Cellulose ultrastructure, understood as organization in the nanoscale, is critical for cellulose biorefining. In particular, cellulose enzymatic hydrolysis is strongly affected by properties as crystal packing and porosity, which modulate cellulose reactivity and accessibility. In this scenario, proper analytical tools are required to probe cellulose ultrastructure. This work presents developments in three analytical techniques: X-ray diffraction, dynamic vapor sorption, and thermoporometry. ln X-ray diffraction, two-dimensional cellulose diffraction patterns are modeled by the Rietveld method. Quantitative inferences about cellulose degree of crystallinity and crystal width are presented. In dynamic vapor sorption, water sorption kinetics is explored in addition to classical equilibrium sorption isotherms. Potential information in sorption characteristic times is highlighted. Thermoporometry (performed with a differential scanning calorimeter) probes cellulose nanometric pores in waterswollen states. A wide spectrum of cellulose samples were studied by these techniques. Despite the diversity of sources and processing routes, it is shown that celluloses obey some general ultrastructural laws (e.g., hydration primarily determined by crystal width). Finally, analytical costs and throughputs are discussed.
Keywords: Diffraction; porosity; sorption; thermoporometry; ultrastructure
Cellulose is the most abundant biopolymer on Earth. It is presently being considered as a renewable resource to replace substantial shares of materials, chemicals, and fuels derived from fossil carbon Conversion of lignocellulosic feedstocks into a wide range of value added products will require strategies beyond those well established in the pulp and paper industry. New strategies to be developed and deployed in large-scale industrial plants will likely benefit from integrated production in the so-called biorefineries.
Limited understanding of cellulose ultrastructure (organization in the nanoscale) is hampering rational optimization of cellulose conversion routes. Enzymatic hydrolysis of cellulose isa clear example. This processis preceded by a physical-chemical pretreatment to make cellulosic substrates more amenable to enzymatic attack. The ways in which pretreatment changes ultrastructure and how these changes impact enzyme action are still far from understood.
For processing that retains native cellulose crystal phase, cellulose ultrastructure can be summarized as follows. Cellulose forms crystallites. In raw lignocellulose, the dominant crystal phase is lβ with possible minor contribution of lα (Atalia & Vanderhart, 1984; Sturcová et al., 2004). Upon processing, crystals are stabilized (Horii et al., 1987; Debzi et al., 1991) in the lβ structure (Nishiyama et al., 2002). These crystals are fibrillar; they are a few nanometers wide and about a hundred of nanometers long (Nishiyama et al., 2003; Samir et al., 2005; Elazzouzi-Hafraoui et al., 2008; Teixeira et al., 2011). Along cellulose microfibrils, crystals are spaced by disordered regions, estimated to be made of 4-5 sugar residues (2.0-2.5 nm). These disordered regions are easier to hydrolyze in acid medium (Nishiyama et al., 2003). Laterally, crystallites are packed into bundles a few tens of nanometers wide (Hult et al., 2001; Fahlén & Salmén, 2005; Zhao et al., 2007). In wetstate, nanometric water-filled pores exist between these bundles; these pores result primarily from separation of cell wall layers during processing (Fahlén & Salmén, 2005; Stone & Scallan, 1968).
The description in the previous paragraph is a general one. Nevertheles, rational development of cellulose conversion routers requires systematic measurement of the aforementioned properties. In this context, the present work presents analytical developments and results from three analytical techniques are X-ray diffraction, dynamic vapor sorption, and thermoporometry. Detailed information about these techniques can be found in publications by our research group (Driemeier & Calligaris, 2011; Driemeier et al., 2011; Driemeier et al., 2012; Driemeier & Bragatto, 2013).
Celluloses were acquired from commercial suppliers, kindly provided by collaborators, or processed in our laboratory. The goal is to have a heterogeneous sample set and look for general trends and material-specific parameters. From Sigma-Aldrich (catalog code in parenthesis) we acquired Fluka cellulose (22183), Sigmacell type 20 (S3504), Sigmacell type 50 (S5504). Avicel PH-101 (11365), Sigmacell type 101 (S6790), and α-cellulose (C8002). In addition, we acquired Celufloc 200 (from Celuflok), Whatman #1 filter paper, and two bleached eucalyptus kraft pulps (from Brazilian mills). A bleached eucalyptus pulp produced in subcritical ethanol-water-Cü, mixture was kindly provided by Dra. M. T. B. Pimenta. Peraceticpulps from sugarcane bagasse were produced in a 1:1 mixture of 8.74 M glacial acetic acid and 21.6 M hydrogen peroxide at 60°C for 15,24, or 48 h. Following the sequence in which they are aforementioned, these 14 materials are here named Fluka, S20, S50, Avicel, S101, Alpha, Floc, FP, Ekp1, Ekp2, Esc, Bpa15, Bpa24, and Bpa48. According to compositional analysis, these samples have negligible noncarbohydrate content and cellulose content between 0.68-1.00 g/g.
X-ray diffraction was performed in fiber geometry, with air-dried cellulose particulates conditioned in capillary tubes, as detailed elsewhere (Driemeier & Calligaris, 2011; Driemeier et al., 2011). Two-dimensional diffraction patterns are collected in mar345 image plates and are analyzed by the Rietveld method, including description of cellulose preferential orientation. An example of experimental and modeled diffraction pattern in presented in Figure 1.
Degree of crystallinity, defined as cellulose crystals mass per sample dry mass, was estimated following previous developments (Driemeier & Calligaris, 2011), which include corrections for incoherent scattering, sample moisture content, crystal texture, blank intensity, and X-ray absorption. Crystallite width was derived by applying the Scherrer equation (see Driemeier et al. 2011; Driemeier & Bragatto, 2013) to the width of 200 diffraction peaks, which are the most intense reflection observed from plant celluloses.
Dynamic vapor sorption
Water vapor sorption analysis is a classical tool to probe the equilibrium moisture contents sorbed in cellulosic materials (Skaar, 1988). In dynamic vapor sorption, water mass is continuously measured as a function of time, allowing inferences about sorption-desorption kinetics in addition to equilibrium contents.
Dynamic vapor sorption was performed in a TA Instruments Q5000 SA gravimetric analyzer. Samples are equilibrated at 50°C in 0.95 relative humidity. Then, relative humidity is decreased in 60 min steps for water desorption. After drying in zero relative humidity for 180 min, relative humidity is increased in 60 min steps for water sorption. Most steps apply a 0.1 change in relative humidity. Equilibrium desorption and sorption isotherms are analyzed with the Hailwood-Horrobin model equations (Hailwood & Horrobin, 1946).
Desorption and sorption kinetics are analyzed by reading the times t1/2, t1/4, t1/8, and t1/16 required to reach, respectively, 1/2, 3/4, 7/8, and 15/16 of the mass change after a relative humidity step. Mass response to relative humidity steps presents a fast and a slow component. The fast component is represented by t1/2. It depends on the analyzed sample mass and, therefore, does not provide information about intrinsic material properties. The slow component is represented by t1/16-t1/8. It does not depend on sample mass and is used to investigate structural differences among samples. Details of experiments and equilibrium and kinetic analyses are given elsewhere (Driemeier et al., 2012).
Thermoporometry was performed in a TA Instruments Q200 differential scanning calorimeter, as detailed elsewhere (Driemeier et al., 2012). Thermoporometry is based on the temperature depression of ice melting due to ice confinement in nanometric pores. Watersaturated samples are prepared in aluminum pans covered with hermetic lids. Samples are frozen to -70°C and then heated step-bystep to 5°C. Measured heat flows are converted to ice mass, which must carefully subtract the contribution of sensible heat. Temperature depression of ice melting ΔT is converted to pore diameter x through x = -2 K/ ΔT, which is a concise form of the Gibbs-Thomson equation, and K, = 19.8 nm K. Thermoporometry results are reported as cumulative freezing bound water (mass of confined ice per unit of dry mass) as a function of pore diameter x (0.8-400 nm).
RESULTS AND DISCUSSION
Figure 2 presents an example of general ultrastructural relation followed by a wide spectrum of celluloses processed from plants (samples described in section 2.1). Water monolayer is an estimate of water mass (per dry mass) in direct molecular contact with the cellulosic solid. Water monolayer was derived from sorption isotherms analyzed by the Hailwood-Horrobin model. Reciprocal crystallite width (1/width) was derived by applying the Scherrer equation to the 200 peak of X-ray diffraction.
The general relation of Figure 2 was explained (Driemeier & Bragatto, 2013) by hydrated polysaccharides, including hemicelluloses and amorphous cellulose, filling regions between bundled cellulose crystallites. Therefore, the primary role of crystallite width in explaining monolayer hydration is associated to the lateral organization of crystallites in bundles.
Figure 3 shows degree of crystallinity versus cellulose content (same set of samples shown in Figure 2). Since both measured parameters are dry-basis mass fractions (units of g/g), the diagonal y=x line corresponds to all cellulose being crystalline. Points are at or below this diagonal line, indicating presence of amorphous (noncrystalline) cellulose in addition to non-cellulosic amorphous (mainly amorphous hemicellulose).
Figure 4 presents wet nanometric porosities of four samples (S101, Alpha, Ekp1, and Bap48), as measured by thermoporometry. Appreciable differences between samples are evidenced. These porosities correlated-positively with rate of enzymatic hydrolysis (Bragatto et al., 2012).
Figure 5 presents characteristic times of water vapor desorption and sorption, comparing several cellulose samples. Characteristic times of desorption are remarkably similar for the wide spectrum of celluloses it was suggested (Driemeier et al., 2012) that desorption times are controlled by evaporation of water pockets inherited from water-swollen states. Sorption times, however, present appreciable variations with relative humidity and cellulose sample. The structural origins of these differences are presently unknown, but it is possible that they reflect the structure of regions interfacing crystallites, where water sorption takes place (Driemeier & Bragatto, 2013).
Finally, it is worth mentioning analytical cost and throughput. Our thermoporometry setup is automated, running four samples per day and consuming only nitrogen gas and disposable aluminum pans and lids. Setup of dynamic vapor sorption is also automated, running one sample per day and consuming only nitrogen gas. X-ray diffraction analysis demands more work. Measurement of 20 samples is done in a 12 hours shift and data analysis demands more than a week of a trained analyst. Nevertheless, automation of data analysis is now under development.
X-ray diffraction, dynamic vapor sorption, and thermoporometry bring complementary ultrastructural information about celluloses. X-ray diffraction is primarily sensitive to cellulose crystallites that form the core of microfibrils. Dynamic vapor sorption probes primarily the first hydration layers surrounding these crystallites. Thermoporometry probes nanometric water-filled pores defined between bundles of cellulose crystallites. Combining these three analytical techniques brings a comprehensive ultrastructural picture of a cellulose sample. Analysis of a wide spectrum of celluloses processed from plants evidenced that material-specific properties as well as general ultrastructural laws can be inferred from these techniques combined.
Financial support from FAPESP (project 2010/05523-3).