Modelling borate formulations through pressure-treated sectional three-layer cross-laminated timber structures by near-infrared spectroscopy

Abstract

The present study examines the borate distribution of water-borne preservative (Timbor and Boracol) formulations into sectional three-layer cross-laminated timber (CLT) structures that were pressure-treated using a full-cell process for exterior exposures to ensure long-term durability. Both commercial preservatives, dispensed through three-layer CLT structures, are formulated with the DOT (from disodium octaborate tetrahydrate) dissolved in a propylene and/or monoethylene glycol solvent, which facilitates the chemical ingredients’ impregnability and correlates with the potential of near-infrared spectroscopy (NIRS) for estimating. This was conducted throughout an orthogonal clear-cut of pressure-treated three-layer CLT structures of Eastern black spruce (Picea mariana var. Mariana) species at different experimental (0, 5, and 14 days) conditions. Cross-sectional and perpendicular cuts of CLT elements described three glued layers of lamination (known as lamella) arranged orthogonally to each other, which were used for measuring depths of borate penetration. As part of evaluating the borate's diffusion gradient in terms of boric acid equivalent (BAE), the edge of wood grain directions of half portions of the three-element CLT structures were sliced every 0.5 cm (^3/16-inch), following borate impregnation. The investigation was undertaken through CLT upper (face grain) and lower (back grain) members, as well as into CLT transverse (radial/tangential) grain directions and the centres (along the longitudinal core grain directions) of inner CLT structures. Validation models achieved statistical values of R² of 0.75 and 0.76 for Boracol and Timbor formulations based on DOT-treated ingredients, respectively. The root mean square error ranged from 0.5611 to 1.2813% BAE, and the average margin of standard deviation of DOT (Timbor and Boracol) formulations varied significantly from 0.6 to 1.49 (±0.0025) through the performance of sample-specific standard error of borate prediction analysis. The potential of NIRS shows some predictive abilities using the projection to latent structures or partial least squares regression method, including the sample-specific standard error of borate prediction analysis for estimating differences in lamellas of CLT three-layer untreated and full-cell pressure-treated with formulations based on DOT ingredients in cases of exterior exposures.

Keywords

Boric acid equivalent evaluation pressure-treated cross-laminated timber (CLT) structures Timbor and Boracol formulations based on disodium octaborate tetrahydrate NIR spectroscopy

Introduction

Most mass timber projects utilising glue-laminated timber systems, such as cross-laminated timber (CLT) configurations, require no protection beyond conventional finishes, as insight into the design and manufacturing considerations involving CLT materials has commonly evolved as one of the most innovative mass-structured concepts for the building industry. Over the last couple of decades, CLT materials have continued to gain popularity in expanding and realising the goals of achieving low-rise, mid-rise, and high-rise residential buildings, as well as non-residential applications.^1,2 However, when the CLT materials are exposed to conditions conducive to attack by wood-decaying fungi, marine borers, or wood-destroying insects, they should be treated with a suitable wood preservative. For example, in their utilisation of ground contact applications, where the CLT materials are subject to permanent or alternate wetting and drying or where high humidity conditions prevail, and because CLT materials consist of several lamellas (known as lamination) of organic compounds stacked crosswise, this is an intended source of issues about abiotic and biotic agents of wood degradation.^3,4 One of the most common treatments safely used to protect mass timber and stacked wood products in exterior residential and other construction projects by the wood preservative industry instinctively remains the two principal types of pressure treatment, such as Full-cell or Bethell processes and Empty-cell or Lueping and Lowry techniques, which usually operate with oil-borne and water-borne formulations.^5,6 In the case of water-borne preservatives, such as formulated borate ingredients, they have already been applied before construction to critical components of buildings to provide manifest and latent protection. When the area becomes wet, borate preservatives can deeply penetrate refractory wood zones, whereas the formulated preservative cannot easily penetrate by pressure treatment procedures.^7–9 These recommended operations of boron preservative treatments are available under pressure processing to CLT structures and are common to protect wood material under overall exposure conditions. Consequently, pressure-treated Full-cell and Empty-cell methods using different borate formulations should potentially be a key result of studying the effectiveness of preservative ingredients and their impregnation through wood physical and chemical attributes, which influence the CLT structures.^10–12

Modern analytical wet chemistry methodologies have overwhelmingly quantitative techniques for monitoring the preservative ingredients in treated wood, as in many cases, a quantitative answer is more valuable than a qualitative one. This requires elaborate analyses for quantifying preservatives, as it is a destructive and time-consuming method.^13,14 Preliminary approaches on non-destructive radiant techniques, such as near-infrared spectroscopy (NIRS) coupled with multivariate statistical analysis, modelled wood moisture content and chemical composition of wood surfaces,^15,16 including prediction of stiffness through veneer products and adhesive bond strength of wood products,^17,18 and to correlate specimens with a chosen assortment of wood species.^19,20 Following the integration investigations concerning borate diffusion modelling on Eastern black spruce (Picea mariana var. Mariana) single-wood products,²¹ the studies will be extended to mass timber structural products using NIRS for novel industrial applications.

The goal of this research was to investigate the potential of NIRS to assess borate distribution and content through an orthogonal cutting process of sectional three-layer CLT structures of Eastern black spruce (Picea mariana var. Mariana) species, which have been full-cell pressure-treated with (Timbor and Boracol) formulations based on disodium octaborate tetrahydrate (DOT) dissolved in a propylene and/or monoethylene glycol solvents, at different experimental (0, 5, and 14 days) conditions, for revealing a better understanding of the significant efficacy effects of the wood preservative chemical ingredients. While the application of NIRS to borate-treated CLT systems is conceptually sound and industrially relevant, this statistical quantification approach should clearly demonstrate a new scientific insight evolved beyond earlier studies^22,23 related to modelling the effect of moisture and boron content in wood products. Borate penetration depths and boron-based preservative concentrations are estimated quantitatively using projection to latent structures (PLS) regression methods, which relate the entire experimental data set to NIRS measurements through sample-specific standard error of borate prediction analysis. Such a study may add industrial relevance through an analytical approach that can distinguish full-cell pressure conditions of various formulated preservatives in structural mass timber products.

Materials and methods

For experimental and laboratory conditions, CLT structures were built up from softwood lumbers of Eastern black spruce (Picea mariana var. Mariana) species glued together in three layers so that the grain directions of each lamella is at right angles (90°) to the one adjacent to it following the requirements described by the LTIC initialism for Laminated Timber Institute of Canada.²⁴

Sawn wood samples for three-layer cross-laminated timber conception

Dimensional lumber of 12 board feet measuring 5.8 cm×10.16 cm×5.49 m (thickness/width/ length) or 2-inch×4-inch×18-foot pieces of nominal sizes were acquired green from a local lumber supply store (Ontario, Canada) to be processed whenever possible for laboratory-conditioned air-dry (stored for 48 h at room temperature) until they attained equilibrium of the optimum moisture content required from 10 to 15% before sawing.²⁵ Lumbers were selected to inspect surface appearances for knots, splits, pitch pockets, and other open defects, to manufacture high-quality visual grades of wood structures. Lumbers were planed and set up to standard dimensions through an orthogonal cutting process at room temperature, as a perpendicular clear-cut was performed crosswise longitudinal grain direction at 90 degrees to the angle of the cutting edge (power saw) made with the longitudinal grain direction. This has generated wood materials of actual sizes of 80 mm or 3^5/32 inches (length and width) face grains, and 35 mm or 1^23/64 inches (thickness) for occurring pairs balanced in grain directions and thickness about the inner wood structures.

Glued-laminated process for three-layer CLT manufacture

The wood materials were manually assembled crosswise at a 90-degree angle between the face grain directions into three elements. These were bonded and glued with a brush treatment process of liquid phenol-formaldehyde (PF) resin in a type-high frequency formulation for cold or medium temperatures supplied by Momentive Specialty Chemicals Canada, Inc. (Edmonton, AB, Canada). The resin formulation was composed of a combination of resorcinol and formaldehyde, whose cure is affected by adding extra formaldehyde, according to the glue manufacturer's recommendation and specification regarding physical and chemical properties for achieving the optimum glue bond using PF between members, as described by Momentive.²⁶

Press-controlled process for three-layer CLT consolidation

The bonded three elements of wood structures were lay-up under a pressing machine (PressMan, type ARC 450, Dieffenbacher North America Inc, Windsor, Ontario, Canada) assisting with a Press Control System (set up at 150 Bar for 30 min in ambient temperature without vacuum activation) to glue become insoluble under practically all conditions of experimental exposure process for typical three-layer CLT structures as permitted by the National Building Code of Canada.²⁷

Three-layer CLT sampling description

A description of CLT structures obtained from three glued layers of wood materials placed in orthogonally alternating orientation to the neighbouring layers within the upper and lower grain sides constituted the wood directional transverses (radial/tangential) called hereafter ‘transverse top and bottom sides’. The central portion (centre or core grain) was described in the longitudinal grain direction. Three-layer CLT structures resulted in quasi-rigid engineered wood products described as a consistency of three lamellas stacked crosswise (typically at 90 degrees) grain directions and glued together on their wise grain faces, which were obtained with the common member actual dimensions of CLT length of up to 80 mm or 3^5/32 inches, a width of up to 80 mm or 3^5/32 inches and a thickness seldom above 105 mm or 4^21/64 inches as schematically represented in Figure 1(a).

Figure 1.

Illustration of Eastern black spruce cross-laminated timber (CLT) configured with three elements: (a) three-layer CLT elements schematically sampled, (b) three-layer CLT sample silicone coated, and (c) three-layer CLT sample split for borate analysis. Each set was replicated three times for experiment reproducibility.

For the experimental context, a total of 45 polished CLT structures were generated by the surface machining process, with actual sizes of about 8 cm×8 cm×11 cm or 3^5/32 inches×3^5/32 inches×4^21/64 inches (length/width/thickness or longitudinal/radial/tangential grain directions).

Full-cell process of three-layer CLT structures

Three-layer CLT conditioning

The full-cell or Bethell process was preferred to the empty-cell treatment because it provides a maximum preservative retention for general exterior exposures, including marine installations and all water-borne salt treatments. Initially, three-layer CLT materials were kept in storage conditioning chamber at 15°C within a relative humidity (RH)> 2% until three sidewise surfaces were sealed with silicone coating except on the top and bottom radial grain surfaces of the 8 × 8 cm² or 3^5/32 × 3^5/32 sq. in. (longitudinal/tangential) areas as well as from one edge grain side of the 3.5 × 8 cm² or 1^3/8 × 3^5/32 sq. in. (radial/tangential) facial transverse top and bottom grain sides and the 3.5 × 8 cm² or 1^3/8 × 3^5/32 sq. in. tangential grain surface at the core centre of the CLT structures to condition the borate diffusion (Figure 1(b)).

Three-layer CLT full-cell processing

The three-layer CLT materials were placed in a treating cylinder to apply a vacuum (at 11 kPa absolute pressure) for about 1 to 2 hours to get a net retention. The pressure cycles varied from 100 to 150 pounds per square inch for the preservatives and continued until the desired retention targets was reached (at 1034 kPa for 2 h). The pressure cycles were then released, and the preservative (Timbor or Boracol) products were withdrawn. A short final vacuum (at 11 kPa for 15 min) was applied to condition the surface as recommended by the American Wood Preservers Association (AWPA) standard A03-05.²⁸

Three-layer CLT structuration

The three-layer CLT materials were full-cell pressure-treated in separate batches using Timbor and Boracol 20-2BD formulations, with only nine (9) units set in individually treated cycles for each turn for laboratory experimental conditions. These formulations contained about 20–60% of preservative active ingredients, such as disodium octaborate tetrahydrate, acronym of DOT, also known as Na₂B₈O₁₃·4H₂O. The Timbor and Boracol, based on DOT, are formulated mixtures of active substances ranging from 19 to 20% w/w (Na₂B₈O₁₃·4H₂O) with 1.0% Didecyl Dimethyl Ammonium Chloride (also known as C₂₂H₄₈ClN). These ingredients were solubilised in propylene and/or monoethylene glycol solvents. All technical advisory and recommendations are found on the technical notices that are commercially available from SANSIN Corporation (Strathroy, ON, Canada) and Sasco Products Limited (Dartmouth, Nova Scotia, Canada). For exterior exposures, both commercial (Timbor and Boracol) formulations based on DOT ingredients are particularly suited to the pressure treatment process for CLT structures, as water-borne substances tend to leach out water-soluble chemicals from treated wood in wet situations, so that the DOT preservative is designed to form relatively insoluble complexes when the wood dries after treatment.

It is noted that approximately six (6) units of three-layer CLT structures were left as untreated control specimens to proceed with distilled water instead of preservative chemical formulations, considering wood swelling and shrinkage effects.

All untreated and full-cell treated three-layer CLT structures were wrapped in black plastic bags to minimise drying during the experimentation conditioning storage for durations starting at 0, 5, and 14 days in a Drytech kiln (USA series 3900MC) chamber at room temperature (reaching 23.5°C) and at a fixed relative humidity (RH >98.9%) as described in AWPA Standard A2-05.²⁹

Borate description analysis in structural three-layer CLT sampling

For evaluating boron into pressure-treated three-layer CLT members, CLT structures were perpendicularly sawn in half to 4 cm×4 cm ×11 cm or 1^37/64 inches×1^37/64 inches×4^21/64 inches (length/width/thickness or longitudinal/radial/tangential grain directions) throughout the centre of the face grain (90 degrees direction), along the lengthwise in both the radials and crosswise longitudinal grain directions of the three-layer CLT structures to analyse the borate impregnation. One-half section was used to estimate the penetration depth, and the other was used to measure the borate gradient along the grain of three-layer CLT elements (Figure 1(c)).

Each pressure-treated three-layer CLT member was used separately for borate penetration depth analysis, along with borate migration accordingly. In both transverse pressure-treated CLT top and back grain faces, the members described area samples of black spruce strips measuring about 40 × 80 mm² or (1^37/64×3^5/32 sq. in.) longitudinally to radially surfaces, and 35 mm or 1^23/64 inches tangentially (widthwise) and 40 mm or 1^37/64 inches along the grain. The inner core (central portion) measured about 35 mm or 1^23/64 inches tangentially (widthwise) and 80 mm or 3^5/32 inches along the longitudinal grain direction (Figure 1(c)).

Borate coverage analysis in structural three-layer CLT sampling

A formulated mixture of curcumin reagent containing about 0.25 g curcumin in a 10 ml concentrated hydrochloric acid, and 10 g salicylic acid in 100 ml ethanol, was sprayed on the split surfaces of three replicate boron-treated members of three-layer CLT to reveal the borate coverage (borate penetration depth) as described in Williams.³⁰ The duration for revealing coloration ranged from 5 to 20 min. In contrast, untreated wood zones are pale yellow, and red/pink coloration develops at a concentration of preservative equivalent to more than 0.2% boric acid, as described by Murphy and Tuner.³¹ The measurements were expressed as a percentage of penetration depth (% borate coverage). In addition, the split sections of the three-layer CLT members were analysed to determine boron retention in wood by an ICP spectroscopy computation, which was expressed through the percentage of boric acid equivalent (BAE). The calculation of elemental boron to the equivalent mass of the boric acid from DOT (Timbor and Boracol) formulations was based on the conversion formula as follows: Concentration of elemental boron measured by ICP (–OES or –MS)×[(Molar mass of boric acid (H₃BO₃), approximately 61.83 g/mol)÷(Molar mass of elemental boron (B), approximately 10.81 g/mol)] = Concentration of H₃BO₃×Conversion factor, approximately 5.72 (61.83÷10.81).³² Both resulting borate measurements were then related to the scanned NIR measurements.

Borate gradient analysis and sampling of three-layer CLT

The gradient of borate diffusion was carried out by analysing thin wood materials cut from a member sample of pressure-treated CLT structures sawn at room temperature to 90 degrees of angles between cutting machine movement and grain directions every 0.5 cm (^3/16-inch) from the pressure-treated CLT upper surfaces, radially for the inner core (centre) grain and longitudinally for both the transverse top and lower grain faces.

All sliced pressure-treated CLT members were conditioned to dry (RH > 2%) at 23.5°C to be ground to powder with a Thomas-Wiley laboratory mill grinder (USA model 4) and screened with a 2-mm mesh Tyler equivalent filter (USA series equivalent N^o18).

To calculate the BAE (H₃BO₃) from ICP spectroscopy analytical measurements, the extraction of borate ingredients was undertaken with about 0.5 g of sawdust in 40 ml of hot (90°C) distilled water for 4 h in a Blue M water bath (USA model MW1165C-Serial No M71211), which concentration was diluted 50 times (about 250 ml of final volume) before the ICP spectroscopy estimating as described in AWPA standard A21-00.³³

Spectral analysis of three-layer CLT

Pressure-treated CLT structures were bagged and kept in conditioned ambient temperature storage in a chamber at 23.5°C under drying conditions (relative humidity [RH] > 2%) before the scanning process. NIR spectral data were acquired from CLT structures following grain directions and borate diffusion at 2-min intervals with a NIR256-2.5-[HL-2000-HP-RS232 ] Vis-Near-infrared Spectrometer (Ocean Optics Inc., 830 Douglas Ave, Dunedin, FL34698 USA) equipped with an optical probe positioned on the top of the sample with a 2-mm or ^5/64-inch diameter beam. NIR spectrometer characteristics are described in the NIRQuest NIR Spectrometers Data Sheet,³⁴ which has a spectral resolution of 5 nm and was manually calibrated for white/dark after every performed data set in triplicate measurements. To ensure reproducibility, pressure-treated and untreated CLT samples were scanned randomly (three replicates) on their surface top and bottom grain sides, particularly on flat zones that bordered glue line canals, wherein scans were undertaken additionally to validate statistical robustness over glue-line effect conditions.

Data set of three-layer CLT

Averages of recorded triplicate data were automatically processed with an Unscrambler^® 9.8 (CAMO Software Inc., 1 North Cir., Woodbridge, NJ 07095-2105, USA) program to NIR reflectance spectra acquired at the standard wavelength 900–2400 nm region, which were smoothed by applying a second derivative 13-point Savitzky-Golay transformation. NIR spectra stored in the variable data set constituents were related by the partial least squares or projection to latent structures (PLS) regression method, which was computed by a concept of test set validation technique to predict data values, where models were established with the sample-specific standard error of borate prediction analysis to reveal the classified DOT (Timbor and Boracol) ingredient effects of CLT members. The Unscrambler ^® program recorded enough scans to be used for model development to classify the whole data as independent calibration and validation sets within two principal component factors determined based on the root mean square error (RMSE) values to provide an optimal prediction for significant effects.^35,36 The ratio of the standard error of performance to the standard deviation (SD) of the reference data of DOT (Timbor and Boracol) ingredients was estimated through the performance of PLS-R models, which was calculated with the relative percent difference (RPD) as described by Miller and Miller.³⁷ Confidence bands of present variability measures (SD, confidence intervals) were associated with specific data replication of the predicted performance following the method described in Faber et al.³⁸ to assess whether a random scanned sample is similar to other scanned treated/untreated samples or to which treatment group it belongs, for example, to identify the scanned area into CLT states of glue canals separations, or unknown scanned zones in the samples for which a projection to latent structures (PLS) regression method has been established.^39,40

Results and discussions

A poor preservative sight in the sapwood cells of wood-concept structures is the most common cause of decay. Nevertheless, inspection of borate distribution could be imperative to reveal factors affecting the penetration and retention of preservative ingredients in wood structures, particularly in cases of failures in the effectiveness.

Borate coverage analysis of sectional three-layer CLT structures in function of treatment duration conditions

Figure 2 represents the % of borate coverage relating to penetration depth and the % of BAE relating to retention in pressure-treated CLT 3 members from the transverses top and bottom, including the central portion of longitudinal grain directions, at the different conditioning (0, 5, and 14 days) durations. Regardless of whether CLT 3 members were DOT-treated with Boracol (Figure 2(a)) or Timbor (Figure 2(b)) formulations, preservative penetration depths have appeared fully retained in surficial CLT 3 members, for the same proportion (about 90% of member surfaces have appeared as covered by borate ingredients) at all treatment durations starting from 0, 5, and 14 days. For pressure-treated wood-concept structures (e.g. three-layer CLT concept), the rate of distribution, retention level, and preservative can affect the specifications, such as the equivalent volume of treated wood, weight, and effectiveness; however, specific conversions may vary depending on the treatment method and wood species. Thus, the wood-concept structures (three-layer CLT) may be conditioned within service life to the preservative ingredients and treatment type. Additionally, with immediate criteria such as the amount of preservative within the zone of retention, penetration depth, and glue canal influences.⁴¹ Similar observations were obtained in a microscopic examination of pressure-treated southern pines by Saadat and Cooper,⁴² who reported that in heavily treated wood cells, many lumens were filled with preservatives, and a slight difference in distribution among preservative ingredients was distinguished. Contrary to zones of lower retention, where the preservative was present as drops or plugs of liquid and in the tips of the earlywood tracheids and parenchyma cells, but not in those at rays. Hence, numerous glue lines or resin canals in CLT configurations and most epithelial cells contained preservatives, as did many tracheids adjoining resin canals, which impregnation takes time due to the slight effect of directing the borate diffusion flow, as the glue line gaps in structural laminated specimens allow borate to diffuse laterally with no action on chemical and mechanical attributes, such demonstration by Colakoglu et al.⁴³ on the effect of preservative boron-based treatment through mechanical properties of beech wood species.

Figure 2.

Penetration depths of boric acid equivalent (BAE) diffusion in Eastern black spruce CLT three-element pressure treated with DOT Boracol (a) and DOT Timbor (b) conditioned at different treatment periods ranging from 0, 5, and 14 days. For each probe, values sharing the same letter (α, β, γ, δ, ε, λ, μ) are not significantly different at the 5% level of confidence. CLT: cross-laminated timber; DOT: disodium octaborate tetrahydrate.

Ultimately, the overall analysis of the percentage of BAE in cross-sectional CLT 3 members significantly increased as a function of treatment duration, reaching maximum values at 14 days. However, the effectiveness of preservatives must be proven with higher experimental conditioning periods for both commercial DOT forms, in actual tests, under weathering exposures through a wide range of experimental screening tests should be enterprise to explain durability of CLT 3 members, including when often dry out conditioning experimental weight losses occurred while handling three-layer CLT materials for potential variables of the relative humidity due to the influence exterior exposure parameters. Combined results of Colak and Colakoglu,⁴⁴ and Lebow et al.⁴⁵ have observed on borate distribution studies from Southern pine (Pinus spp) lumber and beech (Fagus spp) veneers with improved penetration depths that increased proportionally with borate retentions and treatment durations, approximately 1.5% difference above the value of the minimum level required to protect the wood.

Table 1 shows the percentage of BAE related to impregnation through cross-sectional transverses and longitudinal grain directions, whereas three-layer CLT materials, treated with Boracol or Timbor formulations, followed a similar trend (Comparing both Boracol and Timber treatments). Given the differences in chemical composition between the two (Timbor and Boracol) formulations, actions are expected to occur within the mixture ingredients to wood components, particularly in enhancing the DOT substances. Consequently, when full-cell processed, the formulation changes due to parameters such as experimental durations, conditioning exposures, temperatures, pressure, or moisture content, as found by Piao et al.⁴⁶ on pressure-treated Southern pine species to evaluate the effects of surface preparation for borate preservative retention over mechanical strengths.

Table 1.

Understanding the gradient of boric acid equivalent (BAE in %) throughout Eastern black spruce CLT transverses and longitudinal grain directions of pressure-treated Boracol DOT and Timbor DOT specimens.

	Boracol DOT (%BAE)		Timbor DOT (%BAE)
Conditioning (days)	Longitudinal	Transverses	Longitudinal	Transverses
0	1.54	1.35	2.47	2.15
5	1.48	1.6	3.14	3.19
14	1.87	2.49	3.79	5.28

Each conditioning period represents an average of three replications.

CLT: cross-laminated timber; DOT: disodium octaborate tetrahydrate.

The study suggests that other treatment types, instead of pressure treatment processes, should be employed with wider conditioning durations through environmental weathering exposures to require before being boron treated, the use of variable moisture contents, which should be used, for example, borate with finishing's brush treatment process, or boron rod inserted at the in situ active sites of degradation of structured wood material.

Gradient of borate diffusion analysis in CLT 3 members

The gradient of borate diffusion was assessed in black spruce CLT 3 members to progressively describe the distribution as a function of treatment conditioning periods, as shown in Figure 3. The borate distribution flow obtained the same parabolic shapes at treatment duration corresponding to 0-day for DOT (Timbor and Boracol) formulations (Figure 3(a)). This curvature levelled off to a linear or straight shape for the rest of the conditioning periods (Figure 3(b)–(f)). Apart from the CLT surfaces closest to the treatment areas, where concentrations tended to decrease with treatment conditioning duration, the gradients were similar for the DOT (Timbor and Boracol) formulations (Figure 3(b)–(f)). The slight decrease in % BAE around the under core of transverse top and bottom grain faces could be explained by the fact that DOT treatment solutions were impregnated from both (radial and tangential) directions. In addition, the action of the silicone coating halted the migration flows, serving as a barrier to canal diffusion equally along the grain directions. These longitudinal sides forced the borate diffusion to flow through another sense. Also, in addition, the glue line or resin canal did not inhibit borate movement relative to the borate migration in the longitudinal central portion of CLT 3 layers. Examining all percentage values, the borate level in both DOT formulations was above the level to control the decay of about 0.15% BAE. Like the observations of Cabrera and Morell,⁴⁷ who showed that the effectiveness of preservative ingredients through Douglas-fir treated with fire-retarding products, such as borax and boric acid combined with fluoride or ammonium salts, zinc chloride, and chromium compounds, aimed to improve protection against decay fungi and termites for long-term durability. Gezer and co-workers⁴⁸ demonstrated that Southern pine (Pinus spp.) block samples treated sequentially with sodium borate or boric acid formulated with polyethylene glycol ranging from 400 to 600 showed significantly increased resistance to boron leaching from remedial treatment exposures over the same conditioning time frame of 14 days. The reasons for this effect are not clear, but it may reflect a tendency for the treatment process to retain and carry boron deeper into the core of the structural wood block concept. In addition, effectiveness improvement is affected by numerous factors, including the effect of glue type, permeability of the wood species, the form of the timber to be treated, preservative kind and the treatment procedure according to the AWPA, which recommended at least seven satisfactory criteria relating physical and chemical properties of a mixture of compounds concerning wood preservative formulations to be suitable for general long-term use (AWPA Standard A19-93.⁴⁹

Figure 3.

Gradient of borate diffusion in Eastern black spruce CLT three-element pressure-treated with DOT Boracol (a) and Timbor (b) conditioned at various durations starting from 0, 5, and 14 days. Each dot represents an average of three replicate samples. CLT: cross-laminated timber; DOT: disodium octaborate tetrahydrate.

Borate diffusion in three-layer CLT structures by PLS regression analysis

The projection to latent structures or PLS-R modelling of borate retention (Timbor and Boracol in %BAE, only validation date was shown) as a function of treatment conditioning durations and zonations in the three-layer CLT structures is represented in Figure 4. A linear relationship was obtained in the computation of the PLS-R variables for borate diffusion relating to Timbor and Boracol formulations, when borate migration transitioned from its low to its higher level of prediction responses, starting at 0-, 5-, and 14-day conditioning periods (Figure 4(a) and (b)). This linear relationship has resulted from a main effect of DOT (Timbor and Boracol) predictions that was revealed as significant (slopes of about 0.713 and 0.612, respectively) as the observed NIR spectral features (O–H overtones at ∼1450 nm and 1930 nm) obtained with Timbor and Boracol spectral data were already interpreted in the context of the review study of Tsuchikawa and Schwanninger,⁵⁰ clarifying that molecular vibrations are influenced by borate absorption (spectral data results not shown). The related statistics are shown in Table 2. The PLS regressions of DOT (Timbor and Boracol) samples were able to provide reasonably good correlations between the borate measured and predicted values. Correlations (r values) were 0.87 for both the treated DOT (Boracol and Timbor) samples, and R² values of validation models were 0.69 and 0.67, respectively. RMSE (Boracol DOT and Timbor DOT) ranged from 0.5611 to 0.6434% and 1.2813 to 1.5138%, respectively. PLS regression models were significant at p-value < 0.0001 about the main effect found (intercepts for Timbor DOT and Boracol DOT were 0.26 and 0.69, respectively) in the prediction responses, which means that the influence of scanned samples depends on the experimental setting conditions. Thus, results indicate moderate predictive accuracy, which could be explained by confounding effects occasioned in the fractional glue canal designs identified in the multilayer CLT system. Confounding actions in the sense that some effects cannot be studied independently of each other. The glue line design did not affect the flow migration along the CLT concept structures, which is a formalised way of defining how severe the confounding pattern was related to the number of replicate experiments to run. It can be seen from Figure 4 that NIRS may have some predictive abilities to estimate rapidly the borate distribution and retention quantitatively in small three-layer CLT member zones of black spruce samples. This ability to evaluate the distribution of boric acid-containing preservatives in small areas was previously observed by Taylor and Lloyd⁵¹ on southern pine (Pinus spp) sapwood species. They found good correlations for the validation and calibration dataset with R² values ranging from 0.86 to 0.96, and RMSE yielded 0.92 and 2.12%, respectively. However, significant performance in borate diffusion in the longitudinal direction, despite extrinsic and intrinsic influences due to weathering (UV-light, temperature, moisture content, concentration effects, reactivity of the active agent, and adhesive type) exposures, was achieved by Tsunoda⁵² on Cryptomeria japonica D. Don species, including De-Groot et al.⁵³ among Southern Pine (Pinus palustris Mill.), red maple (Acer rubrum L.) or Douglas-fir (Pseudotsuga menziesii (Mirs.) Franco) species to estimate the borate preservative effectiveness with no difference noted at concentration less than 1% BAE seemed sufficient to control biological agents of degradation.

Figure 4.

PLS regression models related to the gradient of boric acid equivalent diffusion (BAE in %) in Eastern black spruce CLT three-element untreated and pressure treated with DOT Boracol (a) and DOT Timbor (b) conditioned at various treatment durations (0, 5, and 14 days). Each dot represents an average of three NIR spectra for calibration and validation (calibration values not shown). CLT: cross-laminated timber; DOT: disodium octaborate tetrahydrate; NIR: near-infrared; PLS: partial least square.

Table 2.

Statistical metrics of the projection to latent structures or partial least squares (PLS) regression of untreated and pressure treated with Boracol DOT and Timbor DOT in Eastern black spruce CLT three layers, related to Figure 4: a) PLS regression-based calibration models and b) PLS regression-based validation models.

Formulations	Slope	Intercept	RMSE (%)	SE (%)	Bias (%)	r	RPD	p-value	n
DOT Boracol
a)	0.755	0.228	0.561	0.563	−7.137^e−07	0.87	2.0	0.0001	175
b)	0.713	0.26	0.643	0.645	−7.916^e−03	0.82	1.8	0.0001	175
DOT Timbor
a)	0.758	0.497	1.281	1.282	1.724^e−06	0.87	2.0	0.0001	166
b)	0.695	0.612	1.513	1.518	−1.58^e−02	0.81	1.7	0.0001	166

Each DOT formulation represents an average of three replications for calibration and validation.

CLT: cross-laminated timber; DOT: disodium octaborate tetrahydrate; n: number of objects; r: correlation coefficient; RMSE: root mean square error; RPD: relative percent difference.

Comparison of the sample-specific standard error of DOT (Timbor and Boracol) prediction analysis of three-layer CLT

Figure 5 represents the sample-specific standard error of borate retention prediction analysis related to Timbor and Boracol (in %BAE) as a function of the treatment conditioning durations and zonations in the three-layer CLT structures. The purpose of using the sample-specific standards error of borate prediction analysis was to reveal better performance for variable responses defining the subgroup in CLT 3 elements by comparing, identifying, and characterising them quantitatively among the scanned zones. Each subgroup or member sample is related to similar predicted values in wood-concept structures. For example, CLT 3 member samples were defined as standard groups necessarily dissimilar from one another. The aim was to investigate the effects on modelling performance to identify the characteristics of each member, as well as to complement the PLS regression method. This was introduced to the more general field of pattern recognition by Faber et al. in.⁵⁴ For example, to see if one or more newly scanned samples at different conditioning durations belong to an already existing group of specifications based on objective factor similarities. It is the scans belonging to one group of CLT samples showing similar rather than individual behaviour. The philosophy behind this classical chemometric technique was to verify first the significant influence of both DOT (Timbor and Boracol) formulations, as represented in Figure 5(a) and (b). When incorporating the treated and untreated (control) predicted values (in % BAE) along the longitudinal grain direction and across CLT edges (transverse top and bottom sides), the comparison for both the DOT (Timbor and Boracol) formulations, the statistical significance of the variances was in the same order of modelling. Ultimately, the observed significance was discussed in terms of Bagheri et al.,⁵⁵ who linked laboratory-scale modelling with the durability of small-scale CLT systems in service environments. Bagheri and co-workers assessed the borate treatment of Douglas-fir species at different retention levels. They found that the statistical prediction on adhesion strengths of mass timber structures was unaffected by borate treatment against biotic agents of degradation. In addition, the computation of statistical residual variance (the difference between observed and predicted values based on the standard F-test computation to check the risk of hypothesis rejections) was related to the main effects, which were confirmed based on the observed differences in response when comparing the average SD of the prediction analysis of the DOT (Timbor and Boracol) dataset. This varied from low to high levels in a similar range for the Boracol DOT and Timbor DOT formulations, of about 0.604 (±0.0025) and 1.49 (±0.0093), respectively. For % BAE, the Boracol DOT and Timbor DOT achieved 2.47% and 6,72%, respectively. These obtained results were weaker than those found on southern yellow or longleaf pine (Pinus palustris Mill), and Douglas-fir veneers (Pseudotsuga spp),^56,57 which had RPD ranging from 5.3 to 12 (±0.05) successively for measured and predicted performances. RPD in both DOT Boracol and Timber formulations ranged from 1.8 to 2 (±0.003) for validation and calibration values. With this approach, the scanned dataset specifically allows for displaying modelling only of the common properties of the three-layer CLT systems.

Figure 5.

Comparison of sample-specific standard error of prediction in Eastern black spruce CLT three-layer untreated and pressure treated with DOT Boracol (a) and Timbor (b) at different conditioning periods starting from 0, 5, and 14 days for laboratory screening exposure test. Each dot represents an average of three NIR spectra, and the error bars were calculated by incorporating the standard deviation (mean SD ranging from 0.6 to 1.5 ± 0.0025) of measurement error into the predicted data values. CLT: cross-laminated timber; DOT: disodium octaborate tetrahydrate; NIR: near-infrared.

Conclusions

Borate-formulated DOT (Timber and Boracol) distribution in pressure-treated sectional three-layer CLT structures was assessed by the potential of NIRS using the projection to latent structures or PLS regression models at different conditioning durations, starting at 0, 5, and 14 days, for further analysis about long-term durability. NIRS prediction-based % BAE achieved a good correlation between measured and predicted variables, ranging from 0.81 to 0.87 for Boracol and Timbor formulations. NIR scans were able to differentiate the slight changes of DOT (Timbor and Boracol) treatment in three-layer CLT elements through their physical and chemical attributes. In this study, the diffusion and quantification of industrial borate formulations based on active borate ingredients in three-layer CLT structures made of Eastern black spruce (Picea mariana var. Mariana) species were using NIRS coupled with multivariate statistical modelling (PLS regression models) in shorter conditioning periods, which involve further development of the environmental experiment to target other wood species, as well as simplify industrial treatment procedures using NIRS assistance. Such an approach remains empirical as the analysis was based on laboratory screening measurements. An operational method that can be utilised in a manufacturing plant environment needs to be developed, and physics-based models are sought to address explanations related to chemical mechanisms involved in borate effectiveness through the structural CLT concept.

Footnotes

Acknowledgments

The authors greatly appreciate the input of Tony Ung, Dr Myung Jae Lee, Dr Sedric Pankras, Dr Viktoriya Pakharenko, Dr Nicolas Tanguy, Dr Bouddah Poaty, and Prof. Ning Yang from the Daniels Faculty's Forestry-Daniels Academic, University of Toronto (U of T), for their support in laboratory technical assistance. Special thanks to Prof. Emeritus Paul A. Cooper from the U of T, who always shared his valuable time and advice on multiple occasions. Your wisdom can be found throughout this reviewed project, especially in helping to bring the Supbioaction business to grow.

ORCID iD

Thierry Koumbi-Mounanga

Funding

The authors received no financial support for the research, authorship, and/or publication of this article.

Declaration of conflicting interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

References

Gagnon

Bilek

EM (Ted)

Podesto

, et al. Chapter 1: CLT Introduction to cross-laminated timber. In: Karacabeyli

Douglas

(eds) CLT handbook: cross-laminated timber. U. S. ed. Washington, DC: US Department of Agriculture, Forest Service, 2013, pp.1–45. ISSN 1925-0495.

Glass

Wang

Easley

, et al. Enclosure. Chapter 10: Building enclosure design for cross-laminated timber construction. In US Edition of the CLT Handbook: cross-laminated timber. Washington, DC: Department of Agriculture, Forest Service, 2013, pp.1–55. ISBN 978-0-86488-553-1.

Morrell

Sexton

Preston

. Effect of moisture content of Douglas-fir heartwood on longitudinal diffusion of boron from fused borate rods. Forest Prod J 1990; 40: 37–40.

Rhatigan

Morrell

Freitag

. Movement of boron and fluoride from rod formations into Douglas-fir heartwood. Forest Prod J. 2002; 52: 38–42.

Drysdale

. Boron treatments for the preservation of wood- A review of efficacy data for fungi and termites. IRG/WP.94-30037. p22. Section 3. Wood protecting chemicals. 25th Annual Meeting, Bali, Indonesia, 1994, 29 May-3 June 1994.

Obanda

Shupe

Barnes

. Reducing leaching of boron-based wood preservatives- A review of research. Bioresour Technol 2008; 99: 7312–7322.

Grace

Yamamoto

. Simulation of remedial borate treatments intended to reduce attack on Douglas-fir Lumber by the Formosan subterranean termite (Isoptera: rhinotermitidae). J Econ Entomol 1994; 87: 1547–1554.

Koumbi-Mounanga

Ung

Groves

, et al. Moisture and surface quality sensing of Douglas-fir (Pseudotsuga menziesii var. menziesii) veneer products. The Forestry Chronicle 2013; 89: 646–653.

Salman

Pétrissans

Thévenon

, et al. Development of new wood treatments combining boron impregnation and thermo modification: effect of additives on boron leachability. Eur J Wood Prod 2014; 72: 355–365.

10.

Blow

Summers

. A laboratory evaluation of fused borate rods for treatment of timber. Int Biodeter 1985; 21: 27–39.

11.

Thevenon

Pizzi

. Polyborate ions’ influence on the durability of wood treated with non-toxic protein borate preservatives. Springer-Verlag Holz als Roh Und Werkst 2003; 61: 457–464.

12.

Freitag

Morrell

. Development of threshold values for boron and fluoride in non-soil contact applications. Forest Products Journal 2005; 55: 97–101.

13.

Awoyemi

Ung

Cooper

. In-treatment cooling during thermal modification of wood in soy oil medium: soy oil uptake, wettability, water uptake, and swelling properties. Eur J Wood Prod 2009; 67: 465–470.

14.

Lebow

Halverson

. Boron diffusion in surface-treated framing lumber. Forest Products Journal 2013; 63: 275–282.

15.

Saadat

. Factors affecting the distribution of borate to protect building envelope components. M.Sc. Thesis and Dissertation in Forestry, Faculty of Forestry, The University of Toronto, Ontario, Canada, 2012, 84pp.

16.

Koumbi-Mounanga

Groves

Leblon

, et al. Estimation of moisture content of trembling aspen (Populus tremuloides Michx.) stands by near-infrared spectroscopy. Eur J Wood Prod 2015a; 73: 43–50.

17.

Meder

Thumm

Bier

. Veneer stiffness predicted by NIR spectroscopy calibrated using mini-LVL test panels. Holz Roh Werkst 2002; 60: 159–164.

18.

Burns

Ciurczak

. Handbook of near-infrared analysis. 3rd ed. p.cm. Pract Spectrosc 2007, 834pp.

19.

Schwanninger

Rodrigues

Fackler

. A review of band assignments in near-infrared spectra of wood and wood components. J Near Infrared Spec 2011; 19: 287–308.

20.

Leblon

Adedipe

Hans

, et al. A review of near-infrared spectroscopy for monitoring moisture content and density of solid wood. The Forestry Chronicle 2013; 89: 595–606.

21.

Koumbi-Mounanga

Morris

Myung

, et al. Prediction and evaluation of borate distribution in eastern black spruce (Picea mariana var. Mariana) wood products. Wood Sc Tech 2015b; 49: 457–473.

22.

Koumbi-Mounanga

Ung

Cooper

, et al. Surface quality sensing of trembling aspen (Populus tremuloides Michx.) veneer products by near-infrared spectroscopy. Wood Mat Sci Eng 2015c; 10: 17–26.

23.

Koumbi-Mounanga

Cooper

Yan

, et al. Prediction of boron content in wood pellet products by near-infrared spectroscopy. Forest Products Journal 2016; 66: 37–43.

24.

Laminated Timber Institute of Canada (LTIC). Timber design manual. P.O. Box2698, K1P5W7, Ottawa. CA Printed version, 1972, 458pp.

25.

Koch

. Utilization of the southern pines. Agriculture Handbook No. 420. Volume II: Processing. U.S. Department of Agriculture Forest Service, Washington, DC, 1972, Printed version, 1663pp.

26.

Momentive^TM. Material safety data sheet. Technical note for industrial use only, 2011, 8p. [online] Available from: http://www.sfm.state.or.us/CR2K_SubDB/MSDS/CASCOPHEN_RBS2345.PDF

27.

National Building Code of Canada. Canadian Wood-Frame House Construction. Central Mortgage and Housing Corporation, Société canadienne d'hypothèques et de logement (SCHL), Ottawa, ON, 1970, Printed version, 200pp.

28.

American Wood-Preservers’Assocociation. (AWPA). Standard methods for the analysis of boron-containing preservatives. AWPA Standard A03-05. AWPA, Selma, AL, 2006.

29.

American Wood-Preservers’Association. (AWPA). Standard methods for the analysis of waterborne preservatives and fire-retardant formulations. AWPA Standard A2-05. AWPA, 2005.

30.

Williams

. The extraction and determination of disodium octaborate in Sitka spruce. Analyst 1968; 93: 1–111.

31.

Murphy

Turner

. A vapor phase preservative treatment of manufactured wood-based board materials. Wood Sci Technol 1989; 23: 273–279.

32.

Cockcroft

. Evaluating the performance of wood preservatives against fungi. J Inst Wood Sci 1974; 6: 2–8.

33.

American Wood-Preservers’Association. (AWPA). Standard methods for the analysis of wood and wood treating solutions by inductively coupled plasma emission spectrometry. AWPA Standard A21-00. AWPA, Selma, AL, 2000.

34.

NIRQuest NIR Spectrometers Data Sheet. NIRQuest256-2.5. Ocen Optics, Halma Group Company, FL, US, 2015. [online] Available at: https://www.rktech.hu/dokumentaciok/OceanOptics/Spectrometers/OEM%20Data%20Sheet%20--%20NIRQuest.pdf

35.

Rosipal

Kramer

. Overview and recent advances in partial least squares. In: Subspace, latent structure, and feature selection. vol. 3940. Berlin: Springer, 2006, pp.34–51.

36.

Esbensen

Guyot

Houmoller

. Multivariate data analysis in practice: an introduction to multivariate data analysis and experimental design. 5th ed. Oslo: CAMO Process AS, 2002, 598pp.

37.

Miller

. Statistics and Chemometrics for analytical chemistry. Ashford Colour Press Gosport, Hants, Great Britain, 5th edn. Pearson Prentice Hall, England, 2005. ISBN 0131291920. Printed version, 268pp.

38.

Faber

Song

Hopke

. Sample-specific standard error of prediction for partial least squares regression. Trends Anal Chem 2003; 20: 1–5.

39.

Boulesteix

Strimmer

. Partial least squares: a versatile tool for the analysis of high dimensional genomic data. Brief Bioinform 2006; 8: 32–44.

40.

Geladi

Kowalski

. Partial least-squares regression: a tutorial. Anal Chim Acta 1986; 185: 1–17 .

41.

Payne

. Plywood construction manual. Published by the Council of the Forest Industries of British Columbia, 2nd edn. Vancouver, Canada, 1971. Printed version, 143pp.

42.

Saadat

Cooper

. Factors affecting the distribution of borate to protect building envelope components from biodegradation. Technical Note 1. Project code: T4-5-C10. Forestry. Faculty of Forestry, The University of Toronto, Ontario, Canada, 2013.

43.

Colakoglu

Colak

Aydin

, et al. Effect of boric acid treatment on mechanical properties of laminated beech veneer lumber. Silva Fenn 2003; 37: 505–510.

44.

Colak

Colakoglu

. Volatile acetic acid and formaldehyde emission from plywood treated with a boron compound. Build Environ 2004; 39: 533–536.

45.

Lebow

Halverson

. Penetration of boron from topically applied borate solutions. Forest Prod J 2010; 60: 13–22.

46.

Piao

Monlezun

Shupe

. Glueline bonding performance of decommissioned CCA-treated wood. Part I: without retreatment. Forest Prod J 2009; 59: 36–42.

47.

Cabrera

Morell

. Effect of wood moisture content and rod dosage on boron or fluoride movement through Douglas-fir heartwood. Forest Products Journal 2009; 59: 93–96.

48.

Gezer

Micheal

Morrell

. Effects of glycol on leachability and efficacy of boron wood preservatives. Wood Fiber Sci 1999; 31: 136–142.

49.

American Wood-Preservers’Association. (AWPA). Sample preparation for determining penetration of preservatives in the wood. AWPA Standard A19-93. Selma, AL: AWPA, 1996.

50.

Tsuchikawa

Schwanninger

. A review of recent near infrared research for wood and paper. (Part2). Appl Spectrosc Rev 2013; 48: 560–587.

51.

Taylor

Lloyd

. Potential of near-infrared spectroscopy to quantify boron concentration in treated wood. Forest Prod J 2007; 57: 414–423.

52.

Tsunoda

. Preservative properties of vapor-boron-treatment wood and wood-based composite. J Wood Sci 2001; 47: 149–153.

53.

De-Groot

Felton

Crawford

. Distribution of borates around point source injections in wood members exposed outside. Res. Note. FPL-RN-0275. Madison, WI: US. Department of Agriculture, Forest Services, Forest Products Laboratory, 2000, 5p.

54.

Faber

Bro

. Standard error of prediction for multiway PLS1: background and a simulation study. Chemometr Intell Lab Syst 2002; 61: 133–149.

55.

Bagheri

Alinejad

Ohno

, et al. Improving durability of cross laminated timber (CLT) with borate treatment. J Wood Sc 2022; 68: 34.

56.

Lloyd

Manning

. Developments in borate preservation technology. Proc. BWPDA, The UK, 1995.

57.

Lloyd

Scoeman

Stanley

. Remedial timber treatment with borates. In: Robinson

Rambo

, editors. Proceedings of the 3rd international conference on urban pests. WMH, 1999. Available from http://www.icup.org.uk/reports%5CICUP452.pdf.