PRINTABLE VERSION (Acrobat PDF, 1MB, 19 pages)
Some 20 years ago, Vincent Bouchard was instrumental in forming a joint marketing venture, then known as Les Tonnelleries des Bourgogne, with a group of coopers from Burgundy and Cognac. As an enologist, Vincent had been eager to contribute to advances in wine making and to foster a greater understanding of barrel usage. In those early years, he advocated experimentation, coercing wineries to look beyond the heavy-handed influence of Limousin oak. Vincent's suggestions on use of tight grain oak from forests of Allier and Vosges, his insight into different toast levels, and the general idea of matching a cooper's style to a wine style were revolutionary at the time.
As time has progressed, so has the sophistication in use of oak barrels for wine aging; and the Bouchard Cooperages team has gained a great deal of experience in matching wine with barrel. Their expertise and their sensory impressions of the different fabrication combinations and permutations are valuable resources. However, in order to verify these impressions and reinforce their recommendations, Bouchard Cooperages (in partnership with the Tonnelleries) have commissioned a novel analytical survey of barrels. The primary goal has been to determine whether "cooper style" can be graphically demonstrated; but of equal interest, has been the challenge of developing a non-destructive procedure for barrel evaluation.
Thanks to ETS Laboratories in St. Helena, California, a tool is now available to the wine production industry for the analysis of oak aroma. (More info) Such analytical techniques have been utilized by researchers to study samples of oak aged wine or extracts prepared from oak particles removed from stave wood. In our study, a sampling method has been developed that employs a water "wash" of the inside of a barrel. This allows one to sample from the entire barrel interior (staves and head) in a manner that leaves the barrel unaltered and intact, thus still available for use in wine aging. It is thought that such a procedure can be employed in a quality control or assurance program, either at the cooperage or within a wine cellar.
Over 120 barrels were selected from the normal production runs of the coopers represented by Bouchard Cooperages, including Tonnelleries BILLON, CADUS, DAMY and VICARD. The barrels were stored for 3 - 4 weeks in a Napa warehouse to equilibrate prior to testing. Besides comparing coopers with each other, certain fabrication variables were also compared: forest origins (Nevers, Allier, Hungarian); air-drying (2yr vs. 3yr), toast level (medium vs. medium-plus) and shape or style (Burgundy export, Bordeaux export, and Bordeaux ferre). Not all variables were examined for all Coopers; but four replicate barrels were included in each test grouping. For all Nevers oak barrels, individual water samples were taken and analyzed; and an arithmetic average was obtained for each test group of four. For the Allier oak barrels, samples from the four barrels were proportionally pooled into a composite sample for analysis.
Basic Sampling Procedure:
Barrels are inspected and rinsed with chlorine-free water (stored in a portable tank at ambient temperature of approximately 60°F) to remove chips and dust particles. A small prescribed volume of fresh water is introduced. The barrel is set on each head, then rolled and rocked on its side in a manner that allows water contact of all staves. The contact time is carefully controlled. The test water is collected and submitted for immediate analysis.
Analysis by ETS Laboratories: (Reference: ETS Technical Bulletin dated 25 April 2001.)
ETS Laboratories utilizes GC/MS (gas chromatography / mass spectrometry) to test for presence of chemical compounds commonly associated with oak aging of wine. These measured compounds serve as markers for certain aromatic attributes. Specifically, these compounds (and associated sensory characteristics) are:
- cis and trans Oak lactones: the two main aroma constituents of raw oak (not "green") often described as fresh oak or coconut. cis Oak lactone is significantly more aromatic.
- Vanillin: A phenolic aldehyde resulting from lignin degradation. It is the main aroma compound of natural vanilla.
- Eugenol: Volatile phenol produced in the oxidative breakdown of lignin during air drying; possessing spicy, clove aromas.
- Guaiacol and 4-Methylguaiacol: volatile phenols, resulting from further thermal degradation of phenolic aldehydes, with smoky aromas. Both compounds have char aromas, but 4-methylguaiacol has spicy characters as well.
- Furfural and 5-Methylfurfural: produced by the degradation of carbohydrates by heat during barrel toasting. These compounds possess aromas of butterscotch, light caramel, and faint almond.
The degradation and formation of the compounds are influenced by both temperature and time (duration); and they vary in their susceptibility to modification. Furthermore, the appearance and disappearance of each compound tends to follow a bell-shaped curve: increasing in quantity at certain heat (toast) levels, and then decreasing at even higher heat levels. This rate of production or destruction varies with each compound. Thus, barrels with vastly different aromatic intensities or qualities can be produced by varying firing time and/or temperature.
Visual inspection: The coloration observed on the barrel interiors was mostly within expectations. Medium toast barrels could be described as light to medium brown with no blisters; while medium-plus toast barrels were a dark chestnut color with occasional blistering. The Cadus slow-blue, medium toast, barrels were exceptional in that they appeared lightest in coloration. No formal measurement of grain tightness was performed, but it was noted that barrels of Allier origin were of tighter grain than those from Nevers; and Hungarian oak was tightest grain of all.
There are a variety of analytical comparisons to be made, and a number of ways to express results. For much of the study, we have chosen to display this information in spider graphs or radar plots, based upon reports used by ETS. (Disclaimer: We are grateful for the analytical work performed by ETS and their help in experimental design and education. Comparisons and interpretations of the data are those of Bouchard Cooperages.)
Please Note: The links below will open separate browser windows for comparing each illustration. You may open and close them as you wish.
Figure 1 represents our best attempt at summarizing the greatest amount of information. This spider graph compares analytical data obtained from barrel samples of each cooper for their most typical barrel style. The data are expressed as a percentage of the average obtained for all analyses for a particular compound. For Billon, Damy, and Cadus (slow blue firing) the plotted lines represent averages of data collected from all Burgundy export barrels at medium and medium plus toast. For Vicard French and Eastern European, Bordeaux export barrels were used, at medium and medium-plus toast. Nevers and Allier forest origins were included in the summary for all coopers except Vicard Eastern European, who specializes in Hungarian Oak.
At first glance, there appears to be distinct analytical differences among the coopers, as evident from the resulting line patterns in Figure 1. Are these differences significant? Figure 2 organizes some of the information into a bar chart. Here we are looking at a summary of information obtained only from Nevers oak extracts from the coopers. The data is further segregated by toast level. Because four individual barrels from each test group were analyzed, the degree of variation (expressed as standard error bars) can be indicated. The great degree of variability in the 4-methylguaiacol and guaiacol content in the Billon barrels makes it difficult to discriminate between M and M+ toast. However, there is no denying that (in this population of barrels) a segregation of coopers can be made based on these compounds. Further, the trend is for decreased amounts of these two smoky or spicy compounds, along with eugenol, to appear at the M+ level versus the M toast level. The Cadus "slow blue" barrels appear to be quite different from the others, especially with respect to the oak-lactones and vanillin. Indeed, for many of the compounds, there does appear to be significant cooper to cooper difference, and/or differences that are dependant upon toast level.
Figures 3 through 5 represent spider graphs specific to each cooper, additionally separated by forest origin and duration of air-drying. The plotted lines compare toast level and/or barrel shape within each graph. In these more detailed views, different patterns again emerge that suggest cooper style. Our hope is that such patterns can be directly linked to specific firing techniques.
Billon: For the selection of barrels used in this study, Billon barrels appear to be clearly different from others in two categories of aroma compounds. Across the board, they measure higher in smoky/spicy/char markers of 4-methylguaiacol and guaiacol, and lower in the oak lactones. Of further interest, is that higher toast level does not translate into greater "smoke" (Figure 3). This is true for other coopers as well. The marker for clove/spice, eugenol, appears to be slightly higher than overall average in Billon barrels, especially at M+ toast.
Cadus: At the opposite end of the spectrum, when viewing the Cadus average plot line on Figure 1, one is struck by the relative lack of smoke/char, and the abundance of oak lactone and vanillin. Closer examination of the detailed graphs (Figure 5) suggests that the summary plot line of Figure 1 is overly influenced by the Cadus "slow blue" style of toasting, and oak lactone and vanillin levels are elevated for that particular style only. It is noteworthy that, when visually examined, the interiors of these barrels were found to be the least colored of the test group.
Damy: A "Damy" style does not appear to emerge as readily from the charts. Indeed, the Damy barrel test groups tend to be clustered closest to the population average for most measured compounds. The Damy sample groups appear to be significantly higher than Cadus in smoke/char markers, but still lower than the Billon samples. In some instances (Figure 4), when comparing toast levels, Damy M+ barrels may be slightly elevated in a compound associated with light caramel/butterscotch: furfural. But without further study and examination of variation within a larger sampling, it is difficult to determine the significance of such observations.
Vicard French: Compared to the average for all barrels, the Vicard test group is notably lower in guaiacol and 4-methylguaiacol, and lower in lactones and vanilla. This is particularly noticeable for M+ toast levels (Figure 6). The remarkable exception is the 3 yr Allier M toasted barrel group. The Vicard barrels differ from others in the furfural and 5-methylfurfural compounds as well. They are not significantly different in concentrations, but in pattern of development. With the three coopers from Burgundy (Figure 10), these markers for caramel/butterscotch increase as you move from M to M+ toast. The opposite is true for Vicard. This is further evidence of firing "style" distinction.
Vicard Eastern European: The shape of the spider graph pattern, with overall diminished levels of guaiacol, 4-methylguaiacol, lactone and vanillin, strongly hints at the association between Vicard French and Eastern European oak. In fact, they are alike in their techniques and equipment employed in barrel firing. Despite the geographic difference in wood source (French vs. Hungarian), the results suggest a relationship in style. The spider graph resulting from the Hungarian oak is notably similar to that of the Vicard Nevers graph pattern. (Figure 7)
HOW TO EXPLAIN COOPER-TO-COOPER DIFFERENCES
There are a number of sources of variation for the stave wood used in barrels, as well as variations in fabrication of the final product. In addition to the gross differences recognized between European and American white oak species, there have been studies demonstrating variation associated with forest origin within France. The two dominant species of European oak, Quercus robur (pendunculate) and Quercus petraea (sessile) can be found segregated and intermingled within forests of France. Pendunculate oak populates the forests of Limousin, while the center of France forest of Troncais is dominated by sessile oak, and the forest of Citeaux in Burgundy is mixed. Several studies have demonstrated that chemical composition (with respect to such compounds as oak lactones and ellagitannins) can vary according to species. But, significant variation from tree to tree within each forest has also been reported. The "raw material" available for modification during toasting process will then vary with source of wood. Indeed, anyone who has looked inside a toasted barrel can often find staves that have darkened to a different degree than neighboring staves; demonstrating a diverse susceptibility to heat.
Thus, differences between coopers may be partly due to the specific origin of trees used for staves within a section of forest. In our study, a comparison between Nevers and Allier forest origins suggests that barrels fabricated from Allier have increased concentration of oak lactones (Figure 8). This was especially true for cooper Cadus. We have observed, on average, that Allier stave wood is of tighter grain than Nevers. One can speculate that the more open-grained, or porous, wood has a greater susceptibility to thermal influence; thus degradation of compounds such as lactones is more likely to occur.
Seasoning plays a role in modification of compounds found in stave wood. During the natural drying process in the yard, exposure to the elements (humidity and air) can degrade ellagitannins, increase levels of methyl octalactone (especially cis-isomer), and promote the oxidative breakdown of lignin to aromatic compounds such as vanillin and eugenol; thus influencing the concentration of chemicals that are later affected by toasting. One then has to wonder if differences in climatic conditions at the cooperage stave yards (e.g. more rain in Burgundy than in Cognac) can play a role. There is also some evidence to indicate that wood is less extractable at 3 year air-dry vs. 2 year. In other research, using American oak, it was reported that certain caramelization products are increased in staves that have been seasoned longer. Unfortunately, in our study, there were only a few possible direct comparisons available between 2 and 3 year air-dried wood; and we can not corroborate this observation (Figure 9).
The techniques employed within each cooperage during the bending/toasting process probably have the greatest impact upon generating a coopers' "style". This is most evident when comparing different toast levels among coopers. In any study of barrels, it is not enough to look at the average of all barrels produced by a certain cooper, or to look at one particular toast level. As seen in this study, looking just at two different levels (M and M+), there can be substantial differences in the analyzed chemical compounds between toast levels. But even more important, the pattern of production or degradation of these compounds differs from cooper to cooper as the toast level changes.
Some preconceptions about barrel toasting, especially notions about toast levels, are contradicted in this study. For example, it is often assumed as you move from M, to M+, to H toast that the amount of smokiness or char increases. In this study of water extracts, most often we see a decrease in smoke/char markers when moving form M to M+. This was particularly noted for cooper Billon. Perhaps increased toast level may be more about deeper heat penetration, and not darker stave color. In examining a toast level summary for all coopers (Figure 10) we also see significantly decreased levels of the oak lactones when moving from M to M+. This is likely due to increased thermal degradation. A slight increase in the caramelization compound furfural is also noted at the higher toast level.
Even stave thickness plays a role. It is recognized by winemakers that stave thickness can influence wine maturation. But there may also be an influence in the initial availability of aroma/favor compounds. In our data collected from Vicard barrels, Bordeaux Ferre (thin stave) barrels have substantially decreased levels of compounds susceptible to thermal degradation (the oak lactones and vanillin) as compared to the thicker staved Bordeaux Export barrels (Figure 11). It is likely that the thin staves heat up faster and/or reach higher temperatures.
This report is the summary of a preliminary study into the use of available analytical techniques to screen oak barrels prior to winery use. We have demonstrated that the non-destructive water wash technique can provide an extract that contains measurable levels of compounds important to aromas associated with oak barrels. We also suggest that discernable toasting styles can be revealed or validated in these types of studies. Only more extensive surveys, employing water and wine with larger numbers of barrels, will determine the significance of these observed differences.
It will be interesting in the next phase of the study, now in progress, to identify those specific practices (or combination of practices) that have the greatest influence on development of aroma compounds at each cooperage. In examination of the firing processes, one can observe differences in the number of burners used, the height of the flame, the amount of air-flow, the use of water, or the use of a lid. These practices do not just differ between fabricators, but within each cooperage when changing from medium to medium+ to heavy toast. For example, when considering the two types firings offered by Cadus (slow red vs. slow blue), one can immediately see the impact that small changes can have upon aroma development. A small change in burners, allowing greater incorporation of air, could be responsible for the observed difference.
Another follow-up study is in progress, employing the cooperation of wineries such as: Cakebread Cellars, Lafond Winery, and Edna Valley Vineyards. Many of the barrels used in this "water rinse" study are to be employed in aging wines (Cabernet Sauvignon, Pinot Noir, and Chardonnay) in a systematic manner. The wines will be analyzed for the same oak aroma compounds, and these results will be compared to the original study. A panel tasting may also be conducted. Perhaps we can determine the correlation, if any, between the water extracted compounds and their presence in wine.
SEE ALSO: PHASE 2 - Follow-up study is now available.
Bouchard Cooperages wishes to thank:
- Gordon Burns, PhD and Eric Hervé, PhD (ETS Laboratories) for assistance in experimental design and analytical services.
- John Montero (Winemaking Consultant) for design and supervision of the study, and for authoring this report.
- C-Line Warehouse for providing space and personnel.
© 2003, Vincent Bouchard