Ultra-low thermal conductivities of hot-pressed attapulgite and its potential as thermal insulation material

Attapulgite (or palygorskite) is a natural hydrated magnesium-aluminum silicate clay mineral with fibrous morphology. As modeled by Bradley,¹ attapulgite is fundamentally built up by ribbons in two silicon–oxygen tetrahedron layers and one octahedral layer of metal cations, in addition to three types of water molecules: hydroxyl (OH) groups bonded to the metal cations, bounded crystal H₂O molecules at the edges of the octahedral layer, and nearly free zeolitic H₂O filling up the empty channels parallel to the ribbons.

Previously, attapulgite's dehydration behaviors^2–7 and structural changes^8–10 during dehydration have been reported. Though the details of each dehydration process vary depending on the specific sample composition, it has been generally agreed that the complete dehydration consists of three types: escape of zeolitic water, loss of crystal water, and condensation of hydroxyl (OH) groups.⁸ Preisinger proposed that the loss of crystal water will lead to the breakdown of the open-channel structure and the formation of a folded structure called “attapulgite anhydrite.”² As temperature (T) goes higher (over ∼700 °C), the fibrous structure will be completely destructed, leading to various high-T silicate phases.^2,3,8

Recently, there is a rising interest in utilizing natural mineral based materials for advanced thermal based applications.^11,12 Attapulgite has been valued in versatile industries for its intriguing properties including large surface area and high absorption capacity, owing to its natural fibrous structure.^13,14 It was also proposed by researchers for adsorption of organics or heavy metals and catalyst support.^15–20 Though with high absorption capability, unlike clays such as montmorillonite, attapulgite does not change much in volume as it absorbs water, advantageous as a barrier clay in applications.¹⁴ 

Despite the intensive work on attapulgite, not much attention has been paid to its thermal properties. As an insulator hydrated silicate, attapulgite is anticipated to have a relatively low intrinsic thermal conductivity. In addition, it was emphasized in previous reports^21,22 that high volume fraction of pores with specific aspect ratio and spatial distribution can effectively impede heat transport in insulating oxides such as zirconia or clay.^23,24 Attapulgite naturally features high porosity, with elongated open-channel microstructures similar to the columnar pores achieved by Hass et al. in low thermal conductivity yttria stabilized zirconia (YSZ).²³ The fibrous microstructures and complex phases of attapulgite, if properly retained in dense bulk state, can allow significant reduction in thermal conductivity.

In this study, we synthesized attapulgite bulk samples from powders by hot-pressing, which has not been previously reported. This technique, easy and widely used for bulk materials, induces an acute and rapid heating reaction which is able to conserve the complicated nano/microstructures of attapulgite while ensuring a relatively complete dehydration. We first examined powder attapulgite from Xuyi County, Jiangsu Province, China. Successively, investigations on the crystal structure and thermal properties of the samples hot-pressed from powders at different temperatures were reported. Our hot-pressed attapulgite samples exhibit an extremely low thermal conductivity. Finally, we discuss the possible application potentials of the hot-pressed attapulgite.

First, X-Ray diffraction (XRD) was performed on the as-received attapulgite powder at room temperature and high-T using a PANalytical multipurpose diffractometer with an X'celerator detector. The dehydration behavior of the powder was analyzed by differential scanning calorimetry (DSC) on a DSC 404C Pegasus Thermal analyzer and thermogravimetric analysis (TGA) on a Hi-Res TGA 2950 Thermogravimetric Analyzer. Microstructural details were studied by scanning electron microscopy (SEM) on a Hitachi 2100U Spectrometer. Second, the as-received powder was hot-pressed into disk samples at 640, 750, 850, and 1000 °C, denoted as HP640, HP750, HP850, and HP1000, respectively. XRD and DSC analyses were performed on all the hot-pressed samples, and the specific heat C_p was calculated from the DSC curve. The porosity p of the samples was estimated by p = 1 − ρ/ρ_t,²⁵ where ρ and ρ_t are the actual and theoretical volume densities.²⁶ The microstructures of the hot-pressed samples were studied by SEM on the freshly broken surface, and further studied by transmission electron microscopy (TEM) and selected area electron diffraction (SAED) on a JEOL-2010F transmission electron microscope. The thermal diffusivity D was measured for the hot-pressed samples on a Netzsch LFA 457 laser flash system, and thermal conductivity κ was calculated from κ = ρDC_p. The annealing experiment was performed in air using a UL standard compact muffle furnace. Thermal expansion coefficient (TEC) was measured on Netzsch Dilatometry 402C at a heating rate of 2 °C/min in air.

The XRD patterns of our attapulgite powder sample measured at different T are displayed in Figure 1. The powder XRD pattern at room temperature has good match with the monoclinic phase of attapulgite, showing a strong (110) peak at 2θ = 8.44°. It also indicates a small amount of impurity quartz (Q:SiO₂). Disappearance of the (110) peak in the XRD patterns measured between 230 °C and 380 °C suggests a transformation from monoclinic into the folded phase which is stable up to 800 °C.⁸ The powder experienced four dehydration stages upon heating up to 800 °C and the TGA/DTGA and C_p data are shown in the supplementary Figure S1.²⁷ The first stage, due to the escape of zeolitic water from the nanofiber channels, occurs from 106 °C. The second and third stages take place at 254 °C and 473 °C, respectively, largely attributed to the loss of crystal water.^4,6,8 The final stage takes place above 694 °C for the dehydroxylation process. The four-stage sequence is also confirmed by the C_p data.

The hot-press temperatures and densities of the samples are summarized in Table I. The C_p-T data of the hot-pressed samples are presented in Figure 2(a1)–2(a4) with the identical C_p axis scale. No obvious endothermic process was observed for samples HP850 and HP1000, implying no significant dehydration because hot-pressing above 800 °C removed most water molecules. The results are also in good agreement with the results by Skauge et al. for dried attapulgite.²⁸ For samples HP640 and HP750, weak peaks were observed upon heating to ∼160 °C, reflecting the loss of small amount of residue zeolitic water. It is the result of the retention of porous structure due to the low hot-press temperature and short hot-pressing duration (5–10 min).

The dehydration processes are accompanied by phase transitions. As shown in Figure 2(b), the loss of most peaks in the XRD patterns of samples HP640 and HP750 implies the destruction of monoclinic phase under hot-pressing. However, in contrast with the XRD patterns of attapulgite powder heated to 640 °C and 750 °C, no obvious indication of folded attapulgite phase could be observed in samples HP640 and HP750, showing different phase transitions from those upon conventional heating.

Microstructural details of these samples were observed by SEM and TEM. The SEM images can be found in the supplementary Figure S2.²⁷ For HP640 and HP750, the fibrous structures are still observable. Differently, samples HP850 and HP1000 show a relatively solid structure with no separate nanofibers.

The almost identical XRD patterns of samples HP850 and HP1000 suggest the formation of enstatite (MgSiO₃) and cristobalite (SiO₂) (indicated by C in Figure 2(b)), consistent with the increased volume density of the samples (Table I). The TEM images of the samples are presented in Figure 3. Figure 3(a) shows that the surviving fibers in HP640 are of random orientations and mainly composed of amorphous phase as shown in the SAED image in the inset of Figure 3(b), consistent with the few crystalline peaks in the XRD pattern. Formation of amorphous phase instead of folded attapulgite phase may be ascribed to the rapid heating/cooling processes and the high pressure during hot-pressing. Two types of pores mainly exist. Large pores between separate fibers were observed, shown in Figure 3(a). In addition, due to the retention of the fibrous structure, the residues of the open-channel microstructures may form very fine pores within the surviving fibers.

For HP850 and HP1000, finer structural features can be identified, as shown in Figures 3(c) and 3(d), for sample HP1000. The sample is composed of highly crystallized regions of 50–100 nm in width, randomly distributed in an amorphous matrix. The FFT analysis of the crystalline regions indicates that they are single crystals of orthorhombic enstatite, as shown in Figures 3(e) and 3(f). No cristobalite phase is identified from the crystal regions of HP1000, which seems contradictory with the presence of the intense cristobalite peak around 22° in the corresponding XRD pattern. Possibility of texturing can be excluded, since the intensity of the peak around 22° does not diminish in the XRD pattern of powder grinded from HP1000 in Figure 2(b) (HP1000 powder). One reasonable explanation is that these seemingly amorphous regions belong to opal, a natural hydrated silica polymorphs, which was reported to exhibit similar XRD patterns to cristobalites due to similar local structures in microcrystallines,^29,30 but lack long-range order. This suspect is proven by two features in our XRD patterns (Figure 2(b)): a broad diffuse band (indicated by the arrow) around 22° in the powder XRD pattern of HP1000, which characterizes the structure of the amorphous opal-A;³⁰ minor tridymite peaks (indicated by arrows) in the XRD patterns of both HP850 and HP1000, which characterize the stacking disorder of opal-C, a major difference between well-ordered opal-C and cristobalite.²⁹ The formation of opal may have occurred in sample HP1000 due to incomplete dehydration during hot-pressing or rehydration during the sample preparation. However, the clarification of the microcrystalline regions remains ambiguous, since it is extremely difficult to distinguish the amorphous opal-A and relatively ordered opal-C in our samples by TEM observation.

Figures 4(a1)–4(a4) display the κ-T data measured from 50 to 600 °C for these hot-pressed samples. All the samples exhibit relatively low thermal conductivities, and the increment of κ for all samples over the measured T-range is small. The κ values of samples HP640 and HP750 are 0.34 W m⁻¹ K⁻¹ and 0.41 W m⁻¹ K⁻¹ at 50 °C, respectively, and they increased to only 0.47 W m⁻¹ K⁻¹ and 0.54 W m⁻¹ K⁻¹ at 600 °C. Such low κ is attributed to the following reasons. First, the samples are highly porous, with porosity 45.7% and 35.2%, respectively. Second, the randomly oriented grain boundaries of the retained fibers and the poor crystallinity induce strong phonon scattering that essentially impede thermal transport. Third, fine pores within the fibers, retained from the original elongated open-channels, has low aspect ratio (width/length) effective in limiting heat conduction, as modeled by Lu et al.²² 

Meanwhile, samples HP850 and HP1000 exhibit slightly higher κ of ∼1.03 W m⁻¹ K⁻¹ and ∼1.85 W m⁻¹ K⁻¹ at 50 °C, ascribed to the lower porosity due to the more compact microstructure. Nevertheless, their κ are still impressively low with respect to many compacted dense solids, even for current oxide thermal barrier materials such as mullite (3.3 W m⁻¹ K⁻¹).³¹ This is mostly attributed to the poor crystallinity of the samples, with only localized crystalline regions in the major amorphous matrix.

To evaluate the intrinsic thermal conductivity in our samples, we estimated κ of the solid regions by a simple analysis using the effective medium approximation.³² Consider the samples as two-component systems consisting of solid regions and pores, and assuming the pore size does not change with T, we have

where f₁ and f₂ are the volume fractions of the solid region and the pores with f₁ = 1 − p and f₂ = p; κ₁ and κ₂ denote the corresponding thermal conductivities. The thermal conductivity of argon gas is used as κ₂ since all the thermal measurements were conducted in argon atmosphere.

A comparison of κ and κ₁ from 50 °C to 600 °C for HP640, HP750, and HP850 are presented in Figures 4(a1)–4(a3). Analysis of HP1000 is not carried out here since the pores are almost negligible. All three samples have a reasonably higher intrinsic κ₁ than the corresponding κ. However, they still feature a relatively low κ₁ below ∼1.7 W m⁻¹ K⁻¹. Interestingly, though with higher κ but lower porosity, HP750 has a κ₁ ∼ 17% lower than that of HP640. This may be attributed to the more severe bending and twisting of the retained fibers in HP750, as indicated by the dashed circles in Figure S2(b).²⁷ The larger inclination angles of grain boundaries and pores lead to increased thermal diffusion path length.

For thermal barriers, the material stability is critical. Our hot-pressed samples show high stability to moisture. First, the κ of HP640 and HP750 were re-measured after submitted in air for 5 months, and a comparison of the κ(T) data is presented in Figures 4(b1) and 4(b2). There is negligible variation in κ in the high-T range. A slightly higher κ around 150 °C in the air-aged samples compared with fresh samples, as indicated by arrows, is induced by the increase in C_p in the low-T range that results from moisture absorption on the surface. In fact, simply after the thermal conductivity measurements, during which the samples were heated up to 600 °C for only a few hours, both samples regain their original mass 5 months ago, as summarized in Table II. Second, our hot-pressed samples also show high thermal stability against high temperature annealing. The samples were heated in air to 400 °C for 24 h and then their κ(T) data were re-measured, as shown in Figures 4(b1) and 4(b2). The two sets of κ(T) data show negligible differences over the whole T-range. In addition, the κ(T) data of the annealed samples no longer show bump-like feature around 150 °C that comes from the surface absorption of moisture.

We also performed the TEC measurement for samples HP750 annealed at 600 °C and 850 °C for 12 h. The average TEC within the temperature range of 25 °C to 600 °C are 5.4 × 10⁻⁶ K⁻¹ and 7.4 × 10⁻⁶ K⁻¹ (Figure S3),²⁷ which are relatively low compared with those of current oxide thermal barrier materials,³¹ as discussed in the supplementary Table SI.²⁷ 

In summary, we are able to obtain very low thermal conductivity in dehydrated attapulgite by hot-pressing attapulgite powders at different temperatures. The rapid heating of hot-pressing process quickly dehydrates the attapulgite powders, forming dehydrated attapulgite with high porosity and complicated microstructures, and promotes the formation of amorphous phase. All these factors contribute to the low thermal conductivity. In addition, our hot-pressed attapulgite also shows good thermal stability in moist environment and under high temperature. These features reveal the great potential of the hot-pressed attapulgite for thermal insulation applications. Further investigations along this line on its possible applicability are definitely deserved in the future.

The work was funded by the U.S. Department of Energy under Contract No. DOE DE-SC0010831. Qinyong Zhang would like to thank the financial support from National Natural Science Foundation of China under Contract Nos. 51372208 and 51572226.