Research Article www.acsami.org
Characterization of Pores in Dense Nanopapers and Nanofibrillated Cellulose Membranes: A Critical Assessment of Established Methods Paola Orsolini,†,§ Benjamin Michen,†,‡ Anja Huch,† Philippe Tingaut,† Walter R. Caseri,§ and Tanja Zimmermann*,† †
Empa − Swiss Federal Laboratories for Material Science and Technology, Applied Wood Materials Laboratory, CH-8600 Dübendorf, Switzerland ‡ Institute for Building Materials and §Multifunctional Materials, ETH Zürich, 8093 Zürich, Switzerland S Supporting Information *
ABSTRACT: Nanofibrillated cellulose (NFC) is a natural fibrous material that can be readily processed into membranes. NFC membranes for fluid separation work in aqueous medium, thus in their swollen state. The present study is devoted to a critical investigation of porosity, pore volume, specific surface area, and pore size distribution of dry and wet NFC nanopapers, also known as membranes, with various established techniques, such as electron microscopy, helium pycnometry, mercury intrusion, gas adsorption (N2 and Kr), and thermoporometry. Although these techniques can be successfully applied to inorganic materials (e.g., mesoporous silica), it is necessary to appraise them for organic and hydrophilic products such as NFC membranes. This is due to different phenomena occurring at the materials interfaces with the probing fluids. Mercury intrusion and gas adsorption are often used for the characterization of porosity-related properties; nevertheless, both techniques characterize materials in the dry state. In parallel, thermoporometry was employed to monitor the structure changes upon swelling, and a water permeance test was run to show the accessibility of the membranes to fluids. For the first time, the methods were systematically screened, and we highlighted the need of uniform sample treatments prior to the measurements (i.e., sample cutting and outgassing protocols) in order to harmonize results from the literature. The need for revising the applicability range of mercury intrusion and the inappropriateness of nitrogen adsorption were pointed out. We finally present a table for selecting the most appropriate method to determine a desired property and propose guidelines for results interpretation from which future users could profit. KEYWORDS: porosity, specific surface area, pore size distribution, mercury intrusion, gas adsorption, thermoporometry, nanofibrillated cellulose, membranes
1. INTRODUCTION
the development of water barrier papers, showing the need for intercalating the nanofibrillated cellulose with layered silicates to achieve better performances in accordance to a fiber−brick composite model, combined with a new native network model and cover fiber composite model:12 when adding layered silicates into a NFC matrix, the water vapor permeance (WVP) is lowered to 2.2 g/(m day Pa) from 6.4 g/(m day Pa) for neat NFC films. The same concept was shown by Wu et al.11 with regards to oxygen-barrier properties. Recent studies14−20 propose the use of NFC for the preparation of neat and composite membranes, targeting mainly ultra- and nanofiltration operations. NFC membranes can easily be prepared by water removal from NFC suspensions; for instance with a vacuum filtration process, the
Nanofibrillated cellulose (NFC) is a biobased material that was pioneered by Turbak et al.1 and subsequently studied over the last three decades. NFC is extracted from plant cellulose pulps or cellulose wastes, making it an abundant, renewable, and sustainable material. It is isolated by wet processing methods, such as high-pressure homogenization or grinding, which lead to a gel-like substance composed of an entangled network of cellulose nanofibers. Such a material is frequently referred to as nanopaper. The properties and applications of NFC nanopapers have been reported extensively in a series of reviews.2−8 One of the main applications can be found in the use of NFC as barrier material for packaging because the material displays low oxygen- and water-vapor-transmission rates9−13 due to the high packing ability of nanofibrillated fibers and the existence of crystalline domains.5,9,10 However, it was proven that by increasing the relative humidity of the nanopaper its barrier ability drops down.10,11 Our group also formerly contributed to © 2015 American Chemical Society
Received: September 4, 2015 Accepted: October 30, 2015 Published: October 30, 2015 25884
DOI: 10.1021/acsami.5b08308 ACS Appl. Mater. Interfaces 2015, 7, 25884−25897
Research Article
ACS Applied Materials & Interfaces fibrils self-assemble upon drying to give a meso- and macroporous structure suitable for separation purposes. The convenience of the process can be found in its simplicity and reproducibility, leading to highly mechanically stable membrane structures in the dry state21,22 with Young’s moduli between 6 and 17.5 GPa and tensile strengths between 100 and 250 MPa.21 NFC membranes possess a more sustainable life cycle and performance comparable to those of membranes made of petroleum-derived materials. A review by Carpenter et al.23 highlights the potential of NFC-derived fabricated membranes in the field of water treatment. NFC gained more relevance over other nanocelluloses types because of better permeances and better mechanical properties.24 Several cellulose-based and -derived materials such as regenerated cellulose (RC), cellulose acetate (CA), cellulose triacetate (CTA), and cellulose nitrate (CN) were among the first materials employed in the development history of membrane processes.25,26 With regards to these cellulose derivatives, NFC-based membranes possess several advantages such as sustainability, the unnecessity of a support layer in the membrane fabrication process and narrower filtration ranges. Similar to cellulose, NFC is insoluble in water but undergoes swelling.27,28 A challenging issue that we address in our work is the characterization of porosity-related properties of NFC nanopapers or membranes in the swollen state compared to those in the dry state. For this purpose, several methods based on different working principles were employed on NFC substrates so far. Permporometry allows for the determination of active pores by forcing different probing fluids through the membranes (e.g., on the basis of liquid displacement, water10,12 and gas permeability,9,11,29 and bubble point measurements). Spectroscopic methods (i.e., proton annihilation30) and scattering methods (i.e., X-ray microtomography8) are also available. In parallel, methods typically employed in membrane science were applied for the characterization of NFC membranes in the swollen state, such as water or solvents flux, permeance, and the molecular weight cut off (MWCO), corresponding to the molecular weight of known compounds whose amount is rejected by 90% through a membrane.19,24 In the present work, we critically assess different established techniques regularly employed for characterizing other porous NFC-based templates (i.e., foams or aerogels) also suitable for fast industrial-quality assessment: electron microscopy, helium pycnometry, mercury intrusion, gas adsorption, and thermoporometry were considered, respectively. Although they are successfully used for other hydrophilic materials such as silica, we analyze under which circumstances they partially fail for NFC membranes because of their more pronounced hydrophilicity, compact structure, and surface chemistry and because of different accessibility and wettability of probing fluids (e.g., Hg, N2, Kr, and H2O). In the present work, we provide information on the density, specific surface area, total pore volume, pore size distribution, and porosity of the NFC-based membranes and compare dry and swollen states. To validate the correct use of the aforementioned techniques, we assess them against inorganic porous silicas used as reference materials with known pore sizes.
volume used for calculations. The American Society for Testing and Materials (ASTM) classifies them as follows:31 theoretical density is derived from the true volume (volume of the solid excluding closed, blind, and open pores) of the material, envelope density is derived from the envelope volume (volume obtained by drawing the contour of a discrete piece of material), skeletal density is derived from the skeletal volume (volume of the solid and the closed and blind pores it may contain), and bulk density is derived from the bulk volume (volume occupied by the solid particles, including all pores and also the interparticle voids). The bulk density changes with the compaction of the material, where the interparticle void space may rearrange by handling the product. 2.2. Mercury Intrusion. The mercury intrusion technique applies to the characterization of pores in the so-called meso and macro range (i.e., between 3 nm and 100 μm). It is based on the use of liquid mercury to probe the pores of solids and largely incompressible substrates. Because the liquid mercury is nonwetting, pressure is needed to force its penetration into the pores. The theoretical development is therefore based on solid/ liquid interactions, the physical properties of mercury, such as surface tension (γ) and contact angle (θ), and mechanical equilibria at the curved interface between the resistive force to penetration of a pore aperture of radius (rp) and the incremental pressure applied for the intrusion (ΔP). The relationship is summarized in the Washburn’s equation (eq 1),32 based on the assumption of a cylindrical pore shape, derived from a modified Young−Laplace’s equation. ΔP =
2γ cos θ rpore
(1)
The pressure needed for the Hg to penetrate the pores apertures is inversely proportional to the pore radii. By increasing the pressure up to 350 MPa, pores down to 3.0 nm can be investigated. In a typical mercury porosimetry experiment, volumes of mercury intruded at each incremental pressure are recorded, whereas the pore radii accessed are calculated through eq 1. The technique provides information about not only the voids present in the substrate (pores) but also interparticle voids present between discrete pieces of the substrate.33 For determining the porosity, the skeletal density measured by helium pycnometry should be known prior to the measurements. Information on the bulk density can be derived after the measurements and further used for determining the porosity of the sample. A hysteresis behavior is normally observed between intrusion (along pressure increase) and extrusion processes (along pressure decrease) because of the pore shape, the tortuosity of the pores, and the substrate surface chemistry. Because impurities might interfere with the wetting behavior, samples must be dried or outgassed prior to the measurements. 2.3. Gas Adsorption. Physical gas adsorption, known as physisorption, is used to probe surface areas and pores. It relies on the enrichment of a gas (adsorptive) on a solid substrate (adsorbent). van der Waals interactions determine the physisorption phenomenon controlling intermolecular forces between the adsorptive and the adsorbent as well as binary interactions among adsorptive molecules. The mathematical formulation of the adsorption process in multimolecular layers applied to porosity, known as BET model, was first derived by Brunauer, Emmett, and Teller in 1938.34 In a typical gas adsorption experiment, an adsorption isotherm is generated by
2. THEORETICAL BACKGROUND 2.1. Density. Different types of densities can be determined for porous materials such as membranes, depending on the 25885
DOI: 10.1021/acsami.5b08308 ACS Appl. Mater. Interfaces 2015, 7, 25884−25897
Research Article
ACS Applied Materials & Interfaces
ing micropores (2 nm). Such samples have a zero intercept. Accordingly, the presence of a positive intercept for the tested sample provides an indication of the existence of micropores. A possible determination of pore volume and pore size distribution from N2 desorption isotherms in the mesoporous range was derived by Barrett, Joyner, and Halenda39 in 1951 and is known as the BJH method. Gas adsorption is carried out until the relative pressure reaches 1; the understanding of the adsorbate condensation process inside pores allows for the characterization of the adsorbent pore size distribution. This method assumes that ideal cylindrical pores are progressively emptied upon relative pressure decrements, exploiting the Kelvin radius rKn for the condensation of vapors inside mesoporous cavities, the thickness reduction Δtn of the adsorbate upon pore emptying, and the areas Ac of the pores from where emptying occurs. The pore size distribution Vpn can be derived by
measuring the amount of gas adsorbed onto the solid at increasing adsorptive pressure and at constant temperature. By decreasing the pressure, the desorption process can be evaluated. To avoid interference with the adsorption process and free all the available adsorption spots at the surface, meticulous cleanings of the samples from any external impurity have to be guaranteed. Normally, the samples undergo severe outgassing protocols at high temperature and reduced pressure. The adsorption isotherms have been classified by the International Union of Pure and Applied Chemistry (IUPAC) into six different types,35 corresponding to different adsorptive/adsorbent interaction strengths and different surface features. Equations 2 and 3 (below) are used to calculate the surface area (SSA):34 p (C − 1) p 1 = a + n (p0 − p) nmC nmaC p0
(2)
SSA(BET) = nmaLam /m
(3)
a
where na is the quantity of adsorptive adsorbed at the relative pressure p/p0, nam is the quantity of the adsorptive needed to cover the adsorbate with one monolayer, and C is a constant varying according to the isotherm shape. The BET specific surface area (SSA(BET)) is then calculated as in eq 3, where L is the Avogadro constant, am is the molecular cross-section area of the adsorptive and m is the mass of the adsorbent. Several gases are available for the adsorption, such as nitrogen (N2), argon (Ar), krypton (Kr), and carbon dioxide (CO2). In the case of N2, am has a value of 16.2 Å2, and the isotherm is generated at the boiling temperature of N2, 77 K. In the range of relative pressures 0.05 < p/p0 < 0.30, the measured adsorption points should result in a straight line in order to satisfy the BET equation (eq 3). However, in some cases the linearity does not extend up to 0.30 p/p0 for samples with homogeneous surfaces of higher energy; in this case, Sing35 advises to report the correct range of linearity. The constant C appearing in both in the intercept and in the slope of the equation depends on the difference between the heat of adsorption of the first monolayer E1, the heat of liquefaction EL, the Boltzmann constant R, and the absolute temperature T. EL is employed because the theory makes the hypothesis that the evaporation−condensation behavior of any additional molecular layer after the first one can be assumed equal to those of the liquid state.34 C ∼ e(E1− E L)/(RT )
j=n ⎛ ⎞2 rpn Vpn = ⎜ ⎟ (ΔVn − Δtn ∑ Acj ) ⎝ rKn + Δtn/2 ⎠ j=1
(5)
The BJH method is commonly used in literature for several materials, and we use it to evaluate the PSD in NFC membranes up to ∼100 nm. 2.4. Thermoporometry. Thermoporometry is a technique that analyzes porous materials in the wet state. The sample is soaked with a probing liquid (e.g. water) that undergoes a shift in freezing and melting temperature (ΔT) when confined in pores with diameters in the submicrometer range. The physical phenomena is explained by the Gibbs−Thomson effect40,41 and allows us to associate ΔT with the pore radius (rp). Brun et al.42 correlated those temperature-phase transition shifts, evident in differential scanning calorimetry (DSC) measurements, to the size and shape of pores. Measuring in an excess of the probing liquid produces the formation of two enthalpy peaks in typical melting or freezing DSC thermograms. One peak is attributed to the excess free water on the surface of the sample (FW) undergoing a phase transition at about 273.15 K, and other peaks at lower temperature are related to water confined or bound within pores (BW). The aforementioned ΔT corresponds to the difference between the peak position of free and bound water and is related to the pore radius. The derivation made by Brun limits the applicability of the thermoporometry to the mesoporous range. Micropores (
