AN EXPLORATORY EVALUATION OF THE PULPABILITY OF Brachystegia spiciformis AND Pericopsis angolensis FROM THE ANGOLAN MIOMBO WOODLANDS

Brachystegia spiciformis and Pericopsis angolensis are two hardwood species found in the Miombo woodlands. The wood features, kraft pulping and strength pulp properties of both species were evaluated in order to determine their potential as raw material for papermaking. Brachystegia spiciformis wood density was 640 kg m-3 and Pericopsis angolensis was 795 kg m-3. Pericopsis angolensis wood has higher cell wall thickness and occluded fibre lumen as remarkable anatomical properties. Runkel ratio, slenderness ratio, and the coefficients of flexibility and rigidity in Brachystegia spiciformis were 1,5; 65,7; 41,2% and 29,4% while in Pericopsis angolensis these values were 17,6; 59,9; 5,4% and 47,3%, respectively. Brachystegia spiciformis has a higher cellulose content, lower hemicellulose and lignin content, and higher S/G ratio than Pericopsis angolensis. In kraft pulping, a higher demand of active alkali was needed for both species, and pulps with high kappa number (24-27) and low pulp yield (40%) were obtained. Pericopsis angolensis pulps reached tensile, tear and burst indexes of 99,6 Nm g-1; 5,9 mN.m2 g-1 and 4,9 kPa.m2 g-1, respectively. Brachystegia spiciformis pulps reached tensile, tear and burst indexes of 100,3 Nm g-1; 10,7 mN.m2 g-1 and 6,1 kPa.m2 g-1, respectively. As a conclusion, Brachystegia spiciformis wood has better pulpability than Pericopsis angolensis wood, according to its pulps properties, despite of the similar pulp yield between both species. Both species may be suitable for unbleached wrapping papers and rigid cardboards manufacturing.


INTRODUCTION
Brachystegia spiciformis and Pericopsis angolensis are two hardwood species from the Miombo woodlands (formation of natural forest, dry and warm predominantly in the South-Central region of Africa).In Angola, this forest occupies an area of 585949 km 2 , which corresponds to 47% of the total country area, being considered mainly as a source for timber production (Figueiredo et al. 2009, Sangumbe andPereira 2014).The wood of B. spiciformis is moderately heavy, slightly hard and difficult to be worked, with no significant difference between heartwood and sapwood colours; while P. angolensis wood is very heavy and hard, the heartwood is coloured green-brown and the sapwood is grey-yellow (Frost 1996, Palgrave 2002).The wood of B. spiciformis and P. angolensis are used for construction, cheap furniture, railway sleepers, utensils and beehives.It is suitable for flooring, joinery, mine props, vehicle bodies, crates, veneer and plywood.It is equally important as a source of firewood and charcoal, being among the preferred species for charcoal making throughout Southern Africa (Louppe et al. 2008).For another applications of P. angolensis wood, Uetimane et al. (2009) studied its anatomical properties in order to facilitate the introduction of the specie to the wood industry; Lhate et al. (2010) determine the chemical composition and relate it to wood durability and machining properties; while Cuvilas et al. (2014) made an evaluation of its wood properties to determine its quality as fuel.Regarding B. spiciformis investigations, there are some related species reports available.Atuanya and Ibhadode (2011) evaluated the chemical composition, microstructure and thermal behavior of Brachystegia nigerica to determine its potential as reinforcement for polymer composites.Other studies have reported the yield and strength properties of B. nigerica veneer products (Olufemi 2012), evaluations about the ethanol production from Brachystegia eurycoma have been made (Afe 2016a, Afe 2016b), and also assessments of the acoustic properties of B. eurycoma talking drums (Noah et al. 2014).
In Angola, the production of pulp and paper dated back to [1950][1951][1952][1953][1954][1955][1956][1957][1958][1959][1960], where plantations of Eucalyptus, Pinus and Cupressus species were established as fibre source for the national cellulose mill (Companhia de Celulose do Ultramar Português, CCUP) (Silva 1971, Delgado-Matas andPukkala 2012).After the 1975 independence, 27 years of civil war struck the country, and an extensive illegal logging of the forests occurred due to the lack of fuel sources, in addition to forest fires that destroyed most of the stablished plantations (Delgado-Matas and Pukkala 2012).However, the forestry activities were relaunched in the 2010's, with the beginning of the rehabilitation of the industrial segments related to the pulp and paper production chains (MINADERP 2011).As a consequence, the knowledge of the Angolan Miombo species needs to be assessed, as well as, the potential industrial application of wood from these species for chemical conversion processes (pulping, cellulose derivatives, among others).This study is aimed to evaluate the characteristics of the wood, kraft pulping and strength pulp properties of B. spiciformis and P. angolensis and their potential as raw material for papermaking.

Site and sampling
Samples of Brachystegia spiciformis Benth and Pericopsis angolensis (Baker) Meeuwen were obtained in Sambo, Angola, located at an altitude of 1704 m above sea (13°15'53,61"S and 16°03'52,28"E).The soil is ferrallitic with sandy loam and pH 4,5.The climate is humid with temperate seasons.The annual rainfall ranging from 1100 to 1400 mm, the average of annual temperature is 20°C and annual relative humidity is 60 to 70%.From one tree of each species (dominant and with straight form) 40 cm logs were taken at 2,5 m from the base of the tree.The age of the specimens was approximately 24 years old for B. spiciformis and 40 years old for P. angolensis.

Wood material
Logs obtained at 2,5 m of height were cut in 3-4 cm thickness disks.From the disks of each species 3,0 x 2,5 x 0,2 cm wood chips where handmade for pulping procedures.A fraction of the chips was also milled in a knife mill and sieved to 45/60 mesh for chemical analysis.For wood quality analysis, sapwood and heartwood sections where manually separated from the disks and wood blocks of 2 cm 3 were taken at each section.

Basic density determination
Wood blocks from sapwood and heartwood of each species were used to determine basic density according to the T258 om-94.Basic density is the ratio of oven dry weight by the green volume, expressed in terms of weight per unit volume.The basic density of wood was calculated according to Heinrichs and Lassen (1970).Basic density measurements of each sample were done in triplicate.

Transversal anatomical characterization
Sapwood and heartwood blocks were macerated with distilled water and glycerine for 7 days.From each block, transversal micro-sections of 30 µm thickness were obtained using a sliding microtome (MICROM H325).Samples were stained with Safranin and Astra blue, dehydrated with ethanol and assembled in a slide using Canada balsam.Images were obtained using a Zeiss microscope (Primo Star) connected to a personal computer and a digital camera (Canon A640) for image capture.Eighty fibres were randomly selected and their cell wall thickness, fibre width and lumen width were measured with a 100x total magnification.Moreover, eighty vessels randomly selected were measured with a 10x magnification for vessel width and mean vessel area determination.From the captures of 10x magnification the average vessel coverage was also determined, corresponding to the percentage of cross-sectional area covered by vessels.All these parameters were measured using AxioVision Software (Zeiss), which had the proper calibration for each capture lens.Similar procedure was used by Aguayo et al. (2010).

Fibre quality analysis
Samples of 0,1 x 0,1 x 0,5 cm were obtained from wood chips.Wood samples were macerated in water and treated using Franklin solution (30% hydrogen peroxide and acetic acid, 1:1 v v -1 ) for 8 h at 70°C.The solution was decanted and the remaining fibrous material was washed with water until a neutral pH was achieved.Average fibre length, fibre width, fines content, and coarseness (defined as fibre mass per fibre length) were determined using L&W Fiber Tester equipment (Lorentzen & Wettre, Sweden).L&W Fiber Tester is an instrument for advanced analysis of fibre dimensions.It has a sample feeder where the pulp sample is introduced to the equipment as a suspension for two-dimensional imaging analysis.It consists in two plates that allow the fibres to move freely in two dimensions where a camera captures fibre images for dimension measurements.200 mg of sample were previously disaggregated in 200 mL of distilled water for 10 min.During the analysis of this suspension, the equipment was set to measure approximately 35000 fibres of each sample, setting as fines to elements of 0 to 0,2 mm of length to ensure that broken fibres and fines are not included in the final averages of fibre measurements.Each sample was analyzed in duplicate.

Lignin content
Two grams of milled wood samples were extracted with 90% acetone for 16 h.Samples were characterized for lignin by acid hydrolysis following the methodology described by Mendonça et al. (2008).In a test tube were weighed 300 mg of sample and added 3 mL of 72% (w w -1 ) H 2 SO 4 .The hydrolysis was carried out in a water bath at 30°C for 1 h with a glass-rod shaking every 10 min.Later, the acid was diluted to 4% (w w -1 ) with 84 mL of distilled water and the mixture transferred to a 250-mL Erlenmeyer flask and autoclaved for 1 h at 121°C (post-hydrolysis).The residual material was cooled and filtered through a porous glass filter number 3. Solids were dried to constant weight at 105°C and determined as insoluble lignin.Soluble lignin was determined by measuring the absorbance of the solution at 205 nm.Each sample was analyzed in triplicate.

S/G ratio of lignin
The determination of the S/G ratio in the lignin of both species was determined by oxidation with copper (II) oxide carried out following the methodology described by Chen (1992).In an 80-mL stainless steel reactor it was added 400 mg of sample (milled and extracted wood), 2 g CuO, 15 mL NaOH 2 M and 100 mg Fe(NH 4 ) 2 (SO 4 ) 2 .6H 2 O. Nitrogen was bubbled inside the reactor which was tightly closed and immersed in an oil bath for 3 h at 170°C.After the reaction, it was added to the reactor 10 mL NaOH 1 M and 10 mL of distilled water.The mixture was acidified to pH 1 with HCl.The content of the reactor was transferred to a centrifuge tube and centrifuged at 2500 rpm for 10 min.The liquid fraction was collected, transferred to a separation funnel and extracted three times with 50 mL of diethyl ether.The organic fraction was further evaporated (40°C) at reduced pressure (450 mbar).The solid residue was dissolved in 1 mL pyridine and 0,5 mL of sample was silylated with 0,5 mL BSTFA.Derivatization products were quantified by GC/FID using the conditions published elsewhere.The amount of syringyl (S) units was the sum of syringaldehyde and syringic acid; and the amount of guaiacyl (G) units was the sum of vanillin and vanillic acid.The amount of S compounds was divided by the amount of G compounds to determine the S/G ratio in lignin.Each sample was analyzed in triplicate.

Cellulose and hemicelluloses determination
The content of cellulose was performed following the Kûrshner-Höffer method, occupied by Carballo et al. (2004).On 0,15 g of extractive-free milled wood, 25 mL reactive mixture of HNO 3 :C 2 H 6 O (1:4) was refluxed in a water bath for 1 h, decanted and a new amount of reaction mixture was added, repeating this operation three times, subsequently, 25 mL of 1% KOH was added for 30 min, the residual material was filtered through porous glass filter number 2. Solids were dried to constant weight at 105°C and determined as wood cellulose.Each sample was analyzed in triplicate.The hemicelluloses were quantified by acid methanolysis according Sundberg et al. (1996).Extractive free wood meal was freeze dried prior to weighing 10 mg into a pear-shaped flask.Samples were subjected to acid methanolysis by the addition of 2 mL of 2 M solution of HCl in anhydrous methanol.Samples were kept in an oven at 100°C for 3 h.After cooling to room temperature, 100 µL of pyridine was added to neutralize the acidic solution as well as 4 mL of methanol (containing sorbitol at 0,1 mg/ mL as an internal standard).To avoid fibres during silylation, 1 mL of the clear sample solution was transferred into another pear-shaped flask and the solution reduced by rotary evaporation at 40°C.Samples were dissolved in 100 µL pyridine.For silylation, 150 µL hexamethyldisilazane (HMDS) and 80 µL trimethylchlorosilane (TMCS) were added prior to thorough shaking of the sample.After 4 h at room temperature, samples were analysed by GC-FID.One µL of a silylated sample was injected via a split injector (260°C, split ratio 1:20) into a 30 m x 0,25 mm i.d.x 0,25 µm film thickness column DB5.The column temperature program was 100°C to 175°C (4°C min -1 ) followed by 175°C to 290°C (12°C min -1 ).The detector (FID) temperature was 290°C.Nitrogen was used as carrier gas.Different peaks were identified by analysing acid methanolysis products of analytical grade sugars (arabinose, xylose, galactose, glucose, mannose, rhamnose, glucuronic acid and galacturonic acid).Calibration curves and factors were determined for each sugar unit in order to calculate the concentration in the samples.Each sample was analyzed in triplicate.

Holocellulose and alpha-cellulose contents
Holocellulose content was determined in extractive-free wood using 250 mg weighed into a 50-mL flask where 5 mL of deionised water, 2 mL of glacial CH 3 COOH and 5 mL of 80% NaClO 2 were added.The flask was closed with a glass cap and was immersed in a water bath at 90°C for 1 h.Subsequently, An exploratory evaluation of the ..: Sangumbe et al. more 2 mL of glacial CH 3 COOH and 5 mL of 80% NaClO 2 were added to the flask, and the reaction was maintained for 1 h at 90°C.The reaction was quenched by cooling the sample in a water bath at 10°C.The solids were filtered through porous glass filter number 2, washed with 500 mL of deionised water, dried at 105°C until constant weight and determined as holocellulose.To determine the alphacellulose content, 100 mg of holocellulose were placed in a 25-mL flask, which was treated with 8 mL NaOH 17,5% (w v -1 ) for 30 min at room temperature with shaking every 10 min.Then, 8 mL of distilled water was added to the solution and the reaction was carried out for another 30 min.The solids were filtered; the sample was washed with 150 mL of distilled water and impregnated with 20 mL of 1 M CH 3 COOH for 5 min.The residue was washed with abundant distilled water and dried at 105°C until constant weight for the quantification of alpha-cellulose (Yokoyama et al. 2002).Each sample was analyzed in triplicate.

Kraft pulping
Kraft pulping was performed in a rotatory digester equipped with four independent 1,5-L reactors (Regmed, Brazil).For each reaction, 100 g of wood chips (dry basis) and cooking liquor with active alkali (AA) concentrations from 14% to 25% and 30% sulfidity (both expressed in NaOH basis) was used.Heating time to the maximum temperature (165°C) was 90 min and the H-factor was 800.The resulting material from each cooking was disintegrated and pulps were screened through a 0,2 mm slot screen.The pulp was centrifuged to 35% consistency and weighted.The exact moisture was determined and the screened pulp yield was calculated.Kappa number was determined according TAPPI T236 om-99.Strength properties were determined in unrefined and PFI-refined pulps (2500 and 5000 rpm) following TAPPI standard methods for sheet formation (T220 sp-00), tensile index (T404 om-92), tear index (T414 om-98) and burst index (T403 om-97).

Data analysis
Statistical analyses of chemical, anatomical and pulp properties were performed using the software SAS system 9.0 (SAS Institute).Unpaired t-test was used to compare the properties between species.

Anatomical characterization of wood
Transversal micrographs showing the cellular anatomy of B. spiciformis wood are presented in Figure 1.It can be observed that the vessels are distributed regularly throughout the xylem and are surrounded by aliform axial parenchyma.However, the proportion of parenchyma surrounding the vessel elements is lower in sapwood (Figure 1A) than in heartwood (Figure 1B).It has been showed that the distribution of aliform and paratracheal axial parenchyma arranged as bands gives place to false rings, which also can be associated to the production of fibres with high cell wall thickness (Trouet et al. 2001, Grundy 2006).In Figure 1C and Figure 1D, it can be observed the fibre anatomy of sapwood and heartwood, where cell wall thickness is higher in sapwood than in heartwood fibres.It must be also noted that fibre and vessels differences from earlywood and latewood of a growth ring could not be detectable, which is in agreement with the studies that report difficulties to determine cross-dating by tree ring Brachystegia analysis (Trouet et al. 2001, Grundy 2006).This is probably due to the ability of the large root system to store water in the dry season (Grundy 2006).Regarding P. angolensis, it can be observed a diffuse distribution of vessels in sapwood (Figure 1E) and in heartwood (Figure 1F).However, in heartwood, a higher number of vessels are filled with extractives deposits.Fibre structure of sapwood (Figure 1G) and heartwood (Figure 1H) showed a high cell wall thickness with occluded fibre lumen in both sections.This feature could cause problems for impregnation and diffusion of reagents inside the lignocellulosic matrix during chemical treatments.Similar anatomical features have been reported by Ali et al. (2008) and Uetimane et al. (2009), who give a more detailed description of the arrangement of xylem cells in P. angolensis.Similar to B. spiciformis observation, earlywood and latewood in P. angolensis could not be detectable.
In Table 1 are presented the features of sapwood and heartwood from B. spiciformis and P. angolensis.The basic wood density of B. spiciformis and P. angolensis sapwood was higher than heartwood basic density, while wood basic density average for B. spiciformis was 640 kg m -3 and for P. angolensis was 795 kg m -3 .For P. angolensis, it has been reported wood density values of 758 kg m -3 (Abbot and Lowore 1999), 865 kg m -3 (Ali et al. 2008), 941 kg m -3 (Uetimane et al. 2009), 920 kg m -3 (Lhate et al. 2010) and 865 kg m -3 (Cuvilas et al. 2014), which is in agreement with our results.Abbot and Lowore (1999) reported for B. spiciformis a wood density of 579 kg m -3 , including also reports for Brachystegia boehmii with 598 kg m -3 , Brachystegia utilis with 598 kg m -3 , Brachystegia longifolia with 548 kg m -3 and Brachystegia floribunda with 676 kg m -3 .Also, it was found reports for B. eurycoma with wood density of 600 kg m -3 (Afe 2016a) and 642 kg m -3 (Noah et al. 2014).All these findings are in agreement with the density value found in this study for B. spiciformis.
Fibre features showed higher values for cell wall thickness in sapwood from both species, being P. angolensis fibre cell walls thicker than B. spiciformis cell walls (Table 1).Lumen width is also different in heartwood and sapwood from both species, but P. angolensis lumen width is occluded as it was already mentioned.Fibre width showed no differences at the different sections of B. spiciformis, which is not the same behaviour for P. angolensis observations.Regarding vessels features, the area of vessels, vessel coverage and vessel width is higher in sapwood than in heartwood of both species; while B. spiciformis showed higher mean area of vessel and higher vessel width average than P. angolensis.The radial variation pattern of anatomical properties in hardwoods has been widely report.It is known that the size of vessels increases as cambial age increases (Hudson et al. 1998, Leal et al. 2003, Carrillo et al. 2015).Same pattern of increasing values from pith to bark is known for fibre width, cell wall thickness, fibre length, wood density and coarseness (Miranda and Pereira 2002, Ohshima et al. 2004, Quilhó et al. 2006, Ramírez et al. 2009, Carrillo et al. 2015).These assertions are in agreement with the results found for the sapwood section, close to the bark, and the heartwood section, close to the pith, in B. spiciformis and P. angolensis.Different letters indicated significant differences between sections of same species and between species (p<0,05).

Derived wood properties
Derived values of B. spiciformis and P. angolensis wood properties are shown in Table 2.These derived wood properties are usually used to predict pulp and paper properties through fibre morphology.In papermaking with hardwood fibres, Runkel ratio lower than 1.0 is desirable for a good paper conformability (Dean 1995, Ona et al. 2001, Ohshima et al. 2005, Azeez et al. 2016) due that fibres are considered as thin walled fibres and good mechanical strength properties are usually obtained (Dutt and Tyagi 2011).Both B. spiciformis and P. angolensis Runkel ratios are greater than 1,0 which indicates that fibres obtained from this wood species could not be suitable for paper production (Xu et al. 2006), however, an index from 1 to 2 have been considered acceptable for papermaking (de Almeida et al. 2016).
Luce's shape factor is related to paper sheet density (Ona et al. 2001, Ohshima et al. 2005).In 14-year-old E. globulus trees, values reported were between 0,297-0,329 (Ona et al. 2001) and0,390-0,440 (Ohshima et al. 2005).However, both B. spiciformis and P. angolensis Luce's shape factor values are higher than the reported for hardwoods and the reason may be associated with the cell wall thickness, since both fibre width and fibre lumen width are used to obtain the cross-sectional fibre wall area in the equation for Luce's shape factor (Ohshima et al. 2005).
The slenderness ratio is an important parameter related to the physical properties of handsheets such as strength, tear index, burst index and breaking length (Tutus et al. 2015).According to those physical properties, the desirable slenderness ratio is between 70-90 in softwoods and 40-60 in hardwoods (Akgul and Tozluoglu 2009).Hence, the ratio obtained in B. spiciformis and P. angolensis was 65,73 and 59,85 respectively.Slenderness ratio in B. eurycoma have been reported 71,23 (Olufunmilayo 2013), while in Eucalyptus species have been reported between 42-66 (Ona et al. 2001, Ohshima et al. 2005, Dutt and Tyagi 2011).

Chemical composition of wood
The chemical composition of wood from B. spiciformis and P. angolensis is presented in Table 3.The results showed that existed important differences in the amount of the main components present in both species.The cellulose, holocellulose, alpha-cellulose and acetone-soluble extractives contents, as well as the S/G ratio, were higher in B. spiciformis than in P. angolensis, while the opposite occurs for hemicelluloses and lignin contents.Interestingly, the S/G ratio of P. angolensis is very low (0,89) indicating that the lignin of this hardwood species has higher amount of G-type, instead of S-type units, which would worth a more detailed characterization of lignin in this species in a future work.This is remarkable due that high S-type units promote the delignification and decrease recondensation during the kraft pulping, facilitating the delignification and bleaching processes (del Río et al. 2005, Pinto et al. 2005, Carrillo et al. 2017).Therefore, B. spiciformis should show a better performance during alkaline pulping procedures due to its lowest lignin content and highest S-type units amount than P. angolensis, in addition to its higher cellulose and alpha-cellulose content.In hardwoods with high pulpability, as E. globulus genotypes, the S/G ratio reported is higher (between 2,0-5,5) while lignin and cellulose contents are in agreement with these values (Guerra et al. 2008, Aguayo et al. 2015).
On the other hand, extractives content affects negatively the pulping procedures, causing increased reagents consumption, inhibition reactions of the delignification process, equipment corrosion and An exploratory evaluation of the ..: Sangumbe et al.

Kraft pulping
Kraft pulping of wood chips from B. spiciformis and P. angolensis was performed at an H-factor of 800 and different active alkali charges (14% to 25%) with the aim to obtain bleachable-grade pulps (Figure 2A).For the most common hardwood species used to produce bleached pulps, such as Eucalyptus, acacia or birch, active alkali charges of 16%-24% are usually enough to generate bleachable-grade pulps with kappa number around 16-19 and pulp yield of 50-55% (Pinto et al. 2005, Aguayo et al. 2010).In the present case, for both species, the demand of active alkali was higher (up to 25%) and pulps with high kappa number (24 for B. spiciformis and 27 for P. angolensis) were obtained, at the cost of a lower screened pulp yield (approximately 40%) for both species (Figure 2B).P. angolensis was harder to pulping than B. spiciformis since it presented higher kappa number during cooking at different alkali charges, associated with lower pulping yield (Figure 2A and Figure 2B, respectively).The main factors that affect the lower pulpability of P. angolensis could be related with the anatomical features.High-density woods (more than 600 kg m -3 ) are usually associated with tylosis formation or extractives deposition in lumens that lead to an irregular delignification of wood chips.Which is mainly due to the poor efficiency of the impregnation with the cooking liquors (Ramírez et al. 2009).Anatomically, P. angolensis wood also presented occluded fibre lumen, which can represent an important drawback for reagent diffusion during pulping procedures.Chemical features such as, extractives amount, lignin content and S/G ratio are also known to affect the delignification rate and alkali consumption.As already mentioned, P. angolensis showed the higher lignin content and higher G-type units (Table 2).G-type lignin contains more resistant linkages than S-type units, making lignin less reactive and resistant to chemical degradation (Boerjan et al. 2003, Pinto et al. 2005, del Río et al. 2005, Rencoret et al. 2008).Unbleached kraft pulps were refined at 2500 and 5000 revolutions in a PFI mill to increase internal and external fibrillation of pulp (Figure 3A).Only B. spiciformis was able to be refined at values higher than 20 o SR (usually required for high strength papers) achieving 24 o SR at 5000 rpm.P. angolensis was harder to refine and only poor fibrillation was obtained at 5000 rpm (10 o SR).The main reason of the poor fibrillation of P. angolensis may be related with the morphological features of its fibres.As was mentioned, its highest cell wall thickness and occluded lumen contribute to the decreasing of fibres elasticity (Runkel ratio of 17,6 and flexibility coefficient of 5,4%), demonstrating that stiff fibres are harder to refine and process.Figure 3B, Figure 3C and Figure 3D shows the strength properties of B. spiciformis and P. angolensis pulps.Tensile, tear and burst indexes of 100,3 Nm g -1 ; 10,7 mN.m 2 g -1 and 6,1 kPa.m 2 g -1 , respectively, were reached for B. spiciformis pulps after refining at 5000 revolutions.
The pulps of P. angolensis after refining at 5000 revolutions reached tensile, tear and burst indexes of 99,6 Nm g -1 ; 5,9 mN.m 2 g -1 and 4,9 kPa.m 2 g -1 , respectively.As was already discussed, the lowest strength properties of P. angolensis were expected, mainly due to the influence of the low slenderness ratio and flexibility coefficients, and high Runkel ratio of its fibres (Table 2).As a reference of strength properties in commercial hardwoods, Guerra et al. (2008) reported for kraft E. globulus pulps tensile indexes of 66-102 Nm g -1 , tear indexes of 6-9 mN.m 2 g -1 and burst indexes 5-8 kPa.m 2 g -1 .
An exploratory evaluation of the ..: Sangumbe et al.In Table 4 are shown some biometric properties of pulp fibres after the refining process as fibre length, coarseness, fines content and kink index.Fibre length is an important descriptor factor of pulp quality, due to its influence to paper strength properties (Ek et al. 2009).Fines content is related with amount of short and thin cells as parenchyma cells and broken fibres, and have been reported that it influences positively the sheet tensile index (Retulainen 1997).The kink index refers to a deformation on the fibre that can be a weak or a breaking point; while coarseness, defined as fibre mass per fibre length, is a good index for predicting pulp properties of fibres and basic density of wood (Via et al. 2004, Mansfield and Weineisen 2007, Carrillo et al. 2015).The fibre length and coarseness of both species decreases after refining, while kink index and fines content increases.These results are expected considering that during refining some fibres are more likely to bend and break, increasing the fines release and kink index of pulp fibres.Particularly, P. angolensis pulps showed the higher values for fibre length, fines content and coarseness, and lower values for kink index during the different refining steps, which is probably due to its higher fibre cell wall thickness.However, despite the physical features of pulp fibres, B. spiciformis showed better strength pulps properties than P. angolensis, which could be considered unexpected according to pulp fibres features.Nevertheless, very-high coarseness and very-high cell wall thickness values can lead to poor conformability and low fibre-to-fibre contact in sheets (Dean 1995), which resulted in poor derived wood properties and low strength properties as in P. angolensis.Different letters indicate significant differences between refined pulps of the same specie (p<0,05).

CONCLUSIONS
P. angolensis and B. spiciformis are species acknowledged by their heavy and hard wood properties.However, P. angolensis wood has higher wood density, higher fibre width and cell wall thickness, and an occluded fibre lumen as remarkable anatomical properties.Regarding chemical features, B. spiciformis has a higher cellulose content, lower hemicellulose and lignin content, and higher S/G ratio than P. angolensis.The kraft pulping procedure of both species required higher alkali charges to achieve bleaching grade pulps, resulting in low pulp yield and high kappa number.The pulps obtained from P. angolensis wood were harder to refine and showed lower strength properties than B. spiciformis, which was attributed to the very-high values obtained for biometric properties and not adequate derived wood properties of wood and pulp fibres of P. angolensis, as wood density, cell wall thickness, Runkel ratio, slenderness ratio, flexibility coefficient and coarseness.It is concluded that B. spiciformis wood has better pulpability than P. angolensis wood, according to its pulps properties, despite of the similar pulp yield between both species.B. spiciformis and P. angolensis may be not suitable for high quality papers, but both species could be useful for unbleached wrapping paper and rigid cardboards.

Table 1 .
Anatomical features of sapwood and heartwood of B. spiciformis and P. angolensis.

Table 2 .
Derived wood properties of B. spiciformis and P. angolensis.

Table 3 .
Chemical composition of wood from B. spiciformis and P. angolensis.