7200
Langmuir 2007, 23, 7200-7208
Characterization by Optical Measurements of the Effects of Some Stages of Champagne Technology on the Adsorption Layer Formed at the Gas/Wine Interface K. Abou Saleh,† V. Aguie´-Be´ghin,† L. Foulon,† M. Valade,‡ and R. Douillard*,† UMR 614 Fractionnement des Agro-Ressources et Emballage INRA/URCA, Centre de Recherche EnVironnement et Agronomique, 2 Esplanade R. Garros, F-51686 Reims and Comite´ Interprofessionnel du Vin de Champagne, 5 Rue Henri Martin, F-51200 Epernay ReceiVed February 15, 2007. In Final Form: March 27, 2007 This study analyzes the effects of some important factors of champagne technology on the ellipticity and Brewster angle microscopy (BAM) of the air/champagne interface in view of using the optical properties of the adsorption layer of base wine to forecast the stability of the champagne bubble collar. Using standard, ultrafiltered, and ultraconcentrated wines it was observed that champagne can lose amphiphilic macromolecules which adsorb on the inner glass wall of the bottle during storage, particles such as dead yeasts can adhere to the adsorption layer, a weak increase of the ethanol content during bottle fermentation can reduce significantly the ellipticity of the adsorption layer, and CO2 has no significant effect on the properties of that layer. Surprisingly, no visible differences of the adsorption layer were noticed between the experimental champagnes of the 2004 vintage of three vine varieties (Chardonnay, Pinot noir, and Pinot meunier). From analysis of all samples it is proposed that the mean value and standard deviation of the ellipticity measured during 30 min after pouring the wine in a Petri dish are physical quantities which satisfactorily characterize the adsorption layer of champagne. When needed, further characterization of the adsorption layer may be obtained by a detailed analysis of the kinetics of ellipticity during the same period and inspection of the BAM images of the interface.
Introduction In a glass poured with champagne or other sparkling wines, bubbles are released from nucleation sites located on preexisting gas cavities trapped in the glass1 and/or in cellulose-fiber structures cast off from paper or cloth and/or remaining from the dry wiping process.2,3 From these nucleation sites bubble trains rise in line during the carbon dioxide discharging process.4 These bubble trains are responsible for formation of a foam ring on the liquid surface at the glass periphery, so-called collar or “collerette”. Bubbles in the collar are subjected to instability phenomena such as resorption, bursting, coalescence, or disproportionation.5-9 It is a general result that these mechanisms are strongly controlled by the properties of the adsorption layers formed at the gas/liquid interface (on the inside of the bubbles and at the air surface of the liquid).5,10 Finally, the coming of bubbles to and the disappearance of bubbles from the collar * To whom correspondence should be addressed. Phone: 33 (0)3 26 77 35 94. Fax: 33 (0)3 26 77 35 99. E-mail:
[email protected]. † INRA. ‡ Comite ´ Interprofessionnel du Vin de Champagne. (1) Liger-Belair, G.; Vignes-Adler, M.; Voisin, C.; Robillard, B.; Jeandet, P. Langmuir 2002, 18, 1294-1301. (2) Liger-Belair, G.; Topgaard, D.; Voisin, C.; Jeandet, P. Langmuir 2004, 20, 4132-4138. (3) Liger-Belair, G.; Voisin, C.; Jeandet, P. J. Phys Chem. B 2005, 109, 1457314580. (4) Liger-Belair, G. J. Agric. Food. Chem. 2005, 53, 2788-2802. (5) Douillard, R. Vign. Champeno. 2000, 6, 50-65. (6) Machet, F.; Robillard, B.; Duteurtre, B. Sci. Aliments 1993, 13, 73-87. (7) Prud’homme, K.; Khan, S. A. Foams: theory, measurements and applications; Marcel Dekker, Inc.: New York, 1995; p 595. (8) Weaire, D.; Hutzler, S. The Physics of foams; Oxford University Press: Dublin, 1999; p 246. (9) Pe´ron, N. Etude de couches d’adsorption de macromole´cules a` l’interface champagne/air: application de la me´thode de la goutte pendante a` la pre´vision des proprie´te´s moussantes. Ph.D. Thesis, Universite´ Pierre et Marie Curie, Paris 6: Paris, 2001. (10) Pe´ron, N.; Meunier, J.; Cagna, A.; Valade, M.; Douillard, R. J. Microsc. 2004, 214, 89-98.
lead to a dynamic quasi-equilibrium state. These delicate phenomena are the hallmark of champagne, and their occurrence is sought after by champagne makers. However, they are not always reproducible. Thus, the question is how to control the stability of the collar? Much work has been performed to answer this question based on the implicit assumption that the foaming properties of the champagne are determined by those of the still wine (or base wine). However, the method often used was an evaluation of the stability of the three-dimensional foam formed by gas sparging.11 To improve the scientific basis of the control of the collar stability an approach has been developed in our laboratory based on the physical mechanisms of bubble and foam stability. A major point in this approach is the properties of the adsorption layer formed at the gas/liquid interface by amphiphilic molecules of the wine. Our general approach is to correlate the bubble and collar stability with properties of the adsorption layer and check how these properties evolve between a base wine and its resulting champagne. The first question was how to get an instrumental measurement of the properties of the adsorption layer? This question is not trivial because the surface tension of a wine is low due to ethanol and its value is hardly affected by formation of an adsorption layer.12-14 Encouraging results were obtained by comparing the surface tension of wines diluted with water with a visual evaluation of the collar.15 However, these results pointed to some limitations: (i) the adsorption layer was formed on a diluted wine whose matrix properties are changed (11) Maujean, A.; Poinsaut, P.; Dantan, H.; Brissonnet, F.; Cossiez, E. Bull. O.I.V. 1990, 61, 23. (12) Pe´ron, N.; Cagna, A.; Valade, M.; Bliard, C.; Aguie´-Beghin, V.; Douillard, R. Langmuir 2001, 17, 791-797. (13) Sausse, P.; Aguie´-Be´ghin, V.; Douillard, R. Langmuir 2003, 19, 737743. (14) Puff, N. Adsorption de prote´ines a` l’interface air/solution hydroalcoolique. Application au champagne. Ph.D. Thesis, INAPG: Paris, 2000. (15) Pe´ron, N.; Cagna, A.; Valade, M.; Marchal, R.; Maujean, A.; Robillard, B.; Aguie´-Be´ghin, V.; Douillard, R. AdV. Colloid Interface Sci. 2000, 88, 19-36.
10.1021/la700453t CCC: $37.00 © 2007 American Chemical Society Published on Web 05/16/2007
Optical Measurement of Champagne Adsorption Layer
by water and (ii) visual evaluation of the collar is close to the organoleptic perception but not accurately quantified. The latter limitation will be dealt with in a forthcoming communication. The first limitation seems to be partly overcome using optical methods (ellipsometry and Brewster angle microscopy (BAM)) to characterize the adsorption layer of unmodified base wines or champagnes. In order to check the relevance of the optical parameters to characterize the adsorption layer of base wines and champagnes, an evaluation is made of their reproducibility in different bottles of a few batches and of the effects of bottle fermentation and vine variety. The previously developed experimental approach using wines with macromolecule concentration adjusted by ultrafiltration was pursued. A brief review of the structure and properties of the adsorption layer is given before the experimental section. The occurrence of an adsorption layer at the gas/champagne interface was revealed using tensiometry and, preferentially, optical techniques: ellipsometry and Brewster angle microscopy (BAM).10 It is formed from macromolecules with a molecular mass between 10 000 and 100 000.15 The surface activity of these macromolecules was quantified by bubble tensiometry in champagne diluted with water.15 Moreover, it has been found, using the BAM technique, that the adsorption layer is inhomogeneous. It often exhibits two-dimensional domains with various patterns and are more or less mobile.10 For instance, 2-D foams with a minor contribution to the Brewster ellipticity can be observed in the case of ultrafiltered champagne. Other 2-D foams organizing condensed films and having a large negative contribution of the ellipticity often occur at the gas/Pinot noir champagne. Moreover, with some champagne samples bright objects seemed to be anchored in condensed films.10 The chemical nature of these macromolecules can be tackled by looking at the composition of champagne wines where macromolecules are mostly proteins (10-30 mg/L)16,17 and polysaccharides (10-400 mg/L)16,18,19 coming from grapes and yeast.20-27 However, the surface properties of one protein coming from grape, grape invertase, are not similar to those of a native champagne.9,12,28 Thus, the nature of the amphiphilic macromolecules responsible for formation of the adsorption layer is still in question. However, the physical approach of this work can be developed irrespective of the detailed chemical knowledge of these macromolecules. (16) Tusseau, D.; van Laer, S. Sci. Aliments 1993, 13, 463-482. (17) Luguera, C.; Moreno-Arribas, V.; Pueyo, E.; Bartolome, B.; Polo, M. C. Food Chem. 1998, 63, 465-471. (18) Moreno-Arribas, V.; Pueyo, E.; Nieto, F. J.; Martin-Alvarez, P. J.; Polo, M. C. Food Chem. 2000, 70, 309-317. (19) Abdallah, Z.; Aguie´-Be´ghin, V.; Douillard, R.; Bliard, C. In Macromolecules and secondary Metabolites of GrapeVine and Wine; Jeandet, P., Cle´ment, C., Conreux, A., Eds.; Lavoisier: Paris, 2007; pp 377-384. (20) Brissonnet, F.; Maujean, A. Am. J. Enol. Vitic. 1991, 42, 97-102. (21) Brissonnet, F.; Maujean, A. Am. J. Enol. Vitic. 1993, 44, 297-301. (22) Marchal, R.; Bouquelet, S.; Maujean, A. J. Agric. Food. Chem. 1996, 44, 1716-1722. (23) Moreno-Arribas, V.; Pueyo, E.; Polo, M. C. J. Agric. Food. Chem. 1996, 44, 3783-3788. (24) Martinez-Rodriguez, A.; Carrascosa, A. V.; Barcenilla, J. M.; Pozo-Bayon, M. A.; Polo, M. C. Food Microbiol. 2001, 18, 183-191. (25) Martinez-Rodriguez, A. J.; Carrascosa, A. V.; Martin-Alvarez, P. J.; Moreno-Arribas, V.; Polo, M. C. J. Ind. Microbiol. Biotechnol. 2002, 29, 314322. (26) Girbau-Sola, T.; Lopez-Tamames, E.; Bujan, J.; Buxaderas, S. J. Agric. Food. Chem. 2002, 50, 5596-5599. (27) Nunez, Y. P.; Carrascosa, A. V.; Gonzalez, R.; Polo, M. C.; MartinezRodriguez, A. J. J. Agric. Food. Chem. 2005, 53, 7232-7237. (28) Puff, N.; Marchal, R.; Aguie´-Be´ghin, V.; Douillard, R. Langmuir 2001, 17, 2206-2212.
Langmuir, Vol. 23, No. 13, 2007 7201
Figure 1. Flow diagram of the preparation of the experimental champagnes of the 2004 vintage.
Materials and Methods Experimental Samples. Base wines resulting from the usual alcoholic fermentation of the “must” in a ventilated vat were produced from three vine varieties grown by CIVC (Commite´ Interprofessionnel du Vin de Champagne, Epernay, France): Chardonnay (BWCH), Pinot noir (BWPN), and Pinot meunier (BWPM) during the 2004 grape harvest (Figure 1). After stabilization at -4 °C followed by frontal filtration with a porosity of 0.6 µm a tangential ultrafiltration of the base wines (70 L batches) was conducted using a Hi-Flow device (Inceltech, Toulouse, France) as previously described.12 The 1.8 m2 ultrafiltration membrane with a nominal molecular mass cutoff of 10 000 was made of hydrophilic polysulfone (30UFIB/ 1/S1.8/10KD, Inceltech). Before running the experiment the device was rinsed twice with 0.8 L of wine. In the same way, after each ultrafiltration the device was rinsed with ultrapure water and ethanol of synthesis grade. During the ultrafiltration the speed of the peristaltic pump was adjusted to achieve a 3× ultraconcentration of the wine and the transmembrane pressure was maintained at 0.8 ×105 Pa. For each vine variety three base wine samples are obtained: standard (Std-BW), ultraconcentrated (UC-BW), and ultrafiltered (UF-BW) base wines with a theoretical relative concentration factor of macromolecules (>10 000) of 1, 3, and 0, respectively (Figure 1). All samples including standard base wine were stored at 4 °C in 0.375 L half bottles for analysis and 0.75 L bottles for the second fermentation. This latter was achieved by adding yeast, sugar, and 0.6 mL/L bentonites (adjuvant 83, Station Œnotechnique de Champagne) in 0.75 L bottles which were closed by a gastight plug. During this additional alcoholic fermentation sugar was changed to ethanol and carbon dioxide. At the end of the bottle fermentation the ethanol concentration was increased to about 1.4% (v/v) and the pressure of CO2 in the bottle was about 6-7 bar. Aging on lees was performed for at least 3 months. The bottles were then subjected to riddling and the lees, the consequent sediment, were expelled before sealing the bottles with capsules. At this stage of manufacturing the number of particles is less than 1 × 106 per mL (CIVC, personal communication). All the base wine and champagne bottles were stored at +4 °C. Champagne samples, containing 10-12 g/L of dissolved CO2,29 were partially degassed for optical reflectivity measurements by opening the bottle 1 h before analysis. Then, the samples were poured in a Petri dish of 60 mm diameter previously treated with chromosulfuric acid and rinsed with ultrapure water. In these (29) Maujean, A. ReV. fr. Oenol. 1989, 120, 11-17.
7202 Langmuir, Vol. 23, No. 13, 2007
Abou Saleh et al. consists of two adjustable arms mounted on a goniometer. The laser arm contains the Nd:YAG (50 mW, λ ) 532 nm) laser, a quarterwave plate, a polarizer, and a compensator. The detector arm contains an analyzer and a photodetector. These two arms were adjusted 1° over the Brewster angle incidence (θB + 1°) where θB is defined by tan θB ) n2/n0
(4)
where n0 and n2 are the refractive indexes of air and wine, respectively. The choice to work at θB + 1° is a compromise between a great sensitivity in ∆ and an ellipticity large enough to be detected.30,31 At the air/base wine interface θB + 1° is 54.30°, and at the air/ champagne interface θΒ + 1° is 54.33°. The ellipticity coefficient of the adsorption layer measured in the Brewster conditions for the substrate, [mac]FB is defined by FjB ) tan Ψ sin ∆
Figure 2. Effect of wine sample confinement on the kinetics of its optical properties. (A) Scheme of the experimental confinement setup. (B) Kinetics of null ellipsometry signal (top) and refractive index (bottom). UC-Ch was poured in a Petri dish in open air (b) or in the confinement setup (9). conditions no nucleation and bubbling interfere with the optical measurements. The Petri dish was placed in a confinement cell (Figure 2A) to avoid ethanol evaporation during the measurement performed in an air-conditioned room at 20 ( 1 °C. The confinement cell was another Petri dish of 90 mm diameter pierced by two holes (0.7 × 1.2 cm) to allow the incident light beam to reach the liquid surface and the reflected one to reach the detector arm of the ellipsometer freely. Null Ellipsometry. Null ellipsometry is a well-established nondestructive optical technique for the characterization of ultrathin layers.30 The difference in polarization between the incident and the reflected beams is usually described by the measured ellipsometric angles, Ψ and ∆, where Ψ reflects the changes in amplitude and ∆ the relative phase shift. These two ellipsometric angles are linked to the two Fresnel reflection coefficients, rp and rs, respectively, in directions parallel and perpendicular to the incidence plane by rp/rs ) tan Ψ ei∆ ) tan Ψ(cos ∆ + i sin ∆)
(5)
In the ideal case of an interface between two transparent media without roughness and without an adsorption layer (Fresnel interface), FjB is zero. Considering interfaces whose optical anisotropy can be ignored, FjB is the sum of two main contributions36 FjB ) Fjr + Fjd
(6)
where Fjr is a roughness positive term which accounts for thermally induced capillary waves and Fd depends on the presence of an adsorption layer. According to Drude’s approach, which is a good enough approximation in the case of molecular layers,37 Fd is negative when the refractive index is larger within the interfacial layer than in the liquid phase. In the case of ultrathin layers, the thickness of which, d, is much smaller than the wavelength, λ, and from the Drude equation,38 the Brewster angle ellipticity coefficient is proportional to the layer thickness and the difference (n1 - n2) when an adsorption layer begins to form (n1 being the refractive index of the adsorption layer). Brewster Angle Microscopy. The technique of Brewster angle microscopy (BAM) uses the properties of the ellipticity coefficient at the Brewster angle, FB, to visualize micrometer-scale heterogeneities due to variations in the thickness or refractive index of the adsorption layer.39 At the Brewster angle of the substrate, the linear p-polarized light is set to 0° and the relation between the two reflectivity coefficients is
(1)
Within null ellipsometry the setting of the polarizer and the analyzer is chosen such that the parallel electromagnetic field of the reflected light rp is null. Only the perpendicular reflection coefficient rs is measured. In this null condition the polarizer and analyzer positions are directly proportional to Ψ and ∆ according to the following equations Ψ)A
(2)
∆ ) 2P + π/2
(3)
where A and P are the positions in degrees of the analyzer and polarizer, respectively. Ψ and ∆ are measured with a precision of (0.003° and (0.01°, respectively.31 Null ellipsometry measurements were performed using an Optrel Multiskop Ellipsometer (Optrel, Berlin, Germany), which has been extensively described elsewhere.32-35 The instrument (30) Azzam, R. M. A.; Bashara, N. M. Ellipsometry and Polarized Light; Elsevier: Amsterdam, 1987; p 539. (31) De Feijter, J. A.; Benjamins, J.; Veer, F. A. Biopolymers 1978, 17, 17591772. (32) Reiter, R.; Motschmann, H.; Knoll, W. Langmuir 1993, 9, 2430-2435. (33) Harke, M. T. R.; Schulz, O.; Motschmann H. H. O. ReV. Sci. Instrum. 1997, 68, 3130-3134.
rp ) irsFB
(7)
The reflectance R ()Ir/I0) of the interface is proportional to FjB2, where Ir and I0 are the reflected and incident intensities, respectively. As a consequence, the uncovered substrate appears dark in the BAM images (rp ≈ 0) whereas all parts covered by adsorption layers appear more or less bright (rp * 0). The setup was the same as that of Null ellipsometry except that the quarter-wave plate and the compensator from the polarizer arm were omitted and a CCD camera was used instead of the photodetector. Moreover, a projective lens and an objective of 10× with a numerical aperture of 0.28 were installed in front the CCD camera. Images were taken with a lateral resolution of 1.2 µm. (34) Benjamins, J. W.; Jo¨nsson, B.; Thuresson, K.; Nylander, T. Langmuir 2002, 18, 6437-6444. (35) Benjamins, J.-W.; Thuresson, K.; Nylander, T. Langmuir 2005, 21, 149159. (36) Meunier, J. J. Phys. 1987, 48, 1819-1831. (37) He´non, S.; Meunier, J. In Modern characterization methods of surfactant systems; Binks, B. P., Ed.; Marcel Dekker, Inc.: New York, 1999; Vol. 4, pp 109-145. (38) Drude, P. Lehrbuch der Optik; Springer Wien: Gottingen, 1889; Vol. 14, p 498. (39) Reiter, R.; Motschmann, H.; Orendi, H.; Nemetz, A.; Knoll, W. Langmuir 1992, 8, 1784-1788.
Optical Measurement of Champagne Adsorption Layer
Langmuir, Vol. 23, No. 13, 2007 7203
In order to isolate the instrument from ambient vibrations, the Multiskop ellipsometer was mounted on a MOD-1 active vibration isolation table (Halcyonics, Goettingen, Germany). Refractive Index Measurement. Refractive indexes of liquid samples were measured with a DUR-W2 refractometer (Schmidt & Haensch, Germany) with an accuracy of (5 × 10-5. Counting of Particles. A frontal filtration was carried out using a PTFE membrane (porosity 0.2 µm) to recover the particles remaining in suspension in champagne bottles. After washing the PTFE membrane two times with 5 mL of filtered champagne, the number of particles was determined with a Beckmann Coulter Z2 (Beckmann, USA).
Results and Discussion Preliminary results pointed to a lack of reproducibility of the optical measurements performed on the champagne interface. An experimental device has been developed and several parameters have been checked to overcome these problems. Finally, the bottle fermentation and vine variety effects have been analyzed on the surface properties, taking advantage throughout this study of the controlled amount of macromolecule obtained by ultrafiltration of the base wine. (1) Experimental Setup. Preliminary optical measurements on champagne samples were carried out in an open cell as previously described.10,15 During these measurements it was observed that the refractive index of the liquid decreases from 1.34298 to 1.34254 ( 5 × 10-5 during the first 30 min after pouring partially degassed champagne in a Petri dish (Figure 2), corresponding to a decrease of 0.3% (v/v) ethanol in the champagne and thus a Brewster angle change from 53.33° to 53.32°. By contrast, when the same measurement was carried out in a confinement cell the refractive index remains constant at 1.34296 during the time course of the kinetics (Figure 2B). Concerning the absolute value of the ellipticity, FjB, of a UC-Ch, the major effect of confining is lowering of the fluctuations of the signal (Figure 2B), which is associated with a decrease of the lateral mobility of the domains observed by BAM. This last effect should be related to a decrease of the convection in the bulk. Moreover, a slight decrease of the absolute value of the mean ellipticity 〈|FjB+1°|〉 was noticed from 7.6 × 10-3 ( 0.7 × 10-3 to 6.4 × 10-3 ( 0.5 × 10-3 when the measurement was performed in a confined cell instead of an open cell. To avoid these effects linked with ethanol evaporation, all forthcoming optical measurements were carried out in the confinement cell. (2) Reproducibility within Batches. With each champagne obtained from UC, STD, or UF base wine recording of the Brewster angle ellipticity for 30 min after pouring gives a kinetics approach of formation of the adsorption layer (Figure 3). In view of our interest in the collar stability, this analysis is usually restricted to 30 min, a time interval consistent with the time champagne remains in a glass and long enough to buffer large ellipticity fluctuations of samples exhibiting domains. In fact, the signal of numerous samples exhibits large fluctuations which correspond to heterogeneities of the mobile adsorption layer (the layer may move quite freely on the interface). Moreover, an increase of the mean value of the signal is also often noticed, indicating that formation of the adsorption layer is in progress during the time course of the kinetics. Nevertheless, the mean value of ellipticity during the 30 min kinetics gives an approximation of the adsorption layer after 15 min, which is also a crude approximation of the amount of adsorption layer occurring during the 30 min of the experiment. That approximation may be tempered by a detailed kinetics analysis of the signal and the BAM images of the interface. The mean ellipticity of the first 30 min 〈|FjB+1°|〉 was determined for UF, STD, and UC champagnes and found to vary in a range
Figure 3. Reproducibility of the ellipticity data at the air/champagne interface of UF, STD, and UC champagnes (from top to bottom, respectively). The time course of the kinetics of the signal was determined for five bottles of each sample (left). The mean value 〈-103 FjB+1°〉 and the standard deviation of the signal were calculated for each kinetics with a sampling of one value every 5 s (360 values for 30 min) (right).
from 1 to 5 for five UF samples but only of 15% for STD or UC samples. Moreover, 〈|FjB+1°|〉 is 5-10 times larger for STD or UC than for UF samples (Figure 3). Since the variability between bottles is much larger for samples with a low macromolecule concentration (UF-Ch), it can be suspected that this effect is related to an adsorption phenomenon on the glass of the bottles. Thus, adsorption of macromolecules was checked on the glass of champagne bottles. (3) Adsorption of Macromolecules or Particles on the Inner Wall of Glass Bottles. In order to know if the adsorption phenomenon on glass is important an analysis was performed of the effects of rocking or stirring champagne bottles on the formation and structure of the adsorption layer at the gas/liquid interface. Complementary experiments were also carried out by washing the glass of the bottles and evaluating the effect of particles in these phenomena. Rocking. Each champagne bottle was analyzed in the following way. At first, 2 × 20 mL of champagne were poured with delicacy into Petri dishes to measure the Brewster ellipticity and make BAM images of the interface for 30 min. Then, the bottle containing 710 mL of champagne was rocked and rotated gently in all directions for 5 min before pouring once more 2 × 20 mL in Petri dishes and recording the surface optical properties. In the case of UF-Ch, the ellipticity of the rocked bottles is, at least, two times larger than the ellipticity of the same bottles before
7204 Langmuir, Vol. 23, No. 13, 2007
Figure 4. Effect of rocking champagne bottles on the ellipticity data. Measurements were performed before (open symbols) and after (filled symbols) rocking the bottles 5 min: Chardonnay UF (4, 2), STD (O, b), and UC (0, 9) champagne.
Figure 5. Effect of rocking bottles on the BAM images of the air/champagne interface. Images taken before (a-c) and after (d-f) rocking with ultrafiltered (a and d), standard (b and e), and ultraconcentrated (c and f) champagnes produced from the Chardonnay vine variety (same samples as in Figure 4). Each image was obtained 15 min after pouring champagne in the Petri dish.
rocking (Figure 4). On the contrary, for STD-Ch and UC-Ch batches there is no significant difference between samples analyzed before or after rocking the bottles. The BAM images obtained from these samples show that the layer is inhomogeneous in the case of the UF-Ch and more homogeneous in the case of STD-Ch and UC-Ch (Figure 5). Rocking does not change the structure of the adsorption layer except in the case of the UF-Ch, where the reflectance is larger after than before rocking. Stirring. After a strong stirring of the bottles in the up and down directions the absolute value of the mean ellipticity at the air/UF-Ch interface changes from the range of 0-0.5 × 10-3 (without stirring) to the range of 1-1.7 × 10-3 (after stirring) (Figure 6). In the case of STD-Ch and UC-Ch samples stirring does not seem to significantly change the surface ellipticity, which possibly decreases in the case of UC-Ch. In conclusion, these rocking and stirring experiments seem to increase, in the wine, the amount of macromolecule material which may have been adsorbed during storage. This effect is especially significant in the case of UF-Ch, where the amount of macromolecules in the wine is small. This suggests that the amount of material which comes in the wine by rocking or stirring is rather constant and on the order of magnitude of the amount which is in the UF-Ch bottles but significantly smaller than the amount which is in the STD-Ch or UC-Ch bottles. The most likely hypothesis is that this material comes from the glass surface of the bottle. However, it may comprise as well (macro)molecules as particles which are particularly visible after bottle fermentation (see, for instance, Figure 5). The effects of particles on the optical properties will be analyzed first, and the occurrence of macromolecule material on the glass wall will be checked afterward.
Abou Saleh et al.
Figure 6. Effect of stirring bottles for 3 min on the ellipticity data obtained at the air-champagne interface: (left) kinetics; (right) mean value of the signal measured for 30 min. Measurements were performed before (open symbols) and after (filled symbols) stirring the bottles: Chardonnay UF (4, 2), STD (O, b), and UC (0, 9) champagne.
Figure 7. Effect of microfiltration on the BAM image of the adsorption layer of standard Chardonnay champagne. Images were obtained 12 (a and c) and 24 min (b and d) after pouring the samples. Frontal filtration was carried out on membranes with a porosity of 0.2 µm.
Particles. Their effect on the optical properties of the adsorption layer was checked using samples enriched or devoid of them by frontal filtration on a PTFE filter of 0.2 µm porosity. Their number in unfiltered STD-Ch was found to be close to 3 × 103 particles/ mL. The adsorption layer formed on unfiltered STD-Ch shows a heterogeneous structure with bright objects (Figure 7), whereas the adsorption layer formed on filtered champagne is more homogeneous without bright objects, which are thus identified with particles larger than 0.2 µm. Moreover, the ellipticity, close to -5 × 10-3, is not significantly different for unfiltered and filtered champagnes. In an other experiment addition of 106 particles/mL (300 times more than in STD-Ch) results in an increase of the absolute value of ellipticity of only 2 × 10-3. The observation of the particles by microscopy with a 100× objective showed that these particles are principally dead yeasts of 4 µm of diameter, which compare favorably with the bright spots observed on BAM images which are also around 4 µm. Thus, the bright spots observed in the adsorption layer should be partly some dead yeasts. All these results on particles show that (i) particles can be present in the adsorption layer formed at the gas/liquid interface but do not increase significantly the ellipticity in the experimental conditions of this work. (ii) As a consequence, the ellipticity increase observed after rocking or stirring champagne bottles should be related to the increase of the concentration of surface active molecules in the wine, resulting from desorption of molecules from the glass. This hypothesis will be checked now by rinsing the glass of emptied champagne bottles. Rinsing of Bottles. Two protocols for rinsing bottles were used: one with ultrapure water (Figure 8A) and the other with
Optical Measurement of Champagne Adsorption Layer
Langmuir, Vol. 23, No. 13, 2007 7205
Figure 9. Ellipticity data at the air/wine interface of five commercial champagnes.
Figure 8. Effect of the rinse of champagne bottles on the ellipticity data at the air/solvent interface. The 750 mL bottles emptied of UF, STD, or UC Pinot noir champagne were rinsed three times with 100 mL of ultrapure water (A) or 100 mL of a 12.5% hydroalcoholic solution (B). The ellipticity data were determined for each rinse. Left: kinetics of (0) native UC-Ch, (2) first rinse, (9) second rinse, (b) third rinse, and (]) pure solvent. Right: mean ellipticity values of UF, STD, and UC champagnes and of the successive rinses of their bottles.
12.5% (v/v) ethanol (Figure 8B). Each bottle was emptied gently and rinsed three times with 100 mL of the rinsing liquid. The surface properties of each rinse were evaluated by null ellipsometry. The value of the ellipticity measured for 30 min is relatively constant (Figure 8A and B). In the case of UC-Ch, the mean value of the ellipticity of the first water rinse is close to -4 × 10-3 at the end of the adsorption kinetics. The ellipticity decreases steadily for the second and third rinses and tends to the value of the ellipticity of ultrapure water (〈|FjB+1°|〉) 5 × 10-4) (Figure 8A). In the case of STD-Ch and UF-Ch, the same phenomenon was observed. The first rinse gives a significant ellipticity, while the second and third rinses exhibit much smaller values. The same operations were carried out with 12.5% ethanol (Figure 8B). In this case only the first rinse of UC-Ch and STDCh bottles gives significant values of 〈FjB+1°〉 near -1 × 10-3 and -2 × 10-4, respectively. All other rinses give values close to that of the pure solvent. The low ellipticity value of the ethanol solution rinses, compared to those with water, suggest that either the rinses with the ethanol solution are not effective or the rinses are effective but ethanol slows down adsorption of macromolecules at the air/liquid interface. An answer to this alternative came from a water rinse performed after a first rinse with an ethanol solution: The ellipticity of the water rinse remains equal to that of pure water (-5 × 10-4). In conclusion, (i) the first rinse allows extracing the majority of molecules or particles which are adsorbed on the inner glass wall of the bottle, (ii) a hydroethanolic solution is more effective than water to desorb this material from the bottle wall, and finally (iii) adsorption of these molecules or particles at the gas/liquid interface is slowed down by ethanol in the bulk. To conclude this part concerned with the reproducibility of the surface properties of experimental champagnes, it should be stressed that champagne is defined as a product with properties coming from both the wine and its packaging: the bottle. The
〈FjB+1°〉 increase after rocking or strongly stirring an UF-Ch bottle can be probably explained by the release of (macro)molecules and particles previously adsorbed on the inner glass wall of the bottle during its storage. In the case of STD-Ch and UC-Ch, this effect is less visible because the champagne samples contain enough macromolecules to form an adsorption layer on the glass of the bottle without significantly lowering their concentration in the bulk. This observation can be relevant for champagnes whose amount of surface active molecules is on the order of magnitude of the amount which can be adsorbed on the glass. The inner area of a bottle is close to 0.06 m2; an adsorption layer of biomacromolecules formed in model systems in a few hours has a surface concentration of 2-4 mg/m2;40 thus, the amount of macromolecules involved in an adsorption layer in a champagne bottle could be on the order of 0.1-0.2 mg or probably more if the bottle has been stored for weeks. As a consequence, in the rinsing experiments with 100 mL of solvent the volume concentration could be on the order of 1-2 mg/L of solvent, a concentration which is significant for formation of an adsorption layer as observed experimentally (Figure 8).14,40 Moreover, this amount seems to be small when compared to the amount of macromolecules expected in a STD-Ch (100-300 mg/L).18,41,42 However, a preliminary check of the surface properties of commercial champagnes showed that their 〈FjB+1°〉 ranges from -1 to -8 × 10-3 (Figure 9). Thus, some commercial samples have surface properties close to those of UF-Ch (〈FjB+1°〉 ≈ -0.5 to -1 × 10-3), where adsorption on the bottle glass is significant. Finally, the surface properties within a single batch of champagne bottles may be influenced by the individual history of each bottle (rocking, stirring, storage temperature, etc.) before it is uncorked and champagne is poured in a measuring device or in a “flute”. (4) Effects of Vine Variety and Bottle Fermentation on the Adsorption Layer. From the previous section it comes that formation of the adsorption layer may be influenced by interaction of the wine with the bottle glass during aging. However, the main properties of the adsorption layer come from the macromolecule composition of the wine, which depends on the vine variety, the vintage, and the wine-making operations. At the present stage of characterization of the adsorption layer a check was made of the surface properties of champagnes obtained from the three vine varieties and the effect of bottle fermentation. The last analysis is of particular interest since wine makers would like to know the foam properties of champagne from the properties of the base wine. (40) Puff, N.; Cagna, A.; Aguie´-Be´ghin, V.; Douillard, R. J. Colloid Interface Sci. 1998, 208, 405-414. (41) Lopez-Barajas, M.; Lopez-Tamames, E.; Buxaderas, S.; Suberbiola, G.; de la Torre-Boronat, M. C. Am. J. Enol. Vitic. 2001, 52, 146-150. (42) Lao, C.; Santamaria, A.; Lopez-Tamames, E.; Bujan, J.; Buxaderas, S.; De la Torre-Boronat, M. C. Food Chem. 1999, 65, 169-173.
7206 Langmuir, Vol. 23, No. 13, 2007
Figure 10. Effect of the vine variety on the ellipticity data of experimental UF (- - - -), STD (- - - -), and UC (----) champagnes of the 2004 vintage: Chardonnay (b), Pinot noir (9), and Pinot meunier (2).
Figure 11. BAM images of the air/champagne interface of experimental UF (a-c), STD (d-f), and UC (g-i) champagnes of the 2004 vintage (same samples as in Figure 10): Chardonnay, left column (a, d, g); Pinot noir, central column (b, e, h); and Pinot meunier, right column (c, f, i).
Effect of Vine Variety. The three vine varieties of champagne Chardonnay, Pinot noir, and Pinot meunier were compared by analyzing their adsorption kinetics (Figure 10) and layer structure (Figure 11). Each measurement was carried out by gently pouring the sample in the Petri dish without rocking or stirring the champagne bottle. The three UF, STD, and UC batches were analyzed. The ellipticity varied from 0 to -4 × 10-3, from -2.5 × 10-3 to -5 × 10-3, and from -4 × 10-3 to -8.5 × 10-3, respectively (Figure 10). For all vine varieties the ellipticity values are directly related to the macromolecule concentration. The kinetics of the UF-Ch exhibit a weak ellipticity and important fluctuations of the signal, as observed in section 2. By contrast, in the case of STD-Ch and UC-Ch the ellipticity is larger and increases more regularly. Surprisingly, there is no significant difference between the three vine varieties when comparing together UF, STD, or UC champagnes. The only visible difference among varieties concerns the Chardonnay UC-Ch, which has a significantly higher ellipticity than the UC of the other varieties. This difference cannot be explained at the present since it is not confirmed in the case of the STD and UF champagnes. BAM images of the surface of STD and UC champagnes exhibit an even structure of the adsorption layer with a uniform reflectance and some bright objects whose size are about 4 µm (Figure 11d-i). The reflectance of the adsorption layer formed on UFCh (Figure 11a-c) is smaller, close to zero. The only common feature with the other batches is the occurrence of the bright
Abou Saleh et al.
Figure 12. Effect of bottle fermentation on the ellipsometry data at the air/wine interface. The experimental wines came from Pinot noir of the 2004 vintage. Filled symbols: base wine. Open symbols: champagne. UF (2, 4); STD (9, 0), and UC (b, O).
spots of 4 µm size. It may be concluded that there is no significant difference between the 2004 experimental champagnes prepared from the three vine varieties. Nevertheless, champagnes prepared from Pinot noir of 1997 and 1998 vintages exhibited systematically 2-D foams within the condensed adsorption layer.10 Thus, the final conclusion should be that the structure of the adsorption layer formed at the champagne/air interface may vary according to factors which have not been identified to date. Effect of Bottle Fermentation. The experimental champagnes were obtained with a period of aging on lees after bottle fermentation of at least 3 months. This period is somewhat shorter than that of standard champagnes, which is at least 15 months. The kinetics of adsorption layer formation has been determined for the UF, STD, and UC experimental base wines and the corresponding champagnes of the Pinot noir vine variety (Figure 12). As previously noticed10 and also observed in this work for Chardonnay champagne (Figure 3), the ellipticity signal 〈|FjB+1°|〉 decreases in the order UC > STD > UF. Moreover, the absolute value of the ellipticity of the base wine is always larger than that of its corresponding champagne. It is also obvious in the case of UF that the signal is less stable for champagne than for the base wine, suggesting a more heterogeneous structure of the adsorption layer. Consistently, BAM observation of the samples showed that BW samples have more homogeneous surfaces than the corresponding champagnes and the UF samples have a much lower reflectance than the other samples (Figure 13) and a much more mobile layer. Thus, the fluctuations of the signal should correspond, at least in part, to bright objects which are mobile and more numerous in the case of champagne than for base wines. Moreover, the differences of reflectance show that the UF layer has a lower surface concentration than the STD or UC ones. An important issue is the reasons for the differences of ellipticity between the base wines and their corresponding champagnes. Clearly, they result from bottle fermentation since no liquor was added after degorging of these experimental wines. Among the chemical composition differences which may have an effect are (i) the ethanol content which is increased 1.4% (v/v) between the base wine and the champagne,11 (ii) the CO2 content which is around 12 g/L in Champagne,14 and (iii) the macromolecules which may have been partly metabolized by the yeasts. To check the effect of CO2 on the ellipticity of the samples, champagne bottles were uncorked and ethanol concentration and ellipticity kinetics were determined at different time intervals while CO2 discharges. There was no significant decrease of ethanol concentration and no change of the 〈|FjB+1°|〉. Thus, the CO2 concentration has no significant effect on the adsorption of macromolecules at the air/champagne interface. To check the effect of ethanol change an experiment was performed by
Optical Measurement of Champagne Adsorption Layer
Langmuir, Vol. 23, No. 13, 2007 7207
Figure 14. Effect of the ethanol content on the ellipticity data at the air/wine interface of base wine and champagne. Native Pinot noir standard base wine (b) and champagne (2) are 10.9% and 12.3% ethanol, respectively. For comparison purposes, the champagne was adjusted to 10.9% with water (4) and the same volume of a 10.9% ethanol solution was added to the base wine (×). Similarly, the base wine was adjusted to 12.3% by adding ethanol (O), and the same volume of a 12.3% ethanol solution was added to champagne (+).
batches of champagne wines and especially with commercial champagnes the surface properties measured on base wines could be helpful to forecast the surface properties of the corresponding champagnes.
Conclusions
Figure 13. BAM images of the adsorption layers formed at the air/wine interface before and after bottle fermentation. The samples are the same as in Figure 12. Only the STD base wine is shown. The images were obtained 15 min after pouring the samples. For the 3-D representation, the gray level (GL) was calculated with the Image-J program.
adjusting the ethanol content of base wine from 10.9% to 12.3% (v/v) and conversely by adjusting the ethanol content of champagne from 12.3% to 10.9% (v/v) (Figure 14). The increase of 1.4% ethanol in base wine decreases the absolute value of the ellipticity of about 1 × 10-3 at the beginning of the kinetics and about 0.5 × 10-3 at the end of the kinetics; the mean ellipticity 〈|FjB+1°|〉 was decreased from 5.1 × 10-3 ( 0.7 × 10-3 to 4.6 × 10-3 ( 0.9 × 10-3. In the same way, the decrease of 1.4% of ethanol in champagne results in an increase of the absolute value of the ellipticity of about 1 × 10-3. Nevertheless, the adjustment of ethanol content does not allow reaching the ellipticity of champagne from base wine or the reverse. The gap of ellipticity is on the order of 1-2 × 10-3. Consequently, another effect, such as the macromolecule nature or concentration, is responsible of the small decrease of the ellipticity between base wine and champagne. This point should be analyzed later when the gross general macromolecule composition of champagne will be established. In conclusion, it can be stated that bottle fermentation of the experimental champagnes results in a significant but slight change in the solvent properties and in the macromolecule composition and/or concentration. These changes should have only marginal effects on the properties of the adsorption layers and the stability of the collar. As long as this conclusion is confirmed with other
Structure of the Adsorption Layer. In this study phase separation was clearly observed only with STD wines, while in a previous study10 this phenomenon seemed quite general with UF, STD, and UC wines and champagnes. Since only a few experimental samples have been studied, no final conclusion should de drawn concerning the structure of the adsorption layer and all the more so concerning the reasons of its variability. Quantitative Approach. The ellipticity kinetics gives a simplified view of the two-dimensional structure of the adsorption layer that may be homogeneous, exhibit domains, and include particles of a few micrometers. Nevertheless, use of kinetics to characterize a sample is a heavy procedure. The mean value of the Brewster angle ellipticity for 30 min (or its absolute value) and its standard deviation seem to be figures which throughout this study give the same classification of samples as the one obtained by visual inspection of the kinetics (Figures 3, 4, 6, 8-10, 12, 14). Consequently, this 〈FjB+1°〉 (or 〈|FjB+1°|〉) value will be used to characterize, at first, the adsorption layer of champagne wines which might be further characterized by a detailed analysis of the kinetics and/or the BAM images. Factors Affecting the Structure of the Adsorption Layer. The present study shows that the structure of the adsorption layer is influenced by the amount of macromolecule and the quality of solvent (ethanol concentration). These preliminary data point to the possible effects of macromolecule structure and solvent quality on the two-dimensional organization of proteins and polysaccharides at the interface between champagne and gas. Work is in progress to specify these effects on the structure of the adsorption layer. Moreover, the small number of samples analyzed to date does not allow drawing conclusions on the effects of vine variety or technology processes on the structure of the adsorption layer. An unexpected result is adsorption of macromolecules on the glass of the bottles. This effect may significantly lower the amount of macromolecules of champagnes when their concentration is low. Concerning bottle fermentation, the results of this work support the idea that it has only a moderate
7208 Langmuir, Vol. 23, No. 13, 2007
effect on the amount and nature of the macromolecules of champagne. This hypothesis derived of experimental wines could be tested on champagnes obtained with various periods of aging on lees or otherwise. Finally, this work has provided new data on the properties of the adsorption layer formed at the gas/champagne interface and pointed to questions concerning the nature of the macromolecules involved in the adsorption layer, effects of vine variety and wine-
Abou Saleh et al.
making practices on these macromolecules, and relations between these macromolecules and the stability of the bubble collar. Acknowledgment. The authors thank the Conseil Ge´ne´ral de la Marne, Conseil Re´gional de Champagne Ardenne, and Ville de Reims for financial support. Thanks also to Jacques Meunier for related discussions. LA700453T
