Article pubs.acs.org/JAFC
Novel Methodology for Measuring Temperature-Dependent Henry’s Constants of Flavor Molecules Shane Avison,*,† Kitty van Gruijthuijsen,§,# Mirela Pascu,§ Alan Parker,† and Igor Bodnár† †
Firmenich S.A., Rue de la Bergère 7, Meyrin 2, CH-1217 Geneva, Switzerland Western Switzerland − Valais, Institute of Life Sciences, University of Applied Sciences and Arts, Route du Rawyl 64, CH-1950 Sion, Switzerland
§
ABSTRACT: A new methodology is presented to measure water−air partition coefficients (Henry’s constants) of volatiles, using APCI-MS. Significant advantages over other Henry’s constant determination methods include the short measurement and sample preparation time and the possibility for simultaneous measurement of multiple volatiles. The methodology is validated by obtaining good agreement with reliable literature values for a series of 2-ketones. The methodology is further explored for eight key volatiles typically found in citrus fruits, including the temperature dependence of the Henry’s constant. Using these data can improve estimates of flavor losses during processing and volatile release during consumption. KEYWORDS: APCI-MS, partitioning, volatiles, citrus
■
INTRODUCTION Flavors are delicate mixtures of volatile compounds, added to a food so that the consumer has the optimal pleasure of eating the product. To get the right flavor profile during consumption, experts such as flavorists create tailored flavors, using hundreds of different molecules. The perception of these molecules depends on the concentration in the headspace as well as the strength of binding of the molecules to aroma receptors in the nose.1 The concentration in the headspace is a consequence of the partitioning of the flavor molecules between the food product and the surrounding air.2 The partitioning can be expressed as the Henry’s constant (or coefficient), which is the ratio between the concentration of volatile in the headspace and the concentration of the volatile in the solvent/solid phase in equilibrium.3 Ka =
P0 cw
methods are time-consuming. This can be due to an elaborate experimental setup Headspace-GC-MS > 2 hours,3−5 long measurement duration i.e. bubble cell > 80 minutes,6−10 adsorption and desorption of the volatiles to a carrier material,3 or to nonspecific gas chromatographic detection,4,5,11,12 which only allows the measurement of a single molecular species at a time. However, in the field of environmental safety and toxicology, an impressive number of volatiles has been directly or indirectly measured. Here, knowledge of Henry’s constants supports the modeling of the fate of toxic waste streams into the environment, where ambient temperatures are relevant. To address the multitude of available chemicals that have to be tested for their environmental impact, models have been proposed to predict Henry’s constants on the basis of the molecular structure13 or derivation from validated models for vapor pressures and activity coefficients.14 Typically, validation of such predictions with experimental data sets focuses on a single temperature of 25 °C.13,14 When water is used as the solvent, the solubility of volatiles is to a large extent determined by the properties of the solvation shell consisting of coordinated water molecules, which in turn is strongly temperature dependent due to the balance between entropy and hydrogen bonds. These effects mean that the predictions can deviate up to several orders of magnitude from the measured values of the Henry’s constant in aqueous solutions.14 As the models are based on molecules that may have an environmental impact, not many flavor molecules appear in these studies. Moreover, temperature-dependent data are required to properly model processes such as spray-drying, processing at elevated temperatures, or flavor release in the mouth. It is therefore crucial to experimentally determine
(1) 0
Ka, Henry’s constant, is expressed in atm/M, P is the vapor pressure of the pure volatile in atm, and cw is the solubility of the volatile in the solvent in M (mol/L). It is related to the dimensionless Henry’s constant, H, according to H=
chs P 0K = 3 a cw 10 RT
(2)
with chs the concentration of volatile in the headspace in g/L or mol/L, c the concentration in the solvent in, respectively, g/L or mol/L, P0 the standard atmospheric pressure, R the gas constant, and T the temperature. The partitioning not only influences flavor perception during consumption but also affects volatile losses during processing. Despite its clear significance for the development of flavors and food, the literature has a surprisingly limited amount of Henry’s constants for commonly used volatile flavor molecules. One of the reasons for the scarcity of literature values for Henry’s constants of flavor volatiles is that typical experimental © XXXX American Chemical Society
Received: October 8, 2014 Revised: June 18, 2015 Accepted: June 20, 2015
A
DOI: 10.1021/acs.jafc.5b01517 J. Agric. Food Chem. XXXX, XXX, XXX−XXX
Article
Journal of Agricultural and Food Chemistry Table 1. Characteristics of the Studied 2-Ketones and Eight Flavor Moleculesa compound validation: 2-ketones 2-butanone 2-pentanone 2-hexanone 2-heptanone 2-octanone 2-nonanone citrus molecules ethyl acetate trans-2-hexenal ethyl butanoate octanal 1-octanol citral citronellal citronellol
M (g/mol)
ρ(g/mL)
cw (g/L)
P0 (Pa)
supplier
c (mg/m3)
72.11 86.13 100.16 114.19 128.21 142.24
0.806 0.808 0.811 0.819 0.820 0.823
76 21 7.8 2.2 0.88 0.17
1.3 × 104 5.3 × 103 1.8 × 103 655 249 86
Acros Sigma-Aldrich Sigma-Aldrich Alfa Aesar Fluka SAFC
8.06 8.08 8.11 8.19 8.20 8.23
88.12 98.16 116.16 128.24 130.26 152.26 154.28 156.30
0.901 0.847 0.880 0.823 0.826 0.891 0.855 0.858
30 5.3 2.8 0.39 0.81 0.085 0.039 0.11
1.3 × 104 629 1.9 × 103 199 13 12 34 2.3
Firmenich Firmenich Firmenich Firmenich Firmenich Firmenich Firmenich Firmenich
9.01 8.47 8.80 8.23 8.26 8.91 8.55 8.58
The density, ρ, solubility, cw, and vapor pressure, P0, are given at 25 °C. P0 and cw values of the citrus flavor molecules are results from, respectively, the MPBWIN and the WSKOW models of EPI Suite at 25 °C.21 a
Henry’s constants for more complex flavor molecules and their temperature dependence. We have developed a methodology to rapidly measure air− water Henry’s constants as a function of temperature. It is suitable for a wide range of H values and is based on equilibrium static headspace analysis (ESHA). A solution of volatiles is equilibrated with air in a closed vessel, and then a sample of the air is led through a deactivated fused silica capillary and heated to 130 °C (to avoid condensation issues) into the ionization source (atmospheric pressure chemical ionization) of a quadrupole mass spectrometer (APCI-MS),15 which is referred to as an MS-Nose. It has a detection threshold of >∼0.01 mg/m3 of volatiles in the headspace.15 Note that a proton transfer reaction mass spectrometer (PTR-MS) could also be used.16 The mass spectrometer is calibrated with a liquid reference. As discussed previously,17 liquid calibration can be a source of uncertainty, but correct practice gives data comparable to those obtained from the internally calibrated phase ratio variation (PRV) method.17,18 With the PRV method, the depletion from the liquid to the gas phase is explicitly used for different liquid−gas ratios, to calculate the partition coefficient. However, as elegant as this internal calibration is, the latter is suited for only a limited number of compounds and temperatures, as significant partitioning into the air, that is, H > ∼0.1, is required for accurate results. To demonstrate consistency with existing methods, we measure Henry’s constants for a series of 2-ketones, for which reliable values are known.3,19,20 We then apply the method to a set of volatiles regularly used to create citrus flavors (herein referred to as citrus flavor molecules).
■
(CAS Registry No. 5392-40-5), citronellal (CAS Registry No. 238577-5), and citronellol (CAS Registry No. 106-22-9) were all obtained internally at Firmenich S.A. All of the molecules and their charachteristics are listed in Table 1. Values for the solubility, cw, vapor pressure, P0, and Henry’s constant, H, were estimated using, respectively, the MPBWIN (v1.42), WSKOW (v1.41), and HENRYWIN (v3.10) model of the Estimation Program Interface (EPI) Suite.21 The molecules were first dissolved in dimethyl sulfoxide (DMSO; CAS Registry No. 67-68-5, Sigma-Aldrich) and then transferred to water. This two-step procedure accelerates dissolution and improves reliability. To determine if the presence of DMSO in any way perturbed the baseline of the molecules being quantified, a DMSO blank was injected into the MS prior to a standard injection. It was observed that DMSO blank did not affect the MS signal. The final concentrations in the aqueous phase, c, are listed in Table 1. They were always