Introduction

Eucalyptus trees are widely used in commercial plantations as raw material for pulp and paper production because of their several desirable features (Carrillo et al. 2017a, b). Consequently, their wood properties have been broadly investigated. Several reports have addressed the morphological, anatomical and chemical characteristics of various Eucalyptus species as well as the respective pulp and paper products (Ramírez et al. 2009; Aguayo et al. 2014; Carrillo et al. 2015, 2017a). Among the different Eucalyptus species, E. globulus has several advantages, including its high wood density and the possibility of manufacturing kraft pulps in high yields, with excellent fiber quality and paper properties (Del Río et al. 2005; Patt et al. 2006). Nevertheless, other Eucalyptus species are also used for papermaking, including E. nitens (and E. nitens × E. globulus hybrids), especially in Chile, because of their frost tolerance (McKinley et al. 2002; Carrillo et al. 2017a) and E. grandis and its hybrids with E. urophylla in Brazil (Dutt and Tyagi 2011).

Bleached Eucalyptus pulp is extensively used in papermaking (printing, tissue and packaging grades) but it is also used for chemical conversion into cellulose derivatives such as viscose, cellulose-acetate, cellulose-xanthate, carboxymethyl cellulose, micro- and nanocrystalline cellulose (Sixta 2006; Clarke et al. 2008; Ek et al. 2009). For these latter applications, the structure of cellulose has a complex but significant influence on the course of chemical reactions towards the production of its derivatives (Klemm et al. 1998; Rojas 2016). During the production of derivatives, cellulose is modified by substitutions at the hydroxyl groups present in the C2, C3 and C6 positions along the anhydroglucose units; in contrast, cellulose is chemically dissolved and then regenerated for viscose production (Ek et al. 2009; Rojas 2016). Thus, bleached pulp properties for production of cellulose derivatives will be determined by the inherent structural features, including inter- and intra-molecular hydrogen bonds, crystallinity and chain organization, which are expected to influence the cellulose accessibility to chemicals and thus its reactivity for further processing and to define the properties of the final products (Klemm et al. 1998; Sixta 2006; Poletto et al. 2014).

Prior investigations have addressed the structural and thermal properties of cellulose from Eucalyptus woods, including the changes associated to conversion from wood to pulp and to enzymatic hydrolysis. Moreover, comparisons with other vegetal fibers, as far as thermal and structural changes, and their behavior under different chemical treatments have been published (Popescu et al. 2007, 2009; Monrroy et al. 2011; Barneto et al. 2011; Sebio-Puñal et al. 2012; Casas et al. 2013; Poletto et al. 2012a, b, 2014; Tonoli et al. 2012, 2016). However, despite its critical role in defining processing conditions and products after given chemo-mechanical and thermal transformations, few information is available about the variations in the structure of cellulose for different Eucalyptus species. Therefore, this work attempts to systematically compare the main structural features in milled wood, holocellulose and alpha-cellulose from seven Eucalyptus species via Fourier transform infrared spectroscopy (FTIR), X-ray diffraction analysis (XRD) and thermogravimetric analysis (TGA). The results are expected to be useful in designing strategies to obtain cellulose from the Eucalyptus genre in order to develop new forest products, especially those derived from dissolving pulps.

Materials and methods

Eucalyptus wood samples

Six-year-old Eucalyptus species were provided by a Chilean forestry company located in the Biobío Region (Southern Chile). The Eucalyptus species, corresponding to Eucalyptus badjensis, E. benthamii, E. dunnii, E. globulus, E. nitens, E. smithii and two hybrids E. nitens × E. globulus, coded En × Eg (1) and En × Eg (2), grew under the same field and plantation conditions. Wood chips from the eight samples were obtained for wood density determination according to the Tappi Standard Method T258 om-94. For chemical characterization, wood chips were milled with a knife mill, sieved to 45/60 mesh and air-dried to 10% moisture content.

Chemical characterization

Two grams of milled wood samples were extracted with acetone:water (9:1) for 16 h in a Soxhlet apparatus (5 cycles/hour) and dried in an oven at 105 °C until constant weight (8 h approximately). The extractives content was determined by weight.

Holocellulose and alpha-cellulose contents were determined according to Yokoyama et al. (2002). Holocellulose was isolated from 250 mg of extractive-free wood samples by adding 5 mL of 80% sodium chlorite and 2 mL of glacial acetic acid for delignification in a water bath at 90 °C for 2 h. The suspension was then filtered through a porous #4 glass filter, washed with deionized water and dried in an oven at 105 °C until constant weight (8 h approximately). The yield of holocellulose was determined by weight, and the samples were stored in a sealed tube. From 100 mg of holocellulose samples, alpha-cellulose was obtained by reaction with 8 mL of 17.5% sodium hydroxide for 30 min at room temperature, followed by reaction with 8.75% sodium hydroxide for another 30 min, by adding 8 mL of distilled water. After reaction, the fiber suspension was filtered through a porous #4 glass filter, washed, soaked for neutralization with 1.0 M acetic acid solution for 5 min, washed again with abundant deionized water and dried at 105 °C until constant weight (8 h approximately). The yield of alpha-cellulose was determined by weight. The samples were stored in sealed tubes for further analysis. Extractive-free wood samples, holocellulose and alpha-cellulose preparations were characterized by their lignin content by acid hydrolysis according to Mendonça et al. (2008), based on hydrolysis with 72% H2SO4 at 30 °C for 1 h. Afterwards, the acid was diluted to 4% (w/w) with distilled water, and the mixture was autoclaved for 1 h at 121 °C. The residual material was cooled and filtered through a porous #4 glass filter. Solids were dried to constant weight at 105 °C and classified as insoluble lignin. Soluble lignin was determined by measuring the absorbance of the solution at 205 nm (Dence 1992). Total lignin was calculated as the sum of insoluble and soluble lignin. The concentrations of glucose and cellobiose in the soluble fraction were determined by HPLC (Merck Hitachi instrument) with and Aminex HPX-87H column operated at 45 °C and eluted at 0.6 mL/min with 5 mM H2SO4 through a RI detector. Glucose and cellobiose served as external calibration standards. The amount of sugar to anhydro monomers were converted using a factor of 0.90 (glucose to glucan) and 0.92 (cellobiose to glucan). The sum of these figures led to the total glucan content in the wood, holocellulose and alpha-cellulose samples. Xylose was converted to xylan using a hydrolysis factor 0.88 (Aguayo et al. 2014). All the chemical analyses were carried out in triplicate.

Fourier transform infrared spectroscopy (FTIR)

FTIR spectra were obtained from a PerkinElmer FTIR spectrometer (PerkinElmer, Inc., Waltham, MA, USA). Five mg of milled wood, holocellulose and alpha-cellulose samples were dispersed in a matrix of KBr, followed by compression to form pellets. The moisture content of the milled samples was lower than 2%. The sample collection was obtained using 64 scans in the range of 4000–400 cm−1 at a resolution of 4 cm−1. The spectra were baseline-corrected and normalized at 2900 cm−1 (CH stretching vibration) (Monrroy et al. 2011) using Spectrum v5.0.1 software (Perkin Elmer, USA). The absorbance intensity of the bands was obtained from a local baseline between adjacent valleys (Oh et al. 2005a; Odabas et al. 2016). FTIR analysis were carried out in duplicate.

Structural characterization

Samples for X-ray diffraction (XRD) analyses were prepared by pressing 50 mg of dried wood, holocellulose and alpha-cellulose powder (45/60 mesh) in a hydraulic press to form pellets. The pellets were placed in a sample holder, and X-ray diffractograms were collected after mounting the sample holder on a SmartLab X-ray diffractometer (Rigaku Co, Japan) with monochromatic CuKα radiation (λ = 0.154 nm) at 40 kV and 30 mA. The intensities were measured in the range of 5° < 2θ < 45°, with scan steps of 0.05°. The apparent XRD crystallinity index (CrI) of the samples was calculated by deconvoluted areas (CrIa) and by peak heights following the Segal method (CrIb). Each sample was analyzed in duplicate.

Deconvolution of XRD patterns were made using PeakFit software (www.systat.com) assuming Gaussian functions for each peak. Iterations were repeated until a maximum F number was obtained. In all cases, the F number was > 10,000, which corresponds to a R2 < 0.997 (Park et al. 2010). The curve fitting was made assuming a broad band at 18.5° 2θ (16° 2θ for cellulose II) as the amorphous contribution to the peak broadening. The CrIa by the deconvolution method (Park et al. 2010) was calculated via Eq. (1):

$$ CrI^{\text{a}} = \frac{{\left( {{\text{A}}_{\text{cryst}} } \right)}}{{{\text{A}}_{\text{total}} }} $$
(1)

where Acryst is the sum of crystalline band areas, and Atotal is the total area under the diffractogram. The CrIb calculated by the method proposed by Segal et al. (1959) was made following Eq. (2):

$$ CrI^{\text{b}} = \frac{{\left( {{\text{I}}_{200} {-}{\text{I}}_{\text{am}} } \right)}}{{{\text{I}}_{200} }} $$
(2)

where I002 is the maximum intensity of the (200) reflection at 22.3° 2θ (20° 2θ for (020) reflection in cellulose II) and Iam is the intensity diffraction of the amorphous band (18.5° 2θ) (16° 2θ in cellulose II). The lateral crystallite size (L) was calculated from the Scherrer Eq. (3) (Nam et al. 2016).

$$ L = \frac{{k \times\uplambda}}{{\upbeta \times {\text{cos}}\uptheta }} $$
(3)

where L is the size of crystallite (nm), k is the Scherrer constant (0.96), λ is the X-ray wavelength, β is the full width half maximum (FWHM) of (200) reflection in radians, and θ is the Bragg angle corresponding to the (200) plane.

Thermogravimetric analysis

The thermogravimetric analysis was carried out on a TGA Q500 (TA Instruments, USA) under constant nitrogen flow (50 mL/min) from 25 to 800 °C at a heating rate of 20 °C/min to obtain the degradation profile of samples. The weight-loss rate was obtained from derivative thermogravimetric (DTG) data. Approximately 10 mg of milled (45/60 mesh) wood, holocellulose and alpha-cellulose samples were used. Each sample was analyzed in duplicate.

Results and discussion

Eucalyptus wood characteristics

The characteristics of the different Eucalyptus wood species studied are shown in Table 1, as the yield and chemical composition from their respective holocellulose and alpha-cellulose isolations. Wood density values ranged from 420 to 484 kg/m3, with E. globulus and E. smithii being the highest and E. badjensis the lowest one. These values are in agreement with whole-tree average densities reported by McKinley et al. (2002) for 8-year-old E. globulus and E. nitens (476 and 440 kg/m3, respectively). Eucalyptus globulus wood showed the lowest content of lignin (23.2%) and extractives (1.5%), the highest glucans content (56.9%) and the highest holocellulose yield (73.5%). Eucalyptus benthamii wood showed the highest content of lignin (27.7%) and extractives (6.7%), the lowest content of glucans (51.2%) and the lowest holocellulose yield (63.6%). Regarding alpha-cellulose isolation, the highest content obtained was 55.5% for E. dunnii, while the lowest content was that for E. badjensis (48.5%). The chemical characterization of the species evaluated was in agreement with other chemical data published for Eucalyptus trees (Cetinkol et al. 2012; Gomes et al. 2015; Carrillo et al. 2017b). Regarding holocellulose samples, the glucans content ranged from 67.7 to 73.5%, the xylan amount was in the range 10.1–14.1%, and the residual lignin between 4.7 and 6.1%. In alpha-cellulose samples, the glucans content was close to 100%, and the residual lignin and xylans were lower than 1%. Insoluble lignin was not detected in holocellulose and alpha-cellulose samples.

Table 1 Basic density and chemical characteristics of the Eucalyptus wood, holocellulose and alpha-cellulose samples
Full size table

Fourier transform infrared spectroscopy (FTIR)

FTIR spectra were obtained for wood, holocellulose and alpha-cellulose samples from the Eucalyptus species studied (Fig. 1). The FTIR spectra can be separated in two regions; the 4000–2700 cm−1 “informative” region, that correspond to OH and CH stretching vibrations; and the 1800–800 cm−1 “fingerprint” region, that is assigned to different stretching vibrations of different groups in wood (Popescu et al. 2007, 2011; Poletto et al. 2012b). In the first informative region, a strong broad band at 3400 cm−1 was noted, which is assigned to different OH stretching modes (Tsuboi 1957; Liang and Marchessault 1959a, b; Marchessault and Liang 1960; Fengel 1992, 1993; Kondo 1997) and a band at 2900 cm−1 that is related with asymmetric and symmetric CH stretching (Tsuboi 1957; Liang and Marchessault et al. 1959a, b; Marchessault and Liang 1960; Collier et al. 1997; Schwanninger et al. 2004). The CH stretching region is rarely analyzed in detail, because cellulose, hemicelluloses and lignin have similar peak shapes at 2900 cm−1 (Schwanninger et al. 2004; Barnette et al. 2012). However, the absorption contribution at 2900 cm−1 is invariant and has been used as an internal standard for relative measurements of the cellulose structural order in samples of pure cellulose or in given sets with same composition (Nelson and O’Connor 1964; Poletto et al. 2014; Odabas et al. 2016). Regarding the OH stretching region (3400 cm−1), the broad band observed for all the Eucalyptus samples resulted from variations in the degree of hydrogen bonding between hydroxyl groups in the cellulose crystal/aggregated states, hemicelluloses, lignin and water (Fengel 1993; Kondo 1997; Oh et al. 2005a, b; Popescu et al. 2011; Cintrón and Hinchliffe 2015). Given that lignin and hemicelluloses were removed in holocellulose and alpha-cellulose preparations, respectively, the relative reduction of absorbance of the hydrogen-bonded OH stretching region was taken as indication of more organized cellulose. Higher spatial order in spectroscopy was also taken to result from sharper features, which have reduced band width because the vibrations contribute at the same frequency. Thus, a reduction in overall area intensity in the OH region was observed.

Fig. 1
figure 1

FTIR spectra for wood, holocellulose and alpha-cellulose of the Eucalyptus samples

Full size image

In addition, for the alpha-cellulose samples of all the Eucalyptus wood species, a change in the band shape and a shift to higher wavenumbers occurred at the maximum absorbance for hydrogen-bonded OH stretching region (Table 2). It has been already demonstrated in cellulosic samples that this behavior in the OH stretching region of the Eucalyptus samples reveals the decrease of the intramolecular hydrogen bonding due to the mercerization process (Fengel et al. 1995; Oh et al. 2005b).

Table 2 OH stretching band shift after milled wood treatment for holocellulose and alpha-cellulose preparations from Eucalyptus samples
Full size table

Concerning to the “fingerprint” region (Fig. 1), the band around of 1740 cm−1 was assigned to C=O stretching vibrations of the carboxyl and acetyl groups in hemicelluloses (O’Connor et al. 1957; Harrington et al. 1964; Faix 1991); the band at 1600 cm−1 was assigned to C=C stretching of the aromatic ring (S) in lignin (Hergert and Kurth 1953; O’Connor et al. 1957; Harrington et al. 1964; Faix 1991) and is overlapped with that at 1642 cm−1, which is associated to adsorbed water in cellulose, hemicelluloses and lignin (Tsuboi 1957; Liang and Marchessault et al. 1959c; Kacurakova et al. 1998); whereas the band at 1510 cm−1 was associated to C=C stretching of the aromatic ring (G) in lignin (Harrington et al. 1964; Faix 1991).

In holocellulose and alpha-cellulose samples, the bands related with lignin (bands at 1600 cm and 1510 cm−1) were absent in the spectra. However, the chemical characterization evidenced residual amounts of soluble lignin in holocellulose samples (between 4.7 and 6.1%), and less than 1% in alpha-cellulose samples (Table 1). The band assigned to hemicelluloses (1740 cm−1) decreased in the alpha-cellulose samples, which was corroborated by the compositional analysis of alpha-cellulose (Table 1). The band at 1430 cm−1 was related to CH2 bending mode of cellulose I (Tsuboi 1957; Liang and Marchessault 1959c; Marchessault and Liang 1960), the band at 1372 cm−1 was assigned to CH bending in cellulose I and cellulose II and hemicelluloses (Tsuboi 1957; Liang and Marchessault 1959c; Marchessault and Liang 1960), the band at 1165 cm−1 was associated with C–O–C asymmetric stretching vibrations in cellulose I and cellulose II (Liang and Marchessault 1959c; Hurtubise and Krassing 1960; Marchessault and Liang 1960; Kataoka and Kondo 1998), and the band at 897 cm−1 was assigned to vibrational mode involving C1-H in cellulose, which is stronger for cellulose II and amorphous cellulose (Tsuboi 1957; Mckenzie and Higgins 1958; Hurtubise and Krassing 1960; Marchessault and Liang 1960; Higgins et al. 1961; Nelson and O’Connor 1964; Schwanninger et al. 2004). It was observed, for all Eucalytpus species, that the relative absorbance of the band at 897 cm−1 increased from wood and holocellulose to alpha-cellulose samples. In contrast, the bands at 1430 and 1372 cm−1 decreased. This is explained if the sample undergoes mercerization (Hurtubise and Krassing 1960; Nelson and O’Connor 1964; Oh et al. 2005a, b). However, we note that the results for the different types of samples (from wood and holocellulose to alpha-cellulose) could be also affected by large variations in their chemical composition.

Relative crystallinity based on FTIR analysis

For the determination of crystallinity, the infrared (IR) crystallinity ratios proposed by O’Connor et al. (1958) and Nelson and O’Connor (1964) were determined. It was calculated the ratio between the bands at 1372 and 2900 cm−1, known as total crystalline index (TCI); and the ratio between the bands at 1429 and 897 cm−1, known as a lateral order index (LOI), which is defined as an empirical crystallinity index. TCI is proportional to the crystallinity degree of cellulose and LOI correlates with the overall degree of order in cellulose (Carrillo et al. 2004; Poletto et al. 2012b). Table 3 includes TCI and LOI ratios obtained for the different Eucalyptus samples. The changes in IR crystallinity ratios for the different Eucalyptus species, after holocellulose and alpha-cellulose isolation, showed different behavior. This is attributed to the different structure of the native cellulose of the Eucalyptus species, as well as lignin, hemicelluloses and extractives structure and content, which may mask the FTIR crystallinity (Poletto et al. 2014; Karimi and Taherzadeh 2016). Thus, the results are not comparable among samples (across columns in Table 3). However, comparisons are possible within each group of samples, given the small variation between the components (within a single column in Table 3).

Table 3 TCI and LOI ratios obtained for wood, holocellulose and alpha-cellulose from Eucalyptus samples
Full size table

TCI ratio in wood samples ranged from 0.304 (E. nitens) to 0.415 (E. dunnii), in holocellulose it ranged from 0.305 (E. badjensis) to 0.496 (En × Eg (2)), while in alpha-cellulose it ranged from 0.247 (E. nitens) to 0.341 (E. benthamii). Regarding LOI ratio, it ranged from 1.100 (En × Eg (2)) to 2.208 (E. dunnii) in wood samples, while in holocellulose samples it ranged from 0.967 (E. smithii) to 2.353 (E. benthamii). In alpha-cellulose samples, the LOI ratio ranged from 0.517 (E. badjensis) to 0.770 (E. nitens). These values are in agreement with the IR ratios published elsewhere for different vegetal fibers (Adel et al. 2011; Poletto et al. 2014; Odabas et al. 2016). However, low values were observed for the LOI ratio in alpha-cellulose samples, which was expected, since the band at 897 cm−1 is stronger for cellulose II and amorphous cellulose. During sodium hydroxide treatment, a degradation of the hydrogen bonds between the cellulosic chains occurs, since the sodium hydroxide penetrated and swelled the cellulose fibers, causing decrystallization and changing the crystal polymorph from cellulose I to cellulose II (Fengel et al. 1995; Fink et al. 1995; Odabas et al. 2016).

X-ray diffraction (XRD)

X-ray diffraction pattern of the Eucalyptus samples are shown in Figs. 2 and 3. The diffraction patterns observed in wood and holocellulose samples (Fig. 2) are consistent with cellulose I lattice. Wood and holocellulose diffractograms showed the 14.8°, 16.5°, 22.3° 2θ reflections assigned to the (1-10), (110), and (200) crystallographic plane, the 18.5° 2θ reflections of the amorphous phase, as well as the 34.5° 2θ reflection assigned to the (004) plane of cellulose I (Wada et al. 2001; Popescu et al. 2011; French 2014). Alpha-cellulose samples showed diffraction patterns consistent with cellulose II lattice (Fig. 3). The 12° 2θ, 20° 2θ and 21.6° 2θ reflections are assigned to the Miller indices of (1-10), (110) and (020), respectively, besides to the 34.5° 2θ reflection assigned to the (004) plane of cellulose II (French 2014).

Fig. 2
figure 2

XRD diffractograms for wood and holocellulose of the Eucalyptus samples

Full size image
Fig. 3
figure 3

XRD diffractograms for alpha-cellulose of the Eucalyptus samples

Full size image

Table 4 shows the apparent lateral crystallites sizes (L) determined by the Scherrer equation, and the apparent crystallinity degrees obtained by deconvolution (CrIa) and the Segal (CrIb) methods for wood, holocellulose and alpha-cellulose from Eucalyptus samples. The results are in agreement with XRD indices published elsewhere (Popescu et al. 2007, 2011; Poletto et al. 2014). It was noticed that the degree of crystallinity by the Segal method gave higher values to those obtained by the deconvoluted areas of XRD patterns, as has been also reported by Park et al. (2010) and Xu et al. (2013). However, the variation of both XRD CrI was similar among the different Eucalyptus samples. Pearson correlation index between both methods were r = 0.66 (p = 0.07), r = 0.74 (p < 0.05) and r = 0.74 (p < 0.05) in wood, holocellulose and alpha-cellulose samples, respectively. The highest CrI for wood was found for E. dunnii and En × Eg hybrids samples, while the lowest values were observed for E. badjensis, E. nitens and E. smithii. In holocellulose samples, the highest CrI was for E. benthamii, E. dunnii and both En × Eg hybrids, while E. badjensis and E. smithii gave the lowest values. For alpha-cellulose, the highest CrI was found for E. smithii, and the lowest for E. nitens and both En × Eg hybrids. Agarwal et al. (2017) reported that crystallinity values depend on the moisture content of the samples and diffractograms should be acquired under similar moisture conditions. For this reason, all the samples were analyzed the same day under similar humidity and temperature conditions. The moisture content of the samples was measured before and after XRD analyses and, in all the cases, the moisture content was lower than 2%.

Table 4 Apparent crystallinity degree (CrI) and lateral crystallite size (L200) of wood, holocellulose and alpha-cellulose from Eucalyptus samples and other species
Full size table

The results obtained from XRD differ from those obtained from FTIR, and no significant correlations could be established between IR and XRD CrI, as has been mentioned in other studies (Oh et al. 2005a; Liu et al. 2012). This is explained by the fact that FTIR is sensitive to the polymer chain’s chemical environment, while XRD is affected by the crystal lattice structure. Thus, XRD is better suited to investigate the crystalline structure, while FTIR can be considered as an indirect method (Park et al. 2010; Fan et al. 2012; Xu et al. 2013; Karimi and Taherzadeh 2016). An increase of both XRD CrI after holocellulose isolation, and a reduction in the case of alpha-cellulose samples were observed. However, the CrI by the Segal method for the alpha-cellulose samples decreased only slightly, in comparison to the CrI deconvolution method values. This is attributed to underestimation of the amorphous portion in cellulose II by the Segal method, since the peak height method fails to measure the amorphous portion of cellulose (Nam et al. 2016). The peak deconvolution method is more appropriate for mercerized samples (Park et al. 2010).

Concerning to the increase of XRD CrI after the holocellulose treatment, sodium chlorite and acetic acid were used to remove lignin. However, in XRD, lignin contributes partly in the same region as cellulose, being similar to amorphous cellulose (Agarwal et al. 2013), contributing to the XRD CrI increasing in holocellulose samples. In addition, the delignification treatment could have also removed the less ordered carbohydrates and some hemicelluloses portions, reducing the amorphous portion of the samples and leading to a higher XRD CrI than that for the wood samples (Rozmarin et al. 1977; Evans et al. 1995; Gumuskaya and Usta 2002; Duchemin et al. 2012). Agarwal et al. (2016) showed that upon hydrothermal treatment and compared to its native state, cellulose becomes more consolidated and partly crystalline. The authors suggested that the native cellulose in wood is not crystalline but organized in aggregates, and that thermal treatment is responsible for the hornification/crystallization. In our case, all the samples were treated at 105 °C until constant weight, and it is possible that a significant portion of the non-crystalline cellulose in wood remains in the original, aggregated state. During the holocellulose preparation (mild acidic conditions and temperature), it is possible that an hornification/crystallization process may occur, contributing to the XRD CrI increase in holocellulose samples. Some authors also reported that non-crystalline thin fragments of cellulose microfibrils contribute to the same regions of XRD as the crystalline cellulose (Li and Renneckar 2011; Su et al. 2015). Thus, other aggregated states of cellulose, not necessarily crystalline, might contribute to the 2θ scattering region of the (200) reflection.

Nam et al. (2016) reported that the Segal CrI depended on FWHM, crystallite size and cellulose polymorph; while Agarwal et al. (2016), based on Raman spectroscopy studies, suggested that the XRD FWHM can be taken as a measure of the existing disorder in cellulose, which can be considered as the degree of lateral order (DOLO), related with the CrI. In Fig. 4 the FWHM was plotted against XRD CrI for wood and holocellulose samples. Both XRD CrI showed a significant correlation with FWHM. Though, CrI-Segal was found to be highly correlated (R2 = 0.987), which is in agreement with Agarwal et al. (2016) and Nam et al. (2016). Accordingly, changes in FWHM are related to change in the crystal size (Nishiyama et al. 2012; Nam et al. 2016), but also may involve changes in DOLO (Agarwal et al. 2016) instead of crystallinity, as could be our case, if it is assumed that the aggregated state of wood and holocellulose samples is not crystalline.

Fig. 4
figure 4

Regression profiles of FWHM (200) versus crystallinity degree (Segal-CrI and Deconvolution-CrI) of wood and holocellulose samples

Full size image

Regarding the lateral crystallite size (L), close relationship with the XRD CrI of wood and holocellulose samples was observed. However, for the alpha-cellulose samples, an increase in the average crystallite size (L020) was observed. A higher CrI and L200 was also observed in unbleached kraft pulp from E. globulus as compared with the respective wood sample (Popescu et al. 2007). In these cases, the variations could be attributed to the removal of lignin, hemicelluloses and less ordered carbohydrates (Evans et al. 1995; Popescu et al. 2007), which could entail a degradation of small crystallites that increase the average lateral crystallite size of the holocellulose samples. However, based on NMR, Hult et al. (2000) reported that cellulose became more crystalline after kraft pulping procedures but not necessarily from the preferential removal of small fibrils or disordered cellulose. On the other hand, the holocellulose treatment and kraft pulping are different procedures, holocellulose isolation is conducted at 90 °C in mild acidic conditions, whereas kraft pulping is performed at 170 °C in alkali conditions. Accordingly, the L200 value reported by Popescu et al. (2007) for E. globulus unbleached kraft pulp was 3.50 nm, while our holocellulose samples showed lower values (2.4–2.9 nm). Therefore, considering the L200 values reported for wood samples (Table 4), and that the crystals are less than three molecules thick, it is plausible that the arrangement may not be as crystals but organized aggregated states. Accordingly, the reasons for the increase in L200 values could be the same as those discussed before (hornification/crystallization process from wood to holocellulose samples).

Duchemin et al. (2012) suggested a co-crystallization of crystalline domains, which is initially hindered by the intercalation of non-cellulosic polysaccharide chains between exterior cellulose chains. Therefore, the removal of polysaccharides can promote a co-crystallization of the cellulose within a bigger crystallite. This phenomenon is expected to be more pronounced for alpha-cellulose samples. Since in this case the isolation via alkaline extraction mostly remove hemicelluloses, and low molecular weight and disordered cellulose chains (Isogai and Atalla 1998). Hence, the remaining fraction is mainly constituted of cellulose and only traces of hemicelluloses in the remaining solid are expected, as was confirmed by the reduction of the 1740 cm−1 band intensity in FTIR and compositional analysis. As the FTIR and XRD data indicated, the treatment of cellulose I with aqueous 17.5% sodium hydroxide resulted in the formation of cellulose II. During this process, the alkali solution penetrates the more disordered or amorphous regions, whereby cellulose chains in the amorphous region rearrange into antiparallel cellulose II while the crystalline regions are hardly affected. As the swelling of cellulose continues, the mobility of the cellulose chains is enhanced. In cellulose I, as the formation of cellulose II proceeds, the crystalline cellulose gradually diminishes in size (Nishimura and Sarko 1987; Dinand et al. 2002; Jiao and Xiong 2014). Newman (2004) observed a co-crystallization of crystalline domains upon wetting and drying of cellulosic pulps. As pointed out before, it is possible that native cellulose in wood is not crystalline (Agarwal et al. 2016) whereas evidence for crystallization for hydrothermally heated wood may be explained by the presence of cellulose nanocrystals (CNCs) upon acid hydrolysis. In the absence of such treatment, wood would not produce CNC. Therefore, as alpha-cellulose treatment included a final neutralization and drying process, cellulose crystallization could explain the increase in L020 of the respective samples.

Different Eucalyptus samples could lead to procedures and products with different requirements towards the manufacture of dissolving pulps and cellulose derivatives, since their reactivity and processing are connected with the supramolecular structure of cellulose (Fink et al. 1995). During chemical transformation, most of the reactants penetrate only the amorphous regions and the reactions take place on the surface of crystallites, leaving the crystalline regions unaffected (Ciolacu et al. 2011). Likewise, cellulose reactivity is affected by the crystallite size, as in mercerization, where the lateral crystallite size governs the number of lattice units that have to be swollen for sodium hydroxide penetration (Fink et al. 1995; Dinand et al. 2002). Therefore, the Eucalyptus samples with lowest CrI and crystallite size, as E. badjensis and E. smithii, may be more reactive than both En × Eg hybrids for chemical transformation during cellulose derivatives manufacture. The importance and influence of supramolecular features of cellulosic raw material have been observed in some studies, i.e., Jin et al. (2016) observed differences in alkali concentration upon mercerization of CNC, those differences were attributed to the different crystallite size of CNC samples, which is affected by the fiber source and the treatments applied in their production. In consequence, the effect of the parameters determined in the present work will be evaluated in a further study with dissolving pulps produced from the same species showed here.

Thermogravimetric analysis (TGA)

Thermogravimetric (TG) and derivative thermogravimetric (DTG) profiles obtained from the Eucalyptus wood samples are shown in Fig. 5. The DTG curves of wood, holocellulose and alpha-cellulose samples of E. globulus are shown in Fig. 6, as an example. Water loss was observed at around 100 °C for all Eucalyptus samples. In wood and holocellulose samples, the thermal degradation took place as a two-step process. In the first step, the degradation of hemicelluloses occurred at 250–300 °C. Then, the second weight loss took place at 300–380 °C, due to the degradation of cellulose (Kim et al. 2006; Carrier et al. 2011; Poletto et al. 2012a; Sebio-Puñal et al. 2012). The shoulder associated to hemicelluloses and the prominent peak corresponding to the maximum decomposition rate associated to cellulose was evident in the DTG curves (Fig. 5b). Since lignin degrades over a wide temperature range (250–600 °C), it does not show a characteristic peak (Popescu et al. 2011; Poletto et al. 2014). For the alpha-cellulose samples, the thermal degradation occurred in a one-step process and a single DTG peak was observed (Fig. 6).

Fig. 5
figure 5

TG (a) and DTG (b) curves for the different wood samples of the Eucalyptus. TG and DTG plots were shifted on y-axis in order to avoid overlap between samples

Full size image
Fig. 6
figure 6

DTG curves for wood, holocellulose and alpha-cellulose samples of E. globulus

Full size image

Each of the three major wood components (cellulose, hemicelluloses and lignin) have their own thermal degradation profile, derived from their composition and structural features. Cellulose, as a relatively long polymer, together with its highly organized/crystalline state, displays a high thermal stability due to the higher energy required to start the degradation (Ornaghi Jr et al. 2014; Poletto et al. 2014). Hemicelluloses, which are random amorphous structures that can be hydrolyzed easily, display a higher activity in thermal decomposition (Yang et al. 2006). Lignin, on the other hand, is composed of three kinds of benzene-propane units that are heavily cross-linked, which endows a high thermal and degradation stability (Yang et al. 2006; Popescu et al. 2011; Poletto et al. 2014). However, each of these components behave differently if they are isolated or if they are combined within the wood structure (Popescu et al. 2011; Sebio-Puñal et al. 2012).

Table 5 shows the thermal degradation temperatures and the amount of residual mass at 800 °C for all the Eucalyptus samples studied. The residual mass at 800 °C is attributed to char. The results are in agreement with thermal degradation temperatures observed in other hardwoods species (Popescu et al. 2011; Poletto et al. 2014). Yue et al. (2012) observed in CNCs a decrease in char formation with alkalization. Our results indicated a reduction in char formation after holocellulose treatment, but after alkalization (alpha-cellulose treatment) char content remained almost constant. Wood samples showed a temperature range for hemicelluloses decomposition (DTG shoulder) of 285–289 °C, whereas the main decomposition of cellulose (DTG peak) was observed between 352 and 369 °C. For the holocellulose samples decreased values were observed: DTG shoulder and DTG peak temperature of 259–264 and 341–349 °C, respectively. This results from the influence of lignin elimination, decreasing the thermal degradation temperatures by about 25 °C for the hemicelluloses shoulder, and about 15 °C for the cellulose DTG peak. DTG temperature range for alpha-cellulose was between 338 and 344 °C, decreasing the temperature of cellulose decomposition in about 6 °C after sodium hydroxide treatment.

Table 5 Thermal degradation temperatures and residue (char) at 800 °C of Eucalyptus samples and other species
Full size table

No significant correlations could be established between DTG peak and IR ratios, but a significant correlation existed between XRD CrI and the DTG peak data for the wood samples (Fig. 7). However, for the holocellulose and alpha-cellulose samples, the CrI did not correlate with the thermal stability of cellulose. Likewise, a significant correlation between the DTG peak temperature and the L020 was observed for alpha-cellulose samples (r = 0.76, p = 0.03). This behavior could be related with the supramolecular structure of cellulose, depending of the presence and abundance of aggregates and/or crystalline states of cellulose. This latter possibility has been a subject of recent efforts (Li and Renneckar 2011; Su et al. 2015; Agarwal et al. 2016).

Fig. 7
figure 7

Regression lines and correlation index of the cellulose decomposition temperature (DTG peak) versus crystallinity degree (CrIa) in wood, holocellulose and alpha-cellulose samples of Eucalyptus. NS not significant

Full size image

Poletto et al. (2012a, b, 2014) studied different lignocellulosic fibers, including one E. grandis sample. The authors reported that higher levels of holocellulose and lignin, lower levels of extractives, CrI and higher L200 were associated with a higher thermal stability. In our study, no correlations were found in wood samples among chemical characteristics of Eucalyptus (extractives, holocellulose and lignin contents) and the thermal degradation of wood components. Ornaghi Jr et al. (2014) reported that in some vegetal fibers a higher CrI does not necessarily imply a higher thermal stability, and that the thermal behavior and the degradation mechanism are not influenced by the lignocellulosic components of the fibers. However, it was evident that among the different samples used (from wood to holocellulose and alpha-cellulose), lignin, hemicelluloses and cellulose indeed affected the thermal stability of the lignocellulosic material.

In this study, higher CrI implies a higher thermal stability for all Eucalyptus wood samples, but this does not apply to holocellulose and alpha-cellulose samples (Fig. 7). This kind of relationships may be quite complex, as a result of the differences in the higher order structural forms, i.e., the fibers may have similar CrI, but the crystallites may be ordered differently (Ornaghi Jr et al. 2014), or they may be in highly organized non-crystalline states (Agarwal et al. 2016). The thermal decomposition of lignocellulosic fibers is a complex process that involves competitive and/or consecutive reactions (Poletto et al. 2012a, 2014). As a consequence, it is difficult to distinguish and model the thermal decomposition behavior of wood components due to its complexity. In wood samples, the temperature of cellulose degradation followed the order: E. smithii < E. benthamii < E. badjensis < E. nitens < En × Eg (2) hybrid < E. globulus < E. dunnii = En × Eg (1) hybrid. In holocellulose samples the cellulose degradation temperature was: E. globulus < E. benthamii < E. badjensis = E. dunnii = E. nitens = E. smithii = En × Eg (1) hybrid < En × Eg (2) hybrid. For alpha-cellulose it was: E. smithii < E. globulus < E. badjensis = both EnxEg hybrids < E. nitens < E. benthamii < E. dunnii (Table 5).

As was observed in FTIR, XRD and TGA data, different structural features of cellulose were attained. The cellulose evaluated after holocellulose and alpha-cellulose treatments cannot be considered to be the same as the one in wood (Agarwal et al. 2016), and it does not become uniform after delignification and alkalization treatments.

Conclusions

The cellulose structure and thermal features of wood, holocellulose and alpha-cellulose from different Eucalyptus species were elucidated by FTIR, XRD and TGA. FTIR results showed differences in the hydrogen-bonded OH stretching region between the cellulosic components isolated from the different Eucalyptus species, as well as, differences in the IR crystallinity ratios. FTIR crystallinity differ from that obtained from XRD, and no significant correlations were evident for these techniques. XRD diffraction patterns accessed in wood and holocellulose samples were consistent with the cellulose I polymorph, while the diffraction pattern obtained for alpha-cellulose samples was consistent with the cellulose II lattice. The highest CrI for cellulose I samples were observed in E. dunnii and En × Eg hybrids, while the lowest values were displayed by E. badjensis and E. smithii. Regarding cellulose II (alpha-cellulose samples), the CrI value was highest for E. smithii and lowest for E. nitens and both En × Eg hybrids. In addition, the highest crystallite size was obtained for E. nitens, E. dunnii and E. benthamii alpha-cellulose samples, while the lowest crystallite size was obtained for E. badjensis wood sample. TGA results showed different thermal degradation profiles for the different cellulosic components isolated from Eucalyptus. A two-step degradation process in wood and holocellulose samples, and one-step degradation process in alpha-cellulose were observed. Furthermore, significant correlations were found for XRD CrI versus DTG peak in wood samples, and DTG peak versus crystallite size in alpha-cellulose samples. Overall, it is demonstrated that cellulose obtained from different Eucalyptus species displays different structural and thermal features. This may be of consideration for the production of cellulose derivatives from such species.