The syntactic foams for this research were fabricated by dispersion of a kind of hollow microspheres in phenolic resins which were synthesized independently. These microspheres were provided by Bengbu Institute of Glass Industrial Design and Research and its properties were listed in Table 1. Formaldehyde solution (37 wt%) and phenol used as the main raw material were supplied by Sinopharm Chemical Reagent Co., Ltd. (Shanghai, China) and Shanghai Titan Scientific Co., Ltd. (Shanghai, China), respectively. Sodium hydrate acted as catalyst while paratoluenesulfonic acid sodium was used as curing agent. They were also purchased from Sinopharm Chemical Reagent Co., Ltd. and Shanghai Titan Scientific Co., Ltd., respectively. All the reagents mentioned above were analytic grade and used without any purification.

The phenol(P) and formaldehyde(F) were mixed in a 500 mL three-necked round bottom flask with a mole ratio of P/F = 1:2. An aqueous solution of NaOH-20% was added into the mixture. The reactions were carried out at 80°C for 4 h with a stirrer, a water bath heating unit, and a reflux condensing tube. After cooling to room temperature, acetic acid solution (36–37 wt%) was added into the reactor. Then the reaction mixtures were left under a vacuum of 0.075 MP at 75°C to remove the by-products such as water and excess formaldehyde. Then we got the phenolic resins used as the binder.

Schematic representation of the syntactic foams was summarized in Figure 1. The microspheres were added to the phenolic resins in multiple steps to avoid agglomeration. Because of the lower density of the microspheres than the binder, the microspheres tend to float to the top surface. This effect was minimized by continuous stirring and injecting the mixtures in a proper mold within a short time, followed by hot compaction for 60 min at 180°C 4 MPa. Then the mold was left in a drying oven to cure for 10–15 h at 150°C. Finally, we took the mold out and cooled it to room temperature. By adding different amounts of fillers to the matrix, syntactic foams with various mass fractions of microspheres (PF/GMs) were thus obtained, named as PF/GM-10wt%, PF/GM-20wt%, PF/GM-30wt%, PF/GM-40wt% composites, respectively. Neat resin plates (PF/GM-0wt%) were fabricated according to the same processing conditions for comparison.

The analyses of fracture morphology were carried out on a FEI Sirion 200 field emission scanning electron microscope (FE-SEM) (FEI company, Hillsboro, OR, United States) at 10 kV. The syntactic foam samples were cut into small pieces and coated with gold to prevent charge buildup.

Thermogravimetric analyses (TGA) were performed with a PerkinElmer Pyris-1 TGA (PerkinElmer Inc., Shelton, CT, United States). The samples of pure PF resins and the syntactic foams were heated from 50°C to 700°C at a heating rate of 10°C min⁻¹ under nitrogen atmosphere.

In order to investigate the heat-transfer properties of the composites we measured the thermal conductivity of the neat resins and the syntactic foams. The thermal diffusivity, λ, was determined by the laser fiash method (Netzsch, LFA-457) at 25°C. The specific heat C_p was measured using differential scanning calorimetry (PerkinElmer, Diamond DSC). The resulting thermal conductivity κ was calculated from the measured thermal diffusivity λ, specific heat C_p, and density ρ according to the equation (Eq. 1): κ = λ ρ C p . (1)

The mechanical tests of tension and compression were performed with different shapes of the foams on a universal testing machine (model CMT4204, SANS Test Machine Co., Ltd., Shenzhen, China). The results presented are an average of five tests. The tensile testing was carried out according to ISO 527:1997 (GB/T1040-2008). The test samples were made into rectangular splines with a length of 100 mm, width of 10 mm and thickness of 4 mm. They were loaded between two stainless steel clamps at a cross-head speed of 2 mm·min⁻¹ at room temperature.

The compression test was carried out in accordance with ISO 604:2002 (GB/T1041-2008) at room temperature. The syntactic foams were machined to blocks of 25 mm³ × 25 mm³ × 25 mm³. The samples were compressed between two stainless steel plates with the cross-head speed of 2 mm·min⁻¹, and the foams were pressed up to 10 mm deflection. The compression modulus, Ec, was calculated by the equation (Eq. 2) (Wouterson et al., 2005): E C = m t A (2) where m is the slope of the initial linear region of the load–deflection curve, and A is the cross-sectional area, and t is the thickness of the syntactic foam.

Figure 2 presented the experimental densities of the neat resin and syntactic foams. It is clear that syntactic foams showed much lower density than that of the neat resin plate. Moreover, the densities decreased with the increasing of microsphere content. Scanning electron microscope (SEM) photomicrographs in Figures 3A–D confirmed the existence of microspheres that homogeneously dispersed in the resin matrix. The glass hollow microsphere is a kind of lightweight materials, so it is quite understandable that the density of the syntactic foams decreased with the increasing of microsphere content. However, it should be noted that densities of the samples cannot be reduced indefinitely because the resin’s capacity for wetting is limited. What’s more, as we added too much glass microspheres, they began to crush as a result of being squeezed by each other during the process of hot compaction. It is clear to see badly crushed microspheres in the matrix everywhere (Figure 3D).

The experimental densities of the samples were obtained from dividing the mass by the volume. Here, the volume includes not only the matrix and microballoons but also some voids existed in the matrix (Figure 3B). Because of the existed voids, experimental densities are lower than the theoretical densities. The voids in the matrix will affect the mechanical properties of the syntactic foams. However, these voids are conducive to heat insulation. The detailed numerical results of the theoretical densities have little impact on subsequent discussions. Thus, we did not further test the theoretical densities of the samples. Generally, being lightweight means material and energy savings. So, it is a key technological requirement for developing materials especially used in the areas of aircraft, spacecraft, and automobiles.

Figure 4 shows the stress-strain curves of the tensile experiment for neat resin and syntactic foam with various contents of microspheres. The tensile strength decreased with the increasing of microsphere loading, while the corresponding tensile modulus increased (Figure 5). There were many researches on the influence of the increasing content of hollow microspheres on tensile modulus of the composites (Bardella and Genna, 2001; El-Hadek and Tippur, 2002; Fine et al., 2003). They concluded that the trend of variation of tensile modulus largely depends on the type of microspheres used. Figure 3B shows a typical SEM fractograph of a tensile specimen with 10wt% hollow glass microspheres. We can see some voids existed among the matrix which will affect the mechanical properties. Luxmoore and Owen concluded that a crack will initiate from an oversized void when a composite is subjected to tensile loading (Luxmoore and Owen, 1978). As a matter of fact, almost all the specimens fractured at a cross-section containing voids. Luxmoore and Owen suggested that the failure of the foam is attributed to the failure of the resin matrix (Luxmoore and Owen, 1978). Introduction of microspheres reduces the resin fraction and increases the inhomogeneity content which leads to poor interfacial strength between the binder and the filler, consequently reducing the tensile strength of the composites. The authors assumed that the matrix served as the load-bearing phase in the composite whereas the hollow microspheres only provided light weight and minimal strengthening effect.

We can see a rough surface of the phenolic resin after brittle fracture and a lot of folded structures were recognized in Figure 3B indicating the plastic yielding of the resin. Apart from that, integral microspheres can be observed on the fracture surface, which demonstrated that when the composites broke under a tensile load the microspheres retained intact. All these facts confirmed the role of the resin matrix serving as the load bearing in the failure of syntactic foam under tensile loading. This conclusion is also in accordance with the results of Luxmoore and Owen’s. Furthermore, it was obvious to see some footprints left by the microspheres, which suggested the presence of filler debonding accompanied with the crack growing over the interface between the resin and the microspheres.

The detail results of compression tests are summarized in Figure 6. Introduction of hollow microspheres leads to the decrease of compression strength (Figure 6A). What’s more, compared with the neat resin, the compression stress of composites with 40wt% microspheres decreased significantly. While the strength and modulus of other foams (with 10wt%, 20wt%, 30wt% microspheres) are similar to each other (Figure 6A, B). The microspheres act as voids and weaken the structure (Wouterson et al., 2007). Figure 7 shows the plots of compression stress–strain curves of neat resin and syntactic foam with various contents of microspheres. As shown in Figure 7 each individual curve exhibits a multistage deformation response when exposed to compressive loading. The initial linear region corresponds to the elastic behavior of the foam. At the end of the linear region a yield point is reached and the load becomes nearly constant followed by a slight decrease, which is called plateau region.

This plateau region attributes to the energy absorption during the process of microballoons crushing (Gupta et al., 2001). To better understand the function of microspheres for compression plateau region, we analyze cross section map of the tested foams (Figure 8). Compared with that of the sample before test, fragments of microspheres are everywhere in the section of the tested foam (marked by red arrows). Glass microspheres are inorganic and rigid. When compressed to a certain extent, they can play a supporting role in the system, which is vital for structural applications. However, when too much microballoons get crushed, further increasing contents result in a gentle downward of the curve. Similar observations were reported by Huang et al. (2010), which is much different from other reports (Goni et al., 1993; Wouterson et al., 2007). This is because that when the strain exceeded a limit value the corners and edges of sample fracture seriously and stress concentrates at void tips. Cracks initiate at these sites and propagate either along microsphere surfaces or through the microspheres (Huang et al., 2010).

In order to investigate the effect of the microspheres on the thermal stability of foams, thermogravimetric analyses (TGA) were performed and the results were shown in Figure 9. For each individual curve, two apparent stages of weight loss can be observed. The first stage (at about 300°C) is the result of the evolution of some small molecules in the end of the chain (such as carboxyl hydroxymethyl, etc.). The second stage is attributed to the major structural (mainly methylene in the resins) decomposition of phenolic resins, which occurs at about 450°C. We compared the temperatures (T₁₀) of which the sample weight loss is 10% and found it shift to higher value with the increasing of microsphere content. The results are listed in Table 2. As shown in Table 2, the final residue content of the synthetic foam at 700°C is higher than that of the neat one. Hollow glass microsphere is a kind of chemical stable materials. So, when heated the microsphere it hardly loosed any weight. Introduction of microspheres reduces the continuity of matrix (Wei et al., 2009), consequently improving the thermal stability as result of hindering heat transmission by the hollow glass microsphere which has excellent heat resistance and heat insulation properties.

Table 3 lists the measured thermal conductivities for the neat resin and the composites. Figure 10 displays the variation of these results graphically. The thermal conductivity of the synthetic foam decreased with increase of hollow glass microspheres, and the thermal conductivity reached a minimum value of approximately 0.129 W (m·K)⁻¹ when 20 wt% of microspheres were added. Generally, there are two ways for heat transferring in materials with voids: heat radiation and heat conduction. The porosity of the foam was improved as the amount of hollow glass microspheres increased, which made the efficient heat transfer surface per unit area decrease. This is equivalent to increasing the thermal resistance of the synthetic foam, thus reducing the efficiency of heat conducting. Then the heat transfer in per unit area was reduced, consequentially reducing the overall thermal conductivity of the material.

However, we observed an increasing of thermal conductivity for PF/GM-30wt% and PF/GM-40wt%. This could be explained by the crushed microspheres. The microspheres dispersed more closely in the resin with the increasing contents as shown in Figure 3C. Then microspheres began to crush as a result of being squeezed by each other in the process of hot compaction, consequently creating effective thermal paths between adjacent microsphere fragments (Figure 3D) embedded in matrix (Kim et al., 2007). So, the thermal conductivity was improved rather than continuously decreasing when the contents of GMs increased to 30wt% and 40wt%. Furthermore, based on the interpretations above, we can assume that the thermal conductivity of the synthetic foam would increase continuously if much more microspheres were added. But back to the fact we could not add the microspheres without limit for there must be a maximum amount of filler which can be fully wetted by the resin matrix in this system. All in all, the best experimental conditions occurred when 20wt% of the hollow glass microspheres were added to the phenolic resin and the thermal conductivity of the composite was improved by ∼56.7% compared with the neat resin.