Grantee Research Project Results

A nanoparticle, according to the National Nanotechnology Initiative, has at least one dimension smaller than 100 nm. The nanocomposites that we have made are referred to as particulate nanocomposites since they consist of a uniform dispersion of rigid nanoscale particles in a matrix of a softer plastic. We used sub-microscopic cellulose and chitin nanofibers and nanocrystals derived from them by acid hydrolysis as the nanoparticles.

 

The cellulose particles, which have transverse dimensions on the order of 10 nm, are derived from wood cells, seed hairs, grasses, certain bacterial capsules, recycled paper and board, or even from municipal solid waste They are produced by a combination of acid hydrolysis, cell disruption, and dispersion methods. Microfibril surface chemistry can be modified, where appropriate, to enhance compatibility of particles with the plastic matrix. In the present work, biodegradable polycaprolactone, cellulose esters, soybean oil polyesters, and polymethylmethacrylate were used as the matrix component of the composite.

 

Our primary objective was to make wholly biobased and biodegradable nanocomposites using nanoparticles of cellulose as the reinforcing particle. These will be compared in terms of thermal and mechanical properties to existing glass filled composites made from petrochemicals. Because the reinforcing particles are microfibrils of cellulose or crystals derived from such microfibrils with transverse dimensions on the order of 10 nm, it is appropriate to classify these materials as nanocomposites. Subsidiary objectives included refining the methods for making the nanoparticles and working towards scale up of the production to small pilot plant levels, measuring the mechanical properties of the resulting composites. Finally, a study of biodegradability in aerobic, compost conditions was made.

 

 

During the course of this work, nanoparticles were made from bacterial cellulose donated by C.P. Kelco, San Diego, CA; a tunicate, Ciona intestinalis, purchased from Woods Hole Biological Laboratories; cotton fiber, obtained from Whatman #1 filter paper; sugar beet, apple pomace, and orange pulp agricultural residues; commercial softwood pulp provided by Weyerhaeuser LLC; and maple wood and its kraft pulp product produced in the SUNY-ESF Bioprocess and Paper Engineering Department. In addition, we also made nanofibers and nanocrystals of chitin, a closely related polymer with N-acetyl-D-glucosamine as the monomer and the same β (1-4) linkage as found in cellulose and obtained from arthropod exoskeletons. In our work, we used shrimp as our chitin source. Our rationale for adding chitin was twofold. First, the particles themselves have different solubility characteristics than cellulose, a fact that might prove useful in making composites with cellulose plastics. Second, the presence of an amine group might provide an easy route to surface modification with proteins. 

 

 

 

While the details vary slightly with the origin of the fibers the process begins with the removal of lignin and hemicellulose for cellulose and deproteinization for chitin. For cellulose on a commercial scale, this is what happens in the conversion of wood or other fiber into pulp. On a laboratory scale, we and others usually use repeated bleaching with sodium chlorite. The cellulose was washed with distilled water repeatedly until a neutral pH was attained. To make nanocrystals, the purified cellulose was hydrolyzed in 65% (w/v) sulfuric acid for 2 hours at 40^oC. Then, the mixture was quenched with ice water, and the cellulose was separated from the bulk acid by vacuum filtration. The hydrolyzed cellulose was rinsed repeatedly with distilled water and filtered until the material reached a neutral pH. The hydrolyzed cellulose was resuspended in distilled water using a CuisineArt blender. The suspension was mechanically disrupted with an APV Gaulin 1000 homogenizer operating at 8000 psi with 10 to 15 recycling treatments. Samples were stored, suspended in distilled water (2-10% w/v)at 4^oC. If a dry product was desired, the cellulose nanocrystal suspension was exchanged in t-butanol and lyophilized. This has the advantage of minimizing the agglomeration typical of lyophilization from water. In many cases, sonication was used in place of or in addition to the homogenization step. To make nanofibers, the acid hydrolysis step was not employed. These conditions were developed in a lengthy study by Maren Grunert (now Maren Roman), which was completed in this laboratory as the funding for this project commenced (Roman and Winter, 2004).

 

Characterization of the nanoparticle fractions includes the use of x-ray diffraction (Rigaku D-MAX 1000 diffractometer mounted on a 12 kW RMU 200 rotating anode generator using Cu Kα ( λ=0.15418 nm) radiation to establish both the crystalline allomorph that exists at any given stage and the crystallite size, which is determined by Scherrer equation analysis. Transmission electron microscopy (JEOL 2000 TEM), with or without uranyl acetate staining, also was used to establish particle dimensions and heterogeneity. Figure 1 shows representative pictures of cellulose and chitin nanocrystals and cellulose nanofibers.

 

Figure 1.  a) Cellulose microfibrils (nanofibrils) from sugarbeet residues converting to nanocrystals upon acid hydrolysis and b) chitin nanocrystals from shrimp.

 

Based on this study and other studies, the conversion of cellulose and chitin to nanofibers and then to

nanocrystals by mechanical dispersion and acid hydrolysis is extremely general and can be accomplished starting from any cellulose- or chitin-rich substrate. The transverse dimensions of such crystals are species dependent. The cellulose images shown in Figure 2 were obtained with uranyl acetate staining and are probably overestimates of particles size, and diffraction measurements suggest appreciably smaller transverse dimensions, < 10 nm. The size of crystallite domains and their morphology did not change measurably during the course of either mechanical or chemical treatments. Nanoparticles from higher plants and tunicates are recognizable as cellulose Iβ, while those from bacteria are cellulose Iα, the pulped samples from maple generated Cellulose II crystals as expected, since Kraft pulps normally are mercerized by the alkali used in the process.

 

Using a 22-liter glass reactor, we demonstrated that it was possible to make the cellulose nanoparticles in 500 g batches, suggesting that the method is scaleable. Solid-state NMR, x-ray diffraction, thermal analysis, and FTIR spectroscopy all indicated that this product was indistinguishable from our previous 5-20 g batches. Subsequently, a group at Royal Technology Institute (KTH) in Stockholm has used mechanical breakdown of endo-cellulase treated wood pulps to do similar preparations on a multi kilogram scale and a Canadian consortium involving Domtar, a commercial pulp and paper company, and Domtar and FPInnovations have announced that they anticipate opening a pilot plant producing 1 ton per day of cellulose nanoparticles.

 

 

It has been stated that all chemistry is surface chemistry and certainly one reason for using nanoparticles as reinforcing agents is the expectation of large area of contact between matrix and particles. Dr. Jacob Goodrich, in his thesis, measured the specific surface area of both cellulose and chitin nanoparticles. A problem with many studies in this area is the use of traditional absorption isotherm measurements using argon or nitrogen. In these studies the test sample needs to be dried and with cellulose nanoparticles this means that the particles re-aggregate with a dramatic loss of specific surface area. For this reason, our approach was to borrow a method widely used in the textile industry and use the uptake of a dye, Congo red, know to form a monolayer on cellulose surfaces as a method of measuring surface areas for particles dispersed in water.² In this work we compare the isolated nanoparticles from three sources with the native fibers from wood pulp and shrimp shell after protein removal. The results are shown in Table 1.

 

Table 1. Dye binding surface area results²

 

Note: [A] is the amount of Congo red (mg) taken up per gram of substrate and the right-hand column is the actual surface area per gram of substrate. As a comparison, we note that at a 300 m²/g specific surface area, the surface area for 25 g of particles, a bit less than one ounce, would be sufficient to cover five regulation American football fields.

 

 

The surfaces of the cellulose nanoparticles were modified by carrying out reactions on the surface hydroxyl groups of the cellulose nanoparticles. The objective of this work was to test the hypothesis that we could improve the compatibility between the particles and the plastic matrices in which they were to be dispersed by adjusting surface chemistry. Derivatives made in this work are listed in Table 2.

 

An interesting result with surface polymerization of ε-caprolactone on cellulose was that the reaction was far less efficient than the parallel solution polymerization of the monomer in the same pot. Because of the difficulty in measuring molecular weights of grafted chains, the assumption in previous work by Dufresnes and others was that the Degree of Polymerization (DP) for chains attached to the nanocrystals was the same as that for

 

Table 2. Surface modifications of nanoparticles

 

 

the polymers not attached to cellulose. This is discussed in Chapter IV of Reference 7 and will be submitted for publication in fall 2010. The polymerization product was fractionated into cellulose free, dichloromethane soluble polycaprolactone and the insoluble caprolactone surfaced cellulose particles. MWn of the polymer, determined by gel-permeation chromatography was 9-10 kDaltons, a DP of ~80-90. The surface modified particles were digested with a mixed endo- and exo-cellulase preparation and the products were examined by Matrix Assisted Laser Desorption Ionization (MALDI) Mass Spectroscopy. Analysis of the MALDI spectrum indicated that the DP of caprolactone chains attached to cellobiose had a maximum DP of 5.^4,7

 

A second and rather general observation with respect to surface modification of cellulose nanoparticle was the disappearance of a shoulder at 62 ppm on the C6 (65-67 ppm) resonance that was related to the extent of reaction and correlated with an increase in intensity in the 72 ppm region of the ¹³C CP/MAS NMR spectrum. There is evidence from selected area microbeam electron diffraction establishing that the surfaces of cellulose I crystals makes the O6 the most accessible hydroxyl group and other groups have suggested that the conformation of the surface O6 is different from that in the interior of a crystallite. This difference is consistent with the occurrence of a shoulder on the C6 resonance and the fact that the shoulder disappears with increasing reaction time strongly supports the idea that surface modification principally affects C6. It also means that the ratio of the area of the shoulder to that of the main peak provides a good estimate of the cross- sectional area of the crystallites, because that area determines the fraction of cellulose chains lying on the surface of the crystallite. The measure is not truly quantitative because resonances from the same functional group in different magnetic environments need not scale exactly.

 

Composites were made using these particles and matrices that included poly (methyl-methacrylate), poly (ε-caprolactone), cellulose acetate butyrate, and a soybean oil polyester produced by ultraviolet radiation induced cross-linking. Efforts focused on making composites by compounding/extrusion, because our goal was commercially viable composites. Draft versions of those papers appear in the dissertations (References 6-8). In general, composites formed by compounding and/or extrusion showed modest, 3-40 fold, increases in modulus at 10% loading with nanoparticles but nothing like the 3000-fold increase observed with 6% w/v nanocellulose filled aqueous solution cast, air dried films. Hydrophobically surface modified cellulose formed weaker composites than unmodified cellulose particles. Although further work is needed to confirm the hypothesis, we and others have concluded that the nanocrystals organize into a network stabilized by hydrogen-bonds between the nanoparticles and water. A group headed by Avik Chatterjee, in our Department, is actively pursuing the theoretical basis for this phenomenon (D.A. Prokhorova and A.P. Chatterjee, Biomacromolecules 2009;10:3259).

 

Soybean oil polyesters, by themselves, form weak polymers; however, when reinforced with 10% w/w nanocrystalline cellulose, they form flexible, elastomeric materials where both the polymer and the filler are biobased and biodegradable. An example of such a material is shown in Figure 2.

 

Figure 2. UV crosslinked, epoxidized soybean oil polymer filled with 10% maleated cellulose nanocrystals.⁸

 

Another potentially useful observation about the composites, particularly when PCL was used as the matrix, is scanning and transmission electron microscopy for transcrystallization of the matrix at the interface with the nanocrystals. Differential scanning calorimetry also revealed increases in the crystallization temperature and degree of crystallinity for cellulose filled PCL as compared to non-filled samples (see p. 149 in the Dissertation by Goodrich).⁷ The utility of this phenomenon with respect to mechanical properties of commercial plastics remains a subject of some debate.

 

One final study was the effect of nanocellulose fillers on biodegradability of the composites. Our studies followed the procedures outlined in ASTM D 6003-96. Test films of regenerated cellulose, PCL, and filled PCL stored in aerated compost were monitored for weight loss. The regenerated cellulose controls were unrecoverable after 21 days. Neat PCL showed a 28% weight loss over 60 days, when filled with 10 bacterial cellulose nanocrystals that increased to a 38% weight loss. When the filling was changed to cellulose nanocrystals with grafted oligo-caprolactone surfaces the weight decreased by 60% in 60 days.

 

Figure 3. Relative weight for filled and unfilled PCL films versus time while stored in aerated compost at room conditions.⁷

 

Although this work was done in triplicate samples, we propose to repeat this study before publishing the results.

Future work, of course, is dependent upon funding. It seems likely that the methods of producing cellulose nanocrystals will become an industrial project. On July 16, 2010 Domtar and FPInnovations jointly announced plans for a facility to produce one ton of nanocrystalline cellulose per day (http://www.fpinnovations.ca/pdfs/BinderEn.pdf ). However, this author believes that it also makes sense to explore the production of such particles as a co-product to bioethanol. Most researchers acknowledge that the conversion of cellulose to ethanol is never 100% and the fraction remaining will be the material most resistant to hydrolysis and is likely to be cellulose nanocrystals. That suggests that the biorefinery process could be tuned to generate different ratios of ethanol and nanoparticles in the same way that the outputs of a petroleum refinery are adjusted to fit product demands. The co-product does have the effect of reducing enzyme cost per gallon of ethanol because the nanocrystals also should be a saleable product with a significantly higher value per unit mass than the ethanol.

 

Many areas of fruitful inquiry do remain. Several laboratories have shown that cellulose crystal particles can be used as templates for the formation of a variety of inorganic crystals including different crystal forms of titania at controlled size, a range of carbides including silicon, and possibly boron. The basic route for carbide formation involves pyrolysis of the nanoparticles to form a carbon replica followed by calcining in the presence of silica, for silicon carbide, to form silicon carbide. Carbides are ceramics with very high melting temperatures and mechanical strength. They find widespread industrial applications ranging from inorganic nanoporous filters for use in harsh chemical environments to bulletproof vests and plates for military applications.

 

Longer term, there is a great deal of interest in Asia in the modification of cellulose nanofibers into conductive fibers and, ultimately, into nanowires for use in large-scale printing of microcircuitry. Biomaterials such as tissue scaffold also are under study in many laboratories, although at present this application is likely to require bacterial cellulose, due to its inherent higher purity. As our work points out, the future of nanoparticulate cellulose as a reinforcing particle for composites will depend upon the ability to effectively disperse the particles in the thermoplastic matrix in a manner that lends itself to commercial processing. The best opportunities will first come in the paper industry itself where the liquid medium is water and the matrix is fibrous cellulose. Most likely, the next opportunities will be emulsion polymers such as acrylates where the polymer is formed in micelles distributed across an aqueous environment.