1. INTRODUCTION

The use of natural plant fibers in composites exists since ancient times [1, 2]. However, the use of lignocellulosic fibers, as are known the natural plant fibers in the engineering area, has decreased over time, mainly due to the emergence of new materials. In the last decades, along the increase of pressure from the society and the appearance of new environmental legislations for a more rational use of natural resources, the vegetable fibers returned to be an alternative for various applications, mainly for the plastic industry for they are renewable natural resources and biodegradable [2, 3, 4].

Nowadays, several types of natural fibers have been investigated to be used in composites, including flax, hemp, ramie, sisal, jute, rice husk, rattan, kenaf, rye, piassava, among others [1, 2, 5, 6].

In this context, bamboo, in recent years, has been receiving considerable attention as a substitute for wood and other fibers, because its mechanical properties are very similar [7].

In contrast of the woods, that presents four layers of microfibrils arranged in different thicknesses and orientation angles in the formation of its cell wall, bamboo is made of several layers which alternate in longitudinal and transversal directions, oriented at angles that enhance their values toward the inside of the wall. The absence of anatomical elements in the radial directions and also the features mentioned above makes this grass a functional gradient material (FGM), in others words, a material in which its properties change gradually or continue along a specific direction. Because of anatomical characteristics differentiated, the bamboo does not suffer abrupt changes in their properties, quite the contrary; the properties vary from point to point according to the actual need [8, 9].

These fibers are considered natural polymers, composed, basically of cellulose, hemicelluloses and lignin, which are constituents that contribute differently amongst each other to the mechanical resistance [2, 10, 11, 12].

The application of vegetable fibers as reinforcement allows the acquisition of composite products, which, besides presenting smaller apparent specific mass and higher porosity, present, still, satisfactory values of traction and impact resistance, bigger control of cracking, besides ductile behavior during break [13, 14, 15].

Cellulosic fibers, such as bagasse from sugar cane, sisal, banana, coconut, wood and jute have been incorporated into thermoplastic and thermoset matrices, as reinforcement or load [3, 16, 17, 18].

However, natural fibers present some disadvantages, for example, the low compatibility with the polymeric matrix, due to the presentation of hydrophilic character, while the polymeric matrix has hydrophobic character. This difference in affinity for water results in a low superficial adherence, which prevents the transfer of effective stress, reducing, consequently, the mechanical properties of the composite [19, 20].

To enhance the adherence of the fiber/matrix interface, the natural fibers have to be modified superficially, so that they can interact with the polymeric matrix [21, 22]. By studying different modification methods of sisal fibers, Singha & Thakur [23]; Kumar et al., [24] e Taherzadeh & Karimi [25], argue that the fibers can be modified effectively by chemical substances, such as isocyanates and silanes, and physical treatments, such as plasma and corona discharge.

The alkaline treatment of fibers is one of the low cost possibilities [26]. The effects of mercerization as a treatment to improve the mechanical properties, especially traction resistance to lignocellulosic fibers, have been extensively studied [27].

In this context, the chemical modification of lignocellulosic fibers by esterification reaction with dianhydrides, after mercerization, presented itself as an attractive method for increasing the compatibility of natural fibers with polymeric matrices of high hydrophobic character.

The objective of this work was to modify chemically refined bamboo pulp with dianhydride benzofenonatetracarboxílic, after mercerization and extraction in organic solvents, aiming your use as strengthening agent in polymeric matrices of hydrophobic nature. The final structure of the sample after esterification is show in figure 1.

The changes in the chemical structure of the surface of pulp were evaluated by means of spectroscopic and chemical analyses, while the thermal stability and the cristallinity were determined through thermogravimetric analysis and spectroscopic observations, respectively.

2.1 Refined Pulp

The refined samples of bamboo pulp from the species Bambusa vulgaris Schrad were provided by CEPASA – Celulose e Papel de Pernambuco S/A, located in the city of Jaboatão dos Guararapes. The refined pulps underwent the industrial pulping process known as KRAFT, with screened yield around 46% and KAPPA number between 45 e 55. At the wash exit, the pH varied from 9 to 10, while the ºSR suffered alterations between 60º and 70º after refining and 25 and 30º before the refining, in a disc refiner. The active alkali varied between 16% and 18%.

The bamboo pulp was washed with distilled water and dried in a climatized room for 15 days, at 23 ºC and 65% of relative humidity, aiming at its equilibrium moisture, according to the TAPPI T 257 cm-00 [28] Standard. Following this, the samples were ground in rotor mills and granulometrically classified in a series of Tyler sieves, following the recommendations of the WS Tyler Standard and ASTM 1921 [29].

The mercerization, performed after the extraction process with solvents, was carried out with 2% of NaOH, at room temperature, for a period of 2 hours, washed in distilled water until the complete elimination of alkali. The samples were, then, dried in an oven at 100 ± 5 ºC. After the mercerization, the refined pulp has undergone the process of extraction of organic solvents, followed by its placement in a round-bottomed flask with benzophenone tetracarboxylic dianhydride, used for the esterification reaction, and triethylamin, as a catalyst, plus acetone, as a solvent, for 48 hours, at 55 °C, which was used until suspension the fibrous blanket. The proportion between the reagents was of  1.0 g of pulp to 2.5 g of BTDA and 1.5 mL of triethylamin. After the reaction was concluded, the pulp was washed in acetone in abundance until the removal of unreacted solids adhered and dried in an oven at 100 ± 5 ºC, to constant mass.

The thermogravimetric, elementar and spectroscopic analysis were performed with samples retained in the 60 and 270 mesh sieves, respectively.

2.2 Characterization

The Spectroscopy in the infrared region (4000-400 cm ^{-1^}) was performed in a Perkin Elmer GX spectrophotometer, operating with the resolution of 16 cm^-1 and 32 scannings, using KBr pellets, compressed at 10-12 Kgf.cm^-2 .

3. RESULTS AND DISCUSSION

Figure 2. a) Absorption spectrum in the IR region, from 4100 to 400 cm^-1, of the BTDA [34]. b) Absorption spectrum in the IR region, from 4100 to 400 cm^-1, of the natural fiber samples (FNAT), industrial unrefined kraft pulp (POKR), industrial refined kraft pulp (PKRR), industrial refined kraft pulp, chemically modified with sodium hydroxide (PHIS), industrial refined kraft pulp, chemically modified with sodium hydroxide + ethanol/toluene + ethanol (PETE) and industrial refined kraft pulp, chemically modified with sodium hydroxide + ethanol/toluene + ethanol + dianhydride (NETEB).

The peak of 1734 cm^-1, related to the acetyl group extraction with organic solvents, confirming the and carboxyl group of the xylanes disappeared with extraction of lignin and extractives from the sample, the treatments of pulping, refining, mercerization, but reappeared during the esterification treatment, indicating the incorporation of ester groups in the pulp.

The fall in the values of lignin, hemicelluloses and extractives was also confirmed by the decrease of the 1636 cm^-1 bands, which represents vibrations in the aromatic skeleton plus stretching (C=O, carbonyl stretching) indicating a decrease in the level of poliosis; 1252 cm^-1 , which represents asymmetric axial deformation of (=C-O-C^-, asymmetric axial deformation ether), common in an aqueous medium where (=C-O^-), is present, as in ester and ether; 1376 cm^-1 , regarding the deformation vibration of the C-H group, due to the removal of polysaccharide of low molecular mass and 1512 and 1431 cm^-1, related to the vibration of aromatic groups of lignin, specific of the guaiacyltype rings.

The decrease of the 3422 cm^-1 band indicates the success of the esterification reaction in the sample and also in the humidity decrease. The widening of the bands of 897 and 1057 cm^-1 point to an increase in the values of alpha-cellulose levels, while the increase of the 1162 cm^-1 band indicates the incorporation of the ester group to the pulp. The little decrease in the 1320 cm^-1 band (guaiacylsyringyl lignin ring) occurred, mainly, due to the treatment of mercerization. The 2916 cm^-1 band, attributed to the symmetric and asymmetric vibrational stretching of the CH2 e CH3 from the aliphatic groups, became smaller, due to the removal of hemicellulose.

3.2 Spectroscopy Analysis in the Infrared Region for Crystallinity Index

The results of the absorbance ratios, representing the indexes found in each type of sample, treated and free of treatment ones, are presented in Figure 3, according to the methodologies adopted by Nelson & O’Connor [30] and Akerholm et al., [31]. The ratio between the heights of the bands at 1372 and 2900 cm^–1 proposed by Nelson and O’Connor as total crystalline index TCI was used to evaluate the infrared (IR) crystallinity ratio. The band 2900 cm^{– 1} is the internal standard correction in the sample and does not depend of changes in crystallinity. The 1372 cm^–1 measures the intensity according to the  change in crystallinity. The band at 1429 cm^-1 associated with the amount of crystalline structure of cellulose, while the band at 897 cm^–1 is assigned with the amorphous region in cellulose. The ratio between the areas of the bands at 1429 and 897 cm^-1 are used as a lateral order index (LOI). Considering the chain mobility and bond distance, the hydrogen bond intensity (HBI) of cellulose is closely related to the crystal system and the degree of intermolecular regularity, that is, crystallinity [35, 36, 37]

In this sense, it is implied that the crystalline region of the native cellulose presented stage β, to the detriment of stage α, when submitted to the processes of industrial pulping and refining and, also, mercerized with sodium hydroxide, as there were gradual raises in the crystallinity indexes of the samples according to removal of parts amorphous without the consequent degradation of the same. These characteristics inherent to cellulose I type β may have possibly occurred in the stage of industrial pulping, as reported by Tserki et al., [38] e Hult et al., [39], which might have happened due to electrostatic interactions amongst cellobiose units that provoked a more favorable alignment to the formation of hydrogen bonds that stabilized the charges of the molecule.

With regard to the decrease in the values of the refined samples’ crystallinity indexes, treated with organic solvents, it is believed that has occurred, according to the utilized methodology a penetration of the solvents in the samples and the entry of new functional groups on the polymer chains, in the case of modification with dianhydride, provoking the rearrangement of the packing of native cellulose chains, aligned in parallel, for cellulose II, where the chains are antiparallel. The material had its chains of polysaccharide expanded and rearranged, which culminated in a greater amount of less organized material, as well as the consequent fall in the pulp’s crystallinity [40]. Siqueira [41] commented that the esterification increases the amorphous part of the sample, reducing therefore, the crystallinity of the same.

In a way, the lignin that acted as a kind of glue that unites the beams was removed with the treatments, easing the penetration of reagents and their access to the hydroxyl groups. Oliveira and Frollini [42] found similar behavior when modifying sisal fibers with ultrasound.

Tserki et al., [38] and Karnitz Junior et al., [43] obtained the same results by treating the surface of other lignocellulosic fibers, using other types of dianhydride.

Figure 3. Crystallinity index of refined and unrefined fibers and pulps of Bambusa vulgaris schrad, with and without chemical modifications. Results obtained indirectly through analysis of absorption spectroscopy in the infrared region, according to Nelson & O’Connor⁴⁵ and Akerholm et al.⁴⁶ FNAT: natural fiber; POKR: unrefined pulp; PKRR: refined pulp; PHIS: refined pulp treated with NaOH at 2%; PETE: refined pulp treated with NaOH at 2% and submitted to extraction with organic solvents; NETEB: refined pulp treated with NaOH at 2% + extraction with organic solvents + esterification with dianhydride.

Figure 4. TGA Curves of modified and not-modified Bambusa vulgaris Schrad samples. 3(a) – Heating rate of 20 °C.min^-1 (POKR, PKRR, PHIS e PETE); 3(b)-Heating rate of 10 °C.min^-1 (FNAT e NETEB). All samples tested in N2 atmosphere under a flow of 100mLmin^-1 .

In general, the fibers presented a similar thermal behavior. The TGA curves displayed in Figure 4 showed an average decrease in mass of approximately 9% at 100 °C, for all the samples of modified and not-modified Bambusa vulgaris schrad. This loss of mass corresponds to the water evaporation of the samples. Between the temperatures of 105 °C and 275 °C, the samples presented little loss of mass, indicating thermal stability up to 275 °C.

It has been also observed a difference with respect to the mass loss of the modified samples and the natural sample from around 330 to 415 °C, and all samples of Bambusa vulgaris Schrad presented greater loss than the natural samples. The difference in percentage of the products derived from the pyrolysis, as the atmosphere used in the test was N2, was attributed to the difference of the natural fiber’s composition, regarding the remaining samples.

Oliveira & Frollini [38] researching the modifications in fibers in order to use them in composites, found, for sisal fibers, a decomposition start temperature at 250 °C. In this study, by chemically treating refined pulps of bamboo, it was found, after a treatment with sodium hydroxide at 2%, for 2 hours, a decomposition start temperature of 275 °C, a value that is higher than the one found in literature for several types of fibers.

in derivative thermogravimetry, the first derivative of the mass change with respect to temperature is recorded as a function of time or temperature. The resulting curve consists of a series of peaks instead of a stepwise curve, if the mass change takes place in more than one step. The maxima of the DTG curve represent the inflection points, where the rate of mass change is maximum. The data are displayed in a more visually accessible manner. Besides, the DTG curve provides ready information about the temperature at which the rate of mass change is maximum. In a TG curve, if the subsequent changes are very close to each other, the different stages cannot be easily distinguished. But, the DTG curve provides sharp maxima corresponding to different stages and hence they can be easily differentiated. It indicates precisely the temperature corresponding to the beginning and the end of a mass change.

The DTG curves confirmed the data found in the TGA analysis, suggesting two degradation stages for all the samples analyzed. In all cases, an initial peak between 28 and 90 °C occurred, which corresponds to the elimination of water in the samples. The lowest percentage of humidity loss in the esterified pulp, when compared to other pulps and also the natural fiber, may be attributed to the greater hydrophobic character, acquired by the fibers modified with BTDA, due to the reaction of esterification. There was a difference of approximately 20%, on average, in the indexes water absorption between the modified pulp (BTDA) and natural fibers (FNAT).

The natural fiber presented a second decomposition peak with maximum at 323.47 °C, attributed to the depolymerization, deacetylation and dehydration of hemicelluloses and rupture of α and lignin β aryl-alkyl-ether bonds, with stabilization at 560 °C and loss of mass of 65.53%. The fiber samples after industrial kraft pulping and refining processes presented main decomposition peaks with maximum at 353.7 ºC and 355.6 °C and stabilization at 589.1 and 598.9 °C, respectively, with mass losses of 74.7 and 72.9%, in that order. The main peaks of mass loss are presented with maximum at 365.9 °C, stabilization at 610.3°C and 81.8% of mass loss for the sample treated and modified with NaOH at 2%, and at 360.8 °C, with stabilization at 586.7 °C and 80.7% of mass loss for the samples treated with organic solvents. The sample modified with dianhydride, through the process of esterification presented a second decomposition peak with maximum at 353.7 °C, stabilization at 569.6 °C and 79.2% of mass loss. Figure 5 shows the results found for the DTG tests.

Figure 5. DTG Curves of modified and unmodified Bambusa vulgaris schrad samples. 4(a)-Heating rate of 20 °C.min^-1 (POKR, PKRR, PHIS e PETE); 4(b)-Heating rate of 10 °C.min^-1 (FNAT e NETEB). All samples tested in N2 atmosphere under a flow of 100mL.min^-1 .

Table 1. Elementary analysis of the fiber, refined and without refining pulps of Bambusa vulgaris Schrad chemically

N(%): Nitrogen content; C(%): Carbon content; H(%): Hydrogen content; S(%): Sulfur content; O(%): Oxygen content; NOTE: Values in brackets refer to variation coefficients VC (%). FNAT: natural fiber; POKR: pulp without refining; PKRR: refined pulp; PHIS: refined pulp treated with NaOH 2%; PETE: refined pulp treated with NaOH 2% and submitted to extractions with organic solvents; NETEB: refined pulp treated with NaOH 2% + extraction with organic solvents + esterification with dianhydride.

Figure 6. Comparative graphic of maximum absorption of water from the culm, culm fiber and unrefined pulps and refined unmodified and chemically modified Bambusa vulgaris Schrad. COLB: culm of bamboo; FNAT: natural fiber; POKR: pulp without refining; PKRR: refined pulp; PHIS: refined pulp treated with NaOH 2%; PETE: refined pulp treated with NaOH 2% and submitted to extractions with organic solvents; NETEB: refined pulp treated with NaOH 2% + extraction with organic solvents + esterification with dianhydride.

It is also possible to observe an increase in the carbon percentage for the pulp modified with dianhydride in relation to the pulp treated with organic solvents. This result confirms the improvement of carbon percentage in the esterified pulp. Another important observation is related to the hydrogen content of refined pulp treated with NaOH, equal to 42.41%. It is observed that in relation to the previous treatment, there was a reduction of about 1.5%. it is estimated that the treatment have broken the extensive hydrogen bonds of the lignocellulosic structure, making the hydroxilic groups more accessible for esterification reactions.

3.5 Analysis of the maximum water absorption

Figure 6 shows the percentage of water absorption of samples of Bambusa vulgaris Schrad in function of time in accordance with the type of process they were subjected. The water absorption is an important feature of composites systens reinforced by vegetables fibers, since it evaluates the potential of these composites for specific final applications. Generally, the water absorption occurs due to the hygroscopic and hydrophilic nature of the natural fibers.

4. CONCLUSIONS

There was a difference of approximately 20%, on average, in the indexes water absorption between the modified pulp (BTDA) and natural fibers (FNAT). The start decomposition temperature also decreased due to the incorporation of the ester group, which caused an improvement of carbon content in the final structure of the lignocellulosic material. Infrared spectra showed bands relative to the ester and carboxylic acid groups, evidencing the exchange of OH by ester groups in the cell wall of the material. Results showed the efficiency of the use of benzophenone tetracarboxylic dianhydride as crosslinker agent in the modification of superficial chemical composition of the bamboo refined pulp, becoming less hydrophilic and more adequate to be used as reinforcement agent in polymers of apolar nature, as has already been done with acetic anhydride in the process known as acetylation. The test of maximum water absorption performed in the samples, confirms definitively that modification with dianhydride was a success. 376.31% of maximum water absorption was the lowest value found among all samples tested.

5. REFERENCES

[1]. Chawla, K. K. “Composite Materials: Science and Engineering”. 1^st Ed. New York, Springer-Verlag, 1987, 292p.         [ Links ]

[2]. Taj, S.; Munamar, A. M.; Khan. S. “Natural Fiber-Reinforced Polymer Composites”. Proceedings of the Pakistan Academy of Science, 2007; 44: p. 129-144.

[3]. Chattopadhyay, S. K.; Khandal, R. K.; Uppaluri, R.; Ghoshal, A. K. “Bamboo Fiber Reinforced Polypropylene Composites and Their Mechanical, Thermal, and Morphological: Properties”. Journal of Applied Polymer Science, 2011, 119: p.16191626.

[4]. Lopez, L. F.; Correal, J. F. “Exploratory study on the glued laminated bamboo Guadua angustifolia kunt as a structural material”. Maderas: Ciencia y Tecnologia, 2009, 11: p.171-182.

[5]. Mengeloglu, F.; Karakus, K. “Thermal Degradation, Mechanical Properties and Morphology of Wheat Straw Flour Filled Recycled Thermoplastic Composites”. Sensors, 2008, 8: p.500-519.

[6]. Dhakal, H.; Zhang, Z.; Richardson, M. “Effect of Water Absorption on the Mechanical Properties of Hemp Fibre Reinforced Unsaturated Polyester Composites”. Composites Science and Technololy, 2007, 67: p.1674-1683.

M. J. “Methods for Pretreatment of Lignocellulosic Biomass for Efficient Hydrolysis and Biofuel Production”. Ind. Eng. Chem. Res, 2009, 48: p.3713-3729.

[30]. Nelson, M. L.; O'connor, R. T. “Relation on certain infrared bands to cellulose crystallinity and crystal lattice type. Part I: Spectra of lattice types I, II, III and of amorphous cellulose”. Journal of Applied Polymer Science, 1964, New Orleans, 8: p.1311-1324.

[31]. Akerholm, M.; Histerstoisser, B.; Salmen, T. “Characterization of the crystalline structure of cellulose using static and dynamic FTIR spectroscopy”. Carbohydrate Research, 2004, 339: p.569-578.

[32]. International Association of Wood Anatomy – IAWA -List of microscopic features for wood identification. IAWA Bulletin, Leiden, 1989, 10: p.226-332.

[33]. Milanez, A. C.; Foelkel, C. E. B. “Processo de deslignificação com oxigênio para a produção de celulose de eucalipto”. In: CONGRESSO ANUAL DA ABCP, 14. 1981, São Paulo. Anais... São Paulo: ABCTP 9 p.37-110. 1981.

[34]. Oliveira, V. A. “Síntese e caracterização de géis de acetato de celullose reticulados com dianidrido piromelítico e dianidrido do ácido 3,3’, 4,4’ benzofenona tetracarboxílico”. 2008. Dissertação (Mestrado em Engenharia e Materiais) -Universidade Federal de Ouro Preto, Ouro Preto, MG. 2008.

[35]. Carrilo, F.; Colom, X.; Suñol, J. J.; Saurina, J. “Structural FTIR analysis and termal characterization of lyocell and viscose-type fibres”. European Polymer Journal, 2004, 40: p.2229-2234.

[36]. Akerholm, M.; Hinterstoisser, B.; Salmén, L. “Characterization of the crystalline structure of cellulose using static and dynamic FTIR spectroscopy”. Carbohydrate Research, 2004, 339: p.569-578.

[37]. Oh, S.Y.; Yoo, D. I.; Shin, Y.; Seo, G. “FTIR analysis of cellulose treated with sodium hydroxide and carbón dioxide”. Carbohydrate Research, 2005, 340: p.417-428.

[38]. Tserki, V.; Zafeiropoulos, F. S.; Panayiotu, C. A. “Study of the effect of acetylation and propionylation surface treatments on natural fibers”. Composites Part A – Applied Science and manufacturing, 2005, 36: p.1110-1118.

[39]. Hult, E.; Larsson, P. T.; Iversen T. “Cellulose fibril aggregation – an inherent property of kraft pulps”. Polymer, 2001, 42: p.3309-3314.

[40]. Sousa, K. S.; Airoldi, C. Termoquímica da Interação de cobre em quitosana quimicamente modificada com anidrido succínicoe etilenodiamina. In: IV SIMPÓSIO IBEROAMERICANO DE QUITINA. Natal, Brasil. 2007.

[41]. Siqueira, G. A. F.. “Produção e caracterização de compósitos fenólicos com fibras de sisal modificadas”. 2006. Dissertação (Mestrado em Engenharia de Materiais) – Universidade Federal de Ouro Preto, Ouro Preto, MG. 2006.

[42]. Oliveira, F.; Frollini, E. “Sisal fibers modified with sodium lignosulphonate: improvement of adhesion between fibers and phenolic type matrices”. In: 26th Polymer Processing Society, Banff. 26th Polymer processing society. 2010.

[43]. Karnitz Junior, O.; Gurgel, L. V. A.; Freitas, R. P.; Gil, L. F. “Adsorption of Cu(II), Cd(II), and Pb(II) from aqueous single metal solutions by mercerized cellulose and mercerized sugarcane bagasse chemically modified with EDTA diahydride (EDTAD)”. Carbohidrate Polymers, 2009, 77: p.643-650.

[44]. Cláudio-da-Silva jr, E. “Chemical pulp beating related to fiber structure”. Thesis (Doctor of Philosophy), State University of New York, Syracuse, New York. 1981.

[45]. Paavilainen, L. “Importance of cross-dimensional fibre properties and coarseness for the characterization of softwood sulphate pulp, Paperi ja Puu”. Paper and Timber, 1993, 75: p.343-351.

[46]. Lindstrom, T.; Carlsson, G. “The effect os chemical environmental on fiber swelling”. Svensk Papperstidining, 1982, 3: p.4-20.

[47]. Nazhad, M.; Paszner, L. “Fundamentals of strenght loss in recycled paper”. Tappi Journal, 1994, 77: p.171-179.

[48]. Brum, S. S.; Oliveira, L. C. A.; Bianchi, M. L.; Guerreiro, M. C.; Oliveira, L. K.; Carvalho, K. T. G. “Síntese de acetato de celulose a partir da palha de feijão utilizando N-bromossuccinimida (NBS) como catalisador”. Polímeros, 2012, 22: p.447 452.

[49]. Abreu, A. L. “Modificação química de Resíduo lignocelulósico para preparação de compósito”. 2011. Dissertação (Mestrado em Engenharia de Materiais) -Universidade Federal de Ouro Preto, Ouro Preto, MG. 2011.

[50]. Gatenholm , P.; Gellerstedt, F. “Surface properties of lignocellulosic fibers bearing carboxylic groups”. Cellulose, 1999, 6: p.103–121.

[51]. Ferreira, P. J. T. “Estudo de pastas kraft de Eucalyptus globulus: Características estruturais e aptidão papeleira”. 2000. Tese (Doutorado em Engenharia Química) -Universidade de Coimbra, Coimbra, Portugal.