Chemical and Physical Blowing Agents in Structural Polyurethane Foams: Simulation and Characterization.
In recent years, the production of polyurethane foams has expanded greatly owing to their enormous and increasing employment in several industrial fields, i.e., car manufacture, furniture, refrigeration, etc. This success is due to the intrinsic capability of polyurethane systems to change physical-mechanical properties as soon as reactants or processes are varied.
The present work deals with structural polyurethane foams, polymers with low-density foamed cores and high-density exterior skin to be used mainly as wood substitutes. Owing to their peculiar "sandwich" structure, these foams reveal good mechanical properties, such as flexural strength, tensile modulus, and elongation, which depend primarily on skin thickness and density, while their lightness is related with the foamed inner layer. A proper choice of blowing agents is the main target of present research; new and alternative blowing agents must be discovered in order to replace traditional and well-known CFCs and HCFCs, which have been banned because of their environmental impact. The aim of this paper is to compare the physical-mechanical properties of specimens prepared with different blowing agents, for the purpose of replacing the above-mentioned dangerous compounds. In our experimental tests, n-pentane and methyl formate are used as physical blowing agents, while, for comparison, water is employed as ch emical blowing agent. Several different structural foam specimens are prepared with the three above mentioned blowing agents, with the employment of standard processing equipment for Reaction Injection Molding (RIM). This equipment makes it possible to prevent the capture of air during the mixing process, a typical drawback of manual mixing techniques. A method for obtaining kinetic data will be developed from the experimental temperature values in order to identify which parameters are most effective for foam structure control, Thereafter, a theoretical model based on previous works (1-7) is presented to simulate the temperature and density profile inside the mold. Simulation data will be correlated with experimental results. Finally, previous specimens will be tested to determine the effective mechanical properties that may be compared at the end.
* Glendion RM 601: a polyether polyol with average functionality equal to 4.3, hydroxyl number = 395 mg KOH/g, viscosity = 2700 mPa[cdotp]s at 25[degrees]C (Enichem S.p.A.).
* Tercarol 1: polyether polyol with average molecular weight equal to 1000, hydroxyl number = 168 mg KOH/g, viscosity = 280 mPa*s at 25[degrees]C (Enichem S.p.A.).
* MDI VM 2085: oligomeric isocyanate of 4-4' biphenyl methane diisocyanate of average functionality = 2.85 and %w NCO = 31 (I.C.I.).
* DMCEA: dimethylcycloexylamina catalyst (Abbot).
* Tegostab 8404: silicone surfactant (Goldsmidth AG).
* Blowing Agents:
-- water = a chemical blowing agent; the reaction with MDI leads to the formation of [CO.sub.2]:
-- R-NCO + [H.sub.2]O [right arrow] -R-NHCOOH [right arrow] -R-[NH.sub.2] + [CO.sub.2][up arrow]
-- n-pentane = a physical blowing agent, nowadays employed as a substitute for CFCs and HCFCs (with O.D.P. = 0) in the production of low density rigid foams
-- methyl formate = a physical blowing agent not yet used, although it is less inflammable than n-pentane. It may combine with water to produce formic acid, which reacts with isocyanate itself (8). Because of the scarce humidity of the system, we will omit these reactions.
The physical and chemical properties of these blowing agents are reported in Table 1 (8-10).
Formulations and Foam Preparation
Several different "sandwich" polyurethane specimens were prepared in a pilot plant. This laboratory system comprises two tanks, one for the polyols mixture (i.e. polyols, surfactant, catalyst and blowing agent) and one for the isocyanate. The temperature in the tanks is kept uniform and constant by means of a circulating fluid and by a stirrer which also prevents the formation of crusts. The appropriately dosed reactants are introduced into a high thermal capacity mold through a high speed mixing-head, which assures utmost homogeneity. After preliminary thermostabilization at 28[degrees]C, the mold is filled with the mixture, closed and set under a press.
The foam formulations used in these experiments, where only the blowing agents are changed, are reported in Table 2. The amount of blowing agent is calculated in order to obtain density values in open molds equal to 0.1 g/[cm.sup.3], in accordance with industrial formulations.
After having prepared all the specimens, structural rigid foams are conditioned at 70[degrees]C for 24 hours before cutting them in order to test their physical-mechanical properties.
Adiabatic temperature rise is the method used to obtain kinetic data, i.e. the reaction rate, activation energy [E.sub.a], the frequency factor A, the heat of reaction [delta][H.sub.r], and the order of reaction n.
The temperature variation with time is measured by using copper-constantan thermocouple and it is recorded using a data acquisition system. The thermocouple is held in the middle of the mixture foaming into an open box (base: 25 X 25 cm, height 30 cm). The energy balance per unit polymer weight may be expressed as:
[c.sub.p] * dT/dt = (-[delta][H.sub.r]) dX/dt - U * (T - [T.sub.o]) * S + [h.sub.v] * dL/dt (1)
This equation takes into account the fact that the temperature increase is the result of exothermic heat generated by the chemical urethane bond formation reaction (also urea bonds if water is used), of heat dissipation through the mold and vaporization and condensation of solvents, if physical blowing agents are involved in the structural foam production process.
Evaporation is assumed to start when the boiling point of the blowing agent in the mixture is reached (cream temperature [T.sub.B]), and terminating at the gel point ([X.sub.gel]). Under this hypothesis, the following conditions may be described:
dL/dt = 0 when: T [less than] [T.sub.B] and X [greater than] [X.sub.gel]
In addition, when T [greater than] [T.sub.B]: dL/dT * dT/dt where dL/dT is the gradient of the experimental curve, obtained by relating the blowing agent mass fraction in the polyurethane mixture with the boiling temperature ([T.sub.B]) of the same mixture. This can be calculated by preparing different formulations with increasing initial amounts of physical blowing agents ([w.sub.B]) and measuring the initial boiling temperature (1). The experimental curves are shown in Fig. 1 and may be correlated with the following equation:
n-pentane: [w.sub.B] = 0.0080 + 14.668 exp (-T([degrees]C)/6.516) methyl formate:
[w.sub.B] = 0.0031 exp (35.751/(T([degrees]C) - 20.236))
The energy balance (Eq 1) for adiabatic conditions is reduced to:
[c.sub.p] . d[T.sub.ad]/dt = (-[delta][H.sub.r]) dX/dt + [h.sub.v] . dL/dt (2)
By integrating Eq 2, the resulting equation using the limits T = [T.sub.0] at t = 0 and T = [T.sub.ad] at a general time t, we obtain the adiabatic temperature profile with time from the experimental one.
[T.sub.ad] = [T.sub.exp] + [[[integral].sup.t].sub.0] U' . ([T.sub.exp] - [T.sub.0] . dt
where: U' = U . S/[c.sub.p] (3)
The heat of reaction (-[delta][H.sub.r]) is given by the following expression, where the evaporation of the blowing agent is also taken into account:
(-[delta][H.sub.r]) = [c.sub.p] . [delta][T.sub.ad] + [h.sub.v] . ([L.sub.0] - [L.sub.f] (4)
Ultimately, by using the general form for the reaction rate (Eq 5), from Eq 2 we obtain the final expression (Eq 6), which carry out all kinetic parameters ([E.sub.a], A, n) related to the three different systems prepared with water, n-pentane and methyl formate. The results are reported in Table 3.
dX/dt = k . [(1 - X).sup.n] (5)
lnk = ln ([c.sub.p] . d[T.sub.ad]/dt - [h.sub.v] . dL/dt/(-[delta][H.sub.r]) . [(1 - X).sup.n]) = lnA - [E.sub.a]/[RT.sub.ad] (6)
For these calculations, an average value of [c.sub.p] = 0.427 cal/g [degrees]C is assumed, in agreement with values given by the Modem Plastics Encyclopedia (11) for cast polyurethane materials.
The value of 'n' is chosen to be equal to 2, because when the left side of Eq 6 is plotted against 1/T, a fairly straight line is obtained (within a satisfactory temperature range) for all the blowing agents.
The value of conversion at gel point ([X.sub.gel]) is obtained experimentally. Figure 2 shows the variation of conversion X with time, which may be easily calculated by using the well-known general form of the kinetic constant (as a function of [E.sub.a] and A) or by integration of Eq 5, respectively, if the reaction order n is chosen equal to 2.
Experiments in Closed Molds
By using the same kinetic parameters of the previous open-mold tests, as an equal amount of blowing agent is employed and the reactants are unchanged, we may obtain the profile of kinetic constant k and conversion x in closed-mold experiments. At first, the temperature variation in the center of the mold is recorded by means of a copper-constant an thermocouple inserted through the side hole into the high capacity aluminum flat mold measuring 210 mm X 130 mm X 10 mm, and used for all the experimental studies reported. The mixture containing polyols, catalyst, surfactant and blowing agents is mixed with isocyanate (using the same pilot plant) and poured into the cavity of the mold, which is locked into position and set under a press.
The results are reported in Fig. 3, which shows the variation of conversion X with time for the three different blowing agents. In closed molds, the maximum value of X is about 20% lower than the X values obtained in open molds because of the choice of a lower initial temperature (28[degrees]C vs. 32[degrees]C) and a smaller amount of reactants poured in the mold.
The mathematical model developed in this paper is a simplified model similar to that proposed by Broyer and Macosko (7). Here, convection effects are ignored because in most of polymer reaction molding processes the filling time is very little compared to the whole period of cure time. In this way the resolution of the equation presented below is greatly simplified. Also molecular diffusion is neglected because of the low value of diffusion coefficient ([less than][10.sup.-9] [cm.sup.2]/s) (7). Thus the mass transfer equation is reduced to the kinetic term. The actual mathematical model is able to predict the temperature change in the mold with isothermal and adiabatic boundary conditions. As known, a large amount of heat is generated in a short time during structural foam polyurethane reaction. Therefore, a temperature gradient will exist inside the mold, with the temperature near the mold walls differing from the temperature in the foamed core, where the influence of heat conduction is not so strong. As a matter of fact, the difference of temperature between the inner and the exterior layer of foamed material is an essential element for the formation of the high skin density. As it seems to be convenient to define a differential material control volume of thickness dz, the heat conduction equation could be written as:
[rho] . [c.sub.p] . (dT/dt) =
[lambda] . [d.sup.2]T/d[z.sup.2] + (1 - [w.sub.B0]) . [rho] . [delta][H.sub.r] . dX/dt + [rho] . [h.sub.v] . dL/dt (7)
From this equation it is quite simple to realize that in any volume element, the temperature changes are the result not only of exothermic polymerization and blowing agent evaporation but also of heat conduction with neighboring layers. In this case, we consider only the temperature gradient perpendicular to the main mold walls as significant (foaming direction z). For simplicity, the specific heat [c.sub.p], the thermal conductivity [lambda], the reaction heat and the density of the system are assumed to be reasonably constant. Actually, thermal conductivity [lambda], according to equation reported by Marciano (3), changes with foam density. We will assume an average value (equal to [10.sup.-4] cal/cm s [degrees]C) for [lambda] related to the average foam density equal to 0.4 g/[cm.sup.3]. It is also assumed that the mold wall temperature is constant, because of the great thermal capacity of the mold.
The energy balance is solved using an implicit finite difference method, according to which Eq 7 is expanded to Eqs 8 and 9:
dT/dt = ([T.sub.m,n+1] - [T.sub.m,n])/[delta]t (8)
[d.sup.2]T/d[z.sup.2] = ([T.sub.m+1,n+1] - 2 * [T.sub.m,n+1] + [T.sub.m-1,n+1]/[delta][z.sup.2] (9)
The whole time period (t - [t.sub.0]) is divided in N little steps equal to [delta]t = (t - [t.sub.0])/N and the distance between the center and the wall of the mold (the mold is assumed to be totally symmetrical) is divided in M steps equal to [delta]z = 0.5/M. [delta]z is considered constant because we have ignored the convective effects in the heat balance equation. In this way, the index n is related to the time coordinate and the index m to the spatial coordinate. A general solution of the differential equation will be the generic temperature T(z,t) = T(m[delta]z, n[delta]t). We employ the implicit form (the 2nd order derivative is written as a function of the (n + 1)th and not the nth time coordinate) in order to choose the integration steps [delta]z and [delta]t independently. The temporal and spatial derivatives written in the above form, if substituted into the energy balance equation, lead to a system of linear differential equation:
[T.sub.m,n] = -A*[T.sub.m-1,n+1] + [1 + 2 . A] . [T.sub.m,n+1] - A . [T.sub.m+1,n+1] - [delta]t.[B.sub.n+1] (10)
A = [lambda] .[delta]t/[rho] . [c.sub.p] . [delta][z.sup.2]
and [B.sub.n+1] = (-[delta][H.sub.r])/[c.sub.p] . k(t) . [[1 - X(t)].sup.2] + [h.sub.v]/[c.sub.p] . dL/dt
The boundary conditions are defined as follows:
[T.sub.m,o] = [T.sub.o] for n = 0 and m = 1,...,M + 1
[T.sub.m+1,n] = [T.sub.o] for m = M + 1 and n = 0,...,N
The system of linear equations, written as a matrix, will be solved with the Gauss method, in particular with the Thomas algorithm, by using an iterative technique: the mth temperature value is calculated by backward substitutions from [T.sub.M+1,n+1] to [T.sub.m,n+1] with m = M,...,1 and the unknowns obtained from previous iteration are substituted into the equations in the subsequent step.
At the gel time we have also calculated the foam density distribution inside the mold according to the equation reported below (3):
[rho] = 1/([w.sub.Bo] - [w.sub.B]) . RT/P . [M.sub.B] + (1 - [w.sub.Bo]/[[rho].sub.pol] + [w.sub.B]/[[rho].sub.B] (11)
The blowing agents in the vapor phase are supposed to act as an ideal gas. Then, the value of pressure, which appears in Eq 11, is assumed to be equal to the blowing agent vapor pressure corresponding to each temperature values at gel time. In this paper, [[rho].sub.pol]. is assumed to be equal to 1.16 g/[cm.sup.3] and this is the same value of skin density assumed for the outer liner.
RESULTS AND DISCUSSION
Figures 4. 5 and 6 show the simulation results of temperature distribution with the three different blowing agents. At 50 s the temperature profile is fairly flat while, after this time, a temperature gradient appears and a maximum value is reached at the foamed inner core (z = 0). In order to check the validity of the model applied here, we need to compare analytical results with experimental data; Figs. 7, 8 and 9 show the temperature increasing at the core, where the experimental values were recorded. The correspondence between real data and values predicted by the model is satisfactory (especially if we keep into account the fact that mean values were assumed for most variables), particularly in relation to the methyl formate blown specimen test. A three-dimensional graph is also reported as an example (Fig. 10), showing the temperature profile as a function of temporal and, at the same time, spatial coordinates.
With the solution of Eq 11, the density distribution at gel time is represented in Fig. 11. Density profiles are only related to tests prepared with physical blowing agents, because Eq 11 is not valid for water-blown systems. Considering the temperature and density distribution inside the mold, the specimens obtained with n-pentane and methyl formate as blowing agents reveal similar behavior with good skin thickness and high skin density: at the moment, from these tests, methyl formate, like n-pentane, appears to be a satisfactory alternative blowing agent.
Then, in order to verify these density values predicted by the model, the skin and core density values are measured by cutting out a rectangular piece of the skin (1 mm thick) and one of the inner layer (5 mm thick). The results reported in Table 4 show that there is good agreement between experimental and theoretic data for physical foaming agents. As foreseeable, the use of [H.sub.2]O as a foaming agent provided foams with a lower skin density, thus greatly affecting the physical-mechanical properties of the specimens.
Characterization of Structural PU Foams
In the final part of this work, the mechanical behavior of structural foams was examined. As known, the mechanical properties of these materials depend quite strongly on their density. So a number of different "sandwich" polyurethane specimens were prepared with the pilot plant by changing the pouring time in order to obtain foams with different densities.
* Superficial Hardness
In accordance with ASTM D 2240 standards, a Shore D instrument was used to measure surface hardness; this instrument comprises a conic tip that penetrates into the surface of structural foams more or less deeply depending on the hardness and compactness of the skin layer. The superficial hardness reported in Fig. 12 is an average value calculated after having tested about ten different areas on the same specimen. Figure 12 shows the clear inferiority of water-blown systems, attested as known by a lower skin thickness. Instead n-pentane and methyl formate reveal a satisfactory outer layer thickness, testified by the fairly high surface hardness values.
* Flexural Strength and Elastic Modulus
Specimens are cut into a rectangular shape and with a particular size in accordance with ASTM D 790 standards. The results of the test are shown in Figs. 13 and 14. In the first Figure, flexural strength values versus average density are reported, while the second Figure shows elastic modulus. From these diagrams, n-pentane and methyl formate turn out to be excellent blowing agents because the rate of increase of the mechanical properties with respect to average density is much greater than foams obtained by using [H.sub.2]O.
* Impact Test
This test is carried out in accordance with ASTM D 256 Standards: measurements of impact strength are made on a Charpy pendulum. The results obtained by changing specimen density are reported in Fig. 15. This chart shows that water is a poor blowing agent when compared to n-pentane and methyl formate; the latter agents managed to reach greater performance ratings, in accordance with previous tests.
In this paper, a study of the effects of employing new blowing agents, as alternatives to harmful CFCs and HCFCs, is presented. With the determination of kinetic data, based on the Marciano model, the most important kinetic parameters for the formation of polyurethane integral skin rigid foams are calculated. A simplified numerical computation, which involves spatial and temporal variations, is also carried out. The distribution of temperature and the density profile through the mold thickness obtained by this theoretical model is in good agreement with experimental values, which reveal the presence of a satisfactory skin thickness when physical blowing agent foamed systems are used (i.e. n-pentane and methyl formate). These data are confirmed by the physical-mechanical characterization of specimens. Every test reveals the highest mechanical properties for the two above mentioned systems, when compared to water-blown foams, that are closely connected with an outer skin layer formation. This indicates the fac t that methyl formate, here tested as a blowing agent for integral skin polyurethane foams, shows as good qualities as the better-known n-pentane, but it is characterized by a lower flammability level.
A = frequency factor [l/mole.s]
[c.sub.p] = average specific heat [cal/g.K]
[E.sub.a] = activation energy [cal/g]
[h.sub.v] = latent vaporization heat of physical blowing agents per unit polymer mass [cal/g]
[delta][H.sub.r] = heat of reaction per unit polymer mass [cal/g]
[L.sub.B] = pondered ratio of liquid blowing agent
[L.sub.B,f]= pondered ratio of liquid blowing agent at [T.ad,max]
[L.sub.B,O] = initial pondered ratio of liquid blowing agent
[M.sub.B] = molecular weight of blowing agent [g/mole]
n = order of reaction
P = pressure in the mold [atm]
R = gas constant [cal/mole.K]
S = heat transfer specific area [[cm.sup.2]/g]
t = temporal co-ordinate [s]
T = experimental temperature [K]
[T.sub.ad] = adiabatic temperature [K]
[T.sub.B] = boiling temperature of blowing agent in the mixture [K]
[T.sub.B,O] = boiling temperature of pure blowing agent at [P.sub.atm] [K]
[T.sub.o] = ambient temperature [K]
[delta][T.sub.ad] = [T.sub.ad,max] - [T.sub.o]
U = global heat transfer coefficient [cal/[cm.sup.2].s.K]
U' = U.S/[C.sub.P] [[s.sup.-1]]
[W.sub.B,O] = initial pondered fraction of blowing agent
[x.sub.B] = molar fraction of blowing agent in liquid phase
X = conversion
z = spatial co-ordinate [cm]
[lambda] = average thermal conductivity of foamed polymer [cal/cm.s.[degrees]C]
[rho] = average density of foamed polymer [g/[cm.sup.3]]
[[rho].sub.pol]. = density of the liquid solution without blowing agent [g/[cm.sup.3]]
[[rho].sub.B] = density of physical blowing agent [g/[cm.sup.3]]
(1.) A. J. Rojas, J. H. Marciano, and R. J. J. Williams, Polym. Eng. Sci, 22,840(1982).
(2.) J. H. Marciano, A. J. Rojas, and R. J. J. Williams, Polymer, 23, 1479 (1982].
(3.) J. H. Marciano, M. M. Reboredo, A. J. Rojas, and R. J. J. Williams, Polym. Eng. Sci, 26, 717(1986).
(4.) L. J. Lee and C. W. Macosko, Int. J. Heat Mass Transfer, 23, 1479 (1980).
(5.) V. K. Gupta and D. V. Khakhar, Polym. Eng. Sci., 39, 164 (1999).
(6.) H. Yokono, S. Tsuzuku, Y. Hira, and M. Gotoh, Polym. Eng. Sci., 25, 959 (1985).
(7.) E. Broyer and C. W. Macosko. AIChE J., 22, 268 (1976).
(8.) M. Modesti, N. Baldoin, and F. Simioni, Eur. Polyrn. J., 34, 1233 (1998).
(9.) Handbook of Chemistry and Physics, 53rd Ed., tables C-407, C-304. D-167. D-157 (1972-73).
(10.) M. Modesti, F. Simioni, and S. A. Rienzi, International Conference: CFCs, The Day After, p. 811, Padova, Italy (1994).
(11.) J. Agranoff, Modern Plastics Encyclopedia, 52, 488, McGraw-Hill, New York (1975-76).
|Printer friendly Cite/link Email Feedback|
|Author:||MODESTI, M.; ADRIANI, V.; SIMIONI, F.|
|Publication:||Polymer Engineering and Science|
|Date:||Sep 1, 2000|
|Previous Article:||The Weldability of Polypropylene Strapping Tape.|
|Next Article:||The Crystallization Kinetics of Polyamide 66 in Non-isothermal and Isothermal Conditions: Effect of Nucleating Agent and Pressure.|