Advertisement
Research Report

Influence of Grain Type and Surface Area on Extraction of French Oak in Model Wine: A Two-Dimensional Kinetic Analysis

Mackenzie Aragon, Robert Coleman, Roger B. Boulton, Thomas S. Collins
Am J Enol Vitic.  2026  77: 0770021  ; DOI: 10.5344/ajev.2026.25052

Abstract

Background and goals An advantage of using barrel alternatives for toasted oak products is that the desired oak volatiles are more readily extracted than with the use of traditional wine or spirits oak barrels. This rapid extraction is typically attributed to the product’s increased surface-to-volume ratio, regardless of grain type (end grain [EG] versus cross grain [CG]). This study examines the effect of surface area (SA) and grain type on the extraction of oak aroma compounds.

Methods and key findings Medium-plus convection toasted French oak (Quercus robur/petraea) staves were sequentially cut in half by length (25.4, 12.7, 6.35, 3.18, and 1.59 cm), resulting in treatments with the same total weight and volume but increased total SA and CG-to-EG ratio. The oak dosage of 20.9 ± 1.4 g/L was extracted in model wine (15% [v/v] ethanol) for 120 days. Headspace solid-phase microextraction gas chromatography-mass spectrometry quantified common oak aroma compounds, including furfurals, guaiacols, and lactones. An average 102.5% increase in total SA resulted in a 90 to 450% increase in concentration, varying by compound and not proportional to the total SA. The results suggest that increased exposure to EG is crucial for understanding extraction, as the additional SA primarily comes from EG, varying from 2.79 to 32.5% of the total SA.

Conclusions and significance A pseudo-first-order kinetic model accounting for both EG and CG best fits the measured data. This novel work models and quantifies the twodimensional nature of oak in the extraction of aroma compounds.

Introduction

Traditionally, oak barrels have been used during the maturation of wine and spirits to add aroma, color, and mouthfeel. Alternative oak products were originally developed to reduce the cost of producing traditional barrels (Caputo 2006, Hines 2015, Marr 2024); unlike barrels, these products are primarily submerged in the liquid of other vessels (i.e., stainless steel tanks). Alternative oak products are produced in a range of shapes, sizes, and formats (e.g., staves, planks, cubes, beans, and powder). Producers have capitalized on the popularity of oak alternatives to create a diverse range of formats and toasts, expanding their portfolios and allowing for customized flavor profiles. Additionally, oak alternative products are more sustainable than traditional oak barrels (Marr 2024) because they allow for more product per cubic meter of the tree to be used compared with conventional barrels, and a greater volume of oak material to be transported with less environmental repercussion. As most of the ecological impact occurs during the harvesting and seasoning stages of processing (Flor et al. 2017, García-Alcaraz et al. 2020), it is critical to use as much of the seasoned material as possible to prevent further waste.

The primary winemaking advantage generally associated with alternative oak products is a more rapid extraction of volatile compounds than can be achieved by traditional oak barrels (Caputo 2006, Hines 2015, Marr 2024). This rapid extraction has been attributed to the increased total surface area (SA)-to-volume ratio (Hines 2015, Marr 2024). However, little work has been done to explore the influence of grain type on the extraction of alternative oak products. Oak wood has two general grain types, cross grain (CG) and end grain (EG) (Record 1911, Glass and Zelinka 2010, De Blaere et al. 2023). CG runs transversely to the vascular vessels (i.e., xylem and phloem) and EG directly exposes the vascular vessels. In a traditional barrel format, the liquid is exposed to the CG, with little to no contact with the EG (negligible contact around the bung, croze, and head of the barrel). This is necessary for liquid retention, just as a tree maintains internal liquid retention when the integrity of the cross section is intact, and leaks sap when the EG is exposed (i.e., through lateral cuts or wounds). Meanwhile, both grain types are directly exposed to the wine, as alternative oak products are not meant to hold liquid (Figure 1). Due to the wide range of formats available, these products vary significantly in EG-to-CG ratios (EG:CG).

A labeled wooden block distinguishes two cross grain faces from one end grain face. The diagram shows a rectangular oak block with visible wood texture. Two arrows labeled Cross grain point to the upper and side faces, where the grain runs lengthwise across the surfaces. An arrow labeled End grain points to the cut end face, which displays the exposed ends of the wood fibers.
Figure 1

An oak alternative product has four sides with cross grain (CG), and two faces are end grain (EG). The EG:CG ratio will vary. Current extraction models account for only one variable. Using a one-dimensional model does not account for the differences in extraction kinetics between the two types of grains exposed in the alternatives.

There is limited research on the kinetics of oak extraction in alcohol solutions or beverages (Delgado-González et al. 2021). In previous research, the oak fragment was evaluated as a single surface or one-dimensional particle, without considering individual contributions from the EG and CG (Psarra et al. 2015). This simplistic approach is common to various fruit extractions, including coffee (Espinoza-Pérez et al. 2007), apple (Hobbi et al. 2021), and vanilla (Rodríguez-Jimenes et al. 2012). As EG has greater permeability than CG (Côté 1963), this one-dimensional model approach may lack a complete understanding of alternative oak products, but it would still be useful for the barrel scenario. Conversely, knowledge of alternative oak extraction would allow for an understanding of both oak formats (barrels and alternatives). The Lagergren model, a pseudo-first-order model, has been used to describe the extraction of color from barrels into grape distillate (Delgado-González et al. 2021). This model is commonly used to describe liquid-solid adsorption systems, in which the concentration of the adsorbed solute is measured over time and the equilibrium concentration of the solute adsorbed onto the solid is calculated (Lagergren 1898, Ho 2004, Yuh-Shan 2004, Tseng et al. 2010). In both this work and the prior work (Delgado-González et al. 2021), the Lagergren model is modified such that the concentration of the solute extracted into the liquid is measured over time, and the equilibrium concentration of the solute extracted from the solid into the liquid is calculated. In turn, this modification enables calculation of the pseudo-first-order extraction rate constant for this liquid-solid system.

This study evaluated the time-course extraction of oak volatile compounds from an alternative oak product in a model wine solution over 120 days. A one-dimensional pseudo-first-order kinetic model accounting for the total SA and a two-dimensional pseudo-first-order kinetic model accounting for the SAs of EG and CG are proposed and fitted to the measured data (modified from the Lagergren solid-liquid kinetic model). This work aims to provide insight into the anecdotal industry observation that oak alternative products result in faster extraction of volatile compounds than that of traditional oak barrels. Lastly, the chemical nature of the individual compounds is compared to their two-dimensional modeled extraction rate constants to further elucidate the nature of wood aroma extraction.

Materials and Methods

Materials

Chemicals used for the volatile analysis, including furfural (CAS#98-01-1, >99%), 5-methyl furfural (CAS#6200-02- 0, >99%), guaiacol (CAS#90-05-1, >99%), 4-methylguaiacol (CAS#93-51-6, >99%), whiskey lactone (CAS#254-357-4, ≥98%), and 4-ethylguaiacol (CAS#2785-89-9, >98%), as well as 𝛾-hexalactone (CAS#695-06-7, >99%) and o-cresold8 (CAS#203645-65-2, >99%) as the internal standards, were obtained from Sigma-Aldrich.

Scanning electron microscopy (SEM)

Oak segments from the same production lot used for extraction (see below) were used for SEM imaging. Segments of ~0.5 cm2 were sputtered using a Cressington High Resolution Sputter Coater and a gold target plate (99.99%), depositing within 0.1 nm of the gold. The conductivity of the gold coating was necessary to prevent charging when exposed to the electron beam in the SEM environment, characterized by the high vacuum and high voltage. Whether the segment was to be used for EG or CG imaging would determine which side would be used to stick a carbon tack on the designated stage. In total, eight segments were used. The tacks were used to ground the specimen and hold it in place during the SEM process. SEM imaging was performed on an Apreo VolumeScope SEM (ThermoFisher Scientific) using an Everhart-Thornley detector (spot size 3.0, working distance 9.6 to 12.6 mm, chamber pressure 3.6 to 7.6 × 10−3 Pa; magnification 1100×). All images were taken at the Washington State University Franceschi Microscopy & Imaging Center.

Samples, sample preparation, and extraction

A commercially available French oak (Quercus robur/ petraea) medium-plus convection-toasted alternative oak from a single production lot was chosen to minimize the differences in toasting but not in the wood anatomy. Traditionally, Q. robur/petraea barrels had to be split to create a leakproof vessel; since alternative formats are intended for different uses, a sawn technique is commonly used. The staves were cut sequentially to length, resulting in five treatments at lengths of 25.4, 12.7, 6.35, 3.18, and 1.59 cm (Figure 2). The number of oak segments used per treatment doubled as the segments were cut (Table 1). This was to ensure that all treatments had the same volume, resulting in an average dosage rate of 20.9 ± 1.4 g/L across all treatments. As more of the EG is exposed, the percentage of exposed EG increases between each treatment by 50%, while the total SA increases by 2.3, 5.3, 12, and 15% when the lengths of the stave segments shorten. This resulted in treatments with EG:CG ratios of 0.03, 0.06, 0.11, 0.24, and 0.48 (Table 1). The extraction process took place in 3.78-L stainless steel vessels (Ozark Trail) using a model wine solution (15% [v/v] ethanol and 5 g/L potassium bitartrate at pH 3.3). All treatments were performed in triplicate. All extracts were stored in cellar conditions (on the research winery floor at temperatures between 20 to 25°C) for 120 days. On days 2, 4, 6, 7, 8, 10, 15, 30, 60, 90, and 120 of the extraction, 20 mL was collected from each vessel for chemical analysis. Samples were stored at -80°C until chemical analysis.

A five level arrangement of oak blocks shows increasing surface area and end grain to cross grain ratio as block length decreases. The diagram arranges oak blocks in five horizontal levels, with each successive level dividing the wood into more and shorter blocks. Surface area, S A, and end grain to cross grain ratio, E G to C G, increase as length decreases. The levels list: length 25.4 centimeters, S A equals 318 square centimeters, and E G to C G equals 0.03; length 12.7 centimeters, S A equals 326 square centimeters, and E G to C G equals 0.06; length 6.35 centimeters, S A equals 343 square centimeters, and E G to C G equals 0.11; length 3.18 centimeters, S A equals 391 square centimeters, and E G to C G equals 0.24; and length 1.59 centimeters, S A equals 461 square centimeters, and E G to C G equals 0.48. Between consecutive levels, S A increases by 2, 6, 14, and 18 percent, while E G to C G increases by 100, 95, 111, and 103 percent. Across the full sequence, S A increases by 45 percent and E G to C G increases by 1600 percent.
Figure 2

Experimental design for sequentially increasing the end grain to cross grain ratio (EG:CG). As the amount of EG increased, the amount of surface area (SA) exposed also increased.

View this table:
Table 1

Average measurements and specifications of the triplicate oak segments used for each treatment from the same production lot. Significance on each column was tested using a one-way analysis of variance and a Tukey multiple-comparison post-hoc test at p < 0.05.

The extracts were diluted 1:10, and 10 mL was added to a 20-mL gas chromatography headspace vial for volatile analysis (Restek). A Gerstel Multipurpose Sampler was used to add 8 μL of internal standard (48.83 μg/L o-cresold8 and 66.752 μg/L g-hexalactone) (Pérez-Coello et al. 1998) in >99% methanol (all from Sigma-Aldrich).

Chemical analysis

Targeted solid-phase microextraction gas chromatography-mass spectrometry (SPME-GC-MS) analysis was conducted with an Agilent 7890B gas chromatograph coupled with a 5977 Agilent mass spectrometer and an Agilent J&W DB-Wax 122-7032 column (29.86 μm × 250 μm × 0.25 μm). The autosampler was operated in headspace SPME mode, equipped with a divinylbenzene/carboxen/polydimethylsiloxane (DVB/CAR/PDMS) 24-ga 2-cm fiber (Supelco), which was conditioned before use according to the manufacturer’s directions. Samples were incubated in the agitator (incubation temperature of 50°C) for 2.00 min at 250 rpm. The SPME fiber was extracted from the sample headspace for 5.00 min before a 3.00-min desorption time in the GC inlet. The fiber was baked out for 15.00 min between every sample using a Gerstel SPME Conditioning Station at 250°C.

The GC inlet was set to splitless mode with an Agilent 5190-4048 liner at a temperature of 250°C. The oven was programmed to start at 40°C (held for 1.00 min), followed by an increasing rate of 25°C per minute to reach 160°C (held for 7.00 min), then an increase of 35°C per minute to reach a final temperature of 240°C (held for 3.00 min). The total runtime was 18.09 min. Ultrapure helium flow gas was used at a constant column flow of 1.2 mL/min.

The MS transfer line temperature was 250°C. Detection was carried out by electron-impact MS in selected ion monitoring (SIM) mode, with an ionization energy of 70 eV (MS source temperature set to 230°C and MS quad temperature set to 150°C). Quantitative determinations were made using SIM. The selected ions for each targeted compound are shown (Table 2). Measurements of each compound were compared to a corresponding calibration made from a known concentration of the pure substance. Linear regression coefficients were between 0.987 and 0.997. The average concentration and standard error of each compound by the EG:CG ratio by time point are provided (Supplemental Tables 1 and 2).

View this table:
Table 2

Targeted compounds and the gas chromatography-mass spectrometry parameters used for the identification and quantification of the oak-derived volatile compounds.

Data analysis

Data were analyzed using quantitative analysis in Agilent MassHunter Quantitative Analysis (ver. 10). Individual analysis of variance (ANOVA) (p = 0.05) and the Tukey multiple comparison post-hoc test were performed on individual sampling days using XLSTAT (Addinsoft).

Kinetic model

The pseudo-first-order kinetic model (modified from Lagergren [1898]) accounting for both one dimension (total SA) and two dimensions (CG and EG SAs) was used to simultaneously fit the five EG:CG treatments. The one-dimensional equation (Equation 1) was solved for two parameters, ktotal_SA and ceq, and the two-dimensional equation (Equation 2) was solved for four parameters, keg, kcg, ceq(eg), and ceq(cg), using the differential evolution parameter estimation routine supplied in the ‘DEoptim’ package (Mullen et al. 2011) within R (R Core Team 2023). The parameter estimation routine minimized the residual sum of squares (RSS) between the model fit and the measured concentration of the aroma compound at time t, (c(t)). The model equations were written in differential form and solved using Euler’s method with a time step of 0.5 days to minimize local truncation error (Coleman et al. 2024).

dc(t)dt=SA(total)Vl×ktotal_SA(ceq−c(t))Eq. 1dc(t)dt=SAegVl×keg(ceq(eg)−c(t))+SAcgVl×kcg(ceq(cg)−c(t))Eq. 2

Equation 1 contains the pre-exponential factor SAtotal/Vl, while Equation 2 has both SAeg/Vl and SAcg/Vl pre-exponential factors to accommodate the EG and CG SAs to liquid volume. These pre-exponential factors allow fitting of the experimental data (accounting for volume removal during sampling) and model prediction at any given oak SA (barrel or alternative) to volume, thus enabling industry application and scaled prediction. The balance of the equation, in the generic k(ceq - c(t)) form, is derived from Lagergren (1898) and represents a one-dimensional analysis. In this study, the one-dimensional analysis evaluates all grain surfaces as equivalent in the form of total SA. Alternatively, the two-dimensional analysis separates and sums the pseudo-first-order solid-liquid extraction equation into EG and CG forms, keg(ceq(eg) - C(t)) and kcg(ceq(cg) - C(t)), respectively.

Results

SEM

Mature Quercus robur, petraea, and alba trees are used to produce both traditional oak barrels and alternative oak products. When a traditional barrel is used, the liquid is only exposed to the CG of the barrel; however, with the use of alternative oak products, the EG is also present. Both grain types were examined for structural architecture using SEM. Multiple samples with each grain type were prepared to ensure clear SEM images could be captured.

An apparent size difference among the vascular pores was observed when comparing grain types in microscopy images of various oak surfaces (Figure 3). The magnification for the EG was 27.2% of what was needed to observe the vascular vessels of the CG. A single pore from the EG was ~300 μm in diameter, compared to the CG, where the average was ~7 μm. The CG required a magnification of 750× or greater to begin to see the vascular network present. The visual analysis of the grain types suggested that structural differences may contribute to differences in extraction rate as the EG:CG ratio changes.

A two-panel scanning electron micrograph compares small cross-grain pores with larger end-grain pores in medium-toasted convection French oak. The two scanning electron micrographs show the porous structure of medium-toasted convection French oak. Panel A shows the cross grain at 1,104 times magnification with a 50 micrometer scale bar. Numerous small oval pores are distributed across the exposed surface and within elongated channels. Panel B shows the end grain at 300 times magnification with a 300 micrometer scale bar. Several large open vessels are surrounded by rough fibrous tissue.
Figure 3

Scanning electron microscope imaging of cross grain (A) pores and structures, as well as the end grain (B) of medium toasted convection French oak. The average pore diameter for the cross grain is ~7 μm compared to that of 300 μm found at the end grain.

Chemical analysis compared to physical dimensions

Variation was expected due to the natural origin of the starting material; however, staves from a single production lot were used to minimize variation as much as possible. The sequential halving of the length of the stave segments resulted in a 50% increase in EG:CG between treatments and increases of 2.3, 5.3, 12, and 15% in SA as the segments decreased in length (Figure 2). There was a 45% increase in SA and a 1580% increase in EG:CG between the largest and smallest oak segments. The amount of CG did not differ significantly between treatments (one-way ANOVA, p > 0.05; Table 1). The intentional doubling of EG between treatments was employed to evaluate the effect of exposure on extraction.

The contribution of EG SA per volume was plotted against the average 120-day concentration of guaiacol, a representative compound (Figure 4). A linear correlation was found between these variables in the EG (SAeg/Vl), with an R2 of 0.978. This was additional evidence supporting the importance of grain type in oak extraction.

A scatterplot graph shows guaiacol concentration increasing linearly with end grain surface area per volume after 120 days. The scatterplot graph plots average guaiacol concentration after 120 days against end grain surface area per volume. The horizontal axis, labeled E G S A per volume in square centimeters per liter, ranges from 0 to 40. The vertical axis, labeled concentration in micrograms per liter, ranges from 0 to 12. Five points occur at approximately 2 and 5.2, 5 and 5.7, 10 and 6.5, 20 and 8.2, and 40 and 10.1. A dashed linear trend line rises through the points, with R squared equal to 0.979.
Figure 4

Average guaiacol concentrations after 120-day extraction versus end grain (EG) surface area (SA) per volume. A positive relationship between the SA of EG and the final concentration of guaiacol was observed.

Time courses for guaiacol and furfural (Figure 5) showed an increased overall extraction rate as the EG:CG ratio rises for all seven oak aroma compounds quantified, which included furfural, 5-methylfurfural, guaiacol, trans-oak lactone, 4-methylguaiacol, cis-oak lactone, and 4-ethylguaiacol. However, the 120-day guaiacol concentration, as noted above, showed a linear response to increases in the pre-exponential factor for EG (SAeg/Vl), which also had a linear relationship with the EC:CG ratio. While guaiacol, 5-methylfurfural, trans-oak lactone, 4-methylguaiacol, cis-oak lactone, and 4-ethylguaiacol had increasing 120- day concentrations with EG:CG, the furfural concentration did not significantly differ with increasing EG:CG using a one-way ANOVA (p > 0.05). The overall extraction rate increased with EG for furfural, yet the 120-day concentration did not; this suggests a similar contribution of furfural from both the EG and CG (see Kinetic Model below).

A four-panel graph compares guaiacol and furfural concentration over 120 days for five end grain to cross grain ratios using measured data and fitted curves. The four vertically arranged graphs plot concentration in micrograms per liter against time from 0 to 120 days. Symbols show measured values, and curves show model fits for end grain to cross grain ratios of 0.03, 0.06, 0.11, 0.24, and 0.48, represented by plus signs, triangles, diamonds, squares, and circles, respectively. Panels A and B show guaiacol concentrations on a scale from 0 to 225 micrograms per liter. In panel A, all five fitted curves rise rapidly before leveling near 145 to 150 micrograms per liter, with higher ratios generally rising faster. In panel B, the fitted curves separate progressively and level near 98, 110, 127, 160, and 200 micrograms per liter as the ratio increases from 0.03 to 0.48. Panels C and D show furfural concentrations on a scale from 0 to 7,000 micrograms per liter. In panel C, all fitted curves rise and approach a similar final concentration near 5,800 micrograms per liter, while higher ratios generally reach the plateau sooner. In panel D, the extraction rate increases markedly with the ratio. The curve for 0.48 rises most rapidly and levels near 6,000 micrograms per liter, followed by 0.24, 0.11, 0.06, and 0.03. By 120 days, the fitted concentrations in panel D range from approximately 5,600 to 6,200 micrograms per liter. All values are approximated.
Figure 5

One-dimensional analysis of average concentration of guaiacol (A) and furfural (C) for each treatment (symbols), with the pseudo-first-order kinetic model fit to the data for each treatment (lines). A two-dimensional analysis of guaiacol (B) and furfural (D) is also shown. Each compound exhibited different behaviors, both in terms of extraction rate and final concentration, as the EG:CG and surface area increased. EG, end grain; CG, cross grain.

Kinetic model

A one-dimensional model incorporating the change in total SA was initially evaluated (Figure 5A and 5C) for the five EG:CG extraction time courses. Given the apparent changes in rate with increased EG:CG ratio as well as the differences in the 120-day concentrations, a two-dimensional kinetic model incorporating EG and CG SAs was used to evaluate the five EG:CG time courses (Figure 5B and 5D). The resulting RSS favored a better fit with the two-dimensional kinetic model, with RSS values of 8.60 × 103 and 6.35 × 106 for guaiacol and furfural, respectively. By contrast, the one-dimensional kinetic model had RSS values that were each nearly a magnitude larger (3.76 × 104 and 4.35 × 107 for guaiacol and furfural, respectively). Additionally, the average coefficient of determination (as defined by Kvålseth [1985]) for guaiacol and furfural in the two-dimensional kinetic model was 0.977 ± 0.005 and 0.977 ± 0.008, respectively, whereas the one-dimensional model yielded 0.833 ± 0.033 and 0.803 ± 0.089, respectively. The comparison presented in Figure 5A and 5B for guaiacol and in Figure 5C and 5D for furfural clearly demonstrates that the one-dimensional model lacks the parameters needed to describe the resulting extraction time-course data.

The two-dimensional model fit the time courses of the seven oak volatile compounds of interest (Table 3). In each case, the equilibrium concentration from EG, ceq(eg), was greater than the equilibrium concentration from CG, ceq(cg), except for furfural. The equilibrium furfural concentrations of ceq(eg) = 5.92 × 103 μg/L and ceq(cg) = 7.46 × 103 μg/L were relatively similar, unlike the remaining compounds, which suggests a ubiquitous extraction of this low molecular weight compound from both the EG and CG.

View this table:
Table 3

Pseudo-first-order kinetic constants of the end grain and cross grain. RSS, residual sum of squares.

Unlike the modeled equilibrium concentrations, the rate constants keg and kcg did not have an obvious relationship with each other. Instead of head-to-head comparisons with each other, the rate constants were graphed against the compound’s molecular weight to the power of -0.6 (Figure 5). Diffusivity for dilute solutions has been defined as (Wilke and Chang 1955):

D=7.4×10−8((xM)12TηV0.6)Eq. 3

where D is the diffusion constant, x is an association parameter of the solvent, M is the molecular weight, T is the temperature, η is the viscosity of the solution, and V is the molar volume. This equation was further derived in molar volumes (Bird et al. 2015). The molar volume of a compound is proportional to the compound’s molecular mass or molecular weight (in g/mol). In turn, the CG plot (Figure 6) showed a weak negative linear relationship, whereas the EG plot showed a strong positive linear relationship with an R2 of 0.88. This relationship suggests diffusional kinetics with the EG contribution, indicating internal diffusion within the oak pieces.

A scatter graph compares end grain and cross grain kinetic constants against molecular weight raised to the power of minus 0.6. The scatter graph plots the kinetic constant, k subscript e g, in liters per square centimeter day against molecular weight raised to the power of minus 0.6 in moles per gram. The horizontal axis ranges from 0.045 to 0.065, and the vertical axis ranges from 0 to 3.0 times 10 to the minus 3. End grain values are shown by circular points, and cross grain values are shown by cross-shaped points. The end grain points remain near zero below 0.05, then increase from approximately 3.0 times 10 to the minus 4 at 0.052 to 2.6 times 10 to the minus 3 at 0.065. The ascending dashed linear fit is labeled y equals 0.135 x minus 0.007 and R squared equals 0.880. Cross grain points range from approximately zero to 5.0 times 10 to the minus 4 and generally decrease as the horizontal-axis value increases. The descending dashed linear fit is labeled y equals minus 0.0144 x plus 0.0011 and R squared equals 0.4031.
Figure 6

Kinetic constants for end grain (EG) and cross grain (CG) are plotted on the y-axis against molecular weight (MW) to the power of -0.6. A correlation is observed between the EG and the MW, indicating that diffusion plays a role in the extraction through the EG.

Discussion

Grain type and SA

When evaluating oak products, it is crucial to consider the toasting method and toast intensity. Untoasted wood has a relatively low volatile content compared to toasted wood. During pyrolysis, lignin, hemicellulose, and other components are degraded, yielding numerous compounds of interest. Each compound behaves differently under prolonged heating; for example, furfural derived from hemicellulose will degrade, whereas guaiacol derived from lignin will accumulate (Collins et al. 2015, Farrell et al. 2015, Le Floch et al. 2015, González-Centeno et al. 2016). Previous studies have shown that the toasting method and protocol have distinct influences on the final volatile composition (Collins et al. 2015, Farrell et al. 2015, Pollon et al. 2023). The concentrations of the measured compounds ultimately determine the kinetic constants in the multidimensional pseudo-first-order kinetic model (modified here from Lagergren [1898]). In this study, a single toast intensity (medium-plus toast) was evaluated to eliminate concentration variation across a range of toast intensities (thermal intensities during toasting) and to focus on kinetic principles before adding more variables to an already complex system. A convection toasted product was selected because the temperature throughout the toasting process is generally more consistent, reducing the gradient experienced in the wood, unlike traditional toasting (i.e., fire toasting) methods. This consistency was apparent in the homogeneous color observed in each dissection of the stave during the creation of the five EG:CG segment trials. Conversely, the entire stave may not be fully thermally penetrated during the traditional fire method (Dupleix et al. 2013, Collins et al. 2015, Farrell et al. 2015, Sikora et al. 2018, Kubovský et al. 2020). The use of model wine in this study is important to consider when evaluating the kinetic model’s results and trends. The neutral background and similar conditions prevent matrix interference, allowing evaluation of oak extraction kinetics. Specific kinetic constant values will vary in wine and other matrices; however, the concepts and trends discussed provide a new perspective on extraction kinetics.

From the time courses of each volatile compound, there were two distinct result types: one where the overall rate increased and the final concentration reached a similar equilibrium concentration (furfural represents this condition in Figure 5C and 5D), and the other where the overall rate and the final concentration increased with increasing EG:CG ratio (guaiacol represents this condition in Figure 5A and 5B).

Furfural is the smallest molecular weight compound studied and is a desirable volatile compound that provides an almond or butterscotch aroma (Boidron et al. 1988). It is formed from Maillard reactions of the pentose sugars found in the hemicellulose, which comprises 20 to 35% of the overall composition of oak (Ralph and Hatfield 1991, Nonier et al. 2006, Collins et al. 2015, Farrell et al. 2015, Le Floch et al. 2015, González-Centeno et al. 2016, Pollon et al. 2023). Hemicellulose is well established as the most thermally sensitive macropolymer in wood compared to lignin and cellulose, resulting in a degradation range of 130 to 345°C (complete degradation) (Shimizu et al. 1972, Ralph and Hatfield 1991, Nonier et al. 2006, Collins et al. 2015, Farrell et al. 2015, Le Floch et al. 2015, González-Centeno et al. 2016, Pollon et al. 2023). Due to the physical characteristics of the starting material, the toasting time and temperature will affect not only the concentration of furanic compounds, but also the kinetics of their extraction (González-Centeno et al. 2016).

As indicated above, furfural had an increase in extraction rate with increasing EG:CG ratio, consistent with the other compounds studied, as would be expected (Record 1911, Ferreira et al. 2006, Crawford et al. 2010, Rodríguez-Rodríguez and Gómez-Plaza 2011, Canas et al. 2013, Tao et al. 2014, Pollon et al. 2023). However, a unique finding was that differences in final furfural concentrations at day 120 were insignificant with all treatments (Figure 5C and 5D). When comparing the modeled ceg and ccg values (5.92 × 103 μ/L and 7.86 × 103 μ/L, respectively), the expected contributions to the final concentration were quite similar, with the difference in concentration between them being less than double. Given the low temperature of formation (low thermoreactivity), the consistent temperature applied with convection toasting, and the moderate temperature applied (avoiding degradation), furfural production appears to have been consistent through the entire piece of the wood. In turn, a barrel and an oak alternative of this type would be expected to reach similar final equilibrium concentrations.

Unlike furfural, 5-methylfurfural is primarily produced from the more crystalline hexose units, which are more thermally stable due to a higher degree of polymerization (Dupleix et al. 2013, Farrell et al. 2015). Furthermore, the boiling points of furfural and 5-methylfurfural may have also influenced their final formation concentration (Farrell et al. 2015). Likely due to the chemical properties of structural components, the extraction kinetics of 5-methylfurfural appeared to behave similarly to those of the lignin breakdown products discussed below.

Guaiacol is a breakdown product of lignin, which is generally associated with smoky/medicinal aroma (Boidron et al. 1988, Ralph and Hatfield 1991, Nonier et al. 2006, Collins et al. 2015, Farrell et al. 2015, Le Floch et al. 2015, González-Centeno et al. 2016, Pollon et al. 2023). Lignin makes up ~16 to 25% of the major structural components in hardwoods (Le Floch et al. 2015) and is comprised of three primary subunits: dimethoxylated (syringyl, S), monomethoxylated (guaiacyl, G), and non-methoxylated (p-hydroxyphenyl, H) phenylpropanoid units. This results in a variety of thermal degradation products. Due to a higher degree of polymerization, the range of possible subunits results in a broad temperature window where thermal degradation occurs (Record 1911, Nonier et al. 2006, Brebu and Vasile 2010, Le Floch et al. 2015).

The concentrations of the EG:CG ratios 0.03 and 0.48 for guaiacol reached an asymptotic maximum of 100 and 201 μ/L, respectively, at 120 days (Supplemental Table 1). Overall, this was a 101% increase in concentration compared to the 45% increase in total SA, the 1590% change in the EG:CG ratio, the 1600% increase in EG SA, and an insignificant change in CG SA. In turn, the 101% increase in concentration cannot simply be attributed to any of the SA contributions or to total SA alone. The total SA pseudo-first-order model (see Results above) failed to fit the five EG:CG extraction time courses. Additionally, the 1600% increase in EG alone cannot explain a 101% increase in concentration (the largest segment versus the small segments), and CG does not significantly change at all. Instead, a more complex extraction understanding was needed to characterize the simultaneous contributions from EG and CG, which were captured by the two-dimensional pseudo-first-order kinetic model. The estimated equilibrium concentration was calculated as 3.29 × 102 μg/L from the EG and 9.12 × 101 μg/L from the CG. The increase in the asymptotic maximum concentration reflects the increase in the EG to CG.

As discussed earlier, previous studies have evaluated a one-dimensional model for oak extraction (Pérez-Coello et al. 1997, Kadim and Mannheim 1999, Ferreira et al. 2006, Canas et al. 2013, Tao et al. 2014, García-Estévez et al. 2015, Li and Duan 2019, Pollon et al. 2023), which assumes the same extraction kinetics throughout the format. When using traditional barrel formats, an assessment like this could be representative due to the negligible exposure to the EG. However, the model’s fit is less suitable for alternative formats. Due to the porous nature of oak wood, liquids can be absorbed into the cell walls and macro voids. Driven by surface tension, capillary action, and additional forces, liquid is drawn along the vessels and fibers into the pits, similar to how sap is moved upward in a living tree (del Alamo-Sanza and Nevares 2018, Fredriksson et al. 2023). A visual inspection of an alternative product revealed differences in pore sizes between the two grain types (Figure 1), and a microscopy image provided scale to the structural differences between the exposed grain types (Figure 3). Exposure to EG allows direct contact with the transverse faces of vessels (up to 500 μm in early wood), tracheids, fiber lumina, and rays, enabling faster capillary action (Glass and Zelinka 2010, del Alamo-Sanza and Nevares 2018, Fredriksson et al. 2023). By contrast, CG is comprised of border pits and cell wall (1 to 5 μm), resulting in slower diffusion through rays or narrow pit membranes (Glass and Zelinka 2010).

This study critically evaluated the extraction of Q. robur/petraea, but Q. alba is also commonly used for these oak products. Structurally, Q. robur contains tyloses within the parenchyma and medullary rays, forming a barrier (Petterson 1984, Chatonnet and Dubourdieu 1998, Kim et al. 2024). Compared to European oak, American oak can contain significantly more tyloses, which increase the structural stability during the milling process. Due to this increased stability, staves from American oak can be sawn rather than split, as with those from European oak (splitting yields fewer staves and more waste). These specific milling techniques are implemented to ensure the oak vessels run parallel to the staves, preventing leakage. However, to produce oak alternatives, all oak types are sawn since the product is not intended to be liquid tight. During the toasting process, the parenchyma that makes up tyloses thermally degrades with hemicellulose ~160 to 220°C (Chatonnet and Dubourdieu 1998, Nonier et al. 2006, Yao et al. 2008, Le Floch et al. 2015, Burlacu et al. 2020, Kubovský et al. 2020, Eder et al. 2021, Hill et al. 2021). Preliminary data indicated that the extraction kinetics of volatile compounds are similar in both American and European varieties of Quercus spp. This may be attributed to milling and structural changes, as well as to the effect of sawing on the integrity of the tyloses. An in-depth comparison of the Quercus spp. is in progress.

Model predictions

Given the fitted two-dimensional kinetic constants and the opportunity to change the pre-exponential factors, SAeg/Vl and SAcg/Vl, any format of oak (variations in grain SA) can be matched in any volume of solution. This would include predicting the concentrations of volatile compounds in barrels, puncheons, and alternatives (e.g., staves, cubes, chips, dust). Unlike barrels, the dosage rate, i.e., quantity of oak to the volume of treated solution (e.g., wine, spirits), can vary in the oak alternatives. An example of this predictive capability is shown (Figure 7). In this example, the total SA of the barrel, stave, and cubes is held constant along with the treated volume (225 L), while the puncheon is adjusted proportionally for its total SA and near double volume to a barrel (500 L).

A two-panel line graph compares predicted guaiacol and furfural extraction over 400 days for cubes, staves, barrels, and a puncheon. The two line graphs plot concentration in micrograms per liter against time from 0 to 400 days for cube, stave, barrel, and puncheon treatments. Panel A shows guaiacol. The cube curve rises fastest and levels near 200 micrograms per liter. The stave curve levels near 115 micrograms per liter, while the barrel and puncheon curves level near 90 micrograms per liter, with the barrel rising slightly faster. Panel B shows furfural. The cube curve rises rapidly and levels near 5,900 micrograms per liter. The stave curve rises more gradually and approaches approximately 6,500 micrograms per liter. The barrel and puncheon curves rise more slowly, cross the cube curve after approximately 200 days, and reach approximately 7,400 and 6,900 micrograms per liter, respectively. All values are approximated.
Figure 7

Prediction and comparison of (A) guaiacol and (B) furfural extraction with 2.54-cm cubes, staves (0.9525 cm × 6.35 cm × 91.33 cm), 225-L barrels with a total of 1.96 × 104 cm2, and a 500-L puncheon with an internal surface area of 3.34 × 104 cm2 over 400 days.

In the resulting model predictions (Figure 7A), the final guaiacol concentrations increase with increasing EG. Barrel and puncheon at 400 days extraction had similar concentrations of ~90 μg/L with a contribution only from CG, while the staves (0.95 cm × 6.35 cm × 91.44 cm format) and the cubes (1 cm × 1 cm × 1 cm) were significantly higher with the increasing EG contribution. This would be expected, as ceq(eg) is greater than ceq(cg) for guaiacol. In turn, furfural (Figure 7B) at 400 days resulted in a similar concentration of ~6000 μg/L; however, the barrels and puncheons were yet to plateau in the extraction curve, an indication of a slower extraction rate of the CG alone. It is worth noting that at 400 days, the puncheon had not reached the same concentration as the barrel. This difference can be attributed to the CG SA-to-volume ratio. Based on this model and given enough time, both the barrel and puncheon would reach ceq(cg). This demonstrates how the model predicts ceq(cg) and ceq(eg) when one or the other has a greater value. Importantly, our experimental total SA-to-volume ratios (Table 1) encompass the range of a 225-L barrel. While scalable, additional trials at lower total SA-to-volume ratios should be conducted to properly interpret the dose rate (5 to 30% new oak equivalency). Additionally, the model presented here is not able to predict an infinitely small dosage rate (SA-to-volume ratio) or an infinitely high dosage rate; nonetheless, within the limits of our experimental conditions, the model should approximate most new oak barrel treatments for commercial wine.

With this knowledge of the model as described above, the percent extraction over time can be calculated and it serves as a valuable tool for winemaking, helping to determine diminishing returns during extraction and/or maximizing extraction extent. The final equilibrium concentration can be calculated at infinite time, c∞. This number then becomes the basis of normalizing the concentration to a unit scale (or percentage scale if easier). Both guaiacol and furfural extractions on this normalized scale show a more rapid extraction rate with increased SA (i.e., larger EG:CG ratio) (Figure 8). While it may be easy to compare the overall rate of extraction for furfural, where the 120-day concentration is not significantly different and ceq(cg) and ceq(eg) are similar, this is not the case for the remaining volatiles. Normalized concentrations over time for these compounds can enable these rate comparisons.

A two-panel line graph compares normalized guaiacol and furfural extraction for five end grain to cross grain ratios over 350 days. The two line graphs plot normalized concentration, C subscript t over C subscript e, from 0 to 1 against time from 0 to 350 days. Five curves represent end grain to cross grain ratios of 0.03, 0.06, 0.11, 0.24, and 0.48. Panel A shows guaiacol. All curves rise steeply and approach 1 by approximately 100 to 140 days. The extraction rate increases progressively from ratio 0.03 to 0.48, although the curves remain closely grouped. Panel B shows furfural. The curves are more widely separated, with ratio 0.48 approaching 1 before 50 days, followed successively by 0.24, 0.11, 0.06, and 0.03. The curve for ratio 0.03 approaches 1 near 280 days. All values are approximated.
Figure 8

Guaiacol (A) and furfural (B) raw extraction curves of all treatments normalized to the maximum extraction concentration. Overall, the segments with the larger end grain-to-cross grain ratio (EG:CG) reached equilibrium/saturation before the other treatments. The contribution of exposed grain types compared to the surface area varied depending on the compound.

Additionally, the normalized values of all quantified oak volatile compounds were plotted together and compared (Figure 9), despite up to multiple orders of magnitude in concentration (e.g., cis-oak lactone and furfural). The normalized extraction time courses show the relative extraction progress (i.e., which compounds extract the fastest or slowest) and the extent of extraction. Again, furfural stood out as unique in that it extracted either the fastest or the slowest, depending on the EG:CG ratio, compared with the other compounds. This can be explained by examining the rate constants with the EG:CG contribution. The rate constant for CG is 29.3 times slower than the EG constant, even though the final liquid concentration would be similar with enough time, given that ceq(cg) and ceq(eg) modeled concentrations are similar. Therefore, an increase in the EG:CG ratio changes the pre-exponential factor, which is multiplied by the significantly fast EG kinetic constant to increase the overall rate of furfural extraction significantly. This provides an understanding of how furfural has the slowest rate of extraction compared to the remaining compounds (on the normalized scale; Figure 9) in the smallest EG:CG ratio segments, and conversely, the highest extraction rate with the largest EG:CG ratio.

A two-panel line graph compares normalized extraction of seven volatile compounds from the longest and shortest oak segments. The two line graphs plot normalized concentration, C subscript t over C subscript e, from 0 to 1 against time from 0 to 350 days for 4-ethylguaiacol, 4-methylguaiacol, guaiacol, cis-oak lactone, 5-methylfurfural, trans-oak lactone, and furfural. Panel A represents the longest segments with an end grain to cross grain ratio of 0.03. The compounds approach 1 at different rates, with 4-ethylguaiacol rising fastest and furfural rising slowest. The other five curves occupy intermediate positions and converge near 1 between approximately 150 and 250 days. Panel B represents the shortest segments with a ratio of 0.48. All curves rise more rapidly than in panel A. Trans-oak lactone and 5-methylfurfural approach 1 first, followed by 4-ethylguaiacol, 4-methylguaiacol, guaiacol, cis-oak lactone, and furfural. Furfural remains the slowest and approaches 1 after approximately 250 days. All values are approximated.
Figure 9

Normalization of the modeled extraction curves for the (A) longest segments (end grain to cross grain [EG:CG] ratio of 0.03) and the (B) shortest segments (EG:CG ratio of 0.48) for all the volatile compounds. The order in which the compounds reach equilibrium varied based on the EG:CG ratio. This ratio determines the overall extraction rate, while the chemical-physical properties of the specific compound determine its maximum extraction concentration.

Comparing compound extraction on a percentage basis over time also provides a winemaking tool for determining the extent of extraction for all compounds. This allows for decision-making based on the acceptable utilization of the alternative oak (measured by percent extracted, irrespective of time) or a limited or desired aging time on the oak (measured by time, regardless of percent extracted). Ultimately, the model could target desired final concentrations of each compound (or recipe concentrations) by adjusting EG SA, CG SA, and time. The EG and CG can be further constrained to known oak alternative product specifications (e.g., staves, planks, cubes, beans, spheres). To produce an accurate model for a specific system, individual experimentation is required. The specific kinetic constants will vary depending on the oak product and wine matrix. Model systems present a way to isolate oak kinetics from additional unknown variables present in wine. The concepts discussed are foundational to oak extraction. As additional variables are added to the matrix, each exposed grain type must be treated separately.

Future work

The experimental trials presented in this research mimic general wine-industry practices with alternative oak products in which the oak is added to a wine vessel (i.e., stainless steel tank), the wine is added, and the oak remains submerged with minimal or no stirring during the extraction/maturation process. Stirring the wine, or in this research, the model wine, may provide additional insights into the fundamentals of oak extraction using a two-dimensional pseudo-first-order kinetic model.

The extraction process of natural products has been described as having four steps: 1) the solvent enters the porous area, 2) the solvent dissolves the solutes, 3) solutes diffuse back to the surface of the particle or boundary layer, and 4) solutes diffuse from the boundary layer to the bulk solvent (Gertenbach 2016). Steps 3 and 4 are the diffusion processes defined by Fick’s Second Law. Gertenbach (2016) states that the internal diffusion (step 3) is generally the rate-limiting step for botanicals. It could be assumed that this is the case for our research and for general winemaking practices, however, others have elucidated both the internal and external diffusion coefficients by evaluating extraction from unstirred and stirred solutions (Espinoza-Pérez et al. 2007, Rodríguez-Jimenes et al. 2012, Setford et al. 2018). Stirred kinetics would minimize or eliminate external diffusion by establishing a mass-transfer condition in the bulk solution, thereby enabling the determination of the internal diffusion coefficient.

As mentioned here and above, the extraction of the oak volatile compounds appears to be diffusion controlled. This Wilke-Chang equation for diffusivity (Equation 3) comprises a temperature term and a viscosity term, therefore, extraction rates are expected to vary with alcohol concentration (e.g., beer, wine, spirits) and/or cellaring and fermenting temperatures.

Conclusion

The solvation of volatile compounds from oak alternatives into model wine is a two-dimensional extraction process with varying rate and equilibrium concentration contributions from both the EG and CG of the oak. These contributions can be modeled by varying the SA of one of the grain types—in this study, the EG. This model allows predicting compound concentration based on SA and solution volume. Furthermore, the model provides an equilibrium concentration that can then be used to normalize the extracted compounds to provide winemaking insights into diminishing returns of the extraction and/or maximizing the extent of extraction. The model developed allows EG:CG ratios to vary depending on the format and dose. In all, this type of kinetic analysis provides a winemaking tool to bridge the gap between traditional barrel maturation and more sustainable wine maturation with oak alternatives.

CRediT Authorship Contributions

MA, RC: Formal Analysis, Visualization, Writing - Original Draft; MA, TC: Data Curation; MA, RC, TC: Investigation; MA, RC, RB, TC: Methodology, Writing - Review & Editing; MA, RC, RB: Validation; TC: Conceptualization, Funding Acquisition, Project Administration, Supervision

Supplemental Data

The following supplemental materials are available for this article in the Supplemental tab above:

Supplemental Table 1 Average concentration of measured volatile phenols at each time point sampled from all treatments. EG:CG, end grain-to-cross grain ratio.

Supplemental Table 2 Standard error of the average concentration of measured volatile phenols at each time point sampled from all treatments. EG:CG, end grain-to-cross grain ratio.

Data Availability

The data underlying this study are available on request from the corresponding author

Footnotes

  • The authors thank Cynthia Coleman for her support and professional insight. This research was funded in part by Scott Laboratories, Petaluma, CA

  • Aragon M, Coleman R, Boulton RB and Collins TS. 2026. Influence of grain type and surface area on extraction of French oak in model wine: A two-dimensional kinetic analysis. Am J Enol Vitic 77:0770021. DOI: 10.5344/ajev.2026.25052

  • By downloading and/or receiving this article, you agree to the Disclaimer of Warranties and Liability. If you do not agree to the Disclaimers, do not download and/or accept this article.

  • Received December 2025.
  • Accepted April 2026.
  • Published online September 2026

This is an open access article distributed under the CC BY 4.0 license.

References

  1. ↵
    1. Bird RB,
    2. Stewart WE,
    3. Lightfoot EN and
    4. Klingenberg DJ.
    2015. Introductory Transport Phenomena. Wiley, NJ.
  2. ↵
    1. Boidron J-N,
    2. Chatonnet P and
    3. Pons M.
    1988. Influence du bois sur certaines substances odorantes des vins. OENO One 22:275-294. DOI: 10.20870/oeno-one.1988.22.4.1263
    OpenUrlCrossRef
  3. ↵
    1. Brebu M and
    2. Vasile C.
    2010. Thermal degradation of lignin – A review. Cell Chem Technol 44:353-363.
    OpenUrl
  4. ↵
    1. Burlacu E,
    2. Nisca A and
    3. Tanase C.
    2020. A comprehensive review of phytochemistry and biological activities of Quercus species. Forests 11:904. DOI: 10.3390/f11090904
    OpenUrlCrossRef
  5. ↵
    1. Canas S,
    2. Caldeira I and
    3. Belchior AP.
    2013. Extraction/oxidation kinetics of low molecular weight compounds in wine brandy resulting from different ageing technologies. Food Chem 138:2460-2467. DOI: 10.1016/j.foodchem.2012.12.018
    OpenUrlCrossRef
  6. ↵
    1. Caputo T.
    2006. Winemaking with oak alternatives. Wines and Vines July. https://winebusinessanalytics.com/sections/printout_article.cfm?article=feature&content=49681
  7. ↵
    1. Chatonnet P and
    2. Dubourdieu D.
    1998. Comparative study of the characteristics of American White Oak (Quercus alba) and European Oak (Quercus petraea and Q. robur) for production of barrels used in barrel aging of wines. Am J Enol Vitic 49:79-85. DOI: 10.5344/ajev.1998.49.1.79
    OpenUrlAbstract/FREE Full Text
  8. ↵
    1. Coleman R,
    2. Nelson J and
    3. Boulton R.
    2024. Methods for parameter estimation in wine fermentation models. Fermentation 10:386. DOI: 10.3390/fermentation10080386
    OpenUrlCrossRef
  9. ↵
    1. Collins TS,
    2. Miles JL,
    3. Boulton RB and
    4. Ebeler SE.
    2015. Targeted volatile composition of oak wood samples taken during toasting at a commercial cooperage. Tetrahedron 71:2971-2982. DOI: 10.1016/j.tet.2015.02.079
    OpenUrlCrossRef
  10. ↵
    1. Côté Jr WA.
    1963. Structural factors affecting the permeability of wood. J Polym Sci C 2:231-242. DOI: 10.1002/polc.5070020122
    OpenUrlCrossRef
  11. ↵
    1. Crawford K,
    2. Benton A,
    3. Grigg A and
    4. Plumb D.
    2010. Oak inserts: Grain orientation effect on extraction of oak flavor compounds. Pract Wine Vine 31:19-26.
    OpenUrl
  12. ↵
    1. De Blaere R,
    2. Lievens K,
    3. Van Hassel D,
    4. Deklerck V,
    5. De Mil T,
    6. Hubau W et al
    . 2023. SmartWoodID—an image collection of large end grain surfaces to support wood identification systems. Database 2023:baad034. DOI: 10.1093/database/baad034
    OpenUrlCrossRef
  13. ↵
    1. del Alamo-Sanza M and
    2. Nevares I.
    2018. Oak wine barrel as an active vessel: A critical review of past and current knowledge. Crit Rev Food Sci Nutr 58:2711-2726. DOI: 10.1080/10408398.2017.1330250
    OpenUrlCrossRef
  14. ↵
    1. Delgado-González MJ,
    2. García-Moreno MV,
    3. Sánchez-Guillén MM,
    4. García-Barroso C and
    5. Guillén-Sánchez DA.
    2021. Colour evolution kinetics study of spirits in their ageing process in wood casks. Food Control 119:107468. DOI: 10.1016/j.foodcont.2020.107468
    OpenUrlCrossRef
  15. ↵
    1. Dupleix A,
    2. De Sousa Meneses D,
    3. Hughes M and
    4. Marchal R.
    2013. Mid-infrared absorption properties of green wood. Wood Sci Technol 47:1231-1241. DOI: 10.1007/s00226-013-0572-5
    OpenUrlCrossRef
  16. ↵
    1. Eder M,
    2. Schäffner W,
    3. Burgert I and
    4. Fratzl P.
    2021. Wood and the activity of dead tissue. Adv Mater 33:2001412. DOI: 10.1002/adma.202001412
    OpenUrlCrossRef
  17. ↵
    1. Espinoza-Pérez JD,
    2. Vargas A,
    3. Robles-Olvera VJ,
    4. Rodríguez-Jimenes GC and
    5. García-Alvarado MA.
    2007. Mathematical modeling of caffeine kinetic during solid–liquid extraction of coffee beans. J Food Eng 81:72-78. DOI: 10.1016/j.jfoodeng.2006.10.011
    OpenUrlCrossRef
  18. ↵
    1. Farrell RR,
    2. Wellinger M,
    3. Gloess AN,
    4. Nichols DS,
    5. Breadmore MC,
    6. Shellie RA et al
    . 2015. Real-time mass spectrometry monitoring of oak wood toasting: Elucidating aroma development relevant to oak-aged wine quality. Sci Rep 5:17334. DOI: 10.1038/srep17334
    OpenUrlCrossRef
  19. ↵
    1. Ferreira V,
    2. Jarauta I and
    3. Cacho J.
    2006. Physicochemical model to interpret the kinetics of aroma extraction during wine aging in wood. Model limitations suggest the necessary existence of biochemical processes. J Agric Food Chem 54:3047-3054. DOI: 10.1021/jf052534v
    OpenUrlCrossRefPubMed
  20. ↵
    1. Flor FJ,
    2. Leiva FJ,
    3. García J,
    4. Martínez E,
    5. Jiménez E and
    6. Blanco J.
    2017. Environmental impact of oak barrels production in Qualified Designation of Origin of Rioja. J Clean Prod 167:208-217. DOI: 10.1016/j.jclepro.2017.08.190.
    OpenUrlCrossRef
  21. ↵
    1. Fredriksson M,
    2. Rüggeberg M,
    3. Nord-Larsen T,
    4. Beck G and
    5. Thybring EE.
    2023. Water sorption in wood cell walls–data exploration of the influential physicochemical characteristics. Cellulose 30:1857-1871. DOI: 10.1007/s10570-022-04973-0
    OpenUrlCrossRef
  22. ↵
    1. García-Alcaraz JL,
    2. Flor Montalvo FJ,
    3. Martínez Cámara E,
    4. Pérez de la Parte MM,
    5. Jiménez-Macías E and
    6. Blanco-Fernández J.
    2020. Economic-environmental impact analysis of alternative systems for red wine ageing in re-used barrels. J Clean Prod 244:118783. DOI: 10.1016/j.jclepro.2019.118783
    OpenUrlCrossRef
  23. ↵
    1. García-Estévez I,
    2. Alcalde-Eon C,
    3. Le Grottaglie L,
    4. Rivas-Gonzalo JC and
    5. Escribano-Bailón MT.
    2015. Understanding the ellagitannin extraction process from oak wood. Tetrahedron 71:3089-3094. DOI: 10.1016/j.tet.2014.10.047
    OpenUrlCrossRef
  24. ↵
    1. Gertenbach D.
    2016. Solid-liquid extraction technologies for manufacturing nutraceuticals. In Functional Foods: Biochemical and Processing Aspects, Volume 2. Shi J et al. (eds.). CRC Press, Boca Raton, FL. DOI: 10.1201/9781420012873
    OpenUrlCrossRef
  25. ↵
    1. Glass SV and
    2. Zelinka SL.
    2010. Moisture relations and physical properties of wood. In Wood Handbook: Wood as an Engineering Material. Gen Tech Rep FPL-GTR-190. pp. 4-10-4-19. U.S. Department of Agriculture, Forest Service, Forest Products Laboratory, Madison, WI. DOI: 10.2737/FPL-GTR-190
    OpenUrlCrossRef
  26. ↵
    1. González-Centeno MR,
    2. Chira K and
    3. Teissedre P-L.
    2016. Ellagitannin content, volatile composition and sensory profile of wines from different countries matured in oak barrels subjected to different toasting methods. Food Chem 210:500-511. DOI: 10.1016/j.foodchem.2016.04.139
    OpenUrlCrossRef
  27. ↵
    1. Hill C,
    2. Altgen M and
    3. Rautkari L.
    2021. Thermal modification of wood—A review: Chemical changes and hygroscopicity. J Mater Sci 56:6581-6614. DOI: 10.1007/s10853-020-05722-z
    OpenUrlCrossRef
  28. ↵
    1. Hines N.
    2015. Oak barrels vs. oak alternatives. Grape Collective. https://grapecollective.com/articles/oak-barrels-vs-oak-alternatives
  29. ↵
    1. Ho Y-S.
    2004. Citation review of Lagergren kinetic rate equation on adsorption reactions. Scientometrics 59:171-177. DOI: 10.1023/B:SCIE.0000013305.99473.cf
    OpenUrlCrossRef
  30. ↵
    1. Hobbi P,
    2. Okoro OV,
    3. Delporte C,
    4. Alimoradi H,
    5. Podstawczyk D,
    6. Nie L et al
    . 2021. Kinetic modelling of the solid-liquid extraction process of polyphenolic compounds from apple pomace: Influence of solvent composition and temperature. Bioresour Bioprocess 8:114. DOI: 10.1186/s40643-021-00465-4
    OpenUrlCrossRef
  31. ↵
    1. Kadim D and
    2. Mannheim CH.
    1999. Kinetics of phenolic extraction during aging of model wine solution and white wine in oak barrels. Am J Enol Vitic 50:33-39. DOI: 10.5344/ajev.1999.50.1.33
    OpenUrlAbstract/FREE Full Text
  32. ↵
    1. Kim D,
    2. Gollihue J,
    3. Poovathingal SJ and
    4. DeBolt S.
    2024. Detailed three-dimensional analyses of tyloses in oak used for bourbon and wine barrels through X-ray computed tomography. Sci Rep 14:17044. DOI: 10.1038/s41598-024-67298-x.
    OpenUrlCrossRef
  33. ↵
    1. Kubovský I,
    2. Kačíková D and
    3. Kačík F.
    2020. Structural changes of oak wood main components caused by thermal modification. Polymers 12:485. DOI: 10.3390/polym12020485
    OpenUrlCrossRef
  34. ↵
    1. Kvålseth TO.
    1985. Cautionary note about R2. Am Stat 39:279-285. DOI: 10.2307/2683704
    OpenUrlCrossRef
  35. ↵
    1. Lagergren S.
    1898. Zur theorie der sogenannten adsorption gelöster stoffe. Bihang Till K Sv Vet-Akad Handl Band 24:1-39.
    OpenUrl
  36. ↵
    1. Le Floch A,
    2. Jourdes M and
    3. Teissedre P-L.
    2015. Polysaccharides and lignin from oak wood used in cooperage: Composition, interest, assays: A review. Carbohydr Res 417:94-102. DOI: 10.1016/j.carres.2015.07.003
    OpenUrlCrossRef
  37. ↵
    1. Li S-Y and
    2. Duan C-Q.
    2019. Astringency, bitterness and color changes in dry red wines before and during oak barrel aging: An updated phenolic perspective review. Crit Rev Food Sci Nutr 59:1840-1867. DOI: 10.1080/10408398.2018.1431762
    OpenUrlCrossRef
  38. ↵
    1. Marr B.
    2024. Oak barrel alternatives: A deep dive into staves and oak chips. Bernard Marr's Wine Guide. https://bmwinecellar.com/oak-barrel-alternatives-a-deep-dive-into-staves-and-oak-chips/
  39. ↵
    1. Mullen KM,
    2. Ardia D,
    3. Gil DL,
    4. Windover D and
    5. Cline J.
    2011. DEoptim: an R package for global optimization by differential evolution. J Stat Softw 40:1-26. DOI: 10.18637/jss.v040.i06
    OpenUrlCrossRefPubMed
  40. ↵
    1. Nonier MF,
    2. Vivas N,
    3. Vivas de Gaulejac N,
    4. Absalon C,
    5. Soulié P and
    6. Fouquet E.
    2006. Pyrolysis–gas chromatography/mass spectrometry of Quercus sp. wood. J Anal Appl Pyrolysis 75:181-193. DOI: 10.1016/j.jaap.2005.05.006
    OpenUrlCrossRef
  41. ↵
    1. Pérez-Coello MS,
    2. Sanz J and
    3. Cabezudo MD.
    1997. Analysis of volatile components of oak wood by solvent extraction and direct thermal desorption-gas chromatography-mass spectrometry. J Chromatogr A 778:427-434. DOI: 10.1016/S0021-9673(97)00286-0
    OpenUrlCrossRef
  42. ↵
    1. Pérez-Coello MS,
    2. Sanz J and
    3. Cabezudo MD.
    1998. Gas chromatographic-mass spectrometric analysis of volatile compounds in oak wood used for ageing of wines and spirits. Chromatographia 47:427-432. DOI: 10.1007/BF02466474
    OpenUrlCrossRef
  43. ↵
    1. Petterson RC.
    1984. The chemical composition of wood. In The Chemistry of Solid Wood. Rowell R (ed.), pp. 57-126. American Chemical Society, Washington, DC. DOI: 10.1021/ba-1984-0207.ch002
    OpenUrlCrossRef
  44. ↵
    1. Pollon M,
    2. Río Segade S,
    3. Giacosa S,
    4. Botto R,
    5. Montanini C and
    6. Rolle L.
    2023. Volatile compound release from oak chips in model wine media: Combined influence of toasting degree, size, time of contact, and ethanol content. J Agric Food Chem 71:13440-13450. DOI: 10.1021/acs.jafc.3c03469
    OpenUrlCrossRef
  45. ↵
    1. Psarra C,
    2. Gortzi O and
    3. Makris DP.
    2015. Kinetics of polyphenol extraction from wood chips in wine model solutions: Effect of chip amount and botanical species. J Inst Brew 121:207-212. DOI: 10.1002/jib.212
    OpenUrlCrossRef
  46. ↵
    R Core Team. 2023. R: A Language and Environment for Statistical Computing. R Foundation for Statistical Computing, Vienna, Austria. https://www.R-project.org/
  47. ↵
    1. Ralph J and
    2. Hatfield RD.
    1991. Pyrolysis-GC-MS characterization of forage materials. J Agric Food Chem 39:1426-1437. DOI: 10.1021/jf00008a014
    OpenUrlCrossRef
  48. ↵
    1. Record SJ.
    1911. Grain and texture in wood. J Forest 9:22-25. DOI: doi.org/10.1093/jof/9.1.22
    OpenUrl
  49. ↵
    1. Rodríguez-Jimenes GC,
    2. Vargas-Garcia A,
    3. Espinoza-Pérez DJ,
    4. Salgado-Cervantes MA,
    5. Robles-Olvera VJ and
    6. García-Alvarado MA.
    2012. Mass transfer during vanilla pods solid liquid extraction: Effect of extraction method. Food Bioprocess Technol 6:2640-2650. DOI: 10.1007/s11947-012-0975-6
    OpenUrlCrossRef
  50. ↵
    1. Rodríguez-Rodríguez P and
    2. Gómez-Plaza E.
    2011. Differences in the extraction of volatile compounds from oak chips in wine and model solutions. Am J Enol Vitic 62:127-132. DOI: 10.5344/ajev.2010.10045
    OpenUrlAbstract/FREE Full Text
  51. ↵
    1. Setford PC,
    2. Jeffery DW,
    3. Grbin PR and
    4. Muhlack RA.
    2018. Modelling the mass transfer process of malvidin-3-glucoside during simulated extraction from fresh grape solids under wine-like conditions. Molecules 23:2159. DOI: 10.3390/molecules23092159
    OpenUrlCrossRef
  52. ↵
    1. Shimizu K,
    2. Teratani F,
    3. Hashi M and
    4. Miyazaki K.
    1972. Effect of the thermal treatment on wood hemicelluloses. Studies on the thermal analysis of arabinogalactan, and O-acetyl- and deacetylated-galactogluco-mannans. Moku Gakkai 18:79-84.
    OpenUrl
  53. ↵
    1. Sikora A,
    2. Kačík F,
    3. Gaff M,
    4. Vondrová V,
    5. Bubeníková T and
    6. Kubovský I.
    2018. Impact of thermal modification on color and chemical changes of spruce and oak wood. J Wood Sci 64:406-416. DOI: 10.1007/s10086-018-1721-0
    OpenUrlCrossRef
  54. ↵
    1. Tao Y,
    2. Zhang Z and
    3. Sun D-W.
    2014. Experimental and modeling studies of ultrasound-assisted release of phenolics from oak chips into model wine. Ultrason Sonochem 21:1839-1848. DOI: 10.1016/j.ultsonch.2014.03.016
    OpenUrlCrossRef
  55. ↵
    1. Tseng R-L,
    2. Wu F-C and
    3. Juang R-S.
    2010. Characteristics and applications of the Lagergren’s first-order equation for adsorption kinetics. J Taiwan Inst Chem Eng 41:661-669. DOI: 10.1016/j.jtice.2010.01.014
    OpenUrlCrossRef
  56. ↵
    1. Wilke CR and
    2. Chang P.
    1955. Correlation of diffusion coefficients in dilute solutions. AIChE J 1:264-270. DOI: 10.1002/aic.690010222
    OpenUrlCrossRef
  57. ↵
    1. Yao F,
    2. Wu Q,
    3. Lei Y,
    4. Guo W and
    5. Xu Y.
    2008. Thermal decomposition kinetics of natural fibers: Activation energy with dynamic thermogravimetric analysis. Polym Degrad Stab 93:90-98. DOI: 10.1016/j.polymdegradstab.2007.10.012.
    OpenUrlCrossRef
  58. ↵
    1. Yuh-Shan H.
    2004. Citation review of Lagergren kinetic rate equation on adsorption reactions. Scientometrics 59:171-177. DOI: 10.1023/B:SCIE.0000013305.99473.cf
    OpenUrlCrossRef
Back to top
Print
View full PDF
Email Article
Citation Tools

Open Access
Influence of Grain Type and Surface Area on Extraction of French Oak in Model Wine: A Two-Dimensional Kinetic Analysis
Mackenzie Aragon, Robert Coleman, Roger B. Boulton, Thomas S. Collins
Am J Enol Vitic.  2026  77: 0770021  ; DOI: 10.5344/ajev.2026.25052
del.icio.us logo Twitter logo Facebook logo Mendeley logo
Save to my folders

Jump to section