by Jeff King, Phillip Irwin, Ike Latham, and Stan Moore, The Dow Chemical Company
Eight years ago, the use of CFC-11 in rigid foam applications was terminated in the U.S. due to environmental concerns arising from its ozone depletion potential. Since that time, rigid foam manufacturers have used the second-generation blowing agents with more and more confidence. In general, the predominant blowing agent for the U.S. market over this period has been HCFC-141b.
Now, however, government regulation requires that manufacturing of HCFC-141b be discontinued in the U.S., beginning Jan. 1, 2003. These regulations require the phasing out of the use of HCFC-141b in rigid foam applications. Many appliance producers have chosen to use HFC-245fa as a blowing agent replacement for HCFC-141b. One reason for this selection is due to the low cabinet energy consumption that is possible with HFC-254fa.
Click to Enlarge
In light of this blowing agent switch and as in the case of the change from CFC-11, experience will be needed to develop and gain comfort with the use of this new blowing agent. It is reasonable to expect differences between HCFC-141b and HFC-245fa. Whereas one is more liquid like, the other behaves more like a gas at typical processing temperatures and pressures. The molecular weights and sizes of the two molecules also differ. A new understanding will be needed to determine how these differences will affect the foam processing and foam properties.
This paper will help lay the foundation for a higher level of understanding on how the use of HFC-245fa will affect the foam properties. In addition, formulation development using HFC-245fa has been extensively investigated, and insights into what components are needed to make acceptable foams have been identified. From this work a database has been generated that allows quick identification of formulations to meet design requirements.
Work described in this paper was performed at the Polyurethanes R&D facility of The Dow Chemical Company in Freeport, TX, U.S. Foams were processed using high-pressure injection foam machines with throughputs of 25-30 lb/min. The foam samples used for the physical property testing were generated from a Brett (Bosch) mold (200 cm x 20 cm x 5 cm) preheated to 50°C.
Foam samples were tested perpendicular to rise direction and in accordance with ASTM 1621. A test specimen for the thermal conductivity was cut from the core of the foam at a pre-set position of the Brett foam sample. The thermal conductivity was measured using a LaserComp Fox 200 at a 75°F mean temperature. The heat flow was perpendicular to the direction of foam flow.
Click to Enlarge
The remaining pieces of the Brett foam sample were put into a cold room at -30°C for an additional 24 hr. After this period, the samples were removed and visually inspected for any dimensional changes that occurred under these conditions. A stringent rating system was used to differentiate samples having small differences in freeze stability. Each Brett foam sample was given a qualitative freeze rating between 1 and 4 (the higher, the better). A rating of 1 was given to foam showing unacceptable dimensional stability. The ratings from 2 to 4 were used to differentiate small dimensional stability differences in samples that most likely would provide acceptable freeze performance, with a 4 rating showing no visible shrinkage.
Testing was done during the foam processing to gain a better understanding of the foaming reaction in HFC-245fa systems. A free-rise foam was produced to measure the string or gel time. In addition, the temperature profile of the free-rise foam was measured over a 45-min period for a selected number of systems. Two parameters were obtained from this time-temperature data: the maximum temperature obtained by the foam mass and the initial rate of temperature rise, dT/dt. This rate was determined by a least-square line fit of the temperature data between 10-25 sec after the foam shot.
Results and Discussion
Optimization of Polyol Blends
An initial experimental screen of eight different polyol blends used in various formulations was made to determine how the various components in the polyol blend would affect the properties of the final foam. These polyol blends were selected to examine a broad range of polyol compositional elements and, in addition, to establish a baseline for the effect of these variations on various foam properties.
Click to Enlarge
Figure 1 shows the average thermal conductivity achieved by the use of each of these polyol blends (designated P01-P08) with the 95-percent confident interval of this average. These averages represent the overall effects of the polyol blend in a variety of foam formulations and not the lowest possible thermal conductivity that is achievable for each blend. Although the lower possible thermal conductivity is not represented here, these averages are extremely useful in establishing the effects of the individual components of the blends on the final properties of the foam. The average thermal conductivities of these systems ranged from 0.133-0.143 btu•in/hr•ft2•°F.
Figure 2 illustrates the average post-demold expansions measured on a 10-percent overpacked Brett with a 3-min demold time using the same polyol blends. Low post-demold expansion is important since it relates directly to productivity for the appliance producer. The post-demold expansion behavior will determine how fast a cabinet can be removed from the molding fixture. It is interesting to observe that there is a bias for higher post-demold expansion at lower thermal conductivity. The bias between post-demold expansion and thermal conductivity for these systems is illustrated in Figure 3. It is doubtful that higher post-demold expansion results in lower thermal conductivity or vise versa, but the two properties may have a common underlying origin.
Another important property for foam systems is the ability to flow, which can be characterized by the minimum fill density. To compare systems with small differences in the water and HFC-245fa levels, the minimum fill density can be adjusted to accommodate the differences in total gas level for the final foam. Figure 4 shows the average of the adjusted minimum fill densities for the initial eight polyol blends. The adjusted minimum fill densities vary over a 12-percent range from 1.81-2.04 pcf. These differences will be reflected in the ability of the foam to flow and fill a refrigerator cabinet.
Table 1 lists the relative behaviors, in comparison to the grand average, of these polyol blends for the previously mentioned foam properties along with their relative freeze stability. The freeze stability is important since it is related to the minimum foam density required in a refrigerator cabinet to produce foam that is dimensionally stable for the lifetime of the appliance. It can be seen from Table 1 that no polyol blend in this first round of experimentation had good overall performance. However, the motivation for this first round of experimentation was not to find the optimum polyol blend, but to understand how the nature of the polyol blend affects the foam properties.
Using the information acquired with the first round of polyols blends, additional polyol blends were evaluated. Thermal conductivity, demold expansion, and flow behavior were significantly improved compared to the initial eight polyols during this second portion of the study. More than 300 foam options were ultimately evaluated.
Relationships between Foam Properties and Polyol Blends
The relationship between average thermal conductivity versus average gel time for the polyol blends was studied in this work. The well-known relationship between faster gel time and lower thermal conductivity holds true for the polyol blends studied here. Faster gel times correlate with a more rapid increase in the viscosity of the reacting polymer system. Higher polymer viscosities will stabilize the foam structure and lead to smaller cell sizes. Thermal conductivity is known to decrease with decreasing cell sizes .
Click to Enlarge
Supporting the premise of a more rapid increase in the viscosity, a good correlation between the gel time and initial temperature rise of the foaming mass was found. A higher initial rate of temperature rise would indicate that the chemical reactions are proceeding at a quicker rate and resulting in a faster growth in polymer molecular weight. Since viscosity is a strong function of molecular weight , the viscosities in systems with higher initial rate of temperature rise would be higher. In light of the above correlations, it is not surprising that the average thermal conductivity is related to the initial temperature rise.
Another relationship that was observed in this study is the correlation between post-demold expansion and the relative length of a polymer network chain. The length of a network chain can be simply defined as the length of a chain between crosslink points. Although there is a distribution of lengths in the polymer network, an average length can be calculated from statistical methods . Although these averages do not represent the entire distribution of chain lengths, the relative comparison of these averages is useful to give a measure of how tight or loose the network structure is in the polymeric material.
Figure 5 illustrates the correlation between post-demold expansion and the relative length of a polymer network chain that was found in this study. This relationship can be explained by recognizing how the modulus of a polymer network increases as the lengths of the interconnected network chains decrease . A higher modulus material should deform less under demold conditions and, thus, lead to smaller post-demold expansions.
It has also been well documented that post-demold expansion is a strong function of demold time . As may be anticipated, this effect was also seen in this study. Figure 6 (Page 6) shows what can be expected with the demold test used in this study as a function of demold time. From this graph, it can be seen that the post-demold expansion measured by the process is decreasing at a rate of approximately 0.003 in/sec in the time frame between 2- 3 min. This number could be used to optimize the time a refrigerator cabinet needs to remain in a fixture leading to possible increases in plant productivity.
The freeze stability behavior of the final foam is very important in the determination of the in-place density needed in the refrigerator cabinet to be dimensionally stable for the lifetime of the appliance. An increase in in-place density will normally increase the freeze stability; however, it will also tend to increase the overall cost of the foam in the refrigerator cabinet. It was hypothesized that the average compressive strength of the foams could be a controlling factor in the freeze stability of the foam . Using a nominal logistic regression , an estimation was made of the probability for the foam falling into one of the four freeze rating categories based on average compressive strength. Figure 7 shows the relationship obtained through this analysis for the samples failing freeze testing (those with a freeze rating of 1). From this figure, one can see that about a third of the samples with compressive strength of 16 psi fail freeze testing, whereas about 15 percent fail at 18 psi.
The relationship between samples failing freeze testing and the overall density of the foam was obtained through the same type of nominal logistic regression. It is no surprise that there was a similar relationship as the one with compressive strength. This is true since the overall density and compressive strength are closely related. This analysis showed that almost half of the samples failed at 2.0 pcf; however, that number dropped to less than 20 percent at an in-place density of 2.1 pcf.
Nominal logistic regression analysis was used to show the effect of overpacking level on the probability for a sample to fail freeze testing. Here a similar trend was found as for the compressive strength and overall density. However, the overpacking level was less of a predictor of freeze stability than the overall density or compressive strength.
Foam flow in an actual refrigerator cabinet will be very different from that in Brett molds. This will cause the freeze behavior in a specific cabinet to vary from what has been found in a Brett. However, once a baseline has been established, there should be a very good correlation between Brett and cabinet data.
Selection of a Foam System
Selection of a foam system based on input requirements can be a difficult and time-consuming task. Many times the formulation that was used as a starting point for the development work is very different from the final system that is developed to meet the foaming requirements. However, using the data generated in this study, the process of formulation development can be streamlined. With this formulation database and by knowing the specific input requirements, one can quickly focus in on the system or systems that have the highest potential for meeting those input requirements.
Click to Enlarge
For example, by assuming the arbitrary input requirements listed in Table 2, it can be clearly shown how this process might work. The database has a total of 334 foam samples where the data is available for these four parameters. Table 2 also lists the total number of these 334 samples meeting the four individual requirements, independent of other properties. However, the four input requirements must be met collectively.
A Venn diagram can be used to show what happens when these constraints are imposed together. When looking at the constraint of thermal conductivity alone, there are 160 samples meeting or surpassing the arbitrary target of less than 0.134 btu•in/hr•ft2•°F and 174 samples that do not meet this criteria. When the dual constraints of thermal conductivity and density are imposed, there are 68 systems that meet or exceed both of these two requirements. When the additional constraint of a freeze rating greater than 2 is added, the pool of samples drops to 17, as demonstrated by Venn diagram in Figure 8. When the final constraint of post-demold expansion is added, the number of samples meeting all four requirements shrinks to just two. Thus, the work now is focused on a smaller number of possibilities where the final system optimization can be performed.
Using an initial polyol blend screen, the effects of components in the polyol blends on foam properties have been studied. With this information, additional polyol blends were developed to meet the requirements of foam systems using HFC-245fa. These additional blends show superior performance in areas of thermal conductivity, post-demold expansion, flow behavior, and freeze stability. These systems represent an excellent starting point for further optimization to give a foaming system that will meet the end foaming requirements of a specific end user.
Click to Enlarge
Several relationships between the polymer structure and foam properties have been observed. These relationships will aid in further development and optimization of a system to meet design requirements. The relative length of a polymer network chain is found to affect the post-demold expansion of the foam. The initial rate of temperature rise is also important in determining the thermal conductivity of the foam. In addition, the effect of demold time on post-demold expansion was studied.
Overall density and average compressive strength were found to be good indicators for the freeze stability of the foam system. Using nominal logistic regression, estimations of the freeze stability were made as a function of both the overall density and compressive strength. Finally, a database has been established to quickly find foam systems that will provide the needed properties of a specific set of design criteria. This database will allow a streamline method to focus on the best potential systems for further optimization.
Click Either to Enlarge
About the Authors
Over the last few years, Jeff King has lead Dow's efforts in developing technical solutions for blowing agent replacements; Phillip Irwin is a development chemist in the Polyurethane's Development Laboratory of The Dow Chemical Company (Freeport, TX, U.S.); Ike Latham is currently working on technology development for rigid polyurethane foam applications; and Stan Moore is a senior technical specialist in rigid foam development and is a technical leader for polyurethane construction and appliance foams in North America.
1. Booth, L. D., “Radiation Contribution as an Element of Thermal Conductivity,” Polyurethanes World Congress, Proceedings of the FSK/SPI, p.85, (1987).
2. Fox, T. G.; Allen, V. R.; “Dependence of the zero-shear melt viscosity and the related friction coefficient and critical chain length on measurable characteristics of chain polymers”, J. Chem. Phys. 41(2), p.344, (1964).
3. Flory, P. J., “Principles of Polymer Chemistry”, Chapter 9, Cornell University Press, Ithaca, N. Y., (1953).
4. Flory, P. J., “Principles of Polymer Chemistry”, Chapter 11, Cornell University Press, Ithaca, N. Y., (1953).
5. Clarke, W. D., “Foam Pressure Monitoring as a Method of Studying Demold Characteristics of Appliance Polyurethane Insulation Foam System”, Polyurethane Marketing & Technology Partners in Progress, Proceedings of the SPI 28th Annual Technical/Marketing Conference, p.194, (1984).
6. Birch, A. J., Parenti, V. Van Duin, K. J., Smits, G. F., Clavel, P., “Iso-/n-pentane Blown Rigid Polyurethane Foam for Appliances: A Realistic and Economically Attractive Alternative to Cyclopentane”, Polyurethanes 95, Proceedings of the Polyurethane 1995 Conference, p.448, (1995).
7. Hosmer, D. W., Lemeshow, S., “Applied Logistic Regression, New York, John Wiley and Sons, (1989).
This paper is an edited version of a technical paper presented at Polyurethanes Conference 2002, Oct. 13-16, 2002 in Salt Lake City, UT, U.S.