Processing and Wood Factors Inﬂuence Medium Density Fiberboard Production from Young Eucalyptus grandis , E. amplifolia , Corymbia torelliana , and Cottonwood Grown in Florida USA

: Fast growing Eucalyptus grandis W. Hill ex K.D.Hill & L.A.S.Johnson ( CT ), and Populus deltoides W.Bartram ex Marshall ( PD ) may be deployed in Short Rotation Woody Crop (SRWC) systems in the lower Southeastern USA, especially in Florida. To evaluate these species for possible use as medium density ﬁberboard (MDF) and other composites, 2.5 m logs of three EG clones, three PD clones, six EA progenies, four CT trees, and one P. tremuloides Michx. ( PT ) tree from northern Wisconsin as a control were characterized for basic wood properties before being chipped, pulped, and pressed into MDF. The chips were thermomechanically pulped (TMP) for a two-phase study of the factors expected to inﬂuence suitability for MDF production: wood characteristics, reﬁning system, resin system, and MDF formation. Phase I used TMP and 4% phenol-formaldehyde (PF) resin to produce 17 MDF species/genotype batches (S/GB). Thickness Swell (TS), Water Absorption (WA), Internal Bonding (IB), Modulus of Elasticity (MOE), and Modulus of Rupture (MOR) were evaluated to: (1) assess within species and within tree variation, (2) relate basic wood properties to MDF potential, and (3) examine repeatability of MDF-making. There was considerable variation among and within species, but only minor within tree variation. Six of the seventeen S/GBs had superior physical and mechanical MDF properties. In Phase II, two of the six better performing Phase I S/GBs were evaluated, along with three average Phase I S/GBs. Phase II compared the effects on IB from using tube and drum blenders for resin application, the inﬂuence of using unscreened versus screened ﬁbers, and the differences of using PF resin at 4% or 6% versus urea-formaldehyde (UF) resin at 8% or 12%. Overall, genetic variation among species, and particularly within these species, affected their potential for commercial MDF. Log speciﬁc gravity (SG), ﬁnes, MDF SG, and ﬁber length inﬂuenced MDF properties, as did reﬁning and MDF-processing variables. Further study of speciﬁc processing requirements can optimize the potential of young EG , EA , PD , and CT genotypes for MDF and other composites.


Introduction
SRWC systems involving the fast-growing hardwoods EG, EA, CT, and PD may be implemented in appropriate portions of Florida and the lower Southeast. EG, CT, and PD are also important plantation species worldwide. On suitable southeastern USA sites and/or with intensive culture, EG, EA, and PD may reach harvestable size in as few as three years [1,2]. EG is the most productive of the three, largely because of a tree improvement program conducted by the US Forest Service from the late 1960s to 1984 [3]. EG is now grown and sold commercially in southern Florida for landscape mulch. EA may be grown from central Florida into the lower southeastern USA, while PD can be grown across much of the US. CT is used as a windbreak for vegetable crops and citrus in central and southern Florida. While these SRWCs have been shown to be suitable for some traditional products and for energy wood [4], little is known about their suitability for a wider range of value-added products.
Accordingly, our general objective was to determine the potential for using these SRWCs as wood-composite products. Previous research on wood-based composites and other similar hybrid composites has laid a foundation for this study. The properties of wood-based composites are known to be a function of wood fiber species, source, and quality [5][6][7][8][9][10][11]. They are also related to composite processing parameters [12][13][14][15][16]. The age of the woody fiber source, especially related to wood juvenility, is also well recognized as being important, which could definitely be an issue with SWRC fiber [11,[17][18][19][20]. Many of these same issues would also likely be a concern in the use of SRWCs as a fiber source for inorganic bonded wood composites [21,22].
Recognizing these critical issues, the specific objectives of our investigations were to evaluate and compare the broad suitability of young EG, EA, CT, and PD for making MDF by evaluating the basic wood properties of MDF produced from defibrillated SRWCs. Then, within-and between-tree variation influencing MDF production was evaluated to assess their potential for use for other wood-composite products.

Materials and Methods
EA, EG, CT, and PD from Florida were assessed for their MDF suitability. The EG, EA, and PD genotypes included three superior EG and three superior PD clones and six top EA progenies, based on statewide genetic tests ( The 2.5 m long logs were harvested and shipped to FPL, where they were cut into 1.2 m lengths with a 50-mm thick mid-log disk removed for anatomical study. The 1.2 m lengths were then debarked, sectioned to chippable sizes, and stored at 4.5 • C. The mid-log disks were then saturated in water, weighed green, and dried at 103 • C to determine wood moisture content (MC) and SG. Each of the resulting 17 MDF S/GBs was chipped (usually the day before being defibrillated by TMP) by presteaming for 10 min and then steaming at 167 • C with 0.621 MPa for~6 min using a 1 kg/min feed rate at 3000 rpm in a Sprout-Bauer Model 121CP (Andritz, Inc., Muncy, PA, USA), a 305-mm thermal mechanical single-disk refiner operated with a digester at 620 Pa and an energy consumption of 200-250 watts/kg when using a 0.152 mm separation between refiner plates (Sprout-Bauer D2B503). The resulting fiber was then immediately dried in a tray drier for approximately 24 h at 104 • C.
For each S/GB, the MC of the dried fiber was determined to assess how much fiber was needed for a 4% blend with PF resin (Dynea/Arclin 13CO85, 50.3% non-volatiles, typical of those used at the FPL [23]) in a tube blender. After blending, the MC was again estimated to derive the amount of the blend necessary to hand-form a 1810 g, 406 × 406 × 12.5 mm MDF board with a target SG of 0.72. The board was then hot-pressed at a constant 82 • C for two minutes; the maximum panel pressure during closing was set at 6.0 MPa and was reduced to 0.11 MPa after the 12.5 mm thickness was reached. The first board of each S/GB was cut-up and inspected to assure the hot-press and blending processes were appropriate for that S/GB. Then, another five MDF boards for each S/GB (nine for PD2/PD3) were subsequently made. Each of the second through to the sixth MDF boards of each S/GB (8-12 for PD2/PD3) were hot-stacked and cooled. They were then included in subsequent physical and mechanical testing in Phase I.
In Phase I, the flexural properties, internal bond strength, and dimensional stability were determined according to ASTM Standard D1037-12 [24] using seven samples cut from each board. All seven test samples were conditioned at 27 • C and 65% relative humidity (RH) for 7 days. MOR, MOE, and board SG and MC were evaluated for two 76.5 × 356 mm samples. IB was derived from three 51 × 51 mm samples. TS and WA were evaluated by immersion of two 152 × 152 mm samples horizontally in a soak tank and weighed after 2 and 24 h at ambient temperature.
Phase II involved studying two of the "best" Phase I S/GBs (Eg2, Pd4) and three nearly average Phase I S/GBs (Ea4, Ct3, Pt1). Using TMP fiber from the earlier pulping process, another three new MDF boards were made using the same methods and processes as Phase I to hand-form and hot press a 1810 g, 406 × 406 × 12.5 mm MDF board with a target SG of 0.72. These Phase II evaluations specifically compared the use of tube or drum blenders, determined the effect of using unscreened or screened fiber, and compared the use of PF resin at 4% or 6% with UF resin (GP789D16, 47% solids, typical of those used at the FPL [25,26]) at 8% or 12%. The various effects of these different MDF processes were determined using from one to five IB samples per S/GB (Table 2). Analyses of variance and covariance examined differences in MOR, MOE, IB, TS, and WA, with panel density as a covariate. MOR, MOE, IB, LE, TS, and WA of MDF panels were related to wood and fiber characteristics such as wood density, pH and base buffering capacity, and fiber coarseness. The analyses also examined the effects of species, genotype, and/or log on MDF panel properties, with a significance level of 5%. Species and S/GB means were tested using Duncan's multiple range test. Finally, a series of progressive weighted-rank analyses of physical and mechanical properties sorted the 17 S/GBs for the MDF properties evaluated in Phase I and the five S/GBs of Phase II.

Phase I
A number of differences were noted for certain wood properties among and within the Florida-grown EG, EA, PD, and CT S/GBs ( Table 3). The EG clones were generally the denser Eucalyptus, while EA was generally less dense based on the limited genotypes and ages represented in this study. Considerable within-species variation for wood properties was evident in each species, suggesting that the deployment of favorable clones would be advantageous in producing currently-used energy products. Similar variation in the characteristics of refined fibers also emphasized the importance of genetic variation in making other products such as value-added wood composites.  Table 5).
The strength properties and dimensional stability of the MDF panels made from the five species varied at many levels ( Table 6). Species differed significantly for TS and WA but not IB, MOE, and MOR. Differences between genotypes/species were significant for IB, MOE, and MOR. The boards/batch were typically highly significant for each property. At the S/GB level, the strength properties and dimensional stability of the MDF panels varied considerably (Table 7). For example, some of the 17 S/GBs were clearly consistently better than others for the MOR-MOE relationship (Figure 1). Similar separations were noted between the 17 S/GBs for the other properties, too.  A rank order analysis compiling ranked performance data from five properties (TS24, WA24, IB, MOE and MOR) ranked the 5 species and 17 S/GBs from best to poorest performance (Table 8). In Table 8, all five sets of the tested properties are evenly weighted using an "Importance Factor" (IF) of 1.0 for each of the five properties. We then repeated the ranking three additional times by first applying weights for TS and WA = 0.5, IB = 0.75,  A rank order analysis compiling ranked performance data from five properties (TS 24 , WA 24 , IB, MOE and MOR) ranked the 5 species and 17 S/GBs from best to poorest performance (Table 8). In Table 8, all five sets of the tested properties are evenly weighted using an "Importance Factor" (IF) of 1.0 for each of the five properties. We then repeated the ranking three additional times by first applying weights for TS and WA = 0.5, IB = 0.75, and MOE and MOR = 1.0, then ranking by weights for TS and WA = 0.6, IB = 0.8, and MOE and MOR = 1.0, and finally weights for TS and WA = 0.75, IB = 1.0, and MOE and MOR = 1.25. All four rank ordered analyses of the ranked scores for all five MDF properties consistently ranked the 6 S/GBs higher than the other 11 groups. Ea6, Pd4, Ea1, Pt1, Ea2, and Eg2 were consistently from 29% to 40% better than the next batch (overall ranked as 7th) and even much better than the others ranked from #8 to #17.
The various minimum property requirements for the evaluation of MDF panels for interior applications are listed in ANSI Standard A208.2-2016 [28]. For MDF, three grade levels are given. None of the 17 S/GBs of MDF panels tested in Phase I met the minimum requirements for MOR of interior-use MDF panels. For MOE, 12 of the 17 groups of MDF panels met the ANSI requirements of the lowest Grade 115 MDF panels, but none met the requirements for the intermediate Grade 130. Only 1 of the 17 S/GBs (Ea6) met the ANSI minimum requirement of IB. We did not compare the WA and TS requirements because no wax nor other water-repellant additives were used in the manufacture of these Phase I MDF panels. While the general results of these comparisons of MOR, MOE, and IB did not generally meet the minimum requirements for commercial MDF, these results are not entirely negative. Recalling that these initial test results are based on first-run fiber and manufacturing processes, it is quite reasonable to assume that future evaluations will have considerably higher material properties, as fiber and manufacturing processes are optimized for fiber prep, resin type, concentration and blender application, and as hot-pressing parameters are improved, more desirable panel density profiles are achieved, and performance enhancing additives are developed and used.

Phase II
Phase II involved comparisons of the two better-performing and three averageperforming S/GBs, as determined in Phase I. Phase II IB testing suggests that G/SBs behave differently to the various MDF resin systems and application rates. MDF made using a tube blender for resin application was better than that made using a rotary drum blender (Figure 2). The percentage of fines when using unscreened fiber clearly negatively influenced MDF properties (Figure 3). It was also clear that the use of UF resin was far more suitable for making better MDF than PF resin (Figures 4 and 5). It is not unreasonable to think that these higher UF resin rates may be needed with these low SG woods, as such fiber density-to-MDF performance ratios are commonly used in commercial MDF manufacture.  Figure 2. Influence of UF resin at 8% or 12%, two blender types, and five S/GBs on IB in Phase II. Figure 2. Influence of UF resin at 8% or 12%, two blender types, and five S/GBs on IB in Phase II. Figure 2. Influence of UF resin at 8% or 12%, two blender types, and five S/GBs on IB in Phase II.    The ANSI A208.2-2016 Standard was used to assess the potential of these experimental MDF panels for commercial interior applications. The IB of Phase II panels were clearly a function of the selection of fiber-and composite-processing variables (Figures 2-5). Only 12 of 17 S/GBs met the ANSI A208.2 requirements for the lowest 115-grade of MDF; none met the next higher 130-grade for MOE. No Phase II S/GB MDF met the MOR requirements, and only 1 of the 17 S/GBs met the ANSI A208.2 requirements for the lowest 115-grade of MDF for IB. It is highly likely that many of these differences can be compensated for by optimizing refining parameters to reduce fines in the fiber content in the pulp and then improving the hot-pressing parameters, resin selection and application rates, and fiber preparation processes so that more desirable panel density profiles and in-service performance properties can be achieved.

Figure 4.
Resin and S/GB influences on IB for tube-blender applied PF resin of 4% or 6% or UF resin at 8% or 12% in Phase II. These preliminary Phase I and II examinations of the potential for 17 S/GBs clearly showed that some have a high potential for further studies to examine their potential for commercial MDF use and possibly other fiber-based composites. More EG, EA, PD, and CT genotypes should be considered. Additional study is needed on thermal pretreatment of the wood [29] and the MDF pressing schedule [30]; especially evaluating the incorporation of nanoclay, zeolite, or cationic starch which could provide distinct benefits for OSB, MDF, and PB board production [31]. Enhancements in IB, WA, and TS and in achieving lower press energy requirements by 10-25%, depending on product specifications, are potentially possible. These additives are commercially available and are amenable to commercial applications. Finally, all additives can be readily incorporated into modern composite board production facilities. These results provide board producers new additive technologies and treatments to enhance board products while reducing energy requirements.
Based on these combined Phase I and II results, it is reasonable to assume that with future work on improving fiber and fiberboard processing, certain EG, EA, and CT genotypes may be suitable for use as a primary or supplement fiber source for commercial MDF and probably other commercial interior-use composite products [32,33], such as particleboard or other types of fiberboards and energy products [27].

Conclusions
This two-phase study of a limited number of genotypes suggests that appropriate young EG, EA, CT, and PD genotypes, with additional fiber and processes experience and improvements in refining, resin, and formation, may be used for wood composites such as MDF: 1.
In Phase I, using 4% PF resin in a tube blender, there was some variation among species, considerable variation within species, minor within-tree variation, some influence of basic wood characteristics on MDF properties TS, W, IB, MOE, and MOR, and a large sampling variation for some MDF properties; 2.
Phase II, involving the six top S/GBs, provided valuable insight into the needed fiber and processes improvements. For example, MDF made with UF resin at 8% or 12% had generally better performance properties than PF resin at 4% or 6%; 4.
Screened TMP fiber produced better MDF than unscreened fiber, and resin application by tube blenders made better MDF than by drum blenders; 5.
Overall, genetic variation among and, particularly, within these species affected MDF performance properties; 6.
Refining and MDF-making aspects have such major impacts on MDF properties that specific processing requirements are needed and must be optimized for future commercial MDF options for appropriate EG, EA, PD, and/or CT genotypes; 7.
EG and EA utilization may, thus, expand from the current mulchwood market to various interior-use wood composites such as MDF and cement board; 8.
These results are encouraging for the development and use of wood composites from SRWCs in Florida and the southeastern USA.