Sealing and Compression Properties of Compressive Gaskets Made of Micrometric Vermiculite Particles

Abstract

Vermiculite gaskets obtained by pressing vermiculite powders in the range of 17.7–80 MPa, were studied for their sealing (leak rate measurements) and compressive properties (compressibility and resiliency). The in plane permeability at roomtemperaturewas found to decrease strongly through increasing elaboration pressure, that reduced both the median pore radius (<30 nm) and the macropore volume fraction (<45%), measured by mercury intrusion. After annealing at temperatures of up to 600 °C, the out of plane permeability (measured at room temperature) was increasing from ~10−20 (at 200 °C) to ~10−24 due to the increase in anisotropy related to the densification and the formation of interlayer bonds. The global leak rate was found to be determined exclusively by the contact leak rate and independent of the material's permeability. The leak rates measured at room temperature were also found to be dependent on the gasket's resiliency values. The global helium leak rate (2.5 × 10−2 atm·cm3·s−1·m−1, for 35 MPa working pressure, under 5 bar helium pressure) was relied neither on the working temperature (25 °C to 800 °C) nor the material porosity for gaskets pressed at 200 °C and 80 MPa. The resiliency (~5%) and compressibility (9%) values of these gaskets were reduced, as heating the materials to 800 °C due to the densification induced by both pressure and temperature, increasing their rigidity.

1. Introduction

Before the 1990s, asbestos was one of the main materials used to build compressive gaskets for high temperature application (up to 800 °C), until its trade and use were banned outright due to the toxicity of inhalation of asbestos dust (Choiron et al., 2011). Graphite can substitute asbestos material in compressive gaskets, thanks to its interesting mechanical properties, but its temperature use is limited to 400 °C– 500 °C because of its oxidation. Glass gaskets are used at high temperature but their main disadvantages lie in their adhesion to the support and the difficulties to disconnect the gaskets from their support (metal or ceramic) after the heating cycle at room temperature (Mahapatra and Lu, 2010). Thus, due to their thermal compatibility, an absence of adhesion to the support at high temperature and a certain compressibility, clay minerals (mainly vermiculite or mica) can be applied advantageously as compressive gasket materials with or without additives. Films of vermiculite or mica (mica paper) can both be formed by roll forming with chemical additives (binder, etc.) in order to prepare (compressive) gaskets. However, the presence of an organic binder can limit the application temperature to the temperature of decomposition of the binder.

Two sorts of mica: phlogopite and muscovite particles are widely applied in the preparation of compressive gaskets (Simner and Stevenson, 2001; Chou et al., 2002, 2003; Bram et al., 2004; Chou and Stevenson, 2004; Fergus, 2005; Chou and Stevenson, 2009). The main advantage of vermiculite upon mica is the possibility of layer exfoliation, which is supposed to give additional elasticity and compressibility to the compressive gaskets. Moreover, vermiculite can withstand temperature heating up to 800 °C, without any main change in its layered structure. Due to its ability to exfoliate (separation of the layers) by thermal shock or chemical reaction, and its thermal compatibility until 800 °C, vermiculite has been extensively used as material for high temperature compressive gaskets in many applications (Hoyes et al., 1998; Fergus, 2005; Batfalsky et al., 2006; Dunn et al., 2006; Hoyes, 2007; Wiener et al., 2007; Rautanen et al., 2009).

For a compressive gasket, the leak depends firstly on the path leak at the contact with the support, and secondly on the permeability through the materials if the contact leak is weak enough. The gasket materials need to be compressed at the initial screwing in order to accommodate the geometrical defects (surface roughness) of the support surface. Thus, the materials should possess the ability to strain in order to adopt the shape of the fine roughness of the sealing support. The gasket may also accommodate the small motion of the support during its life cycle. Thus, elastic mechanical properties are required.

The leaks of a pure mica compressive gasket made of paper (particles of 50 μm average 50 μm size) are usually found in the range of 10−1 –10 sccm·cm−1 (i.e., 1.6 · 10−3 –1.6 · 10−1 atm·cm3 ·s−1 ·cm−1) under compressive stress lower than 6 MPa and at temperature in the range of 25–800 °C. The tightness of the compressive mica gaskets is known to be controlled by the contact between the sealing material and the metallic or ceramic surface support. In order to improve the contact at the interface, flexible soft layers made of melted glass (Bram et al., 2004) or silver (Chou and Stevenson, 2009) on the mica surface have been developed and tested. As an example, the infiltration of phlogopite mica by Bi(NO3)3 or H3BO3 glass has successfully decreased the leak rate to almost 5 · 10−4 sccm·cm−1 (i.e., 8.3 · 10−6 atm·cm3 ·s−1 ·cm−1) after 15 cycles in the range of 100–800 °C (Chou and Stevenson, 2004). The last generation of vermiculite based gaskets developed by Flexitallic Company (“Thermiculite® 866”) was pretended to be prepared without any organic binder (Hoyes, 2007; Hoyes and Rautanen, 2013). They contain few talc particles and can resist up to 800 °C (Hoyes, 2007). Upon 100 mbar of hydrogen and a compressive stress of 4–8 MPa, the leak rate was found to be close to 1–3 · 10−1 atm·cm3 ·s−1 (Rautanen et al., 2009). Upon 15 mbar of a mixture of N2–H2 (50/50 vol.) and a compressive stress of 0.4 MPa, the leak rate per unit length of gasket was found to be close to 10−4 atm·cm3 ·s−1 ·cm−1 (Rautanen et al., 2014).

In a previous work (Nguyen et al., 2014), the possibility to obtain vermiculite materials by uniaxial pressing of sonicated micrometric powders was demonstrated without any binder addition. In this paper, we have investigated the sealing properties of this new vermiculite gasket made of pressed small vermiculite particles (micronic and submicronic) without any binder. The sealing properties were studied in relation with the material texture. The permeability leaks (in plane direction and out of plane direction) and the surface leaks were determined as a function of pressure and the thermal treatment temperature. Moreover, some mechanical properties were studied (compression ratio and resiliency) in order to better understand their impact on the leak rate.

2. Experimental

2.1. Vermiculite powders

The starting vermiculite (Granutec E originating from Yuli China) was purchased from CMMP French Company and was used as received (millimetric plates). Potassium chloride (99%, Chimie Plus) was used to saturate the vermiculite before sonication. Hydrogen peroxide (H2O2, 35%, ACROS) was used to prepare suspensions of vermiculite. After potassium exchange, the average chemical composition of half a lattice cell calculated from elemental analysis was (Si3Al1) (Mg2.62Fe0.32Ti0.06) O10(OH)2K0.61.

The K-vermiculite material was chosen because a weakly hydrated initial material was required in order to improve the stability in further heat treatment and in particular to avoid water release from the gasket during heat treatment. Moreover, previous studies have shown that Kvermiculite can be easily delaminated and micronized by sonication at 20 kHz in hydrogen peroxide solution (Nguyen et al., 2013) and then compacted in solid materials under pressure without any binder (Nguyen et al., 2014).

The vermiculite (0.4 mm mean diameter size, weighted amount of 0.55 g or 3.85 g) was sonicated at room temperature in 55 mL hydrogen peroxide (35%) in a “Rosett” type glass double-jacketed reactor cooled at 25 °C by circulation of a cryogenic fluid, using a Sonotrode (20 kHz, 350 W, Ti ultrasonic probe, Sonics and Materials, 43 mm amplitude, 56 W acoustic power) in order to produce small particles (Nguyen et al., 2013; Ali et al., 2014; Nguyen et al., 2014). The solid/liquid (S/L) ratio and the sonication time were varied in order to modify the particle size distributions. Powders were obtained by sonication of vermiculite dispersions prepared at two different S/L ratios (7% and 1%) during different sonication times (1 h, 5 h and 12 h).

The particle size distributions of the dispersions were measured with a Mastersizer 2000 particle size analyser (Malvern Instruments, range 0.02 μm to 2000 μm).

2.2. Elaboration and structure

The vermiculite powder (about 500 mg) was pressed at room temperature or at 200 °C under various pressures for 3 h (Nguyen et al., 2013, 2014) in the form of cylinders (13 mm external diameter, about 2 mm thickness), either as a full pellet for measurement of the out of plane leak or including a central hole of 5 mm diameter for measurement of the radial leak. For hot-pressing, the powder was first pressed at room temperature for 15 min. The pressure was maintained as the mould was heated at 200 °C (4 °C/min, for 45 min). After 2 h at 200 °C, the pressure was released and the sample was turned out of the mould. The formed materials were then thermally heat treated in the range of 400 °C–1000 °C for 10 h in a muffle furnace (4 °C/min).

2.3. Measurement of total leak rate and in plane permeability leak rate

The gaskets were introduced in the circular groove of a metallic immovable holder, constituting a part of a mould manufactured in a nickel based superalloy (NiCr20Co13Mo4Ti3A, named Waspaloy) and pressed between the two faces of this mould (Fig. 1). A metallic porous cylinder from the same nickel based alloy was introduced into the hole of the gaskets (Fig. 1). The complete mould device and gaskets were then compressed at 35 MPa using an MTS press (Insight 50) for 1 h for the room temperature test or for at least 7.16 h for the test at 800 °C (including 1 h at room temperature, 2.66 h min for the heating ramp, 0.5 h for the plateau at 800 °C, and 3 h or 18 h for the cooling to room temperature).

Before measurements the internal volumes and the porous metallic cylinder were degassed under a primary vacuum. A constant pressure of 5 bar of helium (99.95% purity) was applied in the centre of the drilled gasket (Fig. 1). After the helium injection, the amount of gas crossing radially the drilled gasket was measured using a helium mass spectrometer (ASM% 181 T2H Alcatel) for determining the radial leak rates. In order to measure the permeability in plane leak only, two latex films (12 mm external diameter, 5 mm internal diameter, 0.12 mm thickness) were placed at both the upper and lower gasket faces so that the contact leak was suppressed at the gasket–support interfaces (Fig. 1).

Figure 1

Fig. 1. Schematic section representation of the gasket holder setup for the measurement of the total radial and the in-plane leak rates.

2.4. Out of plane permeability leak rate

The design of the mould device for determining the out of plane leak rate was similar to the one in Fig. 1 except that a full cylindrical pellet (gasket) was introduced in a complete mould device also compressed at 35 MPa using an MTS press (Insight 50).

Latex circular films (12 mm external diameter, 3 mm internal diameter, 0.12 mm thickness) were placed at the two gasket–support interfaces. The axial leak was determined using a Helium mass spectrometer detecting the helium crossing the pellet perpendicularly to the in plane direction.

2.5. Compressibility and resiliency

Compression tests were carried out in the mould device previously described (Fig. 1). The complete mould device including the gasket was compressed using an MTS press (Insight 50) between 4.5 MPa and 35 MPa. The thickness of the gaskets was measured using a video sensor (5 μm precision) from the distance between the upper mobile part and the lower immovable part of the mould as a function of strain (measured by a pressure sensor). A cycle of compression and pressure release (speed rate of 1 mm/min) was applied to the mould device. The initial thickness L0 and the thickness after pressure release L3 were taken at 4.5 MPa (Fig. 2), in order to remove from the strain versus length response, both the contributions of the clearance between the gasket size and mould holder dimension, and of the surface defects (known to occur at low pressure). After the compression, the gaskets were maintained typically for 2 h at 35 MPa in order to estimate the creep (L2 − L1). The values of creep were taken into account to determine the compressibility and resiliency.

The resiliency (R) and the compressibility (C) were determined using the following relations where L0 is the initial thickness at 4.5 MPa, L1 is the thickness after compression at 35 MPa, L2 is the thickness after 2 h at 35 MPa, and L3 is the thickness at 4.5 MPa after the pressure release, respectively:

A typical curve for compression–release cycle is shown in Fig. 2.

A typical curve for compression–release cycle is shown in Fig. 2.

Figure 2

Fig. 2. Evolution of the thickness versus pressure for a cycle of pressing and unloading.

Figure 3

Fig. 3. Schematic representation of the sample holder setup designed for measurement of leak rates and compression test at high temperatures.

2.6. Leak rates and compression properties at high temperature

The sample holder shown in Fig. 1 was adapted to be used in a setup (Fig. 3) designed for measurement of the leak rates at high temperatures (until 800 °C). After introduction of the gasket at room temperature, the device was compressed at 35 MPa using an Instron (500 kN) press. The in plane total leak rates were measured as a function of the temperature following the protocol described in Section 2.3. In typical experiments, the leak rate was first measured at room temperature and then the temperature was increased progressively at a constant speed (5°/min) to 800 °C by heating the furnace shown in Fig. 3. The leak rate was measured simultaneously with the temperature while heating, during the plateau at 800 °C, and while cooling. Compression tests were performed at 800 °C using the procedure described in Section 2.5 at 800 °C after stopping helium injection to determine the compressibility and the resiliency exactly at this temperature.

3. Results and discussion

3.1. Total leak rate evolution with granulometry

The granulometric distribution of the sonicated vermiculite powder used to prepare the gaskets by pressing was varied in order to investigate the effect of the particle size distribution on the total leak rate. Up to our knowledge, the effect of particle size on the leak rate of compressed pure clay mineral was not investigated in detail. A work from Nam et al. (2009) focused on the influence of exchangeable interlayer cations (Na+, Li+, Mg2+, Ca2+, Al3+, Fe2+, and Fe3+) on the barrier properties of self standing films of montmorillonite. They reported a lower permeability of the pure clay-mineral films made of monovalent interlayer cations such as Na+ or Li+, in relation with their surface properties, the swelling of the clay particles and their de-aggregation in water. This de-agglomeration led to few micrometre particles observed in water by laser granulometry for monovalent interlayer cation material instead of a wide distribution (1–400 μm) of size for materials exchanged with other cations. In agreement with this previous work (Nam et al., 2009), the gaskets prepared from raw K-vermiculite with millimetric particles (average diameter of 400 μm) show higher leak rate (about 2 · 10−2 atm·cm3 ·s−1 ) than the more solid (cohesive) gasket materials obtained by pressing micronic size particles (10−3 atm·cm3 ·s−1 leak rates).

From the literature, it is known that barrier materials can be developed by filling a polymer film with impermeable plates of clay mineral (mica, vermiculite, or montmorillonite) aligned in the plane of the film (Ward et al., 1991) and that their permeability leak rate depends on the aspect ratio of the flakes (i.e., the ratio of the average length of the platelets to the average width) and the volumic proportion of flake in the composite but is not dependent on the flake size (DeRocher et al., 2005). Moreover the permeability is related to the orientation of the platelets of clay mineral with respect to the in plane direction of the polymer film (Choudalakis and Gotsis, 2009). However, the equations established for the permeability of the composite are not valid for high volumic fraction of clay mineral and cannot be applied for our gaskets which contain only vermiculite flakes.

K-vermiculite gaskets obtained from 1 h sonicated particles (d50 = 2.5 μm, d90 = 6.9 μm) exhibit an average leak rate close to 5 · 10−3 atm·cm3 ·s−1. Additional reduction of the particle size and widening of the particle size distribution were attained by increasing the sonication time at S/L = 1% to 5 h (d50 = 1.6 μm, d90 = 6.7 μm). A widening of the particle size distribution was also obtained by increasing the sonication time to 12 h (d50 = 4.4 μm, d90 = 13.9 μm) or at S/L = 7% to 5 h (d50 = 2.5 μm, d90 = 11.5 μm). The gaskets prepared from these three granulometries (sonication time of 5 h or 12 h) have exhibited 10−3 atm·cm3 ·s−1 average leak rates. In agreement with our result, Lape et al. (2004) have brought out that the greater polydispersity (larger distribution size) of mica flakes in barrier films made of composite polymer/clay mineral results in a smaller permeability. For example, they have shown that a film with 3 vol.% of 5 μm and 3 vol.% of 50 μm flakes has a diffusional resistance 20% higher than a film with 6 vol.% of only 5 μm flakes.

Among twelve tested gaskets (pressed from sonicated powders exhibiting d50 = 1.6 μm, d90 = 6.7 μm), 75% of them showed leak rates lower than 10−3 atm·cm3 ·s−1. As a conclusion, the leak rates of the gaskets were found to decrease both with the reduction in the particle size to micron dimension, and with the widening of the particle size distribution that might enable compact particle arrangement in which the small particles could fill the larger pores. The dependence of the leak rate with powder particle size might also be related to the decrease of the surface roughness together with the size reduction.

3.2. Influence of heating temperature on the leak rate

The leak rates were studied at 35 MPa stress and 5 bar helium pressure, for the gaskets pressed at 200 °C and 80 MPa, and further heated in the range of 200–1000 °C. The leak rate was found to be constant and equal to 10−3 atm·cm3 ·s−1 for heating temperatures lower than 800 °C. At T ≥ 800 °C, the leak rate attained 10−2 atm·cm3 . Moreover, the time for leak stabilization was clearly dependent on the annealing temperature (Supplementary materials). The higher stabilization times were observed for samples heat treated at 200 °C or pressed at room temperature. The stabilization times were about 30–150 min for samples pressed at 200 °C and only 20 min and 2 min for gaskets heated at 700 °C and 900 °C, respectively.

The leak paths of permeability, determined by the porosity and the texture (pore dimension, pore connection, tortuosity, etc.) are complex 3D paths over the material made up of a lot of segments parallel to the pressed surface and few segments perpendicular to the pressed surface, as observed by SEM on ultra-microtomic sections (Fig. 4). By contrast, the paths for the contact leak depending on the interface nature between the gasket and the support (defects of surface, rugosity, softness ...) are expected to be less tortuous. As a consequence, these less tortuous paths for the contact leak lead to a quicker stabilization of the leak value. This case was observed for the heat treated gaskets (at 700 °C and 900 °C) which have showed quick leak stabilization, due to their inflexibility inducing an impaired contact between the gasket and the metallic support. By contrast, the gaskets pressed at 200 °C and 25 °C were more flexible, and have fitted with the support surfaces so that the interface leak was limited and took longer to stabilize.

Figure 4

Fig. 4. SEM image showing some possible leak paths (red lines). (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

The decrease in the stabilization time of the leak as the heat treatment temperature was increased has brought out that the leak was determined by the contact at the interface, as previously reported for mica compressive gaskets (Chou et al., 2002, 2003).

3.2.1. The contact and permeability leaks

The leaks of permeability were determined from the leak tests operated on gaskets coated at their two interfaces by latex films (two latex films were cut and disposed at each gasket–support interface). Prior to these experiments, the latex efficiency to stop the leak at the interface was checked on a stainless steel gasket, limiting the leak rate to 3 · 10−8 atm·cm3 ·s−1. The contribution of the contact leak was estimated from the difference between the leak rate obtained from the uncoated gasket and the one obtained from the same gasket coated with latex film at their interface. The evolution of the leak rate profiles with time (for the permeability and global leaks) shows a very quick evolution from the starting point of the test and a rapid stabilization (2.5 · 10−3 atm·cm3 ·s−1) for gaskets heated at 400 °C and 600 °C (Fig. III in Supplementary materials). The total leak rate (9 · 10−3 atm·cm3 ·s−1) was the lowest (Supplementary materials) for the sample pressed at 200 °C. The slight decrease might be attributed to the creep, which could re-arrange the particles provoking the filling up of the defects at the interface. The sample pressed at 200 °C, only, displays a typical profile for the permeability leak time dependence, in which the leak increases slowly together with the time. For thermally treated gaskets (at 400 °C and 600 °C), the permeability leak time dependences show a quick increase followed by a weak decrease before stabilization. This was attributed to the formation of cracks observed by optical microscopy at the surface of samples (not shown) tested with latex deposit at their faces. The observed cracks were less than 0.2 mm thick with various lengths. The cracks were found to be radially aligned and have promoted the quick stabilization of the permeability leak. But, they might have been blocked up under stress (by creep), or by the latex films at the interface, leading to a slow and weak decrease of the leak after a long time. The highest flexibility of the nonannealed gasket (pressed at 200 °C) has prevented it from fissuring

The difference between the global leak and the radial permeability leak gives the contact leak (continuous lines in Fig. 5). Fig. 5 shows clearly that the contact leak is increasing together with the heat temperature due to the enhanced stiffness of the prepared gaskets. The radial permeability values estimated from Darcy's law taking into account the radial geometry (dashed lines in Fig. 5) are quite similar for gaskets treated in the range of 200 °C–800 °C in agreement with their almost constant porosity (~30%).

Figure 5

Fig. 5. Evolution, versus the heat treatment temperature of the gaskets pressed at 80 MPa (particles obtained by 5 h sonication at S/L = 7%, d50 = 2.5 μm, d90 = 11.5 μm), of the leak rates (continuous lines) for contact leak (square) and in plane (radial) permeability leak (triangle), and of the permeabilities (dashed lines) measured along the in plane (empty triangle) and the out of plane (empty circle) directions.

Up to our knowledge, the values of the leak rate of permeability or permeability are not given in the common literature about clay mineral based gaskets because the measurements are always affected by a surface leakage which cannot be neglected. Chou and Stevenson (2009) found a 3.2 · 10−3 atm·cm3·s−1 leak rate for mica compressive seals at 800 °C of which surface leak was strongly decreased by a coating of silver. This leak rate value at 800 °C which might be considered as a permeability leak because of the absence of surface leakage is ten times the one found (1.5 · 10−4 atm·cm3·s−1) at room temperature for our vermiculite seal pressed at 200 °C but the disagreement might be attributed to the different working temperatures. The leakage rate values and the dimensions for the circular gaskets mentioned by Hoyes and Rautanen (2013) and Rautanen et al. (2014) indicate that the radial permeability of a vermiculite (i.e., “Thermiculite® 866”) based gasket should be lower than 10−16 m2 under 10 MPa at 1 bar pressure of He, in agreement with our lower radial permeability values (2 · 10−19 m2). Moreover, this radial permeability value agrees with the values reported for natural clays at nitrogen or air pressure of about 5 bar (Yang, 2008) or intrinsic permeability values of CallovoOxfordian clay (Enssle et al., 2011; Shao et al., 2011). This permeability value (~2 · 10−19 m2 ) was higher than the one of the exfoliated graphite pressed at 80 MPa measured in the same conditions (35 MPa, 5 bar He) at 10−20 m2.

3.2.2. In plane and out of plane permeability

The cylindrical symmetry of the porous network of the pressed vermiculites and its anisotropy indicate (Nguyen et al., 2014) that the pores are mainly connected along the radial in plane direction, as for expanded graphite material (Mauran et al., 2001; Celzard et al., 2005). Such an arrangement of the pores is expected to give a strong anisotropy in the permeability values (Shao et al., 2011; Belhouideg and Lagache, 2014). The in plane permeability and perpendicular permeability were measured (see Sections 2.3 and 2.4) for gaskets pressed at 200 °C (and 80 MPa) and for the same gaskets annealed at 400 °C and 600 °C.

Fig. 6 shows the evolution of the in plane and out of plane permeability leaks. The in plane leak rates are almost similar (10−4 atm·cm3·s−1) whatever the gaskets. For the sample pressed at 200 °C, the leak rate increased slowly with time before stabilization. For the annealed gaskets (400 °C and 600 °C), the in plane leak rate values decreased slightly with increasing time after 27 min, before a plateau value was attained due to the creep or cracks blocking up, as explained previously. Indeed, radial cracks were observed using optical microscopy at the surface of the gaskets heated at 400 °C or 600 °C. By contrast, the out of plane leak rate profiles showed a slow increase up to their stabilization. The intrinsic in plane and out of plane permeabilities (κa and κc, respectively) were calculated from the Darcy law (dashed lines in Fig. 5). The leak rates measured axially were found to be lower (10−6 –10−9 atm·cm3·s−1) than the in plane leak rates, as expected for such an anisotropic material. The anisotropy ratio, κa/κc ratio, for vermiculite pressed at 200 °C was found to be equal to 10 in the same order than the value of 8 found for the natural Callovo-Oxfordien clays observed in the underground laboratory of Bure site (Haute-Marne, France) (Shao et al., 2011). The κa/κc ratio values have increased together with the temperature. This increase in anisotropy might be related to the linkage of the vermiculite particles during heating. Indeed, SEM observations have shown that the vermiculite aggregates and pores were mainly oriented along the in plane direction (Nguyen et al., 2014). The connection of the particles in this direction should either increase the tortuosity of the out of plane leak path or block this leak path, without impacting greatly the in plane leak path. The vermiculite gasket permeability anisotropy ratios were found to increase while the materials were densified, as for flexible graphite in which the anisotropy increases together with the density (Han et al., 1998; Mauran et al., 2001; Biloe and Mauran, 2003; Celzard et al., 2005; Wang et al., 2010).

Figure 6

Fig. 6. Evolution versus time of the permeability leak rates measured along in plane (full symbol) and out of plane (empty symbol) directions for gaskets pressed at 80 MPa (particles obtained by 5 h sonication at S/L = 7%, d50 = 2.5 μm, d90 = 11.5 μm), and heat treated at 200 °C (diamond), 400 °C (triangle) and 600 °C (square).

3.3. Evolution of the leak rate with the forming pressure

The radial (in plane) permeability was measured for gaskets pressed in the range of 17.7–80 MPa in order to study its variation with the gasket porosity. Fig. 7 shows that the permeability leak increased together with the porosity, for porosity lower than 31%. As the porosity attains value lower than 31%, the increase in the permeability leak versus porosity tends to be weaker. This result can be explained by the filling up of dead-end pores for a pressure higher than 53 MPa. At this pressure, the pores connected and open on the two opposite surfaces (internal and external) might start to be closed or reduce their size so that the leak would decrease significantly. As previously reported (Nguyen et al., 2014), the material exhibits a pore size distribution including exclusively macropores (~100 nm) and mesopores (10 nm and 30 nm). The increase in the pressure reduced the proportion of macropores and increased the proportion of mesopores because the macropores were transformed in mesopores (Nguyen et al., 2014). Fig. 7 (permeability versus the volumic percent of pores) brought out that above a proportion of mesopores of 54% attained for P >53 MPa, the permeability decreased sharply. In fact, as increasing the pressure, the pore diameters were reduced and attained a median pore diameter (estimated from mercury porosimetry) lower than 30 nm. This reduction has promoted the decrease of the permeability. This might be related to the hindrance of helium diffusion due to the longer, more tortuous, more shrinked, and rarer leak path.

Figure 7

Fig. 7. Evolution of the in plane permeability versus porosity (triangle), versus macroporous volumic fraction (diamond), and versus average pore size (square) for several forming pressures (particles obtained by 5 h sonication at S/L = 7%, d50 = 2.5 μm, d90 = 11.5 μm).

3.4. Relation between mechanical properties and leak rates

The sealing properties can depend mainly on the surface nature but they can also be influenced by the mechanical properties of the seal. Indeed, we have previously shown that the stiffness of the heat treated materials induced an increasing surface leak. The resiliency and the compressibility determined for a series of gaskets heated up to 800 °C (Supplementary materials) have shown that the whole leak rate increased as the compressibility and resiliency of the materials increased together with the heat treatment temperature.

For several gaskets prepared by pressing at 200 °C and 80 MPa from K-vermiculite sonicated particles having the same particle size distribution (d50 = 1.6 μm, d90 = 6.7 μm), an important variability was observed for compressibility and resiliency in the range of 2–30% and 2–16% (Fig. 8), respectively, as the properties depend clearly on the way the seals are introduced into the flange.

Fig. 8 has also shown the evolution of the leak rate as a function of the resiliency and compressibility. The leak rate evolution versus the compressibility did not show a clear tendency. By contrast, leak rates decreased as the resiliency increased. For resiliencies higher than 6%, the gasket exhibited leak rates lower than 10−3 atm·cm−3 ·s−1. This means that the quality of the contact between the support faces and gasket has depended on the resiliency which has promoted the recovering of the defects (defects of flatness) at the interface.

Figure 8

Fig. 8. Evolution of the leak rate as a function of resiliency (diamond) and of compressibility (triangle) for several gaskets pressed at 80 MPa and 200 °C (particles obtained by 5 h sonication at S/L = 1%, d50 = 1.6 μm, d90 = 6.7 μm).

3.5. Sealing properties at high temperatures

The sealing and compression properties were tested between room temperature and 800 °C for gaskets pressed at 200 °C under 80 MPa, originating from particles sonicated for 5 h at two S/L ratios (S/L = 1% and 7%) yielding two distributions of particle size (d50 = 1.6 μm–d90 = 6.7 μm and d50 = 2.5 μm–d90 = 11.5 μm). The radial leak was determined using an especially designed sample holder setup (Fig. 3) and the conditions previously described (Section 2.6). The compressibility and the resiliency were determined exactly at 800 °C (see Section 2.6). Moreover, the compressing mould (Fig. 3) was removed from the sample holder and used for compression test (see Section 2.5) to determine the compressibility and the resiliency before and after the sealing test up to 800 °C.

We have tested samples made of two vermiculite powders characterised by different granulometries (d50 = 1.6 μm and d50 = 2.5 μm). Four samples were prepared from each powder according to the process mentioned in Section 2.2, and then tested for their sealing and compressive properties in the range of 24 °C–800 °C. The leak rate versus temperature (Fig. 9) exhibited exactly the same evolution whatever the sample type. The temperature (Fig. 9, continuous line) was maintained at 25 °C until the stabilisation of the leak, and then increased at 800 °C (plateau) and then decreased. The leaks increased at room temperature and stabilized after less than 30 min. The leak rate kept stable at 10−3 atm·cm3·s−1 as heating to 800 °C, and also after cooling at room temperature. Thus it was concluded that the leak was not dependent on the temperature in the range of 25–800 °C. The ratio between the contact leak and the permeability leak for the as made gaskets (prepared at 200 °C) was equal to about 6 (Fig. 5), indicating that the leak was mainly controlled by the contact. The absence of evolution of leak rate level as heating (up to 800 °C), suggested that the leak was only controlled by the contact with the support and not by the porosity of the gasket materials, and that the contact was not varying with the temperature. Previous studies have shown that gasket porosity was very slightly decreasing as heating from 200 °C to 800 °C (less than 5% decrease of the permeability) so that the permeability leak should have decreased while heating.

Figure 9

Fig. 9. Evolution of the helium leak rate (PHe = 5 bar) as a function of time and temperature in the range of 25 °C–800 °C for gaskets pressed at 80 MPa and 200 °C from micrometre sized vermiculite of different particle distribution sizes: d50 = 1.6 μm–d90 = 6.7 μm (black diamond) and d50 = 2.5 μm–d90 = 11.5 μm (empty square). The temperature evolution versus time is shown by the continuous line.

The same behaviour was found for Thermiculite® commercial gaskets (Rautanen et al., 2009) for which a 6 × 10−2 atm·cm3·s−1·m−1 leak rate was reported in the temperature range of 25–600 °C at 5 bar He, under 20 MPa. Using the same unity, for our gaskets (circumference 40.8 mm) the leak rate value equal to 2.5 × 10−2 atm·cm3·s−1·m−1 is comparable to the one of Thermiculite® commercial gaskets. Moreover, the test of a sample referred to as “Thermiculite® 866” on our test setup gave a similar leak rate than the as made gaskets from pressed sonicated vermiculites.

We have also shown that similar leak rates (10−2 atm·cm3·s−1) were found in the whole studied temperature range both for gaskets pressed at 53 MPa instead of 80 MPa, or pressed at room temperature at 80 MPa though these latter gaskets were more fragile and difficult to handle.

For the heat treated vermiculite gaskets (pressed at 200 °C and then heated at 400 °C, 600 °C and 800 °C), the total leak rates measured at 800 °C are also similar to the rate measured at room temperature confirming the predominance of the contact leak.

More than three heating–cooling thermal cycles in the range of 25–600 °C were applied to the gaskets while the sealing properties were measured. After each test cycle, the upper part of the gasket mould was taken apart easily and no deposit was found both on the surface of the upper gasket mould and on the surface of the gaskets. The leak measured for the first cycles was around 10−3 atm·cm3·s−1, and it increased slightly to 2 × 10−3 atm·cm3·s−1 for both the second and third cycles. This small increase of the leak rate value can be attributed to the increase of gasket rigidity of the surface after first heating at 600 °C, which might increase the contact leak after removing and remounting the upper support.

The values of the compressibility (~9%) and resiliency (~5%) measured at room temperature (Table 1) decreased together with the heat temperature during the sealing test, because the as prepared gaskets pressed at 200 °C were modified by the heat treatment. The heat treatment induced the sintering of materials above 700 °C, the transformation of vermiculite into 3D structure up to 850 °C, and its hardening increase (Ramirez-Valle et al., 2006; Marcos et al., 2009). The 800 °C as prepared gaskets showed higher values of resiliency (~3%) and compressibility (~5%) than the tested gaskets because they have undergone pressure sintering due to extreme conditions of pressure (35 MPa) and temperature (800 °C). This clearly means that the annealing of the gaskets above 200 °C, before their use, is degrading their sealing and mechanical characteristics. Moreover, as heating up to 450 °C, under a working pressure of 35 MPa, the gaskets begun to compress (decrease of their thickness) because of the densification induced by pressure and temperature. Compared to room temperature conditions, the gaskets lost 5% of their initial thickness (about 2 mm) after heating at 800 °C, under 35 MPa.

4. Conclusion

The gaskets prepared from compressed sonicated micrometric vermiculite powder exhibited 2.5 × 10−2 atm·cm3·s−1 ·m−1 helium leak rate under 35 MPa working pressure and 5 bar helium pressure from room temperature to 800 °C. The total leak was the sum of two components: the contact leak and the radial permeability leak. The contribution of the contact (interface) leak to the global leak was shown to be predominant. The tendency of contact leak to increase together with the annealing temperature of the materials (10−2 atm·cm3·s−1 leak rate in gaskets annealed at 800 °C) was explained by the increase of stiffness of the gaskets due to structural transformation of materials (phase transformation of vermiculite: dehydroxylation and 3D structure oxide formation; and sintering) induced by heat already evidenced by Nguyen et al. (2014). The loss of flexibility of the gasket surface prevents the materials to compensate mechanically for the flatness defects of the sample holder in order to insure a tight contact. The contact leak can also be increased by the formation of cracks on the surface of the more rigid materials (gaskets heat treated at T N 600 °C). The permeability leak (10−4 atm·cm3·s−1) for the ~30% porosity gaskets gave a negligible contribution to the total leak (10−3 atm·cm3·s−1). The permeability leak was found to depend more on the pore size distribution than on the porosity. Indeed, we have shown that the main presence of small pores (mesopores) tends to decrease the permeability leak. Thus, the intrinsic in plane permeability value decreases strongly for a forming pressure higher than 53 MPa, which allows the formation of average pore size lower than 30 nm. The analysis of the in plane permeability (perpendicular to the pressing axis) and out of plane permeability (parallel to the pressing axis) temperature dependences has confirmed the anisotropy of the vermiculite gaskets. The out of plane permeability (κa = 1.8 × 10−20 m2) was found to be 10 times lower than the in plane one (κc = 1.8 × 10−19 m2) for gaskets pressed at 200 °C and 80 MPa. The κac ratio of the permeabilities became 16,000 for the same gaskets annealed at 600 °C, which might be explained by the linkage of the vermiculite particles during heating (related to the dehydroxylation of the layer edge), as shown by Nguyen et al. (2014) by scanning electron microscopy analysis, that could limit the gas transport. The values of in plane intrinsic permeabilities were found to be in agreement with the ones of compacted clay soils (Yang, 2008; Enssle et al., 2011; Shao et al., 2011) in the range of 10−19–10−20 m2. At room temperature, we have found a correlation between the decrease of the leak rate and the increase of the resiliency. For resiliency values higher than 6%, gaskets exhibited total leak rates lower than 10−3 atm·cm3·s−1. This means that the quality of the contact between the mould and the seal material partially depends on the flexibility of the materials measured by resiliency. On the other hand, no clear relation was identified between the leak rate and the compressibility. The quality of the contact was not modified by heating from room temperature to 25 °C and after further heating–cooling thermal cycles (3 cycles). The gaskets pressed at 200 °C and 80 MPa possessed maximum values of resiliency (~5%) and compressibility (~9%) at room temperature before they have undergone a thermal treatment. Both values of resiliency and compressibility decreased to almost 1% at 800 °C because at this temperature the vermiculite was transformed into tridimensional structures and the materials started to sinter. These vermiculite gaskets were compatible with the metallic sample holder and were detachable and reusable after a thermal heating cycle of up to 800 °C. All the characterised properties (mechanical and sealing properties, thermal resistance, inertness) of the gaskets made by compressed sonicated vermiculite allow their use in industrial applications such as solid oxide fuel cell sealing, oil chemistry applications and nuclear industry.

Table 1

Evolution of resiliency and compressibility of the gaskets pressed at 80 MPa and 200 °C after a thermal heating and cooling cycle from room temperature to 800 °C.

Gasket type (powder type) Before the sealing test At 800 °C After sealing testing
Compressibility (%) Resiliency (%) Compressibility (%) Resiliency (%) Compressibility (%) Resiliency (%)
Pressed at 200 °C (d50 = 1.6 μm, d90 = 6.7 μm) 8.9 ± 0.8 4.9 ± 1.0 - - 5.1 ± 1.5 2.5 ± 0.5
Pressed at 200 °C (d50 = 2.5 μm, d90 = 11.5 μm) 9.1 ± 1.0 5.5 ± 0.3 ~1.2 ~1.2 4.9 ± 0.8 3.7 ± 0.7
Pressed at 200 °C and treated at 800 °C (d50 = 2.5 μm, d90 = 11.5 μm) 4.9 ± 1.0 2.8 ± 1.0 - - 1.0 ± 0.8 1.0 ± 0.8

Download the original article as a pdf.