Improving Metal Flow in Lost Foam Casting Through Use of Low Thermal Degradation Hot Melt Adhesives

Multiple Authors

Lost foam casting is a manufacturing process capable of producing high volume, high precision castings with undercuts and inverse draft angles. This process uses expanded polystyrene (EPS) patterns that are expended during the casting process. In many applications, EPS patterns are glued together using a hot melt adhesive to achieve complex geometries for high-value castings such as engine blocks. During mold filling, melt flow defects can be caused by hot-melt adhesives thermally degrading slower than the EPS foam patterns. Sun and Littleton observed, using x-rays, that the metal front stalls at glue joints for up to a second, resulting in multiple metal streams through the glue joint that led to fold defects.

Traditional hot melt adhesives are comprised of a mix of polymers, typically wax based, that have a high molecular density. The energy per unit volume necessary to degrade through polymer materials corresponds to the molecular complexity of the material. The molecular complexity can be quantified by the ratio of tetravalent carbon sp3 (σ) bonds to total bonds within the molecule (Eqn. 1). 

Molecular Complexity [%] =sp3 Ctotal C×100

Of the p-block elements, the C-H bond has the second highest bond energy (C-F has the highest), therefore making a molecule with more of them is undesirable for lost foam. The decrease of complex bonding in the adhesive’s components can have positive impacts on the thermal degradation of the hot melt mixture. It is hypothesized that if the fraction of complex sp3 bonds is minimized, then the velocity of the melt front through the glue joints will be more like that of EPS foam because the melt front velocity is proportional to the fraction of complex sp3 bonds.

In the early 2000s, General Motors developed a new lost foam adhesive based on C-9 aromatic monomers (styrene, methylstyrene, vinyltoluenes, indene, and methylindenes) with primary plasticizers that are miscible with aromatic resins and secondary plasticizers that have limited miscibility with the resin. Plasticizers are used to alter properties of the base material, such as elasticity and viscosity. Results from the laboratory study showed the glue thermally degraded 33% faster than other commercial glues; however, during full-scale production trials, issues were identified relating to a strong odor and stringing like melted cheese during application. This ultimately led to the abandonment of the glue.

Over the last few decades, the lost foam casting process has seen a decline in popularity in the U.S., resulting in fewer advances in lost foam technologies including research in adhesives. This work focuses on the continued development of a nontoxic, odorless, and stringless hot melt adhesive that thermally degrades, residue-free, closer to the rate of EPS foam patterns.

Materials and Methods

To find the optimal glue composition that meets the above requirements, several compositions were assessed to understand the effects of various plasticizers on odor, viscosity, and the melt front velocity. 


To reduce the energy required to pyrolyze the polymer, short chain, or “simple,” polymers were preferred. A simple structure was defined as lowest elements per chain and lowest possible bond energies. Poly(styrene-co-methylstyrene) resin [αPS] was selected for the base material with plasticizers: poly(ethylene glycol) dibenzoate [PG], poly[trimethylolpropane/di(propylene glycol)-alt-adipic acid/phthalic anhydride] [AA], and dimethyl phthalate [DM]. A comparison of the molecular chains of poly(ethylene glycol) dibenzoate.

~410) and dimethyl phthalate plasticizers are shown in Figure 1. Dimethyl phthalate does not have the repeat unit’s “n” making it the preferred short chain polymer. A paraffin wax [PW] addition was assessed after the initial screening analysis was conducted.

Eight samples were produced (Table 1). The properties of the samples were compared to a baseline commercial glue commonly used in lost foam foundries consisting of 30-60 wt.% ethylene vinyl acetate, 20-30 wt.% hydrocarbon resin, and 1-20 wt.% polyolefin wax. Each sample composition was mixed in a melting pot with a Variac voltage controller, thermocouple, and a stirring device. The Variac was used to control the voltage to the melting pot to maintain a temperature of 320F (160C). The melting process took about 15 minutes while the αPS was constantly stirred to avoid burning on the bottom. After the αPS was completely melted, the temperature was lowered to 284 ± 18F (140 ± 10C) to minimize vaporization of the plasticizers while staying above the αPS melting point. Once the temperature stabilized, the DM, PG, and AA were added to the mixture forming a sticky putty. 


To test the glue odor, a screening design of experiments (DOE) was conducted using the compositions in Table 1. Once the respective glue mixture was created and at a temperature of 284 ± 18F (140 ± 10C), a researcher would leave the area for roughly 30 seconds to become accustomed to a neutral scent.

The researcher then re-entered the room, wafted the glue, and rated it on a scale from 1-7 where 1 was the best odor and 7 was the worst. This process was repeated by a second researcher.


In the lost foam casting process, the viscosity of the adhesives is carefully controlled during application as low viscosity leads to stringing and high viscosity poor wetting. Viscosity was measured immediately following the mixing with a Saladulce NDJ-9S viscometer. The viscometer was positioned above the melting pot with the spindle in the center of the mixture (Figure 2). For each glue sample the temperature, spindle number, RPM, and viscosity was recorded. The viscosity was targeted to be 400cP ± 100cP.

Thermogravimetric Analysis (Tga)

To quantify the differences in thermal degradation of the samples during casting, thermogravimetric analysis (TGA) was used to study the mass loss over a temperature range. A 20-40 mg sample of each mixture was sectioned off and loaded into an aluminum oxide crucible. The samples were placed on a balance arm in the instrument’s furnace. A Mettler Toledo SDTA851 TGA was used with an inert nitrogen atmosphere and heated at a constant rate of 50F/min (10C/min) up to 1112F (600C) ensuring the complete sample decomposition.


To determine if the new glues improved the melt flow, a head-to-head casting trial was performed. Using the screening result, the top two compositions were selected for comparison with a commercial glue (glues 2 and 8 in Table 1). A pattern consisting of four 200 mm legs with a cross-sectional area of 12.7 mm by 12.7 mm was used for casting trials (Figure 3). EPS foam strips of the same cross-section were cut into 50 mm segments using a band saw and glued together using the three adhesives. The EPS control was a single piece of foam with no glue joints.

Each sample adhesive was heated to 284±18F (140±10C) for application. Application was done by hand by dipping one foam segment into the glue pot so that a thin layer was only present on the gluing surface. Two segments were then immediately joined to maintain the glue temperature.

To ensure the evenness and consistency of the glue joints, the mass of the two segments involved in the application was recorded before and after the glue application. Using the foam joint surface area and the glue densities (Table 2), determined using Archimedes’ principle, the foam joint area, the length of the glue joint was determined. The standard deviations were all within 0.12 mm, and thus the application procedure was consistent across all joints for each glue mixture (Table 2). The average length of each type of glue joint was the same within experimental error. 

After applying the three glue mixtures to three joints (at 50, 100, and 150 mm from sprue well), there were a total of four unique legs in the pattern (solid EPS, glue 2, glue 8, and commercial glue). Three (3) replicates were made of the pattern with randomized leg position. A factorial design of experiments was performed to measure the effects of glue mixture and casting trail variations on the melt front velocity. Each leg was coated in a lost foam ceramic coating to control the permeability during the casting process. The coated pieces were drip dried for 24 hours prior to casting.

To measure the metal flow within each mold, a series of timer circuits were used to record electrical continuity as the mold filled. The timer consisted of an Arduino Uno board, wires placed into the EPS segments (Figure 3), and a ground wire connected at the sprue. The sprue wire and location 7 were placed at the beginning and end of each foam piece respectively, while the remaining locations were placed in pairs 2 mm on either side of a glue joint (Figure 4). 

Three chemically-bonded sand molds were fabricated, and the coated foam patterns placed into the mold. Continuity wire leads exited along the parting line (Figure 3) and were connected to the Arduino. The cope was molded on top of the assembled drag and EPS pattern to ensure that the mold would not separate and leak. All flow timers were checked prior to filling the molds. Aluminum A356 was used for the casting trials at a pouring temperature of 1400 ± 18F (760 ± 10C).

Experimental Results


The odor ratings ranged from moderate (4) to unbearable (7) (Figure 5). Glue 2 had the lowest odor (4), and a second mixture with paraffin wax was assessed to further mitigate the odor. This addition of paraffin wax did not change the molecular complexity of the glue and was added to solely improve qualitative properties. Given the negative impact of dimethyl phthalate on the smell, it was therefore removed for all further testing. The poor rating of the 0% AA glue was from glue 3, which had the maximum level of DM.

Thermogravimetric decomposition curves for EPS, glue 2, and two commercial glues are shown in Figure 6. The objective is to have the glue decompose at lower temperatures (to the left) of the EPS curve. The commercial glues surpass the EPS line as they decompose, resulting in glue remaining after the EPS has fully evaporated.


The progression of the metal front was recorded by metal flow timers (Figure 7). The crosshatched zones are the mean metal flow time in the EPS segments. The black zones represent the mean metal flow time between the glue joints across all three replicates. The metal flow direction is from bottom to top.

The velocity of the melt front through each glue joint was calculated by subtracting the four-millimeter standard distance between glue joints and the melt front position from the metal flow timers. As the molecular complexity of the glue increased, the melt front velocity decreased in agreement with the hypothesis.


The best mixture from the odor screening DOE was glue 2, due to the lack of DM and the overall low amount of plasticizer (Table 1). To dilute the odor of the mixture further, paraffin wax was added, which has a neutral smell compared to the plasticizers and has molecular simplicity to aid thermal decomposition. Adding paraffin to glue 2 produced glue 8, which both have a molecular complexity of 37% (Eqn. 1).

Initial assessment of the molecular complexity hypothesis via the thermal degradation curves was in agreement with expectations (Figure 6). The control curve of EPS indicates optimal zone performance. The commercial glue had a higher thermal degradation temperature than the EPS foam, which was expected given its molecular complexity, and the observed cast defects in production. Glue 2 performed well in TGA with lower thermal degradation temperatures than both the commercial glue and baseline EPS foam. The thermal degradation of glue 2 supports the hypothesis that less molecular complexity will lower the thermal degradation energy.

The aluminum flow time during casting was assessed between a commercial glue and the αPS-based glues (Figure 7). Flow time is decreased in the αPS-based glues as compared with the commercial glue. Slower melt front velocities were observed at the glue joints indicating more energy is lost than when flowing through the same length of EPS (Figure 8). For constant glue joint thickness, the faster melt front through the glue joint with αPS-based glues support the hypothesis that low complexity molecules enable higher melt front velocities. Based on analysis of the factorial design of experiments for the casting trials, the type of glue has a significant impact on the melt front velocity, a p value of 0. The components in the αPS-based glues have a molecular complexity of 37%, while the commercial glue has a molecular complexity of 68%. The stronger, more complex bonds within the commercial glue take more energy to break down.

The long-term thermal stability of glue 2 should be quantified, as typical hot-melt glues have roughly 30%–40% tackifiers, which are thermoplastic resins that are used to modify odor, performance, and thermal stability of the glue. These resins have complex bonds, and they contribute to the high molecular complexity and slow thermal degradation of current industrial glues. To keep the molecular complexity of the glue 2 low, no tackifiers were added, and this may lead to less thermal stability over time. TGA data (Figure 6) shows that thermal degradation does not initiate until around 392F (200C), which is 122F (50C) above the operating temperature, so thermal stability can be assumed for short times. This may be indicative of long-term thermal stability, but TGA should be run at operating temperatures eight or more hours to assess long-term stability.


Defects in lost foam cast parts have been shown to be caused by the higher thermal degradation energies in hot melt adhesives as compared to polystyrene. This work focused on development of a nontoxic, odorless, and stringless hot melt adhesive that thermally degrades, residue-free, at closer to the rate of EPS foam patterns. A polystyrene-based glue was designed to achieve a lower thermal degradation temperature, measured via thermal gravimetric analysis (TGA), while maintaining qualitative metrics such as low odor and a viscosity to avoid stringing during application. Data from the casting trials found the resulting polystyrene-based glue increased the glue joint melt front velocity by 75% compared to a commercial glue. The addition of paraffin wax further increased the melt front velocity and eliminated stringing upon application.