Mesostructural Design and Manufacturing of Open-Pore Metal Foams by Investment Casting

The present paper describes the manufacturing process of open-pore metal foams by investment casting and the mesostructural/morphological evolution resulting from a new technique of modifying the precursor. By this technique, the precursor is coated with a polymer layer whereby a thickening of the struts occurs. Relative densities in the range of 1.85 ≤ ρrel ≤ 25% of open-pore metal foams can be achieved with high accuracy. The samples investigated have pore densities of ρ P = 7 ppi, 10 ppi, and 13 ppi. The relevant processing parameters needed for a homogenous formation of the polymer layer are determined for two different coating materials and the resulting open-pore foam’s mesostructure is characterized qualitatively and quantitatively. The alloy used for investment casting open-pore metal foamsis AlZn11. The microstructural evolution of these foams is evaluated as a function of the mesostructure. Differences in the microstructure are observed for foams with low and high relative densities and discussed in terms of cooling subsequent to investment casting.


Introduction
Open-pore metal foams possess attractive properties due to their highly porous and light weight structure in combination with their base material.Hence, this group of materials is of great interest for different fields of application, such as in the sectors of heat engineering [1,2], biomedical engineering [3,4], electrical energy storage [5,6], or lightweight design [7,8].
Common processing techniques for manufacturing open-pore metal foams are based on sintering or casting [9,10].The method used in this investigation is a modified investment casting process which offers great advantages due to a targeted adaptation/tailoring of the relative density  rel , structure of the cells, geometry, and alloy composition for any specific application.The open-pore metal foam also acts as an exact copy of the precursor-whereby also highly complex component geometries are possible-, and regular structures and pore-size distributions can be achieved [11][12][13].Typically, a reticulated polymer foam [14] is used as a precursor.In many investigations (cf.[15][16][17]) these polymer foams are used for the metal foam production in as-received conditions leading to a relative density of  rel = 2 ± 0.2% for samples in a range of pore densities of 10 ppi ≤   ≤ 40 ppi.However, for some applications such a low metallic portion is not adequate and hence the relative density needs to be enhanced.One common method of doing this is by thickening the struts of the polymer foam by dunking it into liquid wax [16,[18][19][20].However, this method requires a high level of experience to achieve a homogenous structure in a repetitive accuracy and to avoid blocking of the pores [18,19].Furthermore, thickening the struts can just be conducted within narrow confines.
For these reasons an alternative method for a mesostructural design of open-pore metal foams is developed.Using this method any required relative density  rel can be achieved through a homogenous thickening of the struts.The thickening is a thermal-additive process where the struts of the samples are firstly coated with an adhesive layer, followed secondly by powdering them with polymer granules and thirdly treating the samples above the melting temperature   of the polymer granules.In the present study the basics of the modified investment casting process for open-pore metal foams are described briefly.Furthermore, the influence of treatment time  th , treatment temperature  th , diameter of polymer granules   , and quantity of treatment runs  th on the open-pore metal foam's mesostructure and its relative density  rel is investigated.The microstructural evolution resulting in the investment casted open-pore metal foams is identified.(3) thermal treatment of the coated samples in a batch furnace at  th = 160 ∘ C for  th = 14 min (PG1) or  th = 12 min (PG2), respectively, followed by cooling at standard conditions.

Mold Preparation.
The precursor in its dimensions of 50 ⋅ 50 ⋅ 20 mm 3 is mounted on a gate with a feeding system made of wax which represents the pattern in its final state (Figure 1).Thereupon this pattern is placed in a steel cuvette.The molding material is a sulfate-bonded investment, supplied by SRS Ltd. in Riddings, GB, which is mixed with water and stirred to form a ceramic slurry which gets evacuated until reaching a residual pressure of  ≈ 25 mbar to minimize trapped air.After drying for  = 1 h at standard conditions, the mold is heated in an incineration furnace in a multistep cycle, shown in Figure 2. The first step at a temperature of  = 110 ∘ C is conduced to counteract the hygroscopicity.In a second step at  = 240 ∘ C the gating system is dewaxed and in a third step at  = 360 ∘ C the polymeric precursor is pyrolyzed.The final drying and hardening of the mold is achieved at  = 720 ∘ C in a last step followed by furnace cooling.Due to phase transitions and volume changes, these temperatures are kept for at least  = 1.5 h and a heating rate of max./ = 6 K/min is chosen to prevent crack formation.Al and Zn in the mass ratio  Al / Zn = 89/11, are placed in a preheated crucible and inductively melted under vacuum.At a temperature of  = 750 ∘ C, the casting is induced and the mold, which was preheated at  = 650 ∘ C for  = 5 h, is infiltrated by the melt.After  = 3 min, the mold is removed from the casting machine whereupon it is quenched in water ( = 20 ∘ C).In a last step the samples are cleaned by water jet and Tetranatriumethylendiamintetraacetate (C 10 H 12 N 2 Na 4 O 8 ).

Mesostructural Characterization.
The mesostructure of open-pore foams includes the geometric parameters of the foam-like structure such as strut width  st and cell diameter   and general foam parameters including pore density   and relative density  rel .To determine the effects of strut thickening on the foam's mesostructure, the digital microscope "VHX-500FD" from Keyence NV/SA in Mechelen, BEL, is used.The conducted investigations are focused on the degree of wetting of the struts by the polymer layer and the interaction of successively added polymer granules, which is analyzed qualitatively.Furthermore, quantitative correlations of polymer granules diameter  , strut width  st , and relative density  rel are deduced.The relative density, which is defined as the ratio of the foam's mass   and the mass of a solid body   with the same base material in identical outer dimensions, is governed by weighing the open-pore AlZn11 foam sample with a microbalance (type "AB104" from Mettler-Toledo GmbH in Gießen, GER).

Results and Discussion
In the following, the results of the experimental investigations are pointed out.The processing quality is examined to determine the optimum processing parameters for strut thickening and the mesostructural evolution is characterized by quantifying mesostructural parameters.Moreover, the microstructural evolution in consequence of the varied mesostructure is evaluated for the investment casted foam samples.

Processing Quality.
The key factors for the mesostructural design of open-pore metal foams are the processing parameters of the thermal-additive method for thickening the struts.Particularly temperature  th and time  th of thermal treatment affect the quality of the precursor considerably.
As basic conditions for identifying the optimum thermal treatment temperature  th , the chosen temperature needs to exceed the melting temperature  th >   of the polymer granules or coating material, respectively, and needs to be below the temperature where the skeletal structure of the polymer foam collapses,  th <   .According to manufacturer's data and DSC measurements the relevant temperatures for both polymer granules   and the polymer foam   are determined.The melting temperature is   ≈ 93 ∘ C for PG1 and   ≈ 95 ∘ C for PG2.The foam proves to be inherently stable until reaching a temperature of   ≈ 255 ∘ C.
In a series of experiments, thermal treatment in a temperature range of 120 ∘ C ≤  th ≤ 200 ∘ C was conducted.The results show at  th ≲ 155 ∘ C that the desired effect of strut thickening can be achieved.However, the process needs up to half an hour to be finished.At  th ≳ 165 ∘ C, the process can be accelerated which is yet coupled with the negative side effect of mass stratification of the polymer layer due to gravitation and a nonhomogenous relative density within the foam, respectively.The best results are determined to be at  th = 160 ∘ C for PG1 and PG2.
The required duration of the thermal treatment is investigated in another series of experiments.For this purpose several polymer foam samples are prepared using the first and second process step mentioned in Section 2.1 and thermal treatment at  th = 160 ∘ C in a time range of 4 min ≤  th ≤ 14 min is applied, whereby all minute samples are taken and qualitatively analyzed by the digital microscope.The evolution of strut thickening is shown in Figure 3 which demonstrates the formation of a polymer layer by agglomeration of the single powder granules as a function of time leading to the result of a homogenous layer after  th = 14 min for PG1.Using PG2, a homogenous layer is formed after  th = 12 min.

Mesostructural Evolution.
To quantify the mesostructural evolution aroused by the thermal-additive process for strut thickening, all the polymer foam samples are characterized in consideration of the overall mesostructural key parameter: the relative density  rel .It can be observed that  rel is in a linear relationship as a function of the quantity of thermal treatment runs  th , as shown in Figure 4.By comparing the pore densities   of the foam samples, it can be seen that the increase in relative density Δ rel with each treatment run  th rises by a higher pore density   or a smaller cell diameter   , respectively.An increase of pore density by Δ  = 3 ppi leads to an approx.0.325% higher relative density per treatment Δ rel / th .This can be explained by considering the specific surface area of the foam which is available for coating by the polymer powder.Using the example of a simple expression for the specific surface area

Advances in Materials
(see [23,24]) with  1 as a constant and   as the cell diameter and the relation between the relative density and cell diameter with c 2 as another constant and inserting (2) in (3) leads to the relation between the specific surface area and pore density with the consequence of a higher relative density by increasing the pore density.
As another mesostructural parameter, the strut width  st is characterized by measuring the distance between the outer edges of a single strut, as the example depicted by Inayat et al. [23].In each sample batch with identical pore density   and identical quantity of thermal treatment runs  th , at least 150 single struts are measured to achieve a representative mean value, since the mesostructural parameters, such as the strut width, are normally distributed [23].For a sample batch with   = 10 ppi, which was treated by  th = 2 runs with PG1 (resulting in a relative density of  rel = 9.86%), the strut width distribution is shown in Figure 5.The mean value of the strut width is  st = 0.834 mm and the standard deviation amounts to  st = 0.037 mm for this foam.The corresponding strut widths resulting for the other samples coated by PG1 are listed in Table 1 and the ones coated by PG2 are listed in Table 2.
The increase in strut width  st shows, similarly to the relative density  rel , that it is in a linear conjunction with the in which the strut width resulting for any thermal treatment run  st, increases by the product of the polymer granule's particle diameter   and the quantity of thermal treatment runs  th .However, this expression is, in this investigation, only valid for the polymer granule PG1.For the other polymer granule, PG2, the function results.This can be justified by the small particle diameter   of PG2 (compared to PG1) which forms just a thin polymer layer on the foam's surface in conjunction with the fact that the polymer foam in reference conditions or untreated conditions ( th = 0), respectively, exhibits a strut cross section which can be described by a concaved equilateral triangle (Plateau border).Treating the polymer foam by  th = 1, its strut cross sections result in an equilateral triangle (Figure 6(a)) and by  th = 2 the strut cross sections become shaped as a convex equilateral triangle (Figure 6(b)).In all three conditions,  th = 0, 1, and 2, the outer tips of the plateau border remain as the outer edges by applying the above named method for measuring, resulting in an identical strut width  st .In comparison, a strut of a polymer foam treated by  th = 3 is shown in Figure 6(c) at which the strut cross section emerges in a circular-like shape.Furthermore, it can be deduced from Table 1 that a lower pore density   leads to a higher strut width  st for each treatment run.This fact results due to the already higher strut width  st of low-pore density foams in untreated or asreceived conditions as the consequence of a lower amount of struts in total by an approx.identical relative density  rel .

Microstructural Evolution.
The evaluation of the microstructural evolution as a result of the varied mesostructural parameters is investigated for the openpore AlZn11 foams with a pore density of   = 10 ppi.The sample illustrated in Figure 7(a) represents the reference conditions and is hence not strut thickened ( th = 0).The ones shown in Figures 7(b) and 7(c) are strut thickened by a quantity of thermal treatment runs  th = 3 and  th = 5 with PG1.The increase in strut width with each treatment run Δ st / th can be seen clearly through these macroscopic photographs.
Sample preparation and metallographic analysis are applied as mentioned in Section 2.5 whereby the investigated sections of the foam represent a strut orientated parallel to the sample plane.The micrograph illustrated in Figure 8(a) shows a strut in untreated conditions ( th = 0) with two comparatively large grains.In contrast, in the investigated sections of the samples strut thickened with PG1 by  th = 5 (Figure 8(c)), a dominant dendritic microstructure with small grains is visible.The samples treated by  th = 3 (Figure 8(b)) partly exhibit a distinctive dendritic microstructure and there are partial zones at which no casting structure is evident.
These differences in microstructure can be explained by the fact that after solidification in the mold, the cooling rate is higher in foam samples with a higher relative density since a higher volume fraction of AlZn11 is existent and more heat can be dissipated due to the high thermal conductivity  of an Al alloy.In contrast, the thermal conductivity of the plaster is much smaller but its heat capacity   is much higher which leads to a lower heat dissipation in foams with small relative densities  rel .

Summary and Conclusion
In the present study, a new method of thickening the struts is proposed to achieve defined relative densities of investmentcast open-pore metal foams in the range of 1.85% ≤  rel ≤ 25%.This method is a three-step process, where the open-pore polymer foams need to get spray coated by an adhesive layer, afterwards powdered with polymer granules, and thermally treated to achieve a precursor in its required relative density with homogenous struts.Subsequently, the common procedure of investment casting has to be applied.
For evaluating the processing quality of the strut thickening process, two polymer powders with different particle diameters   ≈ 290 m (PG1) and   ≈ 32 m (PG2) are investigated.By varying temperature  th and time  th of thermal treatment, the optimum parameters for a homogenous mesostructure of the foam and for a uniform formation of the polymer layer are identified.The processing temperature is determined to be at  th = 160 ∘ C for both powder types.The formation of the polymer layer occurring by agglomeration of the powder granules is achieved after  th = 14 min for PG1 and  th = 12 min for PG2.
With respect to the mesostructural parameters resulting from this method, the relative density  rel and strut widths  st as a function of the quantity of thermal treatment runs  th are determined for open-pore foams with pore densities of   = 7 ppi, 10 ppi, and 13 ppi.A linear relationship between  rel and  th is observed whereby the increase of  rel with each treatment run is more dominant as the pore density   is higher which is a consequence of the foam's specific surface area   .Furthermore, the strut widths  st also show a linear dependency on the quantity of thermal treatment runs  th in compliance with the requirement of a circular-like shaped strut cross section which can be ascribed by the product of the polymer powder's particle diameters   and quantity of thermal treatment runs  th in addition to the strut width of the foam in untreated conditions  st,0 .
The microstructural evaluation indicates differences in the microstructure of the open-pore AlZn11 foams as a function of the mesostructural parameters.A dominant dendritic structure in foams with high relative densities is

Figure 1 :
Figure 1: Schematic diagram of mold preparation process: (a) modelling in terms of mounting precursor on gate with feeding system, (b) embedding in terms of infiltrating the pattern placed in a steel cuvette by molding material, and (c) burning out in terms of dewaxing the gating system, pyrolyzing the precursor, and hardening the mold.

2. 5 .
Microstructural Characterization.To evaluate the influence of the relative density  rel on the microstructural evolution of the open-pore AlZn11 foams, microscopical studies are carried out.Therefore, the metal foam samples are partitioned at a distance of  = 25 mm to the deadhead by a wet disc grinder.The single sample pieces are embedded in a cold polymerizing plastic (Scandiplex in addition to Varioplex from Scan-Dia Hans-P.Tempelmann GmbH &

Figure 4 :
Figure 4: Relative density  rel as a function of the quantity of thermal treatment runs  th of strut thickened (by PG1) polymer foams with pore densities of   = 7 ppi, 10 ppi, and 13 ppi.

Figure 5 :
Figure 5: Strut width distribution of open-pore foams with a pore density of   = 10 ppi and a relative density of  rel = 9.86%.

Figure 6 :
Figure 6: Cross section of a single strut of a PG2-treated polymer foam with a pore density of   = 10 ppi by a quantity of thermal treatment runs of (a)  th = 1, (b)  th = 2, and (c)  th = 3.