et al.  gave methods of constructing designs that are near optimum. Most commercial statistical packages that treat mixtures include XVERT. McClean-Anderson algorithms are useful for mixture designs where there are only single component constraints . The reader is also encouraged to explore other algorithms that are available for creating efficient experimental designs (c.f. Ref. ).
To begin the experimental design, a constraint equation was derived (c.f. Eq. 2) to compute the number of grams of the epoxy PC 1035 given the percent composition of HFPD, OFP, and R2490.
gramsPC1035 = 1.74 x (grams HFPD) + 1.27 X (grams OFP) + 2.83 X (grams R2490) (2)
The other three constituents of the formulation were constrained according to Table 1. Note the original constraints are only for the lower limits on HFPD, OFP, and R2490. The upper limits, as well as both upper and lower for PC1035 are calculated from the design space.
The proportions of the individual components are shown graphically in Fig. 2. As the sum of the four constituents is a constant (100%) a "mixture formulation" approach was used to design the test matrix. Since PC1035 is also a function of the other three constituents (c.f., Eq. 2) then in this design fixing any two of the three "independent" constituents (HFPD, OFP, R2490) necessarily constrains the remaining third constituent.
Using an XVERT algorithm, the three "extreme vertices" of the test matrix were selected from Formulations 1, 2, and 3 in Table 2. Next, three "midpoints of the long edges" for the three independent constituents were selected and these are shown as Formulations 4, 5, and 6 in Table 2. Finally, the "centroid" of test matrix was selected as also shown as Formulation #7 in Table 2. The combination of the extreme vertices, the midpoints of the edge, and the centroid in the test matrix assured the best estimates of the effects for changes in each of the constituent variables, given the number of formulations in the test matrix.
Experimental aspects used in the preparation of samples are discussed in this Experimental section. The first subsection discusses the calibration of the UV conveyor curing system. Preparation of the polymer films subjected to environmental exposure to solvents is discussed in the Preparation of Test Films for Cationic Resin Formulations subsection followed by Preparation of Adhesion Samples section, which discusses the preparation of adhesion samples.
The procedures for measurement of the total solvent absorption are explained in the Neat Polymer Film Solvent Absorption subsection. TGA and DMA analytical testing procedures of exposed polymer films are briefly discussed in the Thermal Gravimetric Analysis and Dynamic Mechanical Analysis subsections, respectively.
Calibration of the UV Curing F300S Conveyor
A UV PowerPuck[R] UV radiometer (EIT Inc. Sterling VA) was used to calibrate the total dose of UV A, UV B, and UV C radiation for a type H UV source. The belt speed was changed and the total radiation exposure in each of the wavelength bands was recorded and averaged. The peak absorption of the photoinitiators used in this project is at 246 [micro]m.
Preparation of Test Films for Cationic Resin Formulations
A difunctional cycloaliphatic epoxy PC1035 was combined with three mono and difunctional alcohols with varying levels of fluorination according to the experimental design matrix shown in Table 1. To each of the cationic formulations, 3 wt% of the photo-initiator PC2506 was added to the mixture. Once completely dissolved and in a single phase, the formulations were dispensed on prepared 15.2 cm x 15.2 cm x 0.32 cm borosilicate glass plates. Reservoirs 0.51 mm high were created by stacking five strips of Teflon tape.
Once screened to fill the reservoirs, sample films were cured with a nominal UV dose of 10 J/[cm.sup.2]. Polymer films retrieved from the cast plates were divided into solvent absorption specimens, TGA, and dynamic mechanical test specimens.
Preparation of Adhesion Samples
Adhesion to aluminum was a stated requirement for this coating. Seven 5.2 cm x 5.2 cm x 0.64 cm samples were prepared by first grit blasting at a nominal pressure of 0.48 MPa and a standoff distance of 4-8 in. and a 45[degrees] angle to the surface of the coupons. Grit blasted samples were then treated with ATK's patented  sol-gel in situ surface treatment. Adhesion assessment for coating adhesions is typically qualitative (e.g., American Society for Testing and Materials D4145 cross-hatch adhesion test). For this reason and due to project funding constraints, our assessment was also qualitative. If there was any visual indication of poor adhesion, the formulation was given a 0. If there was not apparent debonding, then the formulation was assigned a 1. Once cured, all adhesion samples were placed in environmental aging only in water as specified in Table 3. In addition, once the formulations were cured, the surface of the polymer was required to accept a Polyurethane paint top coat and remain adhered to the surface.
For the cationic curing formulations, an epoxy functionalized organofunctional silane was applied to the surface after cleaning with the sol-gel rinse. Preparation of the adhesion samples first required that a dam around the perimeter of the sample be created with five layers of Teflon tape. This created a reservoir that resulted in a 0.51 mm layer of the formulation. Once dispensed, the coupons were cured with a nominal UV dose of 10 J/[cm.sup.2] and the Teflon tape mask was removed.
After curing, a polyurethane paint (Sherwin Williams F92G601) was mixed and applied and cured at 140 [degrees]F for 45 min. The study focused on the adhesion of the polyurethane top coat. Adhesion could be enhanced with the use of a paint primer. A photograph of the painted cationic adhesion samples is shown in Fig. 3. It is clear from Fig. 3 that all the samples, except for Formulation 2 appear to wet the surface of the polymer coating. Formulation 2 had the highest fluorine content. It was also observed that Formulation 2 did not adhere to the aluminum after cure. This suggests that a weight percent fluorine concentration of 24% represents an upper limit for a formulation.
Neat Polymer Film Solvent Absorption
Polymer films retrieved from the cast plates discussed in the Preparation of Test Films for Cationic Resin Formulations section were carefully weighed using an analytical balance three times. The average weight of the neat polymer film was recorded. The specimen dimensions were also recorded using calipers so that the change in density could be computed after solvent exposure. All samples were packaged in aluminum foil to prevent the accidental handling during transportation. The samples were then removed from the aluminum foil and submitted into environmental aging in triplicate according to Table 3.
After the samples were exposed, they were carefully removed from the solvent, blot dried, and again weighed on an analytical balance again three times. The three measurements from each sample were averaged and a percent weight change per formulation was recorded. The dimensions of the sample were also measured.
Thermal Gravimetric Analysis
Polymer films retrieved from the cast plates discussed in the Preparation of Test Films for Cationic Resin Formulations section were analyzed for thermal stability using TGA. All the formulations were subjected to heating in air to assess the effects of the polymer composition on thermal oxidation. The samples were run in triplicate or duplicate, heated at 5[degrees]C/min and the average percent weight loss was recorded at temperature ranges from 25 to 300[degrees]C and 300 to 400[degrees]
Dynamic Mechanical Analysis
Polymer films retrieved from the cast plates discussed in the Preparation of Test Films for Cationic Resin Formulations section were analyzed mechanical stiffness using a dynamic mechanical analyzer RSA n (Rheometrics Corporation, Piscataway NJ). Each formulation was tested for stiffness at 1 Hz from -50 to +150[degrees]C.
This section discusses the analysis used to make predictions of the behavior of the formulations at points other than those measured. First, the experimental results are compiled and fit to a model. In our case, a simple linear model provided an [r.sup.2] > 0.95. The functional form of the model was expressed in Eq. 3.
property = [3.summation over (i=1)] [B.sub.i] x [component.sub.i] (3)
where [B.sub.i] is an empirical coefficient associated with the response in each solvent system.
i ranges from 0 to 3 for the cationic system.
The average percent weight gain for each of the solvent systems evaluated are shown in Table 4 through Table 7.
%Wieght change = [B.sub.0] x %PC1035 + [B.sub.1] x %HFPD + [B.sub.2] x %OPF + [B.sub.3] x %R2490 (4)
It is interesting to note that the solvent system that consistently induced the largest increase in percent weight gain was JP8. JP-8 is a kerosene-based fuel. In addition to its use on aircraft, it is used in ground-based diesel engines and power generators . JP8 is oily to touch and contains icing, corrosion inhibitors, and lubricants. Because of its low molecular weight hydrocarbon composition, JP8 tends to be readily absorbed into hydrocarbon polymers. Solvents used in this test matrix represent a good sampling of polarities for which the polymers must survive. It is also noteworthy to correlate the formulations that had the highest and lowest weight percent fluorination with the absorption of these solvents. From Tables 4-7, it is observed that Formulation 3 has the lowest percent fluorine at 3.5 wt% and Formulation 2 has the highest at 24.3 wt%. In each case where an organic-based solvent was used, Formulation 2 resulted in the lowest percent weight gain, and Formulation 3 resulted in the highest percent weight gain. Water soaking resulted in a net decrease in weight on the order that is consistent with the 10-wt% offset (in excess) used in the initial formulation to help enhance adhesion.
Release of excess alcohol into the water bath is thought to be responsible for the reduction in percent weight gain for all formulations soaked in water.
Conclusions can be drawn on the effect of coating composition by examining the response surfaces and focusing on the vertices. Response surfaces are shown in Fig. 4a-d. As there are no cross-term interactions in the model, these surfaces reflect the model's predictions only on the proportions of the individual components. As expected, the hydrocarbon diol R2490 had a strong positive effect on the rate of solvent absorption. HFPD and OFP, both fluorinated alcohols, had a negative influence on solvent uptake for varying degrees. In the case of JP8, and Hydraulic fluid, HFPD (See Fig. 4c/Table 6 and Fig. 4/Table 4, respectively), solvent uptake was promoted by HFPD presumably due to the presence of the aromaticity of the phenol groups. The highly fluorinated OFP has a strong negative effect on weight change for all the solvents except for JP8.
As expected, the hydrocarbon diol R2490 has a strong positive effect on the rate of solvent absorption. Model calibration was done based on the responses of the formulations to solvent exposure under the conditions indicated in Table 3. For solvent absorption, calibration was performed using a least-squares fit to the selected sampling of the percent concentration (PC1035, HFPD, OFP and R2490) versus percent weight gain responses in each solvent. With correlation coefficients >0.98 demonstrates a high predictive capability of the model for the property of percent weight gain (c.f. Eq. 4).
Calibration of the model resulted in the coefficients ([B.sub.i]) shown in Table 8. If these coefficients are used in Eq. 3 and a concentration of HFPD, OFP, R2490, and PC1035 (computed from Eq. 3) is input, a predicted response will result. With this analysis, it is now possible to interpolate between the percent composition values to determine the best formulation that yields a percent weight gain closest to zero. A measure of the ability of the model to accurately interpolate between the fit formulation points is the root mean square error (RMSE) for the predictive equations. The RMSE for the different analyses are tabulated in Table 9. For any formulation, the predicted value [+ or -] 2x RMSE) represents the 95% confidence interval. The response called percent weight change will provide a measure of the polar environment within the polymer matrix.
If the constituents of the formulation create a polarity like that of the solvent, then solvent will migrate into the polymer. Weight gain (change), change in density, and swelling will reflect this event.
Even though no paint primer was applied to the surface of the cured polymer coating. Formulation 3 (c.f. Fig. 5a) showed somewhat positive results for adhesion to both aluminum and paint. Poor adhesion was obvious as shown in Fig. 5b and highly correlated with lack of R2490. As summarized in Table 10, adhesion to aluminum was good for all the formulations except formulations that did not contain R2490.
Thermal Gravimetric Analysis
All the formulations were subjected to heating in air to assess the effects of the polymer composition on thermal oxidation. Results of this study are shown in Table 11. As discussed in the Solvent Absorption section, the ability of the linear model to predict results for TGA at points other than the calibration points is related to the RMSE shown in Table 12. The predictive coefficients for weight change at the three temperature zones are shown in Table 12. The effect of composition on percent weight loss at different temperature ranges is not as clear as in the solvent soak analysis. At high temperatures, mean percent weight loss is highly correlated with the presence of fluorine content. High OFP content resulted in the lowest mean percent weight loss, which suggest that OFP is a stabilizing constituent. At low temperatures, the weight loss is believed to be due to the release of excess alcohol that was specified in the original formulation plan. It is believed that most the weight loss at low temperatures is due to the excess alcohol that was designed into the formulation to enhance adhesion. Weight loss at high temperature and is highly correlated with fluorine content. High OFP content resulted in lower average weight loss while high R2490 increased weight loss.
These results suggest that OFP was a positive stabilizing ingredient. Again, effects of changing PC1035 are confounded in the simultaneous changes of the other three variables. In contrast, high R2490 content resulted in the highest mean percent weight loss observations. The response models show that the increasing the concentration of OFP has a slight effect on reducing the change in percent weight loss.
In all formulations, most of the decomposition was complete in the range of 300-400[degrees]C temperature range. This may be enhanced by increasing the cross-link density, possibly by using a lower-molecular weight epoxy terminated siloxane. The current cationic system was formulated to promote paint adhesion by offsetting the stoichiometry by 10% in favor of alcohols. The offset was done because at the time of running this matrix, it was not known that a paint primer existed for the Sherwin William top-coat.
Offsetting to stoichiometry had the unfortunate effect of causing a 10-wt% loss at 232[degrees]C. Paint adhesion should not be a problem with the use of the primer so in Phase II we would propose adjusting the stoichiometry back to equal equivalents which should have the effect of reducing the percent weight loss at 232[degrees]C.
Dynamic Mechanical Analysis
The coatings of interest were intended to serve as a carrier of fillers to the surface of aluminum. Therefore, the coatings were required to bond to aluminum substrates even in the event of large changes in temperature during the service life.
Because of the large coefficient of thermal expansion of aluminum and the potential stiffening effects of integrated fillers, rubbery formulations were sought. This section discusses the thermo-mechanical properties of the surveyed polymer formulations.
Each formulation was tested for stiffness at 1 Hz from -50 to + 150[degrees]C. A summary of the response's glassy modulus, rubbery modulus, and glass-transition temperature is shown in Table 13. The glassy modulus of Formulation 1 is approximately 900 MPa while the Plateau Modulus is approximately 9 MPa. The glass-transition temperature is approximately 30[degrees]C. At room temperature, these formulations are rubbery. The presence of a nearly flat plateau modulus is indicative of a highly cross-linked polymer . The stiffness of these formulations would be expected to increase significantly the addition of fillers for use on aircraft or automotive applications. Therefore, rubbery coating formulations were sought.
This section discusses the analysis used to make predictions of the behavior of the formulations at points other than those measured. Prediction of the responses based on the formulations was performed using a least-squares fit to the selected sampling of the percent concentration (PC1035, HFPD, OFP and R2490) versus percent weight gain responses. While not sampled, based on the response model an optimized formulation for reduced solvent uptake of JP8 would have percent fluorine higher than 24 wt%. A complete optimized formulation considering solvent absorption and adhesion however means that fluorine content must be compromised with R2490 content. The data suggested that at least 6 wt% R2490 was required to promote the adhesion to aluminum. In general, as discussed in the Results section, the fit of the coefficients to the model calibration had [r.sup.2] > 0.95. The usefulness of modeling these data is in the ability to predict the response at points other than the calibration points so that an optimum formulation can be discovered.
An optimized formulation for percent weight change with reduced solvent uptake in the solvents Hydraulic Fluid and Lubricant would have similar compositions with approximately 20 wt% fluorine (c.f. Table 14). A suggested optimized formulation was derived from the experimental design in Table 15. The response model suggested that a formulation optimized for low percent weight change would be like formulation test # 5. Given the need for R2490 to be present for good adhesion to aluminum, it suggests that an optimized formulation for adhesion and solvent absorption would contain approximately 18 wt% fluorine and contain approximately 6 wt% R2490 as indicated in Table 14.
Weight percent fluorine of 13% represents the average of all formulations. As the optimum is higher than the average could suggest that there may be preferential orientation of the fluorine functionality toward the surface of the coating.
CONCLUSIONS FOR CATIONIC FORMULATION
We have attempted to develop a quantitative assessment of the effects of percent fluorination on solvent absorption and adhesion using a cationic curing epoxy siloxane resin PC1035. We screened several formulations in a carefully designed set of experiments using PC1035 and three alcohols with varying amounts of fluorine content; an aliphatic diol neo-pentyl glycol (R2490), a difunctional alcohol HFPD, and a primary alcohol 2,2,3,3,4,4,5,5-octafluoropentanol (OFP).
The goal was to formulate a set of rubbery polymers that would test how percent fluorination affects solvent absorption and adhesion. The solvents chosen were based on military standards for polymer coatings and consisted of hydraulic fluid, lubricant, JP8 jet fuel, and water.
Because our modeling effort is empirically based, the results of the model were restricted to the limits of the formulation trade space. We found that there was a clear trend of decreasing solvent absorption with increasing fluorination for all solvent systems. From weight gain measurements, an empirical linear fit model was performed against formulation number which produced coefficients with high degree of correlation that mathematically describes the behavior of polymer fluorination across the seven-dimensional trade-space. Because the project was limited in funding, the experimental design was limited to a linear model. Nevertheless, there were significant findings from this simple study which we believe will provoke additional studies in this area.
Using the empirical fit, optimized formulations targeting zero weight gain were computed for each solvent system. These are shown in Table 14. For all but JP8, zero weight gain is achievable. Trending suggests that much higher levels of fluorination are required for zero weight gain for JP8. It is clear from Table 15 that the percent fluorination has a significant factor in reducing solvent absorption in the polymer formulations. Additionally, we observed that for JP8 optimized weight gain requires the use of the highest level of fluorination possible. The highest percent fluorination sampled in this study was 24 wt% (formulation test #2).
An optimized formulation for percent weight change with reduced solvent uptake in the solvents hydraulic fluid and lubricant would have similar compositions with approximately 20 wt% fluorine.
The response model showed a clear relationship to reduce solvent-induced weight gain with increasing fluorine concentration. The change was not as strong for specimens with hydraulic fluid and lubricant soaking. An unexpected result occurred with the water soak where all the formulations lost weight. The source for the weight loss is unclear but appears to be associated with OFP concentration and the offsetting of the stoichiometry by 10 % in favor of alcohols to enhance adhesion of the topcoat paint. This relationship is only seen in the water-soaked samples. It is recommended that the formulations be adjusted to equi-equivalents for epoxy to alcohol to improve the thermal mass loss at 232[degrees]C.
Weight gain variability was an unexpected result and warrants further investigation. There was no obvious correlation between percent fluorination of alcohol concentration and weight gain variability. A possible correlation exists of weight gain variability especially with hydraulic fluid and lubricant fluid with glass transitions temperature which is related to cross-link density. We would encourage the use of a larger experimental design which can resolve the effect of time and temperature in the solvent systems to better understand the sources of variability for the weight gain experiments.
We also examined the effect of weight loss and decomposition with percent fluorination. Increasing the level of OFP appears to only slightly stabilize the formulation from thermal oxidation. Nearly all of the formulations decomposed at about the same temperature range of 300-400[degrees]C.
Finally, we sought to optimize the effects of fluorine content to minimize solvent absorption while maintaining good adhesion to primed aluminum. The data suggested that there was a requirement for R2490 content to be at least 6 % to promote adhesion to aluminum. A suggested optimized formulation is shown in Table 15. It shows that the best compromise between adhesion and solvent absorption occurs where the polymer possesses about 17% fluorine and 6% R2490.
The authors would like to thank ATK advanced technologies for the funding to perform this work. Also Mr. Dan Gill and David Lefgren for their patient and careful work in preparing the samples and taking measurements. The authors would like to also acknowledge Mr. Bill Kappele for his helpful discussions on experimental designs.
Russell Allan Crook (iD), Andrew Allen
ATK Space Systems, Launch Systems Group, P.O. Box 707, Brigham City, Utah 84302
Correspondence to: R. A. Crook; e-mail: email@example.com
Published online in Wiley Online Library (wileyonlinelibrary.com).
[1.] D.L. Schmidt, B.M. DeKoven, C.E. Coburn, G.E. Potter, G. F. Meyers, and D.A. Fischer, Langmuir, 12, 518 (1996).
[2.] M.A. Hillmeyer and T.P. Lodge, J. Polym. Sci. A, 40, 1 (2002).
[3.] R.E. Johnson and R.H. Dettre, in Wettability, J.C. Berg, Ed., Marcel Dekker, Inc, New York, NY (1963). Chapter 1.
[4.] M.M. Doeff and E. Lindner, Macromolecules, 22, 2951 (1989).
[5.] F. Sun, D.G. Castner, G. Mao, W. Wang, P. McKeown, and D. W. Grainger, JACS, 118, 1856 (1996).
[6.] P.A.V. O'Rourke Muisener, C.A. Jalbert, Y. Caigen, J. Baetzold, R. Mason, D. Wong, Y.J. Kim, J.T. Koberstein, and B. Gunesin, Macromolecules, 36, 2956 (2003).
[7.] T. Imae, Curr. Opin. Colloid Interface Sci., 8, 307 (2003).
[8.] Y. Huang, Y. He, Y. Wang, Q. Huang, and C. Xin, Polym. Eng. Sci., 58, 1166 (2018). https://doi.org/10.1002/pen.24678.
[9.] S. Liu, J. Wu, J. Chang, and S. Hung, Polym. Eng. Sci., 40, 1256 (2000). https://doi.org/10.1002/pen.11253.
[10.] N. Lebaal, S. Puissant, F. Schmidt, and D. Schlafli, Polym. Eng. Sci., 52, 2675 (2012). https://doi.org/10.1002/pen.23203.
[11.] J.D. Hewes and L.A. Bendersky, High Throughput Materials Research and Development. Growing Effort at N1ST, Appl. Surf Sci. Vol. 189, 196-204 (2000).
[12.] R.D. Snee and D.W. Marquardt, Technometrics, 16, 391 (1974).
[13.] R.A. McLean and V.L. Anderson, Technometrics, 8, 455 (1966).
[14.] W.F. Smith, Experimental Design for Formulation, Society for Industrial and Applied Mathematics Publishing, ASA-SIAM Series on Statistics and Applied Probability, SIAM, Philadelphia, ASA, Alexanddria, VA, p 66 (2005).
[15.] R.H. Hardin, and N.J.A. Sloane, (1994) "Notes on "Gosset", A General-Purpose Program for Experimental Design: Features and Applications" in Proceedings of the International Conference, on Statistics in Industry, Science and Technology, Tokyo, pp. 346-350.
[16.] R.A. Crook, L.W. Poulter, K.J. Schulte, and J.W. Sinclair, U.S. Patent 5,520,768 1996.
[17.] MIL-DTL-83133E technical specifications (includes JP-8)
[18.] J.D. Ferry, Viscoelastic Properties of Polymers. 3rd ed., John Wiley and Sons, New York, NY, 351 (1980).
Caption: FIG. 1. The cationic curing system being evaluated.
Caption: FIG. 2. Illustrates the selection of proportions for the XVERT design.
Caption: FIG. 3. A photograph of all the cationic aluminum adhesion samples with a top coat paint applied. Note the highest fluorinated coating (Formulation 2) exhibited poor adhesion prior to environmental challenge.
Caption: FIG. 4. (a) The effect of increasing polymer composition on the mean %weight change after soaking in hydraulic fluid for 7 days at 66[degrees]C. All formulations gain weight except the Formulation #2 which had the highest fluorine content. Increase weight was highly correlated with the presence of high hydrocarbons. The effects of changing PC1035 are confounded in the simultaneous changes of the other three variables, (b) The effect of increasing polymer composition on the mean %weight change after soaking in lubricant for 7 days at 121[degrees]C. All formulations gain weight except the Formulation #2 which had the highest fluorine content. Increase weight was highly correlated with the presence of high hydrocarbons. The effects of changing PC1035 are confounded in the simultaneous changes of the other three variables. (c)The effect of increasing polymer composition on the mean %weight change after soaking in JP8 for 7 days at 25 [+ or -] 5[degrees]C. All formulations gain weight Increase weight was highly correlated with the presence of high hydrocarbons. The effects of changing PC1035 are confounded in the simultaneous changes of the other three variables. High fluorine content reduced weight gain, (d) The effect of increasing polymer composition on the mean %weight change after soaking in water for 7 days at 49[degrees]C. All polymer formulations lost weight under these conditions. The effects of changing PC1035 are confounded in the simultaneous changes of the other three variables. Formulations with low OFP lost the least amount of weight.
Caption: FIG. 5. (a) In general, paint adhesion was poor because of the lack of paint primer. Coating formulations with R2490 showed good results for adhesion to aluminum after soaking in water at 49[degrees]C for 7 days. Formulation 3 had the best paint adhesion, with some paint blistering, (b) All formulations with R2490 showed good adhesion to prepared aluminum substrates. Better paint adhesion is expected with the use of the paint's primer.
TABLE 1. The fractional range of mixture components. Low High HFPD 0.10 0.37 OFP 0.00 0.32 R2490 0.00 0.19 PC1035 0.58 0.71 TABLE 2. The composition of the formulations used in this study with percent fluorination. Formulation Total PC1035 HFPD OFP R2490 % Fluorination 1 100.0 63.5 36.5 0.0 0.0 12.4 2 100.0 58.0 10.0 32.0 0.0 24.3 3 100.0 71.0 10.0 0.0 19.0 3.4 4 100.0 64.5 10.0 16.0 9.5 13.8 5 100.0 67.3 23.2 0.0 9.5 7.9 6 100.0 60.8 23.3 16.0 0.0 18.4 7 100.0 64.0 19.0 11.0 6.0 13.6 TABLE 3. Solvents, soak time, and conditions used to evaluate solvent compatibility of the neat formulations. Solvent Soak time (days) Soak temperature ([degrees]C) JP 8 jet fuel 7 Ambient Hydraulic fluid 7 66 Lubricant 7 121 TABLE 4. Average weight gain for soaking samples in hydraulic fluid at 66[degrees]C for 7 days. Hydrualic % weight SD Neat formulation Density fluid summary change Tg [degrees]C (gm/cc) Formulation 1 11.99 0.225 30.4 0.98 Formulation 2 -2.30 1.125 -4.5 0.94 Formulation 3 29.22 17.88 -18.8 0.98 Formulation 4 12.66 17.516 -17.9 0.95 Formulation 5 16.50 0.095 8.0 0.97 Formulation 6 5.58 9.836 0.4 0.94 Formulation 7 7.55 9.677 -1.9 0.95 Hydrualic % fluorination PC1035 HFPD OFP R2490 fluid summary Formulation 1 12.4 63.50 36.50 0.00 0.00 Formulation 2 24.3 58.00 10.00 32.00 0.00 Formulation 3 3.4 71.00 10.00 0.00 19.00 Formulation 4 13.8 64.50 10.00 16.00 9.50 Formulation 5 7.9 67.30 23.20 0.00 9.50 Formulation 6 18.4 60.80 23.30 16.00 0.00 Formulation 7 13.6 64.00 19.00 11.00 6.00 TABLE 5. Average weight gain for soaking samples in lubricant at 121[degrees]C for 7 days. Lubricant % weight SD Neat formulation Density fluid summary change Tg [degrees]C (gm/cc) Formulation 1 13.35 7.807 30.4 0.94 Formulation 2 2.20 1.409 -4.5 1.02 Formulation 3 60.77 33.604 -18.8 1.05 Formulation 4 20.81 34.653 -17.9 1.02 Formulation 5 21.65 0.0512 8.0 0.95 Formulation 6 12.73 23.151 0.4 0.99 Formulation 7 11.18 23.112 -1.9 1.01 Lubricant % fluorination PC1035 HFPD OFP R2490 fluid summary Formulation 1 12.4 63.50 36.50 0.00 0.00 Formulation 2 24.3 58.00 10.00 32.00 0.00 Formulation 3 3.4 71.00 10.00 0.00 19.00 Formulation 4 13.8 64.50 10.00 16.00 9.50 Formulation 5 7.9 67.30 23.20 0.00 9.50 Formulation 6 18.4 60.80 23.30 16.00 0.00 Formulation 7 13.6 64.00 19.00 11.00 6.00 TABLE 6. Average weight gain with soaking samples in JP8 jet fuel at ambient temperature for 7 days. JP8 fluid % weight SD Neat formulation Density summary change Tg [degrees]C (gm/cc) Formulation 1 38.43 4.51 30.4 0.92 Formulation 2 15.63 10.911 -4.5 0.91 Formulation 3 49.38 0.589 -18.8 0.93 Formulation 4 35.62 0.093 -17.9 0.95 Formulation 5 36.88 0.759 8.0 0.90 Formulation 6 29.87 1.103 0.4 0.86 Formulation 7 29.87 0.428 -1.9 0.91 JP8 fluid % fiuorination PC1035 HFPD OFP R2490 summary Formulation 1 12.4 63.50 36.50 0.00 0.00 Formulation 2 24.3 58.00 10.00 32.00 0.00 Formulation 3 3.4 71.00 10.00 0.00 19.00 Formulation 4 13.8 64.50 10.00 16.00 9.50 Formulation 5 7.9 67.30 23.20 0.00 9.50 Formulation 6 18.4 60.80 23.30 16.00 0.00 Formulation 7 13.6 64.00 19.00 11.00 6.00 TABLE 7. Average weight gain for samples in water at 49[degrees]C for 7 days. Water fluid % weight SD Neat formulation Density summary change Tg [degrees]C (gm/cc) Formulation 1 -0.95 0.685 30.4 1.03 Formulation 2 -14.99 12.201 -4.5 0.98 Formulation 3 -1.48 0.080 -18.8 0.96 Formulation 4 -7.76 0.200 -17.9 1.07 Formulation 5 -0.86 0.066 8.0 1.08 Formulation 6 -10.74 0.404 0.4 1.06 Formulation 7 -5.94 0.099 -1.9 1.04 Water fluid % fluorination PC1035 HFPD OFP R2490 summary Formulation 1 12.4 63.50 36.50 0.00 0.00 Formulation 2 24.3 58.00 10.00 32.00 0.00 Formulation 3 3.4 71.00 10.00 0.00 19.00 Formulation 4 13.8 64.50 10.00 16.00 9.50 Formulation 5 7.9 67.30 23.20 0.00 9.50 Formulation 6 18.4 60.80 23.30 16.00 0.00 Formulation 7 13.6 64.00 19.00 11.00 6.00 TABLE 8. The value of the coefficients for the predictive models of solvent absorption. Coefficients JP8 soak Water Lubricant Hydraulic fluid B(0) -58.832 9.106 -118.967 -35.218 B(1) 103.404 -15.897 207.358 61.591 B(2) 74.832 -12.021 150.777 44.486 Bo) 168.074 -25.733 338.434 100.71 Correlation 0.993 0.987 0.976 0.994 coefficient TABLE 9. The root mean square error of the numerical fits to the soak data. Solvent system RMSE JP8 1.66 Water 1.26 Hydraulic fluid 1.58 Lubricant 5.81 TABLE 10. Shows the results for visual inspection of adhesion to aluminum. 1 = good; 0 = poor. Adhesion to Formulation Total PC1035 HFPD OFP R2490 Aluminum 1 100.0 63.5 36.5 0.0 0.0 0.0 2 100.0 58.0 10.0 32.0 0.0 0.0 3 100.0 71.0 10.0 0.0 19.0 1.00 4 100.0 64.5 10.0 16.0 9.5 1.00 5 100.0 67.3 23.2 0.0 9.5 1.00 6 100.1 60.8 23.3 16.0 0.0 0.0 7 100.0 64.0 19.0 11.0 6.0 1.00 TABLE 11. The results of the TGA as average % weight loss for the cationic formulations. Most of the decomposition occurs in the 300-400[degrees]C temperature range Formulation 1 Formulation 2 % weight loss 25-300 [degrees]C 19.75 23.54 % weight loss 300-400 [degrees]C 45.03 40.82 % weight loss 400-600 [degrees]C 5.07 2.28 Formulation 3 Formulation 4 % weight loss 25-300 [degrees]C 29.17 26.94 % weight loss 300-400 [degrees]C 43.42 41.36 % weight loss 400-600 [degrees]C 2.47 298 Formulation 5 Formulation 6 % weight loss 25-300 [degrees]C 25.94 23.96 % weight loss 300-400 [degrees]C 46.05 40.59 % weight loss 400-600 [degrees]C 3.30 3.52 Formulation 7 % weight loss 25-300 [degrees]C 24.38 % weight loss 300-400 [degrees]C 40.17 % weight loss 400-600 [degrees]C 3.76 TABLE 12. The RMSE for the TGA models. Temperature range RMSE ([degrees]C) 25-300 1.06 300-400 1.85 400-600 0.41 TABLE 13. The variation in the glassy and plateau moduli and the glass-transition temperatures for each formulation. Formulation # 1 2 3 4 Glassy modulus (MPa) 1,000 20 1,100 1,100 Plateau modulus (MPa) 10 7 1.1 1.4 Glass transition 30.43 -4.5 -18.8 -17.9 temperature ([degrees]C) Formulation # 5 6 7 Glassy modulus (MPa) 110 1,100 1,100 Plateau modulus (MPa) 1.4 1.6 1.5 Glass transition 8 0.4 -1.9 temperature ([degrees]C) TABLE 14. The optimized formulations such that percent weight change is targeted at zero. Optimized formulation Soaking % fiuorination PC1035 HFPD OFP R2490 solution JP8 24.4 58.0 10.0 32.0 0.0 Hydraulic fluid 20.5 60.0 16.1 23.0 1.0 Lubricant 20.5 60.0 16.1 23.0 1.0 Water 5.3 69.5 15.5 0.0 15.0 Predicted % weight gain Soaking Actual Predicted solution JP8 15.6 16.4 Hydraulic fluid 0.5 Lubricant 0.2 Water 0.0 TABLE 15. The optimized formulation for low solvent absorption and adhesion to aluminum. Optimized formulation Soaking % PC1035 HFPD OFP R2490 solution fluorination JP8 24.4 58.0 10.0 32.0 0.0 Hydraulic fluid 20.5 60.0 16.1 23.0 1.0 Lubricant 20.5 60.0 16.1 23.0 1.0 Water 5.3 69.5 15.5 0.0 15.0 Optimized 17.8 62.1 10.0 22.0 5.9 Predicted % Predicted weight weight gain change for optimized formulation for weight Soaking Actual Predicted change and adhesion to solution Aluminum JP8 15.6 16.4 22.3 Hydraulic fluid 0.5 4.0 Lubricant 0.3 7.1 Water -0.3 -10.3 Optimized
|Printer friendly Cite/link Email Feedback|
|Author:||Crook, Russell Allan; Allen, Andrew|
|Publication:||Polymer Engineering and Science|
|Article Type:||Case study|
|Date:||Aug 1, 2019|
|Previous Article:||Relationships Between Crystalline Structure and the Thermal Behavior of Polyethylene 2,5-furandicarboxylate): An In Situ Simultaneous SAXS-WAXS Study.|
|Next Article:||Processability Study of In-Mold Coating for Sheet Molding Compound Compression Molded Parts.|