Physico -Mechanical Properties of Thermally Modified Eucalyptus Nitens Wood for Decking Applications Fizi č ka i mehani č ka svojstva toplinski modificiranog drva Eucalyptus nitens za vanjske podne obloge

• Eucalyptus nitens is a fast growing plantation species that has a good acclimation in Spain and Chile. At the moment it is mainly used for pulp and paper production, but there is a growing market for solid wood products made from this species. Thermal modi ﬁ cation offers a good alternative to produce high quality material to manufacture products with high added value. This study used unmodi ﬁ ed and thermally modi ﬁ ed E. nitens wood from Spanish and Chilean plantations to elaborate external decking and examine if it complies with the necessary properties to be a competitive product. A process similar to ThermoWood ® was applied at the following temperatures: 185 °C, 200 °C and 215 °C. For each modi ﬁ cation and for an unmodi ﬁ ed specimen mass loss, volumetric swelling, anti-swelling ef ﬁ ciency (ASE) and equilibrium moisture content (EMC) were determined. Brinell hardness, dynamic hardness, screw and nail withdrawal resistance, and abrasion resistance according to the Shaker method and the Taber Abraser method were also determined. According to this study, thermally modi ﬁ ed E. nitens from both countries showed high potential to be used as decking material, particularly when modi ﬁ ed at 200 °C.


UVOD
In the last 30 years there has been a steady growth of Eucalyptus nitens plantations in south of Chile and in the region of Galicia in Spain, as it is a species with great adaptability to frost and colder climate conditions. In Chile, there are about 250,000 ha planted as of 2014 (INFOR, 2015), while in Spain no offi cial data are available. Only the current review for the Forestry Plan of Galicia (Xunta de Galicia, 2018), which analyzed the Spanish Forestry Inventory, and the Spanish Forestry Map suggested for E. nitens in this region forests of approximately 40,000 ha, and proposed an increase of 30.000 ha during the next 20 years. However, this number should be higher, as E. nitens plantations have been replacing E. globulus plantations due to its higher resistance against pathogens and cold.
Currently, E. nitens plantation wood is mostly used for pulp and paper or biofuels, but there is an interest to widen the use of this fast growing tree species. Solid wood made out of plantation E. nitens is being sold at the moment in Chile, but it cannot compete with other fast growing species, as the volume of plantations of radiata pine in Chile (INFOR, 2015) allows the species to maintain a competitive low price. As for Spain, there is still no market for dry E. nitens solid wood, but similar species (E. globulus). Thermal modifi cation offers a good alternative to produce high quality material from this species that could be used for decking, claddings, windows, doors, fl ooring, garden products and even saunas or bathrooms (Militz and Altgen, 2014). Most thermal modifi cation processes apply temperatures between 160 °C and 240 °C and limit the oxygen content during the process (Hill, 2006 In Eucalypt species, durability increases after thermal modifi cation, as it has been shown in the resistance against the brown rot fungus Gloeophyllum trabeum for modifi ed E. globulus from durability class 3 (DC 3, moderately durable) to DC 1 (very durable) (González-Prieto and Touza Vázquez, 2009). Changes in mechanical properties of thermally modifi ed Eucalypt species, such as a slight increase of MOE and a decrease in MOR (Table 1), were not a detriment of the potential for using this species for outdoor materials (Esteves et al., 2007a;Calonego et al., 2012;de Cademartori et al., 2015;Knapic et al., 2018;Wentzel et al., 2019), as the MOE and MOR were still higher than those of commonly used WPC boards (Dias and Alvarez, 2017), a material commonly used for decking production (Zeller, 2018).
Currently, tropical woods such as Bangkirai (Shorea laevis), IPE-Lapacho (Tabulea spp.) or European-grown timbers such as Douglas fi r (Pseudotsuga menziesii) and Larch (Larix spp.), and preservative treated pine are the most commonly used materials for decking. In addition to these traditional decking materials in Europe, wood polymer composite (WPC) decking is gaining market share throughout Europe (Zeller, 2018) not at least due to their good capabilities to suppress the moisture uptake and resulting moisture movements, but also due to the thermal conductivity of polymers. Thermal modifi cation improves natural durability and dimensional stability, and decreases the equilibrium moisture content (EMC) of wood (Stamm and Hansen, 1937;Hill, 2006;Esteves and Pereira, 2009). Živković et al. (2008) used thermally modifi ed ash (Fraxinus spp.) and beech (Fagus sylvatica) as fl ooring elements. When compared to unmodifi ed wood, it showed a lower EMC in room conditions and an improvement in dimensional stability.
Thermal modifi cation also leads to an embrittlement of wood coming along with reduced abrasion resistance and the risk of splintering on the wood surface (Kubojima et al., 2000;Phuong et al., 2007). Surface hardness and resistance to abrasion are critical properties in less and non-load-bearing applications. In the case of decking, surface hardness turns into a decisive property for its use (Brischke et al., 2005). Welzbacher et al. (2009) showed that thermally treated beech and larch heartwood (Larix decidua) presented less abrasion and crack formation in relation to the unmodifi ed wood, but showed long term discoloration by weathering. It is also important to consider how to connect the wood when installing the material. For outside use of thermally modifi ed wood, it is recommended to use stainless screws and embedding screw heads (Aytin et al., 2015).
In this study the abrasion resistance, hardness (Brinell and dynamic), screw withdrawal resistance (SWR), maximum swelling, anti-swelling effi ciency (ASE) and equilibrium moisture content (EMC) of thermally modifi ed Eucalypt were determined since they are critical characteristics of outdoor exposed decking.

Materijal
The Chilean E. nitens wood came from 19 year old plantations from the Región del Bío-Bío in Chile, while the wood from Spain came from 16 year old plantations from the north of the province of Lugo in Galicia. The wood specimens were taken randomly from piles of industrially dried wood from general production lines from sawmills in Chile and Spain, which is why the humidity was different at the same laboratory conditions before modifi cation and there was a difference in their respective thickness. Flat sawn slats of 20 mm × 60 mm × 6 50 mm and 30 mm × 50 mm × 650 mm (radial × tangential × longitudinal) size, for the Chilean and Spanish origin, respectively, were prepared from kiln-dried wood free of large knots. Before the modifi cation process, the slats from Chile had an average moisture content of 13 %, whereas the ones from Spain had an average of 15 %. Twelve slats (six for each country of origin) per modifi cation process were used.

Thermal modifi cation 2.2. Toplinska modifi kacija
Thermal modifi cation was performed in a laboratory-scale treatment reactor. The modifi cation followed the ThermoWood ® process (Mayes and Oksanen, 2002) and contained the following steps: The temperature in the vessel was fi rst raised at 12 °C/h to 100 °C and then at 2 °C/h to 130 °C to allow a high-temperature drying of the slats to nearly 0 % MC, before increasing the temperature again at 12 °C/h until reaching the peak temperatures (185, 200 and 215 °C). The peak temperature was hold for 3 h. Afterwards, the temperature was decreased at 20 °C/h until reaching 65 °C, at which the vessel was opened and the specimens were removed. To determine the mass loss (ML) for each modifi ed slat caused by the thermal modifi cation, an adaptation of the procedure reported by Metsä-Kortelainen et al. (2006) was applied. The mass and the corresponding wood moisture content (MC) were recorded for each slat before and immediately after the process.

Determination of oven-dry density 2.4. Određivanje gustoće apsolutno suhog drva
The oven-dry density was determined according to ISO 13061-2 (2014) using fi ve 8.5 mm × 8.5 mm × 35 mm (rad. × tang. × long.) specimens per modifi cation, which were oven dried at (103±2) °C until constant mass. They were then weighed to the closest 0.01 g and their dimensions were measured to the nearest 0.01 mm.

Determination of equilibrium moisture content (EMC), volumetric swelling (S max ) and antiswelling effi ciency (ASE) 2.5. Određivanje ravnotežnog sadržaja vode (EMC), volumnog bubrenja (S max ) i učinka smanjenja bubrenja (ASE)
To measure EMC, S max and ASE, 30 specimens per process run with dimensions of 10 mm × 10 mm × 10 mm (rad. × tang. × long.) were prepared from the modifi ed wood and from unmodifi ed references. The mass and the dimensions of the specimens were measured after each of the following steps: Conditioning at 20 °C/65 % RH until constant weight; oven drying at 103 °C until constant mass; water saturating (specimens were water-impregnated for 30 min at 13 kPa and water soaked for 14 days).
The EMC (in %) at 20 °C and 65 % RH was determined for each sample using Eq. 1: , where m A and m B is the mass (in g) at the end of step A and B, respectively. S max (in %) was measured based on the sample volume at the end of step B and C for each cycle using Eq. 2: , where V B and V c are the sample volumes at the end of step B and C, respectively. ASE (in %) was measured using the S max before and after modifi cation using Eq. 3: where Su max is the maximum swelling of the unmodifi ed sample and Sm max is the maximum swelling of the modifi ed sample, respectively.

Brinell hardness tests 2.6. Tvrdoća prema Brinellu
The Brinell hardness (static hardness) was measured according to DIN EN 1534 (2011) with a universal testing machine (Zwick Roell Z010, Zwick, Ulm, Germany). Ten specimens per modifi cation and country of origin were used. A maximum force of 500 N using a steel ball with a diameter of 10 mm was applied for 25 seconds on specimens of 15 mm × 50 mm × 50 mm (rad. × tang. × long.) for the Chilean wood, and 25 mm × 50 mm × 50 mm (rad.× tang.× long.) for the Spanish wood. The specimens were conditioned at 20 °C/ 65 % RH during seven days until constant weight. The diameter of the residual impression was automatically determined by the testing machine. The Brinell hardness was then calculated according to Eg. 4: , where BH is the Brinell hardness (N/mm 2 ), F is the maximum force used (N), D is the diameter of the steel ball (mm) and d is the diameter of the imprint on the sample (mm).

Dynamic hardness tests 2.7. Dinamička čvrstoća
The dynamic hardness was determined according to Meyer et al. (2011) using specimens of the same quantity and dimensions as for the Brinell hardness tests. An indentation was generated in the surface of the specimen using a steel weight of 500 g that was dropped down on a steel ball from 300 mm of height. Four measurements were conducted on fi ve replicates per material. The dynamic hardness was calculated according to Eq. 5: where DH is the dynamic hardness (N/mm 2 ), m is the mass of the dropping weight (kg), h is the dropping height (m), r is the radius of the imprint on the sample (mm) and g is the gravity acceleration (m/s 2 ).

Resistance to abrasion: Shaker test 2.8. Otpornost na habanje: metoda Shaker
The resistance against abrasion was determined using the Shaker method described by Brischke et al. (2005). Five oven-dry specimens of 8.5 mm × 8.5 mm × 35 mm (rad. × tang. × long.) were placed in polyethylene fl asks (V = 500 mL) together with 400 g stainless steel balls of 6 mm in diameter and shaken in an overhead shaker at 28 revolutions per minute during 72 h. 25 specimens per material were tested. The distances between the opposite corners at oven-dry state were measured of each specimen, before and after the abrasion process. The loss in dimension (%) was determined according to Eq. 6: where ∆ab is the abrasion (%), d b1 is the diagonal 1 before abrasion (mm), d b2 is the diagonal 2 before abrasion (mm), d a1 is the diagonal 1 after abrasion (mm) and d a2 is the diagonal 2 after abrasion (mm). The average of the 5 samples per fl ax bottle was determined for each modifi cation. The resistance against abrasion was determined according to the Taber Abraser method (EN 438-2,  2005). The following modifi cations of the Taber Abra-ser test were made in order to allow testing of solid wood: Specimens (n=4) of 100 mm × 100 mm × 7 mm of a fi nished decking were prepared and conditioned in 20 °C/65 % RH. The tree rings of all specimens had an orientation of 45° to their cutting edges. After weighing and measuring the thickness at four points over the ridges of the decking, the specimens (n=5) were clamped into the Taber Abraser and were abraded with sanding paper S-42 with approx. 72 min -1 for 1000 revolutions. Each wheel had a load of 500 g. Afterwards the decrease in thickness at each of the four abrasion points was determined. The percentage loss in thickness (Δt) was determined as a measure of abrasion according to the following Eq. 7 for each specimen and an average was calculated: where t b is the thickness (mm) before the Taber Abrasion test and t a is the thickness (mm) after the test.

Screw withdrawal resistance tests 2.10. Ispitivanje otpornosti na izvlačenje vijaka
Screw withdrawal resistance (SWR) tests were performed according to EN 320 (2011), but modifi ed as follows: The same quantity and size of specimens used for the Brinell hardness test were used. Screws with nominal dimensions of 4.2 mm × 38 mm were used to penetrate the tangential face until (15±0.5) mm. Afterwards the screws were attached to a bracket to be pulled out at a constant speed of (10±1) mm min -1 . The screw withdrawal resistance corresponds to the maximum force determined to 10 N and was measured according to equation 8: (8) where SWR is the screw withdrawal resistance, N max the maximum force (N) and t the thickness of the specimen (mm).

Statistical evaluation 2.11. Statistička analiza
Statistical analysis was performed using the Pearson Correlation Coeffi cient Test to show the correlations between the mass loss and density with the properties of the modifi ed wood, and an ANOVA test to see if there was a signifi cant difference between the unmodifi ed and the modifi ed wood properties. All statistics were performed using the Statistica Software package Version 13.1 (StatSoft Inc., Tulsa, USA).

Changes in mass, oven dry density, EMC,
swelling and ASE by thermal modifi cation 3.1. Promjene mase, gustoće apsolutno suhog drva, EMC-a, bubrenja i ASE-a zbog toplinske modifi kacije ML increased with rising treatment temperature (  (Calonego et al., 2012). The oven-dry density of the Spanish wood specimens was lower than that of the Chilean ones.
The EMC of the modifi ed wood was reduced in both the Chilean and Spanish specimens (Figure 1a) after all the thermal modifi cations, and so was S max (Figure 1b) (Figure 1c) compared to the modifi cations at higher temperatures. This improvement of the Static hardness (Brinell) of the Chilean material was higher than that of the Spanish material. A Pearson correlation test showed that only the decrease in density was statistically signifi cantly correlated with the dynam-ic and static hardness in the Chilean specimens (Table  3), while the Spanish specimens showed an increase in hardness unrelated to their densities after thermal modifi cation at 200 °C ( Figure 2). There was a slight decrease until 200 °C, and then a clear difference between unmodifi ed specimens and those modifi ed at 215 °C, while the Spanish material showed a noticeable decrease after all modifi cations (Figure 2a). Dynamic hardness decreased with increasing treatment temperature in the Chilean specimens, while at 185 °C and 200 °C similar results were obtained for the Spanish specimens ( Figure  2b). The static and dynamic hardness at all temperatures was lower compared to both WPC (70 N/mm 2 ) and tro-  (Standfest and Zimmer, 2008), while the dynamic hardness was lower than that of European beech modifi ed at 180 °C and with similar mass loss (Meyer et al., 2011).
The abrasion resistance of E. nitens decreased with increasing treatment intensity in all the specimens (Figure 3). Traces of abrasion on unmodifi ed and thermally modifi ed E. nitens at 215 °C after the shaker test can be seen in Figures 4 and 5 for Chilean and Spanish specimens, respectively. The thermally modifi ed specimen (Figure 4c, d and Figure 5c, d) had more severely rounded edges and a slight loss of material due to splintering. The reduced abrasion resistance is likely due to an increased brittleness of the material (Kubojima et al., 2000; Phuong et al., 2007), and the lower density of the material after thermal modifi cation (Esteves and Pereira, 2009). Previous reports indicated that the decrease in abrasion resistance is correlated with the decrease in wood density (Brischke et al., 2014). This also occurred in both our Spanish and Chilean specimens, shown to be statistically signifi cant in Table 3. Wentzel et al. (2019) showed that there was a decrease in the degree of polymerization of the cellulose in thermally modifi ed E. nitens wood, which means that there was a change in the cellulose crystallinity. Estevez and Pereira (2009) suggest that changes in the cellulose crystallinity infl uence the mechanical properties of thermally modifi ed wood, which could also be an explanation of the reduction of abrasion resistance. Compared to other fl ooring materials, the unmodifi ed E. nitens showed similar abrasion resistance as wood-polymer composites (WPC) and tropical species such as Balau (2.9 and 3 %), while E. nitens modifi ed at 200 °C and 215 °C presented similar resistance to abrasion as Douglas fi r and Norway spruce (Picea abies), and namely 6.2 % (Brischke et al., 2014) The abrasion expressed as Δt increased with increasing modifi cation temperature in the Spanish spe-  cimens, while the Chilean specimens showed only small differences between modifi cations ( Figure 6). The measurements were performed on fi nished decking (Figure 7), where the difference in Δt can be seen between the modifi ed specimens at 215 °C (Chile in Figure 7c and Spain in Figure 7d), more material being lost in the Spanish specimens. The screw withdrawal resistance (SWR) decreased with increasing treatment temperature, but did not differ between Spanish and Chilean E. nitens (Figure 8  SWR were previously reported for birch (Poncsák et al., 2006) and wild cherry (Aytin et al., 2015).

Statistical analysis 3.3. Statistička analiza
The correlations presented in Table 3 show that mass loss was positively correlated with ASE and negatively with the resistance to abrasion in the shaker and abraser test, EMC, SWR and dynamic and Brinell hardness. Most correlations were signifi cant. Only density was not correlated with SWR and dynamic and Brinell hardness in the modifi ed wood from Spain, due to a slight increase of the properties in the modifi cation at 200 °C.
As for the ANOVA test, all properties presented signifi cant differences between the unmodifi ed and modifi ed specimens, with the exception of the dynamic hardness of the modifi ed samples from Spain, as can be seen in Table 4.
The results indicate that the characteristics of thermally treated E. nitens from our study, especially the modifi cation at 200 °C for both countries, were similar to other species commonly used for decking, such as Douglas fi r (Brischke et al., 2014), and other thermally modifi ed species, like Larix decidua (Welzbacher et al., 2009), F. sylvatica and Fraxinus spp. (Živković et al., 2008), which means that the modifi ed E. nitens wood could be a probable alternative for decking.

ZAKLJUČAK
Thermally modifi ed E. nitens, from both Chile and Spain, showed very similar characteristics compared to each other. The properties obtained modifying at 200 °C were similar to other species commonly used for decking, showing high abrasion resistance and similar dynamic and static properties. This shows the potential of use of E. nitens as decking material, particularly modifi ed at 200 °C. To further develop the potential of this species as decking material, additional weathering and durability tests should be performed to determine the long-term use of this material.