You are on page 1of 26

Oxygen and Carbon Dioxide Solubility and

Diffusivity in Solid Food Matrices: A Review


of Past and Current Knowledge
Estelle Chaix, Carole Guillaume, and Valérie Guillard

Abstract: Oxygen and carbon dioxide solubility and diffusivity are 2 key parameters to understand gas transfer in food
matrices. Knowledge of these parameters could help to predict gas concentration in modified atmosphere packaging and,
consequently, to predict shelf–life of the product through the development of appropriate mathematical models. The aim
of this review is to present the existing methodologies to quantify O2 and CO2 contents in food, especially in solid food
matrices which is very challenging. There is a focus on how these methodologies could be used to determine gas transfers
kinetics. Data of O2 /CO2 solubilities and diffusivities in food are collected and compared with a specific emphasis on
the food characteristics and factors impacting them. An analysis of the current state of knowledge in solid food matrices
is carried out to tentatively build a general predictive model of the O2 and CO2 solubility and diffusivity extendable to
any kind of food matrix.

Introduction gas exchange between headspace, external atmosphere, and food.


Modified atmosphere packaging (MAP) is used to protect per- Knowledge of overall mass transfers in food/MAP systems and
ishable food and extend its shelf–life (Badran and others 1969). their mathematical modeling is necessary to predict gas concen-
Air in the packaging is replaced by a specific gas or mixture of trations in the headspace and gas gradients within food, and con-
gases (mainly CO2 , O2 , and/or N2 ) that should be close to the sequently, to better comprehend shelf-life of the product (Sandhya
optimal (or recommended) gas concentrations, specific to each 2010). Up to now, the right combination (initial gas content,
food and packaging (Soccol 2003). The objective of MAP is to package surface/volume, mass of food, temperature) and espe-
control the main reactions of food degradation, namely microbial cially the window of packaging permeabilities required that allows
growth, oxidation, or also enzymatic reactions, and thus extending to inhibit/delay microbial growth is principally determined with
the product shelf-life. However, maintaining the appropriate gas the “pack and pray” empirical methodology which is time- and
composition in MAP is not easy since a gradient of gas concen- cost-consuming (Ahvenainen 2003). Even if some modeling ap-
tration is unavoidably created between the external and internal proach has been attempted to take into account the dissolution
atmospheres inducing some mass transfers through the packag- of CO2 within food in predictive microbiology model (for exam-
ing material (that is, permeation). Even when using materials ple, Devlieghere and others 1998), the dynamic of mass transfer
with high barrier properties, which allows the maintenance of through the packaging and within food was never considered.
the original internal atmosphere, gas transfers occur between the However, this is definitively indispensable when, for the same
headspace and the food product (that is, solubilization) and within food product, the packaging materials are changed. For example,
the food (that is, diffusion) leading to changes in headspace partial use of bio materials, and/or lightened material leads to signif-
pressures. The kinetic of this gas dissolution within the food icant change of materials properties and thus permeabilities. In
(which is desired for CO2 to gain the appropriate antimicrobial the field of respiring foods (for example, fruits and vegetables),
activity) strongly depends on the value of diffusivity and the ge- mass balance models have been developed—by coupling math-
ometry of food. If permeation through packaging material could ematical models for gas permeation through the packaging ma-
not be neglected, this diffusivity, in addition with permeation terial (based on Fick’s law) and for gas consumption/production
and solubility, is even more important to predict the dynamic of by the food (based on the Michaelis–Menten-type equation)—
and are already available online such as www.packinmap.com
(Mahajan and others 2007) or www.tailorpack.com (Cagnon
MS 20131480 Submitted 10/17/2013, Accepted 12/20/2013. Authors are with 2012). They can be used to predict either change in gas composi-
UMR 1208 IATE Agropolymers Engineering and Emerging Technologies, Univ. tion of the packaged respiring food or gas permeability values that
Montpellier 2, CIRAD, INRA, Montpellier Supagro, CC 023 Place Eugène Batail-
lon, 34095 Montpellier Cedex 5, France. Direct inquiries to author Guillard (E-mail:
will allow the establishment of the optimal gas concentration. In
guillard@univ-montp2.fr). the field of MAP for nonrespiring food (for example, meat-based
food), such mass balance models, that would take into account


C 2014 Institute of Food Technologists®

doi: 10.1111/1541-4337.12058 Vol. 13, 2014 r Comprehensive Reviews in Food Science and Food Safety 261
O2 and CO2 solubility and diffusivity . . .

both permeation, solubilization, and diffusion, have not yet been tions of O2 and CO2 with an emphasis on factors affecting these
well developed and implemented, despite the increasing interest coefficients and associated mathematical models for prediction
in the development of decision aid tools. This is mainly due to purposes.
the lack of data on solubilization and diffusion for different food
products. Basics of Mass Transfer
Three phenomena govern the gas transfers in the nonrespir- O2 and CO2 diffuse from the surrounding atmosphere to liq-
ing food/MAP system: gas permeation—through the pack- uid and solid foods, where they dissolve. The driving force of
age between the exterior atmosphere and the headspace, gas this phenomenon is the difference in the chemical potential (or
solubilization—between the headspace and the food product, and activity) between the surrounding atmosphere and the medium.
gas diffusion—within the food product; each of them character- O2 and CO2 dissolve in the medium until equilibrium be-
ized by physical constants: permeability, solubility, and diffusivity tween partial pressure in headspace and dissolved concentration
coefficients, respectively. If the knowledge of permeability coeffi- in food is reached. Henry’s law was first demonstrated to de-
cient in classic polymers is well referenced (Massey 2003), data of scribe this relationship in a dilute solution (Henry 1803) and is
O2 and CO2 solubility and diffusivity coefficients in food prod- today extended to any kind of product: at equilibrium and for
ucts are lacking, except in the case of water. As shown in Figure 1, constant temperature and pressure (Eq. 1) the concentration of
there are 70 values for O2 solubility in food/model products (49 dissolved gas A in a solvent (CA ) is proportional to its partial
in liquid, essentially fish and vegetable oils, and 21 in solid matri- pressure (PA ) in the solvent (and consequently in the surrounding
ces) in the available scientific literature (which corresponds to 10 atmosphere)
cited papers on the topic), whereas, as shown in Figure 2 a and b,
more than 115 values are available for CO2 solubility (that is, 11 1
P A = kHA (T) × C A = × CA (1)
cited papers) just in solid food matrices (69 in high–water-content SA
foods and 50 in high–fat foods). Data on CO2 are more common
than for O2 , this might be attributed (i) to the great interest for where kHA (T) is Henry’s law coefficient of the gas A in a solvent
CO2 in MAP because of safety concerns (bacteriostatic activity), at temperature T. Solubility coefficient (SA ) is the inverse func-
whereas O2 in MAP is mainly related to organoleptic changes tion of Henry’s law coefficient, kHA (T) = 1/SA , and expressed in
(but also to inhibit fat oxidations) (Ahvenainen 2003), and (ii) mol.kgfood –1 .Pa–1 (international unit system, but other units can be
the difficulty to control O2 partial pressures (residual O2 , possible found). It represents the maximal quantity of gas A that could be
oxidation, and so on). With regard to diffusivity, very few values dissolved in a solvent, for a given partial pressure of A, and reflects
are available for both the gases (26 values for O2 diffusivity in the compatibility between the gas molecule and the food matrix
food/model matrices, 10 in liquid food and 16 in solid matrices; (Schwartz 2003; Rotabakk and others 2006; Pénicaud and oth-
and 30 for CO2 , essentially in real solid food matrices (25 values) ers 2012). While Henry’s law is widely admitted, its validation is
(Figure 3 and 4). Contrary to solubility, the diffusivity cannot be still mandatory. Up to now, it has been validated for CO2 in water
directly measured from experiments. It is identified from numer- (Carroll and others 1991) and various food products such as cheese
ical treatments of transient kinetic data acquired when the sample and fresh pasta (Fava and Piergiovanni 1992), fish (Sivertsvik and
is subjected to a gradient of gas concentration (or partial pressure). others 2004b), raw meat (Fava and Piergiovanni 1992; Zhao and
Whatever the solubility measurement or diffusivity identification, others 1994), and cooked meat (Sivertsvik and Jensen 2005). In
methodologies for O2 /CO2 assessment and specific set–up for the case of O2 , Henry’s law was validated in water and aqueous
controlling the gas environment are needed but scarcely reviewed, solution (microbial culture media) (Schumpe and others 1982),
especially for solid food. We noted a publication on measurement in various organic liquids (such as acetone or ethanol) (Lühring
of CO2 in liquid or gaseous media (Dixon and Kell 1989a), a and Schumpe 1989), and in soybean oil (Ke and Ackman 1973).
list of diffusion and solubility values for various gases, including Some studies reported a slight deviation from Henry’s law, and
CO2 and O2 , in biological fluids and tissues (Langø and others identified a linear correlation, with a deviation from the ori-
1996), a review on CO2 sensors used in the agri–food industry gin, between partial pressure and CO2 concentration in certain
(Neethirajan and others 2008), a research note on CO2 trans- cheeses (Jakobsen and Jensen 2009), or O2 concentration in veg-
fer in MAP systems and nonrespiring food (Simpson and others etable purees (Samapundo and others 2011). Values of solubility
2009), and a review, published in 2012, which described O2 quan- coefficient encountered in the literature for different food/model
tification methods and their applications to obtain the O2 diffu- matrices are indicated in Figure 1 (for O2 ) and 2a and 2b (for
sivity and solubility coefficient in liquid food product (Pénicaud CO2 ).
and others 2012). Gas diffusivity coefficient is a nonthermodynamic kinetic pa-
In this context, this work proposes an overall presentation of rameter which indicates gas mobility in food matrices. This trans-
methods and/or guidelines devoted to measurements of O2 and/or fer is due to the influence of a gradient of concentration of the
CO2 content in food and/or determination of O2 and/or CO2 considered gas. It is usually described by the well–known Fick’s
solubility and diffusivity in food, and principally in solid food second law (Fick 1855), which connects local changes of cA , vol-
matrices. An emphasis will be put on how these methodolo- ume concentration of the gas A (kg.m–3 ), as function of time
gies could be implemented in experimental setups to monitor (s), and x, the distance between the interface and measurement
O2 and CO2 transfers, and then determine diffusivity coeffi- point (m) with D, the diffusion coefficient (m2 .s–1 ), according
cients, which had never been done before for both gases si- to Eq. 2:
multaneously in a same report. After a brief introduction of ∂cA ∂ 2c A
the basics on gas solubilization and diffusion, this review will =D 2 (2)
focus on gas quantification and solubility/diffusivity determina- ∂t ∂x

262 Comprehensive Reviews in Food Science and Food Safety r Vol. 13, 2014 
C 2014 Institute of Food Technologists®
O2 and CO2 solubility and diffusivity . . .

Figure 1–Oxygen solubility, as a function of temperature, standard deviation, when available, and in brackets methodology used, in various food
matrices. [a] (Battino and Clever 1966); [b] (Ke and Ackman 1973); [c] (Davidson and others 1952); [d] (Battino and others 1968); [e] (Aho and
Wahlroos 1967); [f] (Simpson and others 2001a); [g] (Zaritzky and Bevilacqua 1988); [h] (Fabiano and others 2000); [i] (Samapundo and others
2011); [j] (Sidwell and others 1962). p.v. stands for peroxide values and RH stands for relative humidity.

Considering this kinetic description and unlike to the solubility, is, for example, the one that minimizes the sum of squared er-
the diffusivity cannot be directly measured. For this, a gradient of ror between the 2 datasets. In combination with difficulties to
concentration or partial pressure must be imposed to the sample, set up the experiment, the time–consuming data acquisition and
and the subsequent mass transfer must be measured against time treatment may explain why there are so few diffusivity values avail-
and/or position. The diffusivity is then identified by comparing able in the literature either for O2 (mainly on fruit and vegetable
the theoretical data (predicted by a mathematical model, usually products, Figure 3) or CO2 (presented in Figure 4). The following
the Fickean model) and the experimental data (see section “Exper- presents an overview of the main methodologies used for O2 /CO2
imental set–up of O2 /CO2 diffusivity”). The diffusion coefficient quantification with an attempt, as far as possible, to describe the


C 2014 Institute of Food Technologists® Vol. 13, 2014 r Comprehensive Reviews in Food Science and Food Safety 263
O2 and CO2 solubility and diffusivity . . .

Figure 2–(a) Carbon dioxide solubility, as a function of temperature, standard deviation, when available, and into brackets methodology used, in
various high–water–content food matrices. [a] (Carroll and others 1991); [b] (Geankoplis 1993); [c] (Pauchard and others 1980); [d] (Simpson and
others 2001a); [e] (Fava and Piergiovanni 1992); [f] (Gill 1988); [g] (Sivertsvik and Jensen 2005); [h] (Simpson and others 2009); [i] (Piergiovanni
and Fava 1992); [j] (Rotabakk and others 2010); [k] (Sivertsvik and others 2004b); [l] (Simpson and others 2001b).

pros and cons of each methodology with a special emphasis on Oxygen and Carbon Dioxide Quantification Methods in
the facility of handling and implementation of each set-up, pos- Food Matrices
sibility to carry out direct measurement and feasibility to be used O2 quantification methodologies together with solubil-
in solid food product (which is actually the tricky point). All this ity/diffusivity determination in liquid food/model matrices have
information is summarized for CO2 in Table 1 and 2. The same been recently reported by Pénicaud and others (2012). Several
analysis was already done for O2 by Penicaud and others (2012) methods could be used for determining CO2 concentration in
and was, thus, not repeated in the present work since no update a liquid or gas phase, and they are almost all well described in
was necessary to their work at the time being. the review of Dixon and Kell (1989a) while Neethirajan and

264 Comprehensive Reviews in Food Science and Food Safety r Vol. 13, 2014 
C 2014 Institute of Food Technologists®
O2 and CO2 solubility and diffusivity . . .

Figure 2–(b) Carbon dioxide solubility, as a function of temperature, standard deviation, when available, and into brackets methodology used, in
various high–fat–content food matrices. [a] (Fabiano and others 2000); [b] (Battino and others 1968); [c] (Pauchard and others 1980); [d] (Jakobsen
and Jensen 2009); [e] (Fava and Piergiovanni 1992); [f] (Gill 1988); [g] (Piergiovanni and Fava 1992).

others (2008) listed various types of CO2 sensors used in the cients in solid food will be presented and discussed. This remains
agri–food industry. They pointed out that while several O2 /CO2 the tricky point for mass balance model developments, and will be
quantification methods exist, few directly detect and quantify the further discussed in sections “Experimental Set–Up for O2 /CO2
O2 /CO2 molecule (either partial pressure or concentration). Bias Solubility Determination in Solid Foods” and “Experimental set–
and interferences have to be considered in most of the techniques. up of O2 /CO2 diffusivity,” respectively.
The following section will address a brief recall and update of Quantification methods of O2 /CO2 can be divided into 2
their work (section “Direct measurements of O2 /CO2 ”). The im- main categories: (a) direct measurements that refer to method-
plementation of these methodologies in dedicated experimental ologies that permit to access the molecular O2 /CO2 concen-
setups for assessing both O2 /CO2 solubility and diffusivity coeffi- tration within a given mass of food/model and (b) indirect


C 2014 Institute of Food Technologists® Vol. 13, 2014 r Comprehensive Reviews in Food Science and Food Safety 265
O2 and CO2 solubility and diffusivity . . .

Figure 3–O2 diffusivity values in nonrespiring food or simulants (liquid or solid), as a function of temperature, and some examples of value in fruit.
Adapted and compiled from Pénicaud and others (2010). [a] (Langø and others 1996); [b] (Baird and Davidson 1962); [c] (Barront and others 1993);
[d] (Brown 1955); [e] (Crowe and others 1987); [f] (Davidson and Cullen 1957); [g] (Davidson and others 1952); [h] (Geankoplis 1993); [i] (Grote and
Thews 1962); [j] (Grote 1967); [k] (Lightfoot 1974); [l] (Mannapperuma and others 1991); [m] (Miller and others 2003); [n] (Pénicaud and others
2010); [o] (Sato and Toda 1983); [p] (Schotsmans and others 2004); [q] (Valleguadarrama and others 2005); [r] (Wise and Houghton 1969); [s]
(Zaritzky and Bevilacqua 1988); [t] (Adlercreutz 1986).

measurements that refer to methodologies that permit to ac- measurement of CA in Henry’s law). Few direct methodologies
cess the O2 /CO2 partial pressure or concentration within the are available for quantification of molar quantity of both gases
headspace (or few times the O2 /CO2 partial pressure within the within a matrix: titration or absorbance spectrophotometry after
food/model). chemical reaction, gravimetry, and for CO2 only infra–red mea-
surements.
Direct measurements of O2 /CO2 Chemical reactions. Direct measurements for O2 are known
We call “direct measurement” a method that permits the as colorimetric methods, Winkler test, or gravimetry; neverthe-
molecular quantification of O2 or CO2 in the medium (direct less, the chemical one, the Winkler test, remains the reference to

266 Comprehensive Reviews in Food Science and Food Safety r Vol. 13, 2014 
C 2014 Institute of Food Technologists®
O2 and CO2 solubility and diffusivity . . .

Figure 4–CO2 diffusivity values in nonrespiring food or simulants (liquid or solid), as a function of temperature, and some examples of values in fruit.
[a] (Bertola and others 1990); [b] (Brown 1955); [c] (Carroll and others 1991); [d] (Fabbri and others 2011); [e] (Fabiano and others 2000); [f]
(Geankoplis 1993); [g] (Mannapperuma and others 1991); [h] (Schotsmans and others 2004); [i] (Simpson and others 2001b); [j] (Simpson and others
2001a); [k] (Sirivicha and others 1990); [l] (Sivertsvik and others 2004b); [m] (Sivertsvik and Jensen 2005); [n] (Solomos 1987); [o] (Treybal 1980).

quantify O2 dissolved in water. It exploits the ability of manganese ium dihydroxide (Ba(OH)2 ), KOH, NaOH, or calcium hydroxide
(II) salt to react (a) in a basic medium (presence of hydroxide solu- (Ca(OH)2 ), which is then assayed most of the time by chemical
tion, for example, sodium hydroxide or potassium hydroxide) with titration. One main drawback remains nevertheless unavoidable:
dissolved O2 contained in the sample and (b) to form a brown pre- overestimation of the quantity of CO2 really dissolved in the sam-
cipitate. This further reacts with iodide ions, previously added, in ple due to the leaching of existing carbonate ion (CO3 2– ), bicar-
the presence of acid to produce iodine, which is then quanti- bonate ion (HCO3– ), or carbonic acid (H2 CO3 ) in the sample due
fied by a titrated thiosulfate solution (Battino and Clever 1966). to their unavoidable conversion into CO2 during the acidification
This extraction and titration is difficult because it must be done step. To correct the measured CO2 content by the amount of
without O2 exchange between the sample and the surrounding initial CO3 2– , HCO3– , or H2 CO3 initially present in the sample,
atmosphere. it is necessary to titrate a sample which has not been in contact
In the second example of direct chemical measures, the CO2 is with CO2 before (except the unavoidable contact with a small
trapped in and reacts with a scavenger, mainly liquid–based: bar- amount of atmospheric CO2 ). In the case of use of Ba(OH)2


C 2014 Institute of Food Technologists® Vol. 13, 2014 r Comprehensive Reviews in Food Science and Food Safety 267
O2 and CO2 solubility and diffusivity . . .

Table 1–Different methods used in the literature to obtain CO2 concentration, with emphasis on measurement phase.

Measurement mediuma
Methods Precision Principle Reference for description Gas Liquid Solid
Direct Infra-red FTIR- ATR Sample absorption of (Brantley and others 2000) x x x
spectrometry IR radiation in situ
Raman Sample absorption of (Beutel and Henkel 2011) x x X
radiation and shift
of energy in
function of
vibrational modes
NDIR Desorption of sample, (Watten and others 2004) x
absorption of IR
radiation by CO2
headspace
Gravimetric Weight change by (Dixon and Kell 1989) x x
absorption/release
of CO2
Indirect Gravimetric Piezoelectric Weight change in (Fanget and others 2011) x x x
resonant frequency
due to adsorption of
CO2
Chromatography Katharometer (TCD) Differences in thermal (Girard and Boyaval 1994) x
conductivity
Mass spectroscopy Characterization by (Dixon and Kell 1989) x
the mass to charge
ratios and relative
abundances
Manometric Increase/Decrease (Sivertsvik and others 2004a) x
headspace pressure
above an acidified
sample
Volumetric Increase/Decrease in (Dixon and Kell 1989) x
volume of a gas
sample after
chemical removal of
CO2
Chemical titration A standard alkaline (Gill 1988) x
solution is
neutralized after
absorbing CO2
Amperometric (Clark Type) Electron transfer due (Ishiji and others 1993) x
to reduction of CO2
in aqueous solution
and measure of
electric flow
Potentiometric (Severinghaus type) pH changes in buffer (Puligundla and others 2011) x
solution induced by
CO2
Colorimetric Color change of pH (Dixon and Kell 1989) x x
indicator (solution
or materials)
Fiber optic probes Changes in light (Neethirajan and others 2008) x
transmittance of pH
indicator
Enzymatic Spectrophotometric Absorbance change (Forrester and others 1976) x
by consumption of
NADH to NAD+ by
the enzyme
(function of CO2 )
a Refers to the phase where the CO is quantified.
2

solution, the “aging” between the beginning and the end of the powerful experimental procedure. It permits to measure the O2
experiment induces a change in its molality. To assess the aging of or CO2 solubility, but also the O2 or CO2 diffusivity when the
the Ba(OH)2 solution, it is thus necessary to titrate the Ba(OH)2 mass uptake is monitored. One of the main requirements neces-
solution before and after the experiment. sary for this methodology is the accuracy of the balance. Currently,
Gravimetry. Another methodology used for direct quantifica- some commercial equipment (that is, Cahn microbalance, Hiden
tion of O2 or CO2 inside solid food products (but that could be Isochema gravimetric microbalance, and so on) enables to perform
also used in certain cases for a liquid) is gravimetry. The princi- such an experiment with a sensitivity less than 0.1 μg (Shiflett and
ple of this method is to sorb O2 or CO2 in a matrix of known Yokozeki 2005).
and precise mass. At equilibrium, the sample mass increases due Infra–red measurement. Infra-red measurement does not ap-
to the incorporation of gas and may thus be used to calculate ply for the quantification of the O2 molecule. The principle
the O2 or CO2 solubility. The matrices must be devoid of O2 or of infra–red measurement is based on the energy absorption of
CO2 . That could be achieved during the desorption step under a the CO2 molecule and its asymmetrical stretching, at a known
flux of gas free of O2 or CO2 (dynamic mode) or under vacuum wavelength (2.4 μm) (Neethirajan and others 2008). For mea-
(static mode). The gravimetric method is a simple, direct, and surement of gaseous CO2 , infra–red spectroscopy–based captors

268 Comprehensive Reviews in Food Science and Food Safety r Vol. 13, 2014 
C 2014 Institute of Food Technologists®
O2 and CO2 solubility and diffusivity . . .

are the most commonly used. An alternative method is gas chro- Chromatography measurement. As for O2 , gas chromatogra-
matography, generally coupled with a thermal conductivity de- phy is the most widely used technique in the laboratory to mea-
tector or mass spectrometry (section “Chromatography measure- sure CO2 content in a gaseous phase. The main interest of this
ment”). Both methodologies can permit the quantitative and technique is the possibility to simultaneously measure different
qualitative analysis of gaseous CO2 (% CO2 ). To obtain CO2 gases (for instance, O2 and CO2 ). After separation, molecules are
quantification via the intensity of the stretching band, the infra– detected, generally by katharometer (measure of thermic conduc-
red sensor prerequires calibration which is a time–consuming pro- tivity), but also by flame ionization (FID) or by mass spectrometer,
cess (Yasuda and others 2012) but usually is made by suppliers. the most frequently used. If the gas chromatography apparatus has
Infra-red sensors are conventionally used for gaseous detection a low cost, detectors are more expensive. Furthermore, the analysis
and control in the laboratory and industry because most com- of O2 /CO2 content in samples (liquid or solid) by gas chromatog-
mercial CO2 analyzers are based on this technology; for example, raphy is time–consuming because it requires an extraction step
we can obtain headspace analyzers such as “oxygen and carbon (Ruyssen and others 2013) and periodic calibration (Neethirajan
dioxide headspace gas analyzer” from Illinois Instruments (Johns- and others 2008). As with quantification by infra–red measure-
burg, Illinois, U.S.A.), “CO2 Gas sensor” from Vernier (Beaver- ment, gas chromatography applied to the system food/package is
ton, Oregon, U.S.A.), “CheckMate” from PBI Dansensor (Ring- also an intrusive and destructive measurement, even if volumes of
sted, Denmark), or “Pac Check” series from Mocon (Minneapolis, samples are low (about 10 μL).
Minnesota, U.S.A.). (PBI Dansensor (Ringsted, Denmark) has re- Manometric and volumetric methods. Manometric and volu-
cently been bought by Mocon (Minneapolis, Minnesota, U.S.A.). metric approaches are the oldest methodologies developed to in-
Calibration time is reduced by self–calibration of commercial de- vestigate sorption of gases in solids. The quantification of O2 /CO2
vices, which display good sensitivity and accuracy (from ± 2% to absorbed or desorbed by a sample is obtained from the difference
0.8% for the most precise). Nevertheless, they require a large gas in total pressure (manometric method) or volume (volumetric
sample volume (around 5 to 10 mL) for analysis and interferences method)—at a constant volume (manometric method) or pres-
could appear with water vapor (dilution effect) (Mills and Hodgen sure (volumetric method)—in the gas phase, in contact with the
2005). When used for headspace analysis in a sealed package, the sample, between the beginning and the equilibrium. This implies
measure can be destructive, by the direct insertion of a syringe, or the following main assumptions: the gas follows (a) ideal gas law
not, by using a septum. However, in any case, the method is con- and (b) a compressibility factor equal to 1 (Sivertsvik and oth-
sidered as destructive because of the too large sampling leading to ers 2004a). Moreover, one difficulty is to keep constant the exact
considerable change in the headspace volume. Infra–red headspace known volume of the closed container used in the manometric
analyzers remain the cheapest gas analyzers compared to other sys- method or the pressure in case of the volumetric method. Both
tems such as gas chromatography, gravimetry techniques, and so volumetric and manometric methods are approved by the AACC
on. Intl. Organization for Standardization Measurement (respectively,
Fourier transform infra–red–attenuated total reflection (FTIR– AACC Method 22–11.01 and 22–14.01) with the advantages that
ATR) spectroscopy is a powerful tool for studying molecules ab- they (a) are easy to implement and (b) do not require sophisti-
sorbed into solid matrices. The spectrum of dissolved CO2 is well cated high–tech equipment, especially for the volumetric method
characterized in mid–infrared and can be identified by the ab- (Keller and Staudt 2005). However, possible bias can occur in both
sorbance in the spectral region 2300 to 2350 cm–1 . It is widely cases from (a) an inaccurate evaluation of the container volume
used in the field of polymer and membrane science to measure or pressure (b) an unwanted release of volatile molecule from the
CO2 transfer within thin films or plane sheets (Pasquali and others sample, and (c) uncontrolled leaks of the container (Dixon and Kell
2008; Karimi 2011). The solid sample is in contact with the ATR 1989b). Moreover, we noted that it is necessary to have a signif-
crystal and must be homogeneous and flat, so that the “thick film icant pressure or volume modification to perform some accurate
approximation” theory can be applied (Ferrari 2009). This type solubility calculation. Thus the volume of the container used for
of spectrometer offers precise detection of CO2 at ppb levels, and this experiment, as well as the mass of the sample, must be adapted
the possibility to measure multiple molecules simultaneously (Mc- according to the expected amount of dissolved O2 /CO2 in the
Donagh and others 2008). Furthermore, the monitoring of CO2 assay.
sorption kinetics as a function of time in the sample can be easily pH–based sensor. Among the main indirect measurements for
performed and thus determination of diffusivity could be achieved CO2 , pH–based electrodes are one of the most current. The first
(Karimi 2011). However, as for all spectroscopic methodologies, CO2 sensor based on potentiometric measurement was developed
FTIR–ATR should be first calibrated using another technique in by Severinghaus and Bradley in 1958. This electrode was initially
order to gain some quantitative values of dissolved CO2 . Another used to measure dissolved CO2 in blood. The measuring principle
problem occurs in view of CO2 dissociation in aqueous media into is based on pH–value modification of a buffer solution trapped in
different carbonate species, and the molecules associated with CO2 the probe, following CO2 dissolution. Dissolved CO2 is hydrated
becoming visible at different wavelengths (Wijnja and Schulthess into carbonic acid, bicarbonate ion, and carbonate ion, with lib-
1999). eration of protons, which are detected by the probe (Eq. 3).

Indirect measurements CO2 + H2 O ↔ H2 CO3 ↔ H + + HCO3− ↔ 2H + + CO32−


For the measurement of gaseous CO2 , infra–red spectroscopy– (3)
based captors are the most commonly used. Alternative methods Severinghaus–type probes were then miniaturized: from 1 to
for both O2 /CO2 content are (a) gas chromatography that needs 1.5 mm of diameter in 1974 (Caflisch and Carter 1974) to be-
an extraction of the sample and a gaseous calibration to quantify, tween 50 and 300 μm of diameter (Irimia-Vladu 2006). Their
(b) manometric and volumetric methods. Both methodologies can main advantages are simplicity of handling and good detection
allow the quantitative and qualitative analysis of gaseous O2 /CO2 limit (around 10 μmol) (Janata 2009). Drawbacks are, first, a
(% O2 /% CO2 ). slow response time (time-lag due to CO2 diffusion in the probe),


C 2014 Institute of Food Technologists® Vol. 13, 2014 r Comprehensive Reviews in Food Science and Food Safety 269
Table 2–Main advantages and drawbacks of methodologies commonly used to quantify CO2, with emphasis on the facility of their use for solid food products (“D” stands Direct and “I” stand Indirect).

Methods Precision Cost Advantages Drawbacks Example of matrix studied


Food (if available), italic (if nonfood)

D Infra-red spectrometry FTIR- ATR ++ – Possibility of monitoring the kinetic of – Necessity to have an intimate contact Olive oil (Hennessy and others
adsorption between sample and crystal (for FTIR- 2009)
– Possibility to measure multiple molecules ATR technique)
Raman simultaneously – Calibration required Sea water (Dunk and others 2005)
O2 and CO2 solubility and diffusivity . . .

+++
– Difficulty to analyze the spectral signature
of CO2 and its dissociated species (visible
at different wavelengths)
NDIR + – The cheapest gas analyzers – Large gas sample volume required Cheese (Vivier and others 1996)
– Easy to handle (for commercial apparatus) – Possible interferences with water vapor Fresh scallops (Simpson and others 2007)
– Fast methodology – Difficult implementation for measuring Water, Cheese, Whey, (Pauchard and others 1980)
CO2 in solid food Butterfat, Olive oil
D Gravimetric +++ – Simple, easy to adapt for solid – Cost Olive oil (Yokozeki and Shiflett 2011)
food – Lack of molecular specificity
I Gravimetric Piezoelectric +++ – Possibility of monitoring the kinetic of Polymers (Aubert 1998)
adsorption

270 Comprehensive Reviews in Food Science and Food Safety r Vol. 13, 2014
I Chromatography Katharometer (TCD) ++ – Common technique in laboratory for gas – Time-consuming (due to the desorption step Cheese (Girard and Boyaval 1994)
phase analysis that could be long) Tomato (Bertola and others 1990)
– Accurate – Difficult implementation for measuring CO2 Potatoes (Sirivicha and others 1990)
– Low gas volume required for measurement in solid food.
Mass Spectroscopy +++ Cheese (Tammam and others 2001)
I Manometric + to ++ – Reference method (AACC Method 22 14.01) –Difficulty to keep constant the exact known Cooked meat product (Sivertsvik and others 2004b)
– Easy to implement volume of the closed container,
– Possible bias can occur due to lack of
inaccuracy in container volume evaluation,
– Unwanted release of volatile molecule and Raw meats, Cheeses, (Fava and Piergiovanni 1992)
uncontrolled leaks of the container Pastas, Cooked cold


– Necessity to have a significant pressure meats
modification to perform some accurate
solubility calculation Pesto sauce (Fabiano and others 2000)

C 2014 Institute of Food Technologists®



Table 2–Continued

Methods Precision Cost Advantages Drawbacks Example of matrix studied


Food (if available), italic (if nonfood)
I Volumetric + to ++ – Reference method (AACC Method 22 11.01) – Possible bias can occur due to lack of Cod (Rotabakk and others 2007)
– Easy to implement inaccuracy in container pressure
– Could not require sophisticated high tech measurement

C 2014 Institute of Food Technologists®


equipment – Unwanted release of volatile molecule and
uncontrolled leaks of the container
– Necessity to have significant volume
modification to perform some accurate
O2 and CO2 solubility and diffusivity . . .

solubility calculation
Chicken fillets (Rotabakk and others 2010)
I Chemical titration – – Reference method –Possible overestimation of the quantity of CO2 Meat/Cheese/Egg (Gill 1988)
– Easy to handle due to leaching of carbonates from the food
– The less expensive methodology product: necessity to realize a blank sample
– Easy to adapt for measuring CO2 in solid food – Labor cost
– Time-consuming
Cheese (Jakobsen and Bertelsen 2004)
Egg (Keener and others 2001)
I Amperometric (Clark Type) – /+ – One of the more current indirect measurement – Slow response time (time-lag due to CO2 Blood plasma (Fasching and others 2003)
for CO2 in liquid phase diffusion in the probe)
– Miniaturized system possible – Invasive method (problem for bulky probe)
(micro-electrodes) – CO2 is not directly detected but its ionic forms
indirectly
I Potentiometric (Severinghaus type) – /+ – Simplicity of handling, good detection limit – Lack of stability and shorter life for Fish fillet (Wang and Ogrydziak 1986)
(around 10 μmol) micro-electrodes compare to
macro-electrodes
– disruptions due to other dissolved species that
induce pH change
I Colorimetric – to + – Cheap method – Not suitable for continuous measurement Soils (Rowell 1995)
– Could not require sophisticated high tech – Time consuming
equipment – Calibration curve
I Fiber optic probes – /+ – Nondestructive, noninvasive – Relies on pH change: CO2 is not directly Pears (Ho and others 2007)
– Better sensitivity to dissolved CO2 than detected and disruptions due to other
ph-based electrode (Severinghaus type) dissolved species that induce pH change
could occur
I Enzymatic Spectro-photometric – to + – Easy to handle (for commercial apparatus) – Could be time-consuming (biological Cheese (Crow and Martley 1991)
– Could not require sophisticated high tech molecule)
equipment – Unsuitable for continuous use
– Specific for CO2 – Calibration curve

Vol. 13, 2014 r Comprehensive Reviews in Food Science and Food Safety 271
O2 and CO2 solubility and diffusivity . . .

second, the fact that it is an invasive method (bulky probe), and, than 5) in which the main form of carbon dioxide is gaseous CO2
then, CO2 is not directly detected but its ionic forms indirectly. (Figure 5). This system was compared with the Severinghaus–type
Some micro–electrodes have been developed to overcome the sensor, and it displayed better sensitivity to dissolved CO2 in water
main drawback of their size, but they are unstable and have a (Long and others 2012). A major constraint of these O2 /CO2 sen-
shorter life compared to macro–electrodes (Zhao and Cai 1997). sors is that the spot should be placed inside a transparent container.
Commercial Severinghaus–type electrodes exist, for example, “V– Measurement in the gas phase is possible, but the sensor needs a
SignTM Sensor” by Artemis (Kent, United Kingdom) for partial sufficient relative humidity for the possible pH variation detection
pressure CO2 measurement in medical application or “Sonde CO2 by the sensor spot. We may suppose that O2 /CO2 partial pres-
InPro 5000” by Mettler Toledo (Paris, France) design for biotech- sure measurement in high moisture solid matrices could also be
nological and fermentation controls. possible.
Generally, the measure of CO2 content with a pH–measurement
system is affected by other dissolved species that induce pH
changes, for example, volatile acid or basic gases, modification Experimental Set–Up for O2 /CO2 Solubility Determi-
of acid present in the medium with addition of CO2 , or im- nation in Solid Foods
pact of salt conditions (osmotic pressure), but also by biologi- Determination of O2 /CO2 solubility requires implementation
cal processes that may lead to changes in CO2 content and/or of an experimental set–up to (a) bring the sample into equilib-
pH (acid production by microorganisms, proteolysis, and others) rium with an atmosphere of controlled O2 /CO2 partial pressure
(Mills and Hodgen 2005; Neethirajan and others 2008; Oludare and (b) assay the concentration of dissolved gas into the sample.
and others 2009). To avoid pH interference, recent developments Henry’s law is accepted for both gas solubility determinations;
on optical–based CO2 –sensors tend to directly quantify CO2 or the experimental set–up must allow obtaining headspace partial
its ionic forms in a medium instead of protons (section “Optical pressure and concentration of dissolved gas into the solid food. It
sensors”). is worth noting that implementation for O2 solubility assessment
Optical sensors. Recent O2 /CO2 sensors based on optical mea- is more difficult than for CO2 , because of the risk of “O2 sam-
surements allow nondestructive measurements and, in some cases, ple contamination” by air during the assay: 20.358 ± 1.197% for
also noninvasive ones. A variety of optical sensors exist to detect O2 and 0.793 ± 0.680% for CO2 in Europe (Fonseca and others
CO2 or ionic forms, exploiting different detection methodologies, 2000). The difficulty of measuring dissolved O2 and monitor-
such as colorimetric, infra–red, fluorescence, or pH–based ones ing O2 transfer in food without contamination from atmospheric
(Burke and others 2006). For example, in this last case, a sens- O2 has prevented the implementation of routine/standard meth-
ing element (pH–sensitive fluorescent indicator or organometal- ods. The following section will focus on the use of the above–
lic complexes) reacts with gaseous CO2 and/or its ionic forms, mentioned methodologies (section “Oxygen and Carbon Diox-
which induces a modification of its absorbance or reflectance. This ide Quantification Methods in Food Matrices”) in experimental
change is quantified by an amber light–emitting diode, which re- set–up for gas quantification, and mainly the interest of the indi-
lates modification of the emitted signal to the CO2 concentration rect ones to access O2 /CO2 solubility in solid food, and also the
(Neethirajan and others 2008). Optical O2 /CO2 sensors are still presentation of destructive and nondestructive food set–up.
essentially lab–made, but some commercial products are appear-
ing such as for CO2 “FCO2 –R,” a nondestructive probe (Ocean Equilibrium set–up
Optics, Duiven, The Netherlands) and “CO2 sensors,” nonde- As mentioned in section “Basics of Mass Transfer,” to obtain
structive and noninvasive (PreSens, Regensburg, Germany), or the solubility value, the headspace/food system must be at equi-
for O2 “Optical Oxygen Analyzer,” noninvasive (Oxysense, Dal- librium. That implies there is no gaseous exchange between the
las, Texas, U.S.A.) and “Non-Invasive Oxygen Sensors” (PreSens, headspace (partial pressure is constant) and the matrices (intrinsic
Regensburg, Germany). This latter (a small dot of 5 mm diam- concentration is also constant). The experimental set–up must al-
eter) is based on luminescence quenching and up to now ap- low reaching a steady state, while permitting the quantification of
pears to be the most promising method to obtain a measure of both partial pressure and concentration of gases at the equilibrium
CO2 partial pressure in situ for liquid and viscous food matri- (the end of the experiment). From a general point of view, we
ces. This luminescence methodology relies on the permeation of could divide the experimental set–up into 2 categories: (a) static
CO2 molecules through a membrane protecting a gel with a lu- methods, in which gaseous mass balance of headspace/food sys-
minescence complex. It particularly relies on the gel–induced pH tem is maintained constant over time and (b) dynamic methods,
variation combined with Förster resonance energy transfer. More where a gas flow rate is imposed in the headspace, and so the
precisely, CO2 molecules are trapped into the gel phase contain- mass–balance increases with time.
ing 2 dyes (ruthenium(II) polypyridyl complexes and a colorimet- On the one hand, static (or batch) methods are easier to imple-
ric indicator, m–cresolpurple) and decrease the pH value of this ment: for example, a food sample which was sealed in a container,
medium. This variation induces a change in the proton transfer, with gaseous concentration, volume and/or pressure of headspace
between the ruthenium (II) polypyridil complex (proton donor) could be measured without opening it. To obtain equilibrium, gas
to the gel–immobilized colorimetric indicator (proton acceptor). transfer occurs between the headspace and food, until the latter is
This transfer induces a spectral shift of the colorimetric indicator saturated: concentrations in food increase while gas partial pressure
with a concomitant change of the proton donor, and also a vari- of headspace decreases with time. In such conditions, final partial
ation of Förster resonance energy transfers. This variation can be pressure is not user–defined and depends on the gas solubility in
correlated to the amount of CO2 in the gel medium (Liebsch and food.
others 2000). It should be pointed out that the membrane is not On the other hand, dynamic experiments enable to force the
permeable to CO2 ionic forms (for example, HCO3 – ) but only to equilibrium with a continuous flow. This means the gas concen-
CO2 gaseous molecules. Consequently, to obtain CO2 solubility, tration in the food increases while gas partial pressure or concen-
this sensor can only be used for acid products (pH–value lower tration in the headspace is maintained constant and equal to that

272 Comprehensive Reviews in Food Science and Food Safety r Vol. 13, 2014 
C 2014 Institute of Food Technologists®
O2 and CO2 solubility and diffusivity . . .

Figure 5–Fraction of dissolved carbon dioxide, carbonic acid, bicarbonate, and carbonate in water, as a function of pH (adapted from Daniels and
others 1985).

which the user has defined. This experimental set–up guarantees O2 /CO2 partial pressure in the gas phase in contact with the
the saturation of the food product at the initial partial pressure. product after a desorption step to extract O2 /CO2 content. How-
Both are used in the literature. For example, Fabiano and others ever, implementation of the set–up with direct measurements of
(2000) used a batch set–up that implies the samples were placed O2 /CO2 content is possible with the gravimetry method.
inside a closed chamber at a constant temperature filled with a Solubility measurement based on a direct quantification of
gas mixture, whereas Ke and Ackman (1973) used a flow of air to gases. Gaseous solubility with gravimetric apparatus is easy to
saturate oils with O2 . handle. The food sample is enclosed in a hermetic chamber, on
a support, that relies on high–precision mass balance. It is nec-
Solubility acquisition for solid food essary to totally desorb the O2 /CO2 initially contained in the
The second step after equilibrium is the quantification of sample (under vacuum or by nitrogen flow); and then the sam-
O2 /CO2 dissolved into food using a direct assay of the food or an ple is submitted to O2 or CO2 flow until equilibrium (constant
indirect measurement by measuring the O2 /CO2 partial pressure mass).The perfectly controlled conditions of humidity, tempera-
in the headspace that is in equilibrium with the food. ture, and O2 /CO2 partial pressure in the surrounding medium of
Among the list of O2 quantification methods available and listed the product is required to obtain an accurate measurement. Gen-
above only a few were implemented in an experimental set–up erally, the use of the gravimetry method to measure O2 variation
used to obtain O2 solubility in solid food (Figure 1): lumines- in food products has been made to assess the antioxidant capac-
cence detection was used in a viscous matrix (vegetable puree) ity of oils (Pénicaud and others 2010). Most of the references of
(Samapundo and others 2011), a manometric set–up was used gravimetric measurements for CO2 solubility are for membranes
with lard (Davidson and others 1952) and a polarographic method which are either bio–based, such as wheat gluten films (Pochat-
with freeze–dried food (Sidwell and others 1962). All of the O2 Bohatier and others 2006) or synthetics (Aubert 1998; Wong and
solubility methods for solid food are indirect measurements. others 1998).1 Indeed, among the 120 values of CO2 solubility
As regards CO2 , for solid matrices only indirect measurements measured in common synthetic polymers listed in the review of
of CO2 were implemented in an experimental set-up for solubil- Tomasko and others (2003), the majority of data (around 100
ity measurement. Among all methodologies listed in Table 1 for values) were obtained by gravimetry, confirming the prevalence
quantifying CO2 , most of them have been used in a dedicated of this methodology in the fields of polymer studies and mem-
experimental set-up to quantify CO2 in solid food. However, the brane science. Few measurements have been carried out on food
complexity of their implementation for CO2 quantification in matrices, for example, on CO2 solubility in olive oil (Yokozeki
solid food depends on the method as shown in Table 1. Examples and Shiflett 2011). Some bias could be introduced due to the
of solid food or nonfood matrices investigated are given in the last
column of Table 1. 1 Aubert (1998) uses a specific gravimetric/piezoelectric method, with quartz
Whatever the gas considered, for solid matrices, indirect meth- crystal microbalance (Phelps Electronics) at 40◦ C, in 10 different polymers;
ods are the most commonly used and, in this case, measurement Wong and others (1998) used a Cahn electronic microbalance at temperatures
of O2 /CO2 is generally achieved via the quantification of the ranging from 120 to 300◦ C in 3 different polymers.


C 2014 Institute of Food Technologists® Vol. 13, 2014 r Comprehensive Reviews in Food Science and Food Safety 273
O2 and CO2 solubility and diffusivity . . .

loss of volatiles by the sample (if the desorption step is made by tained in cheese after release of gas. However, like the chromato-
using vacuum) and/or due to buoyancy. Therefore, some correc- graphic method, an infra–red set–up could be used to measure
tions should be performed. Moreover, it appears very difficult to gas absorption or release (Watten and others 2004). Both sys-
manage O2 /CO2 sorption measurement in wet samples because tems need a gaseous volume to quantify O2 /CO2 concentration,
controlling humidity usually generates some variations in the mass and monitoring sorption or desorption could imply extraction of
of the sample that add too much noise to the O2 /CO2 sorption volatiles.
kinetic leading to too high uncertainty on the measured solubility, Manometric/volumetric method. The manometric method has
especially when materials sorb low amounts of O2 /CO2 . These been used by several researchers (Fava and Piergiovanni 1992;
main drawbacks explain why the gravimetric method is widely Piergiovanni and Fava 1992) who measured the solubility CO2 in
used in the study of solubility in polymers and not yet in food about 20 food matrices such as raw and cooked meats, cheese, and
matrices. pasta (Figure 2a and b). More recently, using the same method-
Few direct methods exist to quantify O2 /CO2 (section “Direct ology, Fabiano and others (2000) analyzed the CO2 content in
measurements of O2 /CO2 ”), and more difficulties exist to im- pesto sauce. Sivertsvik and co–authors measured CO2 solubility
plement them to obtain O2 /CO2 solubility in solid food, mainly in 5 different fishes (Sivertsvik and others 2004b) and meat (ham
due to the nonhomogeneity of solid matrices. Experimental set– and 2 types of sausage) (Sivertsvik and Jensen 2005). For O2 solu-
up essentially used indirect methodologies to quantify O2 /CO2 bility, this determination in olive oil used the manometric method
solubility. (Davidson and others 1952; Rodnight 1954). It seems to be a
Solubility measurement based on indirect quantification of bit trickier to monitor volume change while maintaining con-
gases. In food fields, the experimental set–up generally used to stant pressure than to detect pressure change while maintaining
assess and calculate the solubility value at partial pressure is the constant volume in a rigid container. Indeed, just 1 volumetric
static method (section “Equilibrium set–up”). This is typically determination was found (Rotabakk and others 2007) and the
used by indirect methodologies to obtain solubility, through the authors reported CO2 solubility in chicken by measuring the vol-
monitoring of changes in pressure or volume of the headspace ume change of flexible pouches (Figure 2a). Gaseous release or
(that is, the manometric method by Sivertsvik and others (2004a)) absorption could be measured.
or partial pressure directly with noninvasive measurement (that is, Chemical titration. After gaseous absorption, the O2 /CO2 des-
Li and others 2008). Among all the data found in the literature, orbed from the food matrices by acidic solution is trapped in and
some of them presented solubility values that measured using de- reacted with a scavenger (mainly liquid–based), which is then
vices such as the Clark electrode or O2 /CO2 luminescence–based assayed, most of the time by chemical titration. With the mano-
sensors immersed in the liquid. Knowing that these devices mea- metric methods, chemical titration is one of the most preferred
sure only partial pressure and that their internal calibration is based methodologies to quantify CO2 in solid food: see, for example,
on Henry’s coefficient for water, the quantification done was not the work of Gill (1988) and Jakobsen and others (2004, 2009).
strictly speaking a solubility value but was equivalent to a measure Chemical O2 quantification by the Winkler test is possible in
of solubility in water! aqueous solution, but with solid food it is difficult, essentially due
Quantification methods commonly used to assess solubility in to multiple components that can react with the chemical additive
solid food are (a) chromatography–based and infrared–based ones used for the assay (that is, iodine).
to assess changes in partial pressure/concentration of the target gas Among all existing techniques (Table 1), chemical titration of
(O2 /CO2 ) in the headspace, (b) manometric or volumetric meth- CO2 scavengers appears currently the most versatile and low–cost
ods to assess change in total pressure or volume in the headspace. one to determine CO2 solubility for any kind of product (liquid
One additional indirect methodology for solid food products is (c) or solid); but does not permit on–line measurements because it is
a quantification of gas after entrapment by a scavenger, generally a destructive, with a step of desorbing and trapping CO2 that could
chemical assay and considered as a direct method for liquid, namely exceed 24 hours depending on the nature of the sample (liquid or
chemical titration for CO2 (Gill 1988) or polarographic method solid).
for O2 (Sidwell and others 1962). In this latter case, assessment Different scavengers can be used: Ba(OH)2 for solids, such
could be done directly in the scavenging phase using a pH–based as for meat (Gill 1988) or KOH and NaOH for liquids, such
electrode. For example, Wang and Ogrydziak (1986) extracted as for wine, AOAC 28.1.46 (Horwitz and Latimer 2005). In
CO2 from fish flesh after blending of the sample in an acidic buffer soil science, NaOH and Ca(OH)2 are commonly used to mea-
solution, and the measured CO2 directly in the liquid phase using sure CO2 production by solid matrices, according to a soda–
a pHelectrode. Other indirect methods are referenced for CO2 lime method (Keith and Wong 2006). The author both used
in Table 1; nevertheless, the most used methodologies are among Ba(OH)2 solution (in excess) which reacts with gaseous CO2 to
the above–mentioned list. For O2 , the use of (d) luminescence– form barium carbonate (BaCO3 ), a solid compound. The CO2 is
based methods is a promising way to obtain data of O2 being consumed by Ba(OH)2 , the overall equilibrium of the re-
solubility. action encourages the release of CO2 content into the matrices
Chromatography/infrared headspace method. Chromatogra- until no CO2 remains in the sample. The residual quantity of
phy and infra–red measurement could both be used to mea- Ba(OH)2 is then titrated against a standard hydrochloric acid so-
sure a variation of gaseous concentration of headspace at the lution (HCl). Ba(OH)2 solution (in excess) reacts with gaseous
gaseous solubility determination in food. The set–up is easy to CO2 to form BaCO3 . An example of a possible experimen-
handle, it could be, for example, a hermetic jar with volume tal set–up for performing such experiment is given in Figure 6.
and pressure controlled such as that used with the respiration Furthermore, the accuracy of titrimetric methods is not well as-
measurement of vegetables (Guillaume and others 2010). The sessed. Some authors criticized its inaccuracy (coefficient of vari-
major use of chromatography was a quantification of O2 /CO2 ation around 10%) (Girard and Boyaval 1994), whereas other au-
extracted from saturated food matrices, such as Girard and Boy- thors disagreed (coefficient of variation around 4%) (Keener and
aval (1994) and Pauchard and others (1980) measured CO2 con- others 2001).

274 Comprehensive Reviews in Food Science and Food Safety r Vol. 13, 2014 
C 2014 Institute of Food Technologists®
O2 and CO2 solubility and diffusivity . . .

Figure 6–Example of experimental set-up to measure carbon dioxide content in liquid or solid food, and associated reactions (adapted from Gill 1988).

Oxygen luminescence sensors. According to the review of gineering for the measurement of O2 /CO2 solubility and further
Pénicaud and others (2012), the use of indirect luminescence– diffusivity.
based methods for measuring O2 content and O2 transfer in solid Finally, we note that the gaseous sorption or desorption in ma-
food matrices is the most promising way to obtain data of O2 trices of food may last for minutes, hours, or days (in the case of
solubility and diffusivity coefficients. They permitted the devel- a big sample for example) and it is possible that thermodynamic
opment of nondestructive methods which have been used with equilibrium is not reached in the time scale of food shelf–life
various matrices (foods, polymers, human tissues, and so on) for (Keller and Staudt 2005). The implementation of methodologies
monitoring O2 profiles in the headspace of packed food (Smiddy to obtain these data must take into account such slow kinetics
and others 2002a, b; O’Mahony and others 2006; Hempel and and the biases that may appear during the time of the experiment
others 2013), for determining O2 permeability of polymers and (discussed later on in sections “Interfering reactions” and “Fac-
packaging materials (Siró and others 2010; Diéval and others tors Impacting and Predictive Modeling of Oxygen and Carbon
2011), for in situ measuring O2 level and diffusion of O2 in hu- Dioxide Solubility”).
man cellular tissues (Nock and others 2009; Nock and Blaikie
2010), for characterization of O2 sorption in film polymer (Li and Typical values of O2 /CO2 solubility in solid food
others 2008, 2011), and for assessment of O2 dissolved in foods The main difficulty with comparing O2 /CO2 solubility values
(Smiddy and others 2002b; Nevares and others 2010; Pénicaud in food is the variety of units that are found in the literature. For
and others 2010). However, the main bottleneck in measuring O2 example, some publications used the Ostwald coefficient (ratio of
solubility is the almost inevitable O2 consumption by the food the volume of gas absorbed per unit volume of the solvent) or the
that can occur due to (a) microorganism respiration, (b) chemical Bunsen coefficient (volume of gas reduced to 0 ◦ C and to 1 atm ab-
reactions such as oxidation, and/or (c) metabolism of the prod- sorbed by unit volume of solvent at a gas partial pressure of 1 atm).
uct itself (in the case of respiring products). The experimental To employ units from the Intl. System of Units (mol.kg–1 .Pa–1 ),
set–ups used for measuring O2 solubility and diffusivity should assumptions must be made, essentially on the density of the fluids
take into consideration these constraints (section “Interfering (liquid or gas phases). Sometimes, gaseous partial pressure imposed
reactions”). at the sample is not indicated and, according to the methodological
Implementation of the aforementioned method is easier in liq- details, we assume generally an atmospheric concentration for O2 ,
uids than in solids. To date, only 2 values of O2 solubility in or 100% of CO2 . These exhaustive databases for solid products
“dense” matrices with luminescence probe are known from Sama- (Figure 1, 2a, and 2b, but also 3 and 4) could allow comparing all
pundo and others (2011) on potato and potato–spinach puree of the values existing in the literature, and also methodologies.
(Figure 1). However, these values are questionable because this Analysis of the data in the literature (Figure 1) shows that O2 sol-
methodology enables to quantify O2 partial pressure in media, not ubility varied between 8.8×10–10 mol.kg–1 .Pa–1 (obtained for pre-
a concentration, even if a conversion is made with O2 solubility cooked freeze–dried chicken at 20 ◦ C (Sidwell and others 1962))
in water. and 1.1×10–7 mol.kg–1 .Pa–1 (for sunflower oil at 20 ◦ C). It de-
The listing of the different methodologies currently used for pends on the temperature (for example, for water, at 0 ◦ C; sol-
assessing dissolved O2 /CO2 concentration in solid matrices re- ubility is 2.3×10–8 mol.kg–1 .Pa–1 while it decreases to 9.8×10–9
veals a strong lack of nondestructive and/or noninvasive O2 /CO2 mol.kg–1 .Pa–1 at 50 ◦ C (Battino and Clever 1966)); but also on
methods (Table 1). Development of sensors able to react with the composition of the product: at 20 ◦ C, O2 solubility in wa-
gaseous O2 /CO2 and/or dissolved O2 /CO2 would be essential ter is 1.4×10–8 mol.kg–1 .Pa–1 (Battino and Clever 1966), while it
to achieve a significant breakthrough in the domain of food en- is 5.3×10–8 mol.kg–1 .Pa–1 in herring oil (Ke and Ackman 1973),


C 2014 Institute of Food Technologists® Vol. 13, 2014 r Comprehensive Reviews in Food Science and Food Safety 275
O2 and CO2 solubility and diffusivity . . .

or 5.4×10–9 mol.kg–1 .Pa–1 in freeze–dried chicken, uncooked, 0% piration of microorganisms, when fresh fruits and vegetables are
relative humidity (Sidwell and others 1962). Furthermore, we note contaminated (which is commonly the case for those produced in
O2 was found to be more soluble in olive oil and sunflower oil than fields). Then to obtain true values of O2 /CO2 solubility in fruits
in water, respectively, 5.4 and 7.9 times more soluble at 20 ◦ C, this and vegetables, respiratory activity in both fresh food and microor-
can be explained with the lower affinity of oxygen molecules for ganisms must be stopped. This could be done by using preserva-
polar products (water) than for nonpolar products (oils) (Pénicaud tion treatments commonly encountered in various food industries
and others 2010). (for example, thermal, alcohol conservation) and sanitation pro-
As already mentioned, CO2 solubility has been more studied cedure (UV exposition or chlorine treatment) respectively. But,
than O2 and, consequently, more data are available in the lit- preservation treatments would certainly lead to textural/chemical
erature. We separated the data collected in 2 figures: Figure 2a changes in the core of the food and consequently modify the prod-
representing CO2 solubility in high–water content food, and Fig- uct. As a consequence they are never applied to fresh fruits and
ure 2b in high–fat-content food. In all of there solid matrices, vegetables and it is still a tricky task to manage product respira-
CO2 solubility was found to vary between 4.9×10–8 for grated tion while measuring O2 /CO2 solubilities. Regarding sanitation,
parmesan cheese at 7 ◦ C to 7.8×10–7 mol.kg–1 .Pa–1 for a snel- textural and chemical changes might also occur but remain at the
lina cheese at 7 ◦ C (Fava and Piergiovanni 1992). Analysis of data surface of the food; then such procedures are commonly applied
gathered in Figure 2a and b did not reveal any impact of the na- prior to the determination of respiration in fresh fruits and veg-
ture of the food on this coefficient. For example, CO2 solubility etables. Despite the difficulty to directly measure gas solubility
varies from 1.8×10–7 to 3.7×10–7 mol.kg–1 .Pa–1 for a pasta type (and diffusivity) in such products, their estimation would be pos-
product at 7 ◦ C (Fava and Piergiovanni 1992), from 2.5×10–7 sible from global respiration measurement, if cell respiration and
to 5.5×10–7 mol.kg–1 .Pa–1 for a meat product with temperature skin resistance were accurately characterized and a mass balance
4.9×10–8 to 3.6×10–7 mol.kg–1 .Pa–1 for cheese products (Fava formalism established for gas exchange in the fresh and sanitized
and Piergiovanni 1992; Jakobsen and Jensen 2009). However, as produce (cell respiration, skin resistance to gas transfer, gas solu-
for O2 , CO2 solubility was found higher in fat products than in bility, and gas diffusivity) (Fonseca and others 2002). Respiration
aqueous ones: at 22 ◦ C, CO2 solubility was found, respectively, also exists in “nonrespiring” products, but not as pronounced as in
1.6 and 1.8 times more soluble in olive oil and grape seed oil fresh fruits and vegetables. It relies on the cell activity (mitochon-
than in water (Pauchard and others 1980). We note that, in water, drial consumption of O2 ) in postmortem muscle (Zhao and others
values of CO2 solubility are around 25 times higher than that of 1994). This respiration rate of meat is turned into an asset in the
O2 at 25 ◦ C (Carroll and others 1991; Geankoplis 1993); and 10 case of vacuum–packing where residual O2 is then consumed by
times higher at 25 ◦ C in fat products (olive oil), which means that the product, and it eventually increases the CO2 concentration in-
the solubilization of CO2 would be more visible than that of O2 . side, a function of the choice of permeability properties of the film
(Blakistone 1999). However, the amount of O2 consumption and
Interfering reactions CO2 production by meat respiration is neither well known nor
In addition to gas solubilization, real food products are subjected quantifiable. Few studies exists, for example, Gašperlin and others
to complex and kinetic reactions (biological or chemical) in which (2001) estimated the O2 consumed at 6.34 m%O2 .g–1 food .h–1 for
either O2 or CO2 is consumed or released, which is not the case meat (in the case of raw meat) and 3.50 m%O2 .g–1 food .h–1 for
(or to a lesser extent) in water and polymeric membranes. Acqui- heated meat (at 60 ◦ C). The authors overestimated the values of
sition of the true solubility coefficient (the same for diffusivity) mitochondrial respiration because they did not take into account
is difficult and most of the time values reported for real foods the consumption due to oxidation and to microbial metabolism.
refer to an apparent solubility coefficient (and an apparent diffu- In the case of heated meat where we can suppose that microor-
sivity) but are not indicated as such. Indeed, steady state condition ganisms and mitochondrial respiration were both inhibited, the
for solubility determination takes a longer time when these reac- amount 3.50 m%O2 .g–1 food .h–1 could represent O2 consumption
tions occur and in addition is considered as a pseudo–equilibrium due to oxidation. Meat respiration is essentially due to high respi-
state. For instance, neglecting gas consumption (for example, O2 ), ratory activity of the mitochondria contained in meat cells (Kim
the solubility coefficient (SA ) is usually overestimated during the and others 2011), which decreases with time. An assumption could
experiment (Eq. 1). The next section will present the main inter- be made with consumption and production of both these gases by
fering kinetics reactions through illustrations and the ways, when cells, especially by mitochondrial metabolism, more easily acces-
possible, to overcome then and tentatively assess true solubility. sible in vitro. It is possible to characterize meat tissue O2 rate
Food metabolism. Aerobic respiration is the common metabolic consumption after isolation of the mitochondria (Gao and others
pathway for most living organisms to provide sufficient energy to 2013). Even if respiration parameters are accurately determined,
maintain their cellular integrity. This oxidation process requires O2 there is still an uncertainty about them. This uncertainty of the
to degrade glucose and it produces CO2 . It is widely studied in solubility value is amplified by a correction applied to the amount
fresh and fresh–cut fruits and vegetables because of its importance of sorbed O2 /CO2 measured in the product to calculate the O2
in MAP: it interplays with gas transfers through the packaging and CO2 quantity due to solubilization.
to permit the establishment of a steady atmosphere. The respira- Oxidant capacities. Beyond the aforementioned variability in
tory activity is determined either by monitoring O2 consumption O2 solubility induced by the presence of fat and/or protein, O2
or CO2 production, and modeled by using Michaelis–Menten– is also very reactive with unsaturated fat and proteins. Oxidation
type equations, considering, or not, the possible CO2 inhibition reactions induce O2 consumption that disturbs the solubility mea-
on product respiration (Fonseca and others 2002). Whatever the surement (overestimation of the “true” solubility). Unfortunately,
method used to determine this activity, it relies on a global assess- it is very difficult at this moment in time to assess O2 consumption
ment including, of course, cells’ respiration but also skin resistance for 1 mole or 1 g of fat or protein due to the complexity of ox-
to gas transfer, gas solubility, and gas diffusivity in the product. idation and oxygenation mechanisms, and then little information
In addition, this global assessment also takes into account the res- is found in the literature. In the past, Frayn (1983) has studied fat

276 Comprehensive Reviews in Food Science and Food Safety r Vol. 13, 2014 
C 2014 Institute of Food Technologists®
O2 and CO2 solubility and diffusivity . . .

and protein oxidation reactions in vivo, using a global approach act with O2 dissolved in food. As for protein, the determination
focusing on O2 consumption and CO2 production during the res- of O2 consumption by antioxidants is not a sinecure, mainly due
piration phenomena. According to his results, protein oxidation to the complexity of oxidation mechanisms and the necessity to
would lead to the consumption of 0.04 mol of O2 and the produc- have recourse to simplified equations of reaction to determine the
tion of 0.03 mol of CO2 per gram of protein. For fat, one gram number of moles of O2 that react with one mole of antioxidant.
of a studied fat (PSOG C55 H104 O6 ) would consume 0.09 mol of It is thus very difficult to take into account oxidation reactions
O2 . However, to our knowledge, there is no equivalent study to while measuring O2 solubilities and to make some corrections of
quantify O2 consumption due to the fat or protein oxidation in the measured amount to retrieve the “true” solubility. A solution
other food products. Most of the studies focused on the oxidation to overcome this difficulty would be to correlate the O2 con-
mechanism and oxidation rate. Evaluation of O2 consumption by sumption to the total antioxidant capacities of the food product
fat and protein oxidation on the basis of a mechanistic analysis is studied. These antioxidant capacities are indeed more easily as-
very tricky. Indeed, oxidation of lipids and proteins is complex sessed by experimental measurement than the oxidation reaction
with the occurrence of a chain reaction where the role of O2 is rate (Pérez-Jiménez and others 2008; Serpen and others 2012).
not always well understood. For instance, more than 90 different This was never attempted but could be a good future research
reactions were listed by Roman (2012) in her study on oil oxi- path. In any case this future research should pay attention to the
dation during thermal treatment. O2 takes part in only a few of natural variability in antioxidant content within different classes of
them, generally in the propagation step of oxidation (Roman and product.
others 2013). Microbial metabolism. The microbial population of a food
Oxidation rates of fat and protein are compatible with the product can affect O2 and CO2 concentration due to the res-
time scale of solubility measurement (around 24/48 h to obtain piration metabolism of aerobic microorganisms. This effect is usu-
equilibrium in food). For instance, kinetic rates of oxidation by ally observed in the monitoring of headspace gas concentration
singlet oxygen are around 0.5 and 1.5 .105 L.mol–1 .s–1 for fat of packed food. For anaerobic microorganisms, CO2 production
and around 1.3 and 5.107 L.mol–1 .s–1 for amino acid (Choe and could also be significant, such as, for example, production of CO2
Min 2005). To compare these kinetic rates, the half–life time is by yeast or fermentative bacteria (Eliot and others 2006). Aero-
commonly used, which is the time required to divide the initial bic or anaerobic respiration of microorganisms could both occur
concentration by 2 (Eq. 4). during the measurement experiments of CO2 or O2 solubility if
no precautionary measure for avoiding this is taken. Indeed, the
1 time span of the experiment is long enough to be significantly
t 12 = (4)
k × [B]0 affected by the microbial respiration. According to the few publi-
cations that have focused on the quantification of microorganism
For one mole of product, oxidation of amino acids occurs faster respiration in real food/packaging systems (only 2 papers were
than lipid oxidation. Obviously, in real food, lipid and protein found in the literature dealing with the respiration of Pseudomonas
oxidation kinetics depend on the content of each molecule: for (Thiele and others 2006) and O2 /CO2 consumption/production
example, in the case of a product with a high lipid content (>30%) by a consortium of bacteria (Simpson and others 2001a). This phe-
and 10% protein (as in liver pâtés) lipid oxidation was faster than nomenon is not negligible for microbial counts above a threshold
protein oxidation, but in minced cod muscle, with low lipid con- value. Indeed, Simpson and others (2001a) have demonstrated
tent of 0.6%, protein oxidation occurs faster than lipid oxidation (theoretical and experimental work) that for microorganism con-
(Lund 2007). centrations more than 105 CFU.g–1 , headspace gas composition
In the case of raw meat products, O2 can react, for example, was substantially modified and could reach an anaerobic medium
with myoglobin to produce metmyoglobin, a reaction responsible in a short time (less than 3 d). This phenomenon would be even
for color changes during storage (Eq. 5) (George and Stratmann more amplified if the oxidation rates of reactions overlaid with the
1952; Sáenz and others 2008). microorganisms’ respiration were higher than 10–6 s–1 (Simpson
and others 2001a). In spite of this preliminary work that revealed
Mb 2+ + O2 ↔ Mb O22+ (5) that microorganism respiration would be far from being negligi-
ble, O2 /CO2 consumption/production due to microbial reactions
According to Eq. 5, 1 mole of myoglobin would react with is still neglected when measuring O2 /CO2 solubility coefficients.
1 mole of O2 . Considering a myoglobin concentration of Therefore, this O2 /CO2 solubility values measured by these au-
486 mg.100 g–1 as found by Babji and Abdullah (1998) in beef, thors could be distorted by microorganism respiration. However,
0.0285 mmol of myoglobin would react with the same quantity of microorganism growth could be avoided by ultraviolet exposure or
O2 . Considering only the myoglobin reaction, 100 g of beef would adding an antimicrobial compound such as sodium azide, ethanol,
consume around 0.9 mg of O2 , which is far from being negligible. sorbic acid, and so on (Sanches Silva and others 2007; Pénicaud
Indeed, starting from the value of O2 solubility found in water at and others 2011).
20 ◦ C (around 0.8 mg O2 per 100 g, Figure 1), the consumption
of O2 by myoglobin would be equivalent to the quantity of O2 Factors Impacting and Predictive Modeling of Oxygen
dissolved in the 100 g of product at equilibrium with atmospheric and Carbon Dioxide Solubility
O2 pressure. Let us now consider the total protein content in a Both O2 and CO2 solubility depend on environmental factors
sample of meat (20 wt%, around 40–fold more than myoglobin such as temperature or pressure and on intrinsic parameters such as
(Babji and others 1989). It is obvious that total O2 consumption composition and structure of the matrices. Intrinsic characteristics,
due to oxidation of proteins would overcome the O2 solubility such as the effect of salinity and sugar concentration on O2 solu-
value. Furthermore, in addition to proteins, food products may bility, have been well described (Langø and others 1996; Pénicaud
contain different antioxidant components, such as vitamin C or E, and others 2012), especially for aqueous solutions. As for O2 , CO2
carotenoids, polyphenols, or enzymes like catalase, which inter- solubility seems deeply impacted by salt content (Schumpe 1993)


C 2014 Institute of Food Technologists® Vol. 13, 2014 r Comprehensive Reviews in Food Science and Food Safety 277
O2 and CO2 solubility and diffusivity . . .

and its effect has been well documented for NaCl in the liter- solubility values would decrease by 2 orders of magnitude when
ature (Duan and Sun 2003; Portier and Rochelle 2005). Some the temperature increases from 0 to 20 ◦ C.
parameters have an impact on both O2 and CO2 solubility, oth- Other models exist to predict O2 and CO2 solubility as a func-
ers have a specific action on one gaseous species. Fat and protein tion of temperature. For example, Carroll and others (1991) used
contents seemed to greatly impact both O2 and CO2 solubilities, an empirical model to predict CO2 solubility as a function of
especially because O2 and CO2 are classically more soluble in a fat temperature, based on a simple extension of Henry’s law model,
than in an aqueous phase. In addition, in the specific case of O2 , valid from 0 to 160 ◦ C. Generally, these empirical models often
it is consumed by oxidation reactions that could deeply disturb couple the effect of temperature to that of salt or sugar addition.
the measurement of O2 solubility (section “Oxidant capacities”). The most famous ones are the equation of Benson and Krause
Finally, in the case of CO2 solubility, as easily deduced from Eq. 3, (1984), that relates the effect of temperature and salt content to
the pH value of the matrices has been noticed as a particularly O2 solubility using a polynomial relationship, and that of Sadler
relevant parameter affecting CO2 dissolution (Zeng 1995). and others (1988) that allows to estimate O2 solubility in fruit
In the following, the impact of these main parameters on gas juices according to a linear law including temperature and ◦ Brix
solubility will be considered with special emphasis on current (in the range of 4 to 40 ◦ C).
bottlenecks in the knowledge of the relationship between food An analysis of O2 solubility data against temperature allows
nature/structure and the value of solubility. Mathematical mod- to point out that in fish oil an unaccounted atypical profile was
eling approaches to predict gas solubilities in food will then be obtained (Figure 1): solubility increased with increasing the tem-
addressed. perature to 40 to 60 ◦ C, and then fell when the temperature was
raised to 80 ◦ C. The same atypical profile was obtained for CO2
Temperature solubility in the fat tissue of meat (Figure 2b): it increased with
The solubility of O2 and CO2 decreases when the temperature temperature until a maximal value was reached at 10, 14, and
increases according to a van’t Hoff–type expression (Eq. 6). 22 ◦ C for fat tissue of lamb, pork, and beef, respectively, and then
  decreased at higher temperatures (Gill 1988). This unexpected be-
−Hs havior in relation to the temperature (“bell–like curve”) was more
S A = S0 exp (6)
RT related to a possible antagonistic effect between temperature and
viscosity: increasing temperature decreases viscosity which should
where S0 is the preexponential factor, HS is the apparent heat favors gas diffusion and then solubilization but at the same time
of sorption (or enthalpy of sorption, enthalpy of solubilization, or increasing temperature directly decreases solubilization (van’t Hoff
calorimetric enthalpy), R is the universal gas constant, and T is the law) (Ke and Ackman 1973). A lack of knowledge still remains
absolute temperature. However, the main problem with using this today about the impact of temperature on O2 and CO2 solubility
expression is to access to the value of enthalpy. This value includes in food products.
the heat of dissolution, condensation, and mixing in the phase. It is
well known for water and other aqueous solutions; for instance, the Salt and sugar content
enthalpy of sorption HS for O2 was found to be equal to –18.4 The solubility of O2 and CO2 in aqueous salt solutions usually
and –15.1 kJ.mol–1 in distilled water and NaCl solution (0.5 M) at decreases when the salt concentration increases. This impact of
15 ◦ C, respectively (Mirhej 1962). O2 solubility in wine would be salinity and sugar content on O2 solubility has already been re-
affected by temperature in the same manner as in water (Chiciuc viewed by Battino (1981) and recently completed by Geng and
2010). For CO2 the enthalpy of sorption was found to be equal to Duan (2010). The effect of salinity was usually coupled with that
–19.4 kJ.mol–1 in water (Carroll and others 1991). On the basis of of temperature. For high salinity values, the temperature effect
the aforementioned values obtained in aqueous solutions and using on O2 and CO2 solubility is extremely small (Millero and Huang
a mixing law, Simpson and others (2004) estimated the enthalpy 2003; Debelius and others 2009; Verberk and others 2011). The
of gas sorption in gelatin to be –13.2 kJ.mol–1 for O2 and –19.7 variation of O2 solubility by adding salt was correlated to the na-
kJ.mol–1 for CO2 , pointing out that the temperature would have ture of the salt: for example, addition of 1 mol.kg–1 of NaCl leads
a similar effect on O2 and CO2 dissolution. If no experimental to a decrease in O2 solubility of 26%. This decrease is 38% with
values of enthalpy for O2 and CO2 sorption in real food products MgCl2 , 51% with MgSO4 , and 53% with Na2 SO4 (Millero and
were reported in the current literature, it is nevertheless possible others 2002). In the case of CO2 solubility, the addition of salt has
to estimate it, by taking the gas solubility value assessed at various less impact than on O2 solubility: for example, the addition of 1
temperatures (Figure 2) as input data for Eq. 6. For instance, from mol.kg–1 of NaCl decreased the CO2 solubility 14% (Duan and
CO2 solubility values measured by Sivertsvik and Jensen (2005) Sun 2003).
and Jakobsen and Jensen (2009a), the enthalpy for CO2 sorption Similar to salt, increasing sugar concentration tends to decrease
was estimated in the present work to be between –16.25 kJ.mol–1 O2 solubility in food, with a decrease in magnitude which de-
and –12.73 kJ.mol–1 in cheese (for temperatures ranging from 0 pends on the nature of the sugar (and more specifically on its
to 25 ◦ C) and between –13.76 kJ.mol–1 and –12.37 kJ.mol–1 in carbon number). For example, the addition of 1 mol.kg–1 of ri-
meat (for temperatures ranging from 0 to 8 ◦ C). Such estimated bose caused a decrease in O2 solubility of 24%, this decrease was
values are very close to those measured in water. For comparison, of 29% with glucose, 38% with maltose, and 47% with raffinose
the enthalpy of O2 sorption was experimentally found to be equal (Mishima and others 1996, 1997). Sugar alcohols, such as mannitol
to –31.6 kJ.mol–1 , and enthalpy of CO2 sorption to be –13.7 or sorbitol, have the same effect as sugars on O2 solubility (Whit-
and –17.7 kJ.mol–1 in hybrids of poly(amide–6–b–ethylene oxide) combe and others 2005). In the same way, but to a lesser extent,
(Kim and Lee 2001). From literature values and values calculated CO2 solubility decreases with increasing sugar concentration. This
in this work, it is obvious that enthalpy of sorption is around – effect also depends on the nature of the sugar: the addition of 1
20kJ.mol–1 , whatever the product considered. As a consequence, mol.kg–1 of glycerol caused a decrease in CO2 solubility of 7% and

278 Comprehensive Reviews in Food Science and Food Safety r Vol. 13, 2014 
C 2014 Institute of Food Technologists®
O2 and CO2 solubility and diffusivity . . .

0.8 mol.kg–1 of glucose or sucrose to a decrease of 15% (Risch- ther be extrapolated to another pH range nor to a product other
bieter and others 1996). The impact of salt and sugar addition on than meat. Currently, the prediction of the effect of pH on CO2
O2 /CO2 solubility in real food products is less referenced than for solubility could only be assessed case by case and in an empirical
aqueous solutions and not at all, to our knowledge, in solid food manner.
matrices.
Some mathematical modeling has been attempted in order to Effect of fat and protein content on O2 /CO2 solubilities
predict O2 solubility in aqueous–based media according to their If the impact of fat and protein contents was clearly identified
salt and sugar composition. In addition to the empirical correla- on CO2 solubility, it was never clearly stated to what extent it
tions of Benson and Krause (1984) and Sadler and others (1988) would affect this solubility values. For example, recently, it was
(see above), Schumpe and others (1982) predicted O2 solubility found that fat content had a low impact on CO2 solubility in
by taking into account the contribution of each ion species dis- meat (Jakobsen and Bertelsen 2004), whereas it was significant for
solved in the solution using a semiempirical UNIFAC model to cheese products (Jakobsen and Jensen 2009). An earlier study by
calculate the chemical acitivity of each species present. Gros and Fava and Piergiovanni (1992) described a relationship between the
others (1999) went further than Schumpe and others (1982) by solubility coefficient and food composition (including protein and
considering, in addition to salt, the activity of each functional fat content) for a meat product, salami, and fresh pasta. This rela-
group of macromolecules on O2 and CO2 solubility. Ji and oth- tionship was based on a multiple regression model relating CO2
ers (2007) proposed to use an equation derived from statistics- solubility to the pH (ranging from 4.93 to 6.54), the moisture
associated fluid theory (SAFT; Economou 2002) to predict the content (between 24 and 76.8%), the water activity (ranging from
effect of sugar content (glucose, fructose, sucrose, and maltose) 0.870 to 0.999), the protein content (7.7 to 31%), and the fat
on O2 solubility in liquids by taking into account molecular content (0.8 to 38.9%) of the product analyzed. This statistical
interactions. model was a first step toward an understanding of the relationship
between solubility and food composition. However, experimental
pH (for CO2 solubility) validation was successful only for meat products and pasta and not
Gaseous CO2 dissolves in a liquid phase and then dissociates for dairy products. One explanation of this discrepancy between
into various dissolved species according to Eq. 3. This dissocia- theoretical and experimental values would be that, during exper-
tion depends on the pH of the liquid phase (Figure 5). Therefore, imental measurement of CO2 solubility, the authors did not take
the pH of a food matrix strongly effects the quantity of CO2 into account possible proteolysis, lipolysis, and microbial growth
able to dissolve within. For instance, CO2 absorption by a buffer in the food, which would have randomly influenced the amount
solution at pH 5.5 is lower than that by water (pH 7) because of CO2 absorbed by the matrices (section “Interfering reactions”).
the hydration reaction (CO2 which turns into HCO3 – ) is higher Following this preliminary modeling attempt, Jakobsen and Jensen
at pH 7 than pH 5.5 (Lı́vanský 1982). In real food products, (2009) developed a model to predict CO2 solubility in semi hard
the same effect of initial food pH on the quantity of CO2 able cheeses based on the weight fraction of water (ww ) and fat (wf ) in
to dissolve in the product was noticed (Gill 1988; Jakobsen and the 2–phase cheese system, temperature (T), and the CO2 solubil-
Bertelsen 2004). Therefore, decreasing the pH value leads to a ities in, respectively, pure water (SCO2/W ) and pure fat (butterfat,
decrease in CO2 solubility (less CO2 is able to dissolve in the SCO2/F ) (Eq. 7)
product) and increasing the pH–value leads to an increase in CO2
solubility. For example, Gill (1988) found that for pH increasing ww × SCO2/W (T) + w f × SCO2/F (T) (7)
linearly from 5.4 to 6.9, CO2 solubility increased approximately by
1.5×10–7 mol.kg–1 .Pa–1 for each pH unit increase. However, such Equation 7 was found suitable for predicting CO2 solubility
behavior was not confirmed with fish products (Sivertsvik and in semihard cheeses and the range of temperature from 0 to
others 2004b). It is worth noticing that in all the aforementioned 20 ◦ C was investigated by Jakobsen and Jensen (2009). How-
studies, only the initial pH in the food matrices was measured ever, the authors noted that for temperatures below 10 ◦ C, pre-
and not its evolution as a function of time and CO2 dissolution. dictions made considering only the water phase in Eq. 7 were
However, it is fair to say that CO2 dissolution leads to pH changes accurate enough to predict solubility in the entire cheese. The
to a greater or lesser extent depending on the buffer properties main drawback of this equation is that, of course, it is not extrap-
of the food. To be as rigorous as possible, the addition of CO2 olatable to other cheese and food products than those studied by
into a matrix must be considered from the acidification point of the authors.
view: when CO2 dissolves, it acts as an acid on the pH value. A To our knowledge, except for the aforementioned empirical
theoretical mathematical approach exists to model the effect of pH approaches, there are no mechanistic models aimed at formalizing
on CO2 transfer rate in liquids and vice versa (Shanableh 2009). the relationship between O2 /CO2 solubility values and fat content
In this work, the amount of CO2 is represented by the concen- or protein content of a product. Therefore, it is thus currently
tration of all its species present in the solution (CO2 , HCO3 – , very difficult to predict O2 /CO2 solubilities in a food product on
and CO3 2– ) as a function of time. According to the authors, the the basis of its composition. To overcome this lack of scientific
contribution of the different carbonic species on CO2 solubility data, specific studies aimed at the impact of food composition on
is far from negligible. However, this type of model does not ex- O2 /CO2 solubilities should be undertaken.
ist for real food and its applicability to such complex products is
questionable. Indeed, the buffering properties of real food depend Determination of O2 /CO2 Diffusivity
on its composition and then would limit the versatility and ex- Experimental set–up of O2 /CO2 diffusivity
trapolation of such models to other kinds of products than the O2 /CO2 diffusivity is not directly measured: obtaining such in-
one for which it was developed. The only relationship between formation requires a set of experimental data and a mathematical
pH change and CO2 solubility was the one established by Gill model. To obtain O2 /CO2 diffusivity, a gradient of O2 /CO2 con-
and others (1988). It was an empirical relationship that could nei- centration (or partial pressure) has to be imposed within the food


C 2014 Institute of Food Technologists® Vol. 13, 2014 r Comprehensive Reviews in Food Science and Food Safety 279
O2 and CO2 solubility and diffusivity . . .

Figure 7–Example of experimental set-up to measure O2 kinetic for diffusivity identification in liquid and viscous food products (adapted from
Pénicaud and others 2010).

and then the O2 /CO2 diffusion has to be monitored (using one of the indirect measurement methodologies listed above at different
the aforementioned methodologies for measuring O2 /CO2 par- time intervals.
tial pressure or concentration). For example, calibration to obtain
CO2 concentration with FTIR–ATR is very time–consuming
(absorbance crossed with a quantitative method) but it would Typical values of O2 /CO2 diffusivity
detect a kinetic of dissolved CO2 in solid matrices. Experimental From 0 ◦ C to 37 ◦ C, O2 diffusivity varied from 0.2 × 10–9
set–up permits direct monitoring of CO2 content in a sample, with m .s with agar 2%w/v (Miller and others 2003) to 53 × 10–9
2 –1

an analysis of spectral band variation. In the same way, an example m2 .s–1 with apple (Schotsmans and others 2004) (Figure 3). If
of experimental design used with measurement of O2 partial pres- we excluded fruit cellular tissues that probably displayed a porous
sure for determining O2 diffusivity is presented in Figure 7. The texture that enhanced O2 transfer: then the range of values for
diffusivity is then identified by adjusting a mathematical model of O2 diffusivity was only from 0.2 × 10–9 m2 .s–1 with agar 2%w/v
diffusion (mainly based on Fick’s second law) to the experimental (Miller and others 2003) to 2.9 × 10–9 m2 .s–1 with water (Grote
kinetics of O2 /CO2 transfer using an algorithm of optimization 1967); such low variation indicates that the nature of the product
(for example, Gauss–Newton, Levenberg–Marquardt, and so on). only influenced the O2 diffusivity slightly. Whatever the tem-
Fick’s second law can be solved either analytically (Crank 1975) perature, the variation is very slight leading us to conclude that
or numerically, using finite difference, finite element, and finite temperature has a low impact on O2 diffusivity: O2 diffusivity is
volume, approaches more and more used in the food applications 1.7 × 10–9 m2 .s–1 on water for 10 ◦ C (Langø and others 1996),
(Wang and Sun 2003). In both cases the geometry of the sam- 20 ◦ C (Grote and Thews 1962), and 37 ◦ C (Lightfoot 1974). As
ple and the boundary conditions imposed on the sample must be regards the data available (Figure 3), O2 diffusivity would be lower
strictly controlled in order to ensure the accuracy of the mathe- in oils and fats than in aqueous products.
matical modeling. From 0 ◦ C to 25 ◦ C, CO2 diffusivity varied from 0.2 × 10–9
We noted that in some particular cases (vegetable tissues), m .s with egg yolk (Fabbri and others 2011) to 151 × 10–9
2 –1

O2 diffusivity was obtained using permeation measurement m2 .s–1 with apple (Solomos 1987) (Figure 4), and if we excluded
(De Wild and Peppelenbos 2001; Schotsmans and others 2004; fruit cellular tissues, the range of variations for CO2 diffusivity
Valleguadarrama and others 2005). In this last case, a thin sheet of was only from 0.2 × 10–9 m2 .s–1 with egg yolk (Fabbri and others
material is placed between 2 atmospheres maintained at constant 2011) to 11 × 10–9 m2 .s–1 with sausage (Sivertsvik and Jensen
and different O2 partial pressure. After a time period, the surfaces 2005), whatever the food matrices. The temperature has a low
of the sheet come into equilibrium with the diffusing sources, impact on CO2 diffusivity. For example, in ham, CO2 diffusivity
thus, developing a constant gradient of surface concentrations, is 3.7 × 10–9 , 3.9 × 10–9 , and 3.8 × 10–9 m2 .s–1 , respectively, for
leading to steady state conditions of diffusion. This state could be 0, 4, and 8 ◦ C (Sivertsvik and Jensen 2005). CO2 diffusivity would
expressed for a plane sheet by Fick’s first law (Eq. 8) be higher in meat products than in fish products (Figure 4).
In both cases (O2 and CO2 ), the lack of available data on dif-
∂C A fusivity values for solid food matrices prevented us to estimate the
J A = −D (8)
∂x activation energy on diffusivity. In theory, the increase of tem-
perature should increase the diffusion of gases. However, Simp-
where JA is the gaseous diffusion flux (mol.m.kg−1 .s−1 ), D is the son and others (2004) estimated the activation energy in gelatin
diffusion coefficient (m2 .s−1 ), CA is the concentration of dissolved at 20.3 kJ.mol–1 for O2 diffusivity, and 14.5 kJ.mol–1 for CO2
gas A (mol.kg−1 ), and x is the distance between the interface and diffusivity, taking into account the water content of the matrix.
measurement point (m). For comparison with another field of science, activation energy
Diffusivity can be estimated by measuring the flux of the diffus- for diffusion through polyimide membranes was 17.6 kJ.mol–1
ing gas, with known surface concentrations and thickness of the and 16.8 kJ.mol–1 for O2 and CO2 , respectively (Lin and Chung
material sheet, which can be done experimentally by using one of 2001). Similar activation energies were found for diffusion through

280 Comprehensive Reviews in Food Science and Food Safety r Vol. 13, 2014 
C 2014 Institute of Food Technologists®
O2 and CO2 solubility and diffusivity . . .

Figure 8–CO2 diffusivity (m2 .s–1 ) as a function of food composition (square: water, diamond: fat, and triangle: protein fraction) taking into account
values from the literature in a range of temperatures from 0 to 8◦ C. Light color dot: data of various food products: water (Carroll and others 1991),
hake (Simpson and others 2001a), sausage and ham (Sivertsvik and Jensen 2005) and dark color dot: same data plus anglerfish, cod, salmon, tuna,
wolf-fish (Sivertsvik and others 2004).

polyurethane membrane, 28.2 kJ.mol–1 and 35.6 kJ.mol–1 for O2 as the viscosity measurement of meat is not achievable, Zaritzky
and CO2 diffusion, respectively (de Sales and others 2008) and and Bevilacqua (1988) used the viscosity of water at the experi-
lower values were determined for a synthetic membrane made of mental temperature to calculate O2 diffusion in muscle tissues, and
poly (2-ethoxyethyl methacrylate-co-2-hydroxyethyl methacry- this is definitively not realistic. These authors did not carry out
lamide), 15 kJ.mol–1 and 19 kJ.mol–1 for, respectively, O2 and CO2 experimental validation of their calculations. This Wilke–Chang
(Tiemblo and others 2004). For liquids, there are more data avail- equation is initially derived from the Stokes–Einstein equation
able, but too high a variability prevented us from calculating any (Eq. 11), which is also valuable for predicting O2 or CO2 diffu-
activation energy for O2 or CO2 diffusivity. Indeed, if we ex- sivity, and which relates the diffusivity to viscosity of the medium
amined the set of data reported for water by Langø and others and temperature as in Eq. 10
(1996) about O2 diffusivities, and we can see that the range of
variation at 25 ◦ C (from 1.6 × 10−9 to 3.5 × 10−9 m2 .s−1 ) is R× T 1
as large as that obtained for the min–max span of temperatures Dp = (11)
NA 6 × π × μ × a
studied. The same observation could be made for CO2 diffusivity
(values ranging from 1.57 × 10−9 to 2.0 × 10−9 m2 .s−1 in water where D is the diffusion coefficient of the migrant (with a radius
at 25 ◦ C) (Figure 3 and 4). Unver and Himmelblau (1964) have of a); T is the absolute temperature, μ is the dynamic viscosity, R
developed an empirical model to relate diffusivity of CO2 in water is the gas constant, and NA is Avogadro’s number. This equation,
to temperature, between 6 and 65 ◦ C (Eq. 9) originally written for fluids and spherical particles undergoing
Brownian motion would remain validated for smaller molecules
DCO2 = (α + βT + γ T 2 ) × 10−9 (9) such as O2 or CO2 (McCabe and others 2005). The Wilke–Chang
equation (Eq. 10) and the Stokes–Einstein equation (Eq. 11) are
where D is a coefficient of diffusion of CO2 (m2 .s–1 ) and α, β, and available for O2 diffusion in water, organic liquids, and salt and
γ are constants with values of 0.95893, 0.024161, and 0.00039813, glucose solutions (Schumpe and Luehring 1990; Jamnongwong
respectively. and others 2010), and also available for CO2 diffusion in liquid
mixtures (water or methanol with various viscosities) (Hikita and
Predictive models for O2 /CO2 diffusivity
others 1985; Frank and others 1996; Iglauer 2011).
An alternative to experimental determination of diffusivity is the
Apart from the Wilke–Chang and the Stokes–Einstein equa-
use of a predictive mathematical model based on the characteristics
tions, mathematical models for predicting diffusivity are scarce. In
of the food structure and composition as main inputs. For example,
order to identify if any empirical correlation between food com-
to estimate O2 diffusivity in gelatin (Simpson and others 2004) and
position and diffusivity would exist, we analyzed diffusivity data
meat (Zaritzky and Bevilacqua 1988), both teams have adapted
from the literature in relationship with the composition of the
an empirical correlation, the Wilke–Chang equation, previously
food concerned (that is, water, fat content, and protein content).
established in another field of science (Wilke and Chang 1955).
This exercise was done on CO2 diffusivity only (lack of data on
This equation (Eq. 10) correlates diffusivity of the migrant (D) as
O2 prevented us from performing a representative analysis) using,
a function of the viscosity (μ) and the absolute temperature (T).
in a first attempt, data taken from the studies on water (Carroll
D×μ and others 1991), hake (Simpson and others 2001b), sausage and
= constant (10) ham (Sivertsvik and Jensen 2005), and presented in light color dot
T
in Figure 8. From this set of data, CO2 diffusivity seems corre-
It was valuable initially for infinite dilute solution only and its lated with water and fat content, but not with protein content
extrapolation to solid product could be questionable. For instance, contrary to previous studies made on cell tissues (Geers and Gros


C 2014 Institute of Food Technologists® Vol. 13, 2014 r Comprehensive Reviews in Food Science and Food Safety 281
O2 and CO2 solubility and diffusivity . . .

2000) or hemoglobin solutions (Gros and Moll 1971), where the ample, material science and engineering where the relationship
authors indicate a decrease in CO2 diffusivity with the increase between the structure of dense matrices and mass transfer proper-
of protein concentration. In order to extend such correlations, ties has already been formalized in some cases).
other data from the literature on fish products (anglerfish, cod,
salmon, tuna, and wolfish in Sivertsvik and others 2004a) were Acknowledgment
added and the result is displayed in dark color dot in Figure 8. This work was supported by the French Ministry of Education
Correlations put in evidence in light color dot in Figure 8 are no and Research (MAP’OPT project).
longer valid when the range of data is enlarged to fish products,
illustrating the difficulty of finding a simple and universal correla- Conflicts of Interest
tion between CO2 diffusivity and food composition. Future work The authors have declared no conflict of interest.
should be carried out in order to deepen predictive models, which
would avoid a time–consuming experimental approach to assess
diffusivity within solid food. Nomenclature
As for solubility measurements, interfering reactions inducing −A: Radius of spherical particle (m)
unwanted and nonmastered O2 /CO2 consumption/production A: Any species of gas (O2 , CO2 , N2 )
could also lead to incorrect assessment of the diffusivity. For AACC International: American Association of Cereal Chemists
instance, in the case of O2 , the occurrence of oxidizable com- International
pounds in the food tends to disturb measurements. O2 consump- AOAC: Association of Official Agricultural
tion within the food product may create a significant delay in Chemists
the establishment of the equilibrium and, consequently, lead to an ATR: Attenuated total reflection
underestimated O2 diffusivity value. In the specific case of CO2 , Ba(OH)2 : Barium dihydroxide
interfering reactions are linked mainly to the enzymatic activity BaCO3 : Barium carbonate
of carbonic anhydrase (EC 4.2.1.1) (Simpson and others 2009). cA : Volume concentration of dissolved gas A
These pH regulatory enzymes catalyze the hydration of CO2 to (kg.m−3 )
bicarbonate (HCO3 – ) and, for example, can contribute to dou- CA : Concentration of dissolved gas A
bling the CO2 transfer in a buffer solution at a pH range of 6.9 to (mol.kg−1 )
7.8 (Geers and Gros 2000). Ca(OH)2 : Calcium dihydroxide
CFU: Colony-forming unit
Conclusion CO2 : Carbon dioxide
In this paper, the different methodologies used to assay O2 and CO3 2− : Carbonate ion
CO2 within solid food products are commented on and discussed D: Diffusion coefficient (m2 .s−1 )
with special emphasis on their implementation in an experimen- FID: Flame ionization detector
tal set–up to assess solubility and diffusivity. Literature data on FTIR: Fourier transform infra-red
O2 /CO2 solubility and diffusivity are listed and discussed. Data H+ : Proton (Hydrogen(1+))
are scarce, especially for O2 solubility and diffusivity due to, mainly, H2 CO3 : Carbonic acid
the difficult implementation of O2 quantification methodologies H2 O: Water
in the experimental set–up of solubility and diffusivity measure- HCl: Hydrochloric acid
ment. Inventory and analysis of this collection have confirmed HCO3 − : Bicarbonate ion (hydrogen carbonate)
some facts: O2 is less soluble in food than CO2 : salt and sugar JA : Gaseous diffusion flux (mol.m.kg−1 .s−1 )
concentrations decrease O2 /CO2 solubility values but obviously k: Kinetic rates of oxidation (L.mol−1 .s−1 )
not the diffusivity values whatever the gas considered. This analysis kHA (T) Henry’s law coefficient for gas A
has also revealed some less known facts such as a lack of repeatabil- (Pa.kg.mol−1 )
ity (that is, for O2 diffusivity in water), some unexpected behavior KOH Potassium hydroxide
as a function of temperature (that is, for O2 and CO2 solubility MAP Modified atmosphere packaging
within oils and also fat products). The role of interfering reactions Mb2+ Myoglobin
(that is, oxidation of fats and proteins, microorganism respiration, MbO2 2+ Metmyoglobin
and so on), on O2/CO2 solubility /diffusivity values (even if it is MgCl2 Magnesium dichloride
well known to influence these parameters) was scarcely quantita- MgSO4 Magnesium sulfate
tively related to the amount O2 /CO2 consumed by the product N2 Nitrogen
during measurement and have usually been uncontrolled during NA Avogadro’s number
solubility and diffusivity measurements. A particular focus of this Na2 SO4 Disodium sulfate
paper is the predictive modeling of solubility/diffusivity as a func- NaCl Sodium chloride
tion of external (mainly temperature) and intrinsic food charac- NaOH Sodium hydroxide
teristics (for example, composition, viscosity, and so on). If some O2 Oxygen
empirical correlations exist to predict the effect of salt and sugar p.v. Peroxide values
concentrations on O2 /CO2 solubility and the effect of viscosity PA Partial pressure of gas A (Pa)
on O2 /CO2 diffusivity, these relations remain validated only for pH Potential hydrogen
aqueous solutions, and extrapolation to real food (for example, pKa Logarithm of the acid dissociation constant
solid product) is questionable. Further work should be conducted ppb Part(s) per billion
in the near future to increase our knowledge on the relationship PSOG Palmitoyl-stearoyl-oleoyl-glycerol
between food structure/composition and O2 /CO2 diffusivity by R Universal gas constant (8.31446
using some approaches taken from other fields of science (for ex- J.K−1 ·mol−1 )

282 Comprehensive Reviews in Food Science and Food Safety r Vol. 13, 2014 
C 2014 Institute of Food Technologists®
O2 and CO2 solubility and diffusivity . . .

R.H. Relative Humidity Bertola N, Chaves A, Zaritzky N. 1990. Diffusion of carbon dioxide in
S0 Pre-exponential factor of van’t Hoff ex- tomato fruits during cold storage in modified atmosphere. Int J Food Sci
Technol 25:318–27.
pression (mol.kg−1 .Pa−1 )
Beutel S, Henkel S. 2011. In situ sensor techniques in modern bioprocess
SA Solubility of gas A (mol.kg−1 .Pa−1 ) monitoring. Appl Microbiol Biotechnol 91:1493–505.
SAFT Statistical associating fluid theory Blakistone BA (ed). 1999. Principles and applications of modified atmosphere
SCO2/F (T) CO2 solubility in fat matrices model packaging of foods. Boston, Mass.: Springer. p 293.
(mol.kg−1 .Pa−1 ) Brantley NH, Kazarian SG, Eckert CA. 2000. In situ FTIR measurement of
SCO2/W (T) CO2 solubility in pure water carbon dioxide sorption into poly (ethylene terephthalate) at elevated
(mol.kg−1 .Pa−1 ) pressures. J Appl Polym Sci 77:764–75.
t Time (s) Brown G. 1955. Operaciones ba´sicas de la ingenieri´a qui´mica. Barcelona:
Manuel Mari´n & Cia.
T Absolute temperature (K) Burke CS, Markey A, Nooney RI, Byrne P, McDonagh C. 2006.
t1/2 Half-life time (s) Development of an optical sensor probe for the detection of dissolved
UNIFAC Universal functional activity coefficient carbon dioxide. Sensors Actuators B Chem 119:288–294.
wf Weight fraction of fat (%) Caflisch CR, Carter NW. 1974. A micro pCO2 electrode. Anal Biochem
wt% Weight solute/weight total product 60:252–7.
ww Weight fraction of water (%) Cagnon T. 2012. Transferts dans les syste`mes emballagealiments:
structuration a` fac¸on de mate´riaux multicouches pour l’emballage sous
x Distance between the interface and mea- atmosphe`re modifie´e des produits frais [PhD Thesis]. France: Montpellier
surement point (m) SupAgro. p 313.
α Constant for diffusivity empirical model Carroll JJ, Slupsky JD, Mather AE. 1991. The solubility of carbon dioxide in
β Constant for diffusivity empirical model water at low pressure. J Phys Chem Ref Data 20:1201–9.
γ Constant for diffusivity empirical model Chiciuc I. 2010. Etude des paramètres affectant le transfert d’oxygène dans les
vins [Ph.D Thesis]. France: Univ. Bordeaux 1. p 297.
μ Dynamic viscosity (kg.m−1 .s−1 )
Choe E, Min D. 2005. Chemistry and reactions of reactive oxygen species in
HS Apparent heat of sorption of gas A foods. J Food Sci 70:142–9.
(J.mol−1 ) Crank J. 1975. The mathematics of diffusion. Oxford, U.K.: Clarendon Press.
%w/v Weight/volume percent Crow VL, Martley FG. 1991. An enzymic assay for CO2 in cheese. J Dairy
[B]0 Initial concentration of the reactive sub- Res 58:521–5.
stance B (mol.L−1 ) Crowe M, O’Connor P, Masterson B. 1987. Oxygen diffusion in natural oils

Bx Degree(s) Brix and pure triglycerides. Biochem Soc Trans 15:260–1.
Daniels JA, Krishnamurthi R, Rizvi SSH. 1985. A review of effects of carbon
dioxide on microbial growth and food quality. J Food Prot 48:
532–7.
References Davidson J, Cullen E. 1957. The determination of diffusion coefficients of
AACC Intl. 22-11.01. Approved methods of analysis. 11th ed. Method sparingly soluble gases in liquids. Trans Inst Chem Eng 35:51–60.
22-11.01. Measurement of gassing power by the pressuremeter method. St.
Paul, Minn.: AACC Intl. Davidson D, Eggleton P, Foggie P. 1952. The diffusion of atmospheric gases
through fats and oils. Q J Exp Physiol Cogn Med Sci 37:91–105.
AACC Intl. 22-14.01. Approved methods of analysis. 11th ed. Method
22-14.01. Measurement of gassing power by volumetric method. St. Paul, Debelius B, Gómez-Parra A, Forja JM. 2009. Oxygen solubility in
Minn.: AACC Intl. evaporated seawater as a function of temperature and salinity. Hydrobiologia
632:157–65.
Adlercreutz P. 1986. Oxygen supply to immobilized cells: 5. Theoretical
calculations and experimental data for the oxidation of glycerol by Diéval J, Vidal S, Aagaard O. 2011. Measurement of the oxygen transmission
immobilized Gluconobacter oxydans cells with oxygen or p-benzoquinone rate of co-extruded wine bottle closures using a luminescence-based
as electron acceptor. Biotechnol Bioeng 28:223–32. technique. Packag Technol Sci 24:375–85.
Ahvenainen R. 2003. Novel food packaging techniques. Washington, D.C.: Dixon NM, Kell DB. 1989a. The control and measurement of CO2 during
CRC Press. p 589. fermentations. J Microbiol Methods 10:155–76.
AOAC 28.1.46 Official methods of analysis of AOAC Intl. 2005. 18th ed. Dixon NM, Kell DB. 1989b. The inhibition by CO2 of the growth and
Method 28.1.46. AOAC official method 988.07 carbon dioxide in wines metabolism of micro-organisms. J Appl Bacteriol 67:109–36.
titrimetric method. Gaithersburg, M.D: AOAC Intl.Official Method Duan Z, Sun R. 2003. An improved model calculating CO2 solubility in
2005.988.07. pure water and aqueous NaCl solutions from 273 to 533 K and from 0 to
2000 bar. Chem Geol 193:257–71.
Aubert JH. 1998. Solubility of carbon dioxide in polymers by the quartz
crystal microbalance technique. J Supercrit Fluids 11:163–72. Dunk RM, Peltzer ET, Walz PM, Brewer PG. 2005. Seeing a deep ocean
CO2 enrichment experiment in a new light: laser Raman detection of
Babji AS, Ooi P, Abdullah A. 1989. Determination of meat content in
dissolved CO2 in seawater. Environ Sci Technol 39:9630–6.
processed meats using currently available methods. Pertanika 12:33–41.
Economou I. 2002. Statistical associating fluid theory: a successful model for
Badran A, Woodruff R, Wilson L. 1969. Method of packaging perishable
the calculation of thermodynamic and phase equilibrium properties of
plant foods to prolong storage life. United States Pat 3:450–543.
complex fluid mixtures. Ind Eng Chem Res 41:953–62.
Baird MHI, Davidson JF. 1962. Annular jets—II Gas absorption. Chem Eng
Eliot SC, Vuillemard J-C, Emond J-P. 2006. Stability of shredded mozzarella
Sci 17:473–80.
cheese under modified atmospheres. J Food Sci 63:1075–80.
Barront FH, Harte B, Giacin J, Hernandez R, Segerlind L. 1993. Finite
Fabbri A, Cevoli C, Cocci E, Rocculi P. 2011. Determination of the CO2
element computer simulation of oxygen diffusion in packaged liquids.
mass diffusivity of egg components by finite element model inversion. Food
Packag Technol Sci 6:311–21.
Resea 44:204–8.
Battino R. 1981. Oxygen and ozone. In: Battino R, editor. IUPAC solubility
data ser. Vol. 7. Oxford, U.K: Pergamon Press. p 41–55. Fabiano B, Perego P, Pastorino R, Del Borghi M. 2000. The extension of
the shelf-life of “pesto” sauce by a combination of modified atmosphere
Battino R, Clever HL. 1966. The solubility of gases in liquids. Chem Rev packaging and refrigeration. Int J Food Sci Technol 35:293–303.
66:395–463.
Fanget S, Hentz S, Puget P, Arcamone J, Matheron M, Colinet E, Andreucci
Battino R, Evan R, Danforth F. 1968. The solubilities of seven gases in olive P, Duraffourg L, Myers E, Roukes ML. 2011. Gas sensors based on
oil with references to theories of transport through the cell membrane. J Am gravimetric detection—a review. Sensors Actuators B Chem 160:804–
Oil Chem Soc 45:830–3. 21.
Benson BB, Krause D. 1984. The concentration and isotopic fractionation of Fasching R, Kohl F, Urban G. 2003. A miniaturized amperometric CO2
oxygen dissolved in freshwater and seawater in equilibrium with the sensor based on dissociation of copper complexes. Sensors Actuators B
atmosphere. Limnol Oceanogr 29:620–32. Chem 93:197–204.


C 2014 Institute of Food Technologists® Vol. 13, 2014 r Comprehensive Reviews in Food Science and Food Safety 283
O2 and CO2 solubility and diffusivity . . .

Fava P, Piergiovanni L. 1992. Carbon dioxide solubility in foods packaged Irimia-Vladu M. 2006. Chemical sensing employing pH sensitive emeraldine
with modified atmosphere. 2: correlation with some chemical-physical base thin film for carbon dioxide detection [PhD Thesis]. Auburn Univ.
characteristics and composition. Ind Aliment 31:424–430. p 194.
Ferrari M-C. 2009. Mass transport in polymers [PhD Thesis]. Italy: Univ. di Ishiji T, Takahashi K, Kira A. 1993. Amperometric carbon dioxide gas sensor
Bologna. p 162 based on electrode reduction of platinum oxide. Anal Chem 65:2736–9.
Fick A. 1855. On liquid diffusion. J Memb Sci 100:33–8. Jakobsen M, Bertelsen G. 2004. Predicting the amount of carbon dioxide
Fonseca SC, Oliveira FAR, Lino IBM, Brecht JK, Chau KV. 2000. absorbed in meat. Meat Sci 68:603–10.
Modelling O2 and CO2 exchange for development of perforation-mediated Jakobsen M, Jensen PN. 2009. Assessment of carbon dioxide solubility
modified atmosphere packaging. J Food Eng 43:9–15. coefficients for semihard cheeses: the effect of temperature and fat content.
Fonseca SC, Oliveira FAR, Brecht JK. 2002. Modelling respiration rate of Eur Food Res 229:287–94.
fresh fruits and vegetables for modified atmosphere packages: a review. J Jamnongwong M, Loubiere K, Dietrich N, Hébrard G. 2010. Experimental
Food Eng 52:99–119. study of oxygen diffusion coefficients in clean water containing salt, glucose
Forrester RL, Wataji LJ, Silverman DA, Pierre KJ. 1976. Enzymatic method or surfactant: consequences on the liquid-side mass transfer coefficients.
for determination of CO2 in serum. Clin Chem 22:243–5. Chem Eng J 165:758–68.
Frank MJW, Kuipers JAM, van Swaaij WPM. 1996. Diffusion coefficients Janata J. 2009. Principles of chemical sensors. Atlanta, Ga.: Springer. p 340.
and viscosities of CO2+ H2O, CO2+ CH3OH, NH3+ H2O, and NH3+ Ji P, Feng W, Tan T, Zheng D. 2007. Modeling of water activity, oxygen
CH3OH liquid mixtures. J Chem Eng Data 41:297–302. solubility and density of sugar and sugar alcohol solutions. Food Chem
Frayn KN. 1983. Calculation of substrate oxidation from gaseous exchange 104:551–8.
rates in vivo from gaseous exchange. J Appl Physiol 55:628–34. Karimi M. 2011. Diffusion in polymer solids and solutions. In: Markoš J,
Gao X, Xie L, Wang Z, Li X, Luo H, Ma C, Dai R. 2013. Effect of editor. Mass transf in chem eng process. Rijeka, Croatia: InTech.
postmortem time on the metmyoglobin reductase activity, oxygen p 306.
consumption, and colour stability of different lamb muscles. Eur Food Res Ke PJ, Ackman RG. 1973. Bunsen coefficient for oxygen in marine oils at
Technol 236:579–87. various temperatures determined by an exponential dilution method with a
Gašperlin L, Zlender B, Abram V. 2001. Colour of beef heated to different polarographic oxygen electrode. J Am Oil Chem Soc 50:429–35.
temperatures as related to meat ageing. Meat Sci 59:23–30. Keener KM, LaCrosse JD, Babson JK. 2001. Chemical method for
Geankoplis CJ. 1993. Transport processes and unit operations. 3rd ed. determination of carbon dioxide content in egg yolk and egg albumen.
Englewood Cliffs, New Jersey: Prentice Hall. p 921. Poult Sci 80:983–7.
Geers C, Gros G. 2000. Carbon dioxide transport and carbonic anhydrase in Keith H, Wong S. 2006. Measurement of soil CO2 efflux using soda lime
blood and muscle. Physiol Rev 80:681–715. absorption: both quantitative and reliable. Soil Biol Biochem 38:1121–31.
Geng M, Duan Z. 2010. Prediction of oxygen solubility in pure water and Keller JU, Staudt R. 2005. Chapter 2. Volumetry/manometry. In: Gas
brines up to high temperatures and pressures. Geochim Cosmochim Acta adsorpt equilibria exp methods adsorpt isotherms. Springer US, Boston:
74:5631–40. Springer. p 79–116.
George P, Stratmann CJ. 1952. The oxidation of myoglobin to Kim JH, Lee YM. 2001. Gas permeation properties of
metmyoglobin by oxygen. 2. The relation between the first order rate poly(amide-6-b-ethylene oxide)–silica hybrid membranes. J Memb Sci
constant and the partial pressure of oxygen. Biochem J 51:418–25. 193:209–25.
Gill CO. 1988. The solubility of carbon dioxide in meat. Meat Sci 22:65–71. Kim HK, An DS, Yam KL, Lee DS. 2011. Package headspace composition
changes of chill-stored perishable foods in relation to microbial spoilage.
Girard F, Boyaval P. 1994. Carbon dioxide measurement in Swiss-type Packag Technol Sci 24:343–52.
cheeses by coupling extraction and gas chromatography. Lait 74:389–98.
Langø T, Mørland T, Brubakk A. 1996. Diffusion coefficients and solubility
Gros G, Moll W. 1971. The diffusion of carbon dioxide in erythrocytes and coefficients for gases in biological fluids and tissues: a review. Undersea
hemoglobin solutions. Pflug Arch Eur J Phy 324:249–66. Hyperb Med 23:247–72.
Gros J, Dussap C, Catté M. 1999. Estimation of O2 and CO2 solubility in Li H, Ashcraft DK, Freeman BD, Stewart ME, Jank MK, Clark TR. 2008.
microbial culture media. Biotechnol Prog 15:923–7. Non-invasive headspace measurement for characterizing oxygen-
Grote J. 1967. The oxygen diffusion constant in lung tissue and water and its scavenging in polymers. Polymer 49:4541–5.
temperature dependency. Pflug Arch Gesamte Physiol Menschen Tiere Li H, Tung KK, Paul DR, Freeman BD. 2011. Effect of film thickness on
295:245–54. auto-oxidation in cobalt-catalyzed 1, 4-polybutadiene films. Polymer
Grote J, Thews G. 1962. Requirements for the oxygen supply of heart 52:2772–83.
muscle tissue. Pflug Arch Gesamte Physiol Menschen Tiere 276:142–65. Liebsch G, Klimant I, Frank B, Holst G, Wolfbeis OS. 2000. Luminescence
Guillaume C, Schwab I, Gastaldi E, Gontard N. 2010. Biobased packaging lifetime imaging of oxygen, pH, and carbon dioxide distribution using
for improving preservation of fresh common mushrooms (Agaricus bisporus optical sensors. Appl Spectrosc 54:548–59.
L.). Innov Food Sci Emerg Technol 11:690–6. Lightfoot E. 1974. Transport phenomena and living systems: biomedical
Hempel A, O’Sullivan MG, Papkovsky DB, Kerry JP. 2013. Use of optical aspects of momentum and mass transport. New York: Wiley. p 495.
oxygen sensors to monitor residual oxygen in pre- and post-pasteurised Lin W-H, Chung T-S. 2001. Gas permeability, diffusivity, solubility, and
bottled beer and its effect on sensory attributes and product acceptability aging characteristics of 6FDA-durene polyimide membranes. J Memb Sci
during simulated commercial storage. LWT – Food Sci Technol 50:226–31. 186:183–93.
Hennessy S, Downey G, O’Donnell CP. 2009. Confirmation of food origin Lı́vanský K. 1982. Effect of temperature and pH on absorption of carbon
claims by Fourier transform infrared spectroscopy and chemometrics: extra dioxide by a free level of mixed solutions of some buffers. Folia Microbiol
virgin olive oil from Liguria. J Agric Food Chem 57:1735–41. 27:55–9.
Henry W. 1803. Experiments on the quantity of gases absorbed by water, at Long C, Anderson W, Finch C, Hickman J. 2012. CO 2 Measurement in
different temperatures, and under different pressures. Philos Trans R Soc microfluidic devices: facilitating biological application using a flow-through
London 93:29–42+274–6. cell. Genet Eng Biotechnol News 32:42–3.
Hikita H, Ishimi K, Ueda K, Koroyasu S. 1985. Solubility and diffusivity of Lühring P, Schumpe A. 1989. Gas solubilities (H2, He, N2, CO, O2, Ar,
carbon dioxide in aqueous slurries of kaolin. Ind Eng Chem Process Des CO2) in organic liquids at 293.2 K. J Chem Eng Data 34:250–2.
Dev 24:261–4.
Lund MN. 2007. Protein oxidation in meat during chill storage [PhD
Ho QT, Verlinden BE, Verboven P, Vandewalle S, Nicola BM. 2007. Thesis]. Denmark: Univ. of Copenhagen. p 203.
Simultaneous measurement of oxygen and carbon dioxide diffusivities in
pear fruit tissue using optical sensors. J Sci Food Agric 87:1858–67. Mahajan PV, Oliveira FAR, Montanez JC. 2007. Development of
user-friendly software for design of modified atmosphere packaging for fresh
Horwitz W, Latimer G. 2005. Official methods of analysis of AOAC Intl. and fresh-cut produce. Innov Food Sci Emerg Technol 8:84–92.
18th ed. In: Horwitz W, Latimer G, editors. Gaithersburg, Md.: AOAC Intl.
Mannapperuma JD, Singh RP, Montero ME. 1991. Simultaneous gas
Iglauer S. 2011. Dissolution trapping of carbon dioxide in reservoir formation diffusion and chemical reaction in foods stored in modified atmospheres.
brine – a carbon storage mechanism. In: Nakajima H, editor. Mass transf – J Food Eng 14:167–83.
Adv asp. Rijeka, Croatia: InTech. p 233–62.

284 Comprehensive Reviews in Food Science and Food Safety r Vol. 13, 2014 
C 2014 Institute of Food Technologists®
O2 and CO2 solubility and diffusivity . . .

Massey LK. 2003. Permeability properties of plastics and elastomers: a guide Portier S, Rochelle C. 2005. Modelling CO2 solubility in pure water and
to packaging and barrier materials. 2nd ed. Norwich, N.Y.: Plastic Design NaCl-type waters from 0 to 300 ◦ C and from 1 to 300 bar Application to
Library. p 615. the Utsira Formation at Sleipner. Chem Geol 217:187–99.
McCabe M, Maguire DJ, Lintell NA. 2005. The anomalous Einstein-Stokes Puligundla P, Jung J, Ko S. 2011. Carbon dioxide sensors for intelligent food
behaviour of oxygen and other low molecular weight diffusants. Adv Exp packaging applications. Food Control 25:328–33.
Med Biol 566:143–9. Rischbieter E, Schumpe A, Erlangen D, Wunder V. 1996. Gas solubilities in
McDonagh C, Burke CS, MacCraith BD. 2008. Optical chemical sensors. aqueous solutions of organic substances. J Chem Eng Data 41:809–12.
Chem Rev 108:400–22. Rodnight R. 1954. Manometric determination of the solubility of oxygen in
Miller CW, Nguyen MH, Rooney M, Kailasapathy K. 2003. Novel apparatus liquid paraffin, olive oil and silicone fluids. Biochem J 57:661–3.
to measure oxygen diffusion in gel-type foods. Food Aust 55:432–5. Roman O. 2012. Mesure et prédiction de la réactivité des lipides au cours du
Millero FJ, Huang F. 2003. Solubility of oxygen in aqueous solutions of KCl, chauffage d’huiles végétales à haute température [Ph.D Thesis]. France:
K2SO4, and CaCl2 as a function of concentration and temperature. J Chem AgroParisTech. p. 172.
Eng Data 48:1050–4. Roman O, Heyd B, Broyart B, Castillo R, Maillard M-N. 2013. Oxidative
Millero FJ, Huang F, Laferiere L. 2002. The solubility of oxygen in the reactivity of unsaturated fatty acids from sunflower, high oleic sunflower and
major sea salts and their mixtures at 25 ◦ C. Geochim Cosmochim Acta rapeseed oils subjected to heat treatment, under controlled conditions.
66:2349–59. LWT – Food Sci Technol 52:49–59.
Mills A, Hodgen S. 2005. Advanced Concepts in Fluorescence Sensing Part Rotabakk BT, Birkeland S, Jeksrud WK, Sivertsvik M. 2006. Effect of
A: Small Molecule Sensing. In: Geddes C, Lakowicz J, editors. Topics in modified atmosphere packaging and soluble gas stabilization on the shelf life
Fluorescence Spectroscopy. US, Boston: Springer. of skinless chicken breast fillets. J Food Sci 71:124–31.
p 119–62. Rotabakk BT, Lekang OI, Sivertsvik M. 2007. Volumetric method to
Mirhej ME. 1962. Studies on dissolved molecular oxygen in pure and sea determine carbon dioxide solubility and absorption rate in foods packaged
water [PhD Thesis]. Canada: The Univ. of British Columbia. in flexible or semi-rigid package. J Food Eng 82:43–50.
p 217. Rotabakk BT, Lekang OI, Sivertsvik M. 2010. Solubility, absorption and
Mishima K, Matsuo N, Kawakami A. 1996. Measurement and correlation of desorption of carbon dioxide in chicken breast fillets. LWT – Food Sci
solubilities of oxygen in aqueous solutions containing galactose and fructose. Technol 43:442–6.
Fluid Phase Equilib 118:221–6. Rowell M. 1995. Colorimetric method for CO 2 measurement in soils. Soil
Mishima K, Matsuo N, Kawakami A. 1997. Measurement and correlation of Biol Biochem 27:373–5.
solubilities of oxygen in aqueous solutions containing ribose and raffinose. Ruyssen T, Janssens M, van Gasse B, van Laere D, van der Eecken N, de
Fluid Phase Equilib 134:277–83. Meerleer M, Vermeiren L, Van Hoorde K, Martins JC, Uyttendaele M.
Neethirajan S, Jayas DS, Sadistap S. 2008. Carbon dioxide (CO2) sensors for Characterisation of Gouda cheeses based on sensory, analytical and
the agri-food industry – a review. Food Bioprocess Technol 2:115–21. high-field 1H NMR spectroscopy determinations: effect of adjunct cultures
Nevares I, del Alamo M, Gonzalez-Muñoz C. 2010. Dissolved oxygen and brine composition on sodium-reduced Gouda cheese. Int Dairy J
distribution during micro-oxygenation. Determination of representative 33:142–52.
measurement points in hydroalcoholic solution and wines. Anal Chim Acta Sadler GD, Roberts J, Cornell J. 1988. Determination of oxygen solubility in
660:232–9. liquid foods using a dissolved oxygen electrode. J Food Sci 53:1493–6.
Nock V, Blaikie R. 2010. Spatially resolved measurement of dissolved Sáenz C, Hernández B, Alberdi C, Alfonso S, Diñeiro JM. 2008. A
oxygen in multistream microfluidic devices. IEEE Sens J 10:1813–9. multispectral imaging technique to determine concentration profiles of
Nock V, Blaikie RJ, David T, Hendy SC, Brown IWM. 2009. Oxygen myoglobin derivatives during meat oxygenation. Eur Food Res Technol
control for bioreactors and in vitro cell assays. AIP Conf Proc p 67–70. 227:1329–38.
Oludare S, Olabiyi A, Macdonald O. 2009. Some biochemical and de Sales JA, Patrı́cio PSO, Machado JC, Silva GG, Windmöller D. 2008.
organoleptic changes due to microbial growth in minced beef packaged in Systematic investigation of the effects of temperature and pressure on gas
aluminium polyethylene trays and stored under chilled condition. Life Sci J transport through polyurethane/poly(methylmethacrylate) phase-separated
7:47–51. blends. J Memb Sci 310:129–40.
O’Mahony FC, O’Riordan TC, Papkovskaia N, Kerry JP, Papkovsky DB. Samapundo S, Everaert H, Wandutu JN, Rajkovic A, Uyttendaele M,
2006. Non-destructive assessment of oxygen levels in industrial modified Devlieghere F. 2011. The influence of headspace and dissolved oxygen level
atmosphere packaged Cheddar cheese. Food Control 17:286–92. on growth and haemolytic BL enterotoxin production of a psychrotolerant
Bacillus weihenstephanensis isolate on potato-based ready-to-eat food
Pasquali I, Andanson J-M, Kazarian SG, Bettini R. 2008. Measurement of products. Food Microbiol 28:298–304.
CO2 sorption and PEG 1500 swelling by ATR-IR spectroscopy. J Supercrit
Fluids 45:384–90. Sanches Silva A, Cruz JM, Sendon Garcıa R, Franz R, Paseiro Losada P.
2007. Kinetic migration studies from packaging films into meat products.
Pauchard JP, Flückiger E, Bosset JO, Blanc B. 1980. CO2 Löslichkeit, Meat Sci 77:238–45.
Konzentration bei Entstehung der Löcher und Verteilung in
Emmentalerkäse. Schweiz Milchw Forsch 9:69–73. Sandhya. 2010. Modified atmosphere packaging of fresh produce: current
status and future needs. LWT – Food Sci Technol 43:381–92.
Pénicaud C, Guilbert S, Peyron S, Gontard N, Guillard V. 2010. Oxygen
transfer in foods using oxygen luminescence sensors: influence of oxygen Sato K, Toda K. 1983. Oxygen uptake rate of immobilized growing Candida
partial pressure and food nature and composition. Food Chem 123: lipolytica. J Ferment Technol 61:239–45.
1275–81. Schotsmans W, Verlinden BE, Lammertyn J, Nicolaı̈ BM. 2004. The
Pénicaud C, Broyart B, Peyron S, Gontard N, Guillard V. 2011. Mechanistic relationship between gas transport properties and the histology of apple. J
model to couple oxygen transfer with ascorbic acid oxidation kinetics in Sci Food Agric 84:1131–40.
model solid food. J Food Eng 104:96–104. Schumpe A. 1993. The estimation of gas solubilities in salt solutions. Chem
Pénicaud C, Peyron S, Gontard N, Guillard V. 2012. Oxygen quantification Eng Sci 48:153–8.
methods and application to the determination of oxygen diffusion and Schumpe A, Luehring P. 1990. Oxygen diffusivities in organic liquids at
solubility coefficients in food. Food Rev Int 28:113–45. 293.2 K. J Chem Eng Data 35:24–5.
Pérez-Jiménez J, Arranz S, Tabernero M, Dı́az- Rubio ME, Serrano J, Goñi Schumpe A, Quicker G, Deckwer W. 1982. Gas solubilities in microbial
I, Saura-Calixto F. 2008. Updated methodology to determine antioxidant culture media. Adv Biochem Eng 24:1–38.
capacity in plant foods, oils and beverages: extraction, measurement and Schwartz S. 2003. Presentation of solubility data: units and applications. In:
expression of results. Food Res Int 41:274–85. Fogg PGT, Sangster J, editors. Chemicals in the Atmosphere – Solubility,
Piergiovanni L, Fava P. 1992. Carbon dioxide solubility in foods packaged Sources and Reactivity. Chichester, England: Wiley. p 452.
with modified atmosphere. 1: measure methods. Ind Aliment 31: Serpen A, Gökmen V, Fogliano V. 2012. Total antioxidant capacities of raw
424–30. and cooked meats. Meat Sci 90:60–5.
Pochat-Bohatier C, Sanchez J, Gontard N. 2006. Influence of relative Shanableh A. 2009. Carbon dioxide transfer with chemical equilibrium
humidity on carbon dioxide sorption in wheat gluten films. J Food Eng reactions: an alternative mathematical approach. Am J Eng Appl Sci
77:983–91. 2:726–34.


C 2014 Institute of Food Technologists® Vol. 13, 2014 r Comprehensive Reviews in Food Science and Food Safety 285
O2 and CO2 solubility and diffusivity . . .

Shiflett MB, Yokozeki A. 2005. Solubilities and diffusivities of carbon dioxide Tiemblo P, Laguna M, Garcia F, Garcia J, Riande E, Guzmán J. 2004. Gas
in ionic liquids: [bmim][PF 6] and [bmim][BF 4]. Ind Eng Chem Res transport properties of poly (2-ethoxyethyl methacrylate-co-2-hydroxyethyl
44:4453–64. methacrylamide). Macromolecules 37:4156–63.
Sidwell CG, Salwin H, Koch RB. 1962. The molecular oxygen content of Tomasko DL, Li H, Liu D, Han X, Wingert MJ, Lee LJ, Koelling KW. 2003.
dehydrated foods. J Food Sci 27:255–61. A review of CO2 applications in the processing of polymers. Ind Eng Chem
Simpson R, Almonacid S, Acevedo C. 2001a. Development of a Res 42:6431–56.
mathematical model for MAP systems applied to nonrespiring foods. J Food Treybal RE. 1980. Mass-transfer operations. 3rd ed. McGraw-Hill, New
Sci 66:561–7. York: McGraw-Hill Book Co. p 800.
Simpson R, Almonacid S, Acevedo C. 2001b. Mass transfer in pacific hake Unver AA, Himmelblau DM. 1964. Diffusion Coefficients of CO2, C2H4,
(Merluccius australis) packed in refrigerated modified atmosphere. J Food C3H6, and C4H8 in water from 6◦ to 65◦ C. J Chem Eng Data 9:428–31.
Process Eng 24:405–21. Valleguadarrama S, Espinosasolares T, Saucedoveloz C, Penavaldivia C. 2005.
Simpson R, Almonacid S, Acevedo C, Cortès C. 2004. Simultaneous heat Oxygen diffusivity in avocado fruit tissue. Biosyst Eng 92:197–206.
and mass transfer applied to non-respiring foods packed in modified Verberk W, Bilton D, Calosi P, Spicer J. 2011. Oxygen supply in aquatic
atmosphere. J Food Eng 61:279–86. ectotherms: partial pressure and solubility together explain biodiversity and
Simpson R, Carevic E, Rojas S, Simpson BR. 2007. Modelling a modified size patterns. Ecology 92:1565–72.
atmosphere packaging system for fresh scallops (Argopecten purpuratus). Vivier D, Compan D, Moulin G, Galzy P. 1996. Study of carbon dioxide
Packag Technol Sci 20:87–97. release from Feta cheese. Food Res Int 29:169–74.
Simpson R, Acevedo C, Almonacid S. 2009. Mass transfer of CO 2 in MAP Wang MY, Ogrydziak DM. 1986. Residual effect of storage in an elevated
systems: advances for non-respiring foods. J Food Eng 92:233–9. carbon dioxide atmosphere on the microbial flora of rock cod (Sebastes
Sirivicha S, Johnson AT, Douglas LW. 1990. Diffusion of carbon monoxide, spp.). Appl Environ Microbiol 52:727–32.
carbon dioxide, and sulfur dioxide in potato tissue. Trans ASABE Wang L, Sun D-W. 2003. Recent developments in numerical modelling of
33:189–98. heating and cooling processes in the food industry—a review. Trends Food
Siró BI, Plackett D, Sommer-Larsen P. 2010. A comparative study of oxygen Sci Technol 14:408–23.
transmission rates through polymer films based on fluorescence quenching. Watten BJ, Boyd CE, Schwartz MF, Summerfelt ST, Brazil BL. 2004.
Packag Technol Sci 23:301–15. Feasibility of measuring dissolved carbon dioxide based on head space partial
Sivertsvik M, Jensen JS. 2005. Solubility and absorption rate of carbon pressures. Aquac Eng 30:83–101.
dioxide into non-respiring foods. Part 3: cooked meat products. J Food Eng Whitcombe MJ, Parker R, Ring SG. 2005. Oxygen solubility and
70:499–505. permeability of carbohydrates. Carbohydr Res 340:1523–7.
Sivertsvik M, Jeksrud WK, Vågane Å, Rosnes JT. 2004a. Solubility and Wijnja H, Schulthess CP. 1999. ATR–FTIR and DRIFT spectroscopy of
absorption rate of carbon dioxide into non-respiring foods Part 1: carbonate species at the aged γ -Al2O3/water interface. Spectrochim Acta
development and validation of experimental apparatus using manometric Part A Mol Biomol Spectrosc 55:861–72.
method. J Food Eng 61:449–58.
de Wild HP, Peppelenbos HW. 2001. Improving the measurement of gas
Sivertsvik M, Rosnes JT, Jeksrud WK. 2004b. Solubility and absorption rate exchange in closed systems. Postharvest Biol Technol 22:111–9.
of carbon dioxide into non-respiring foods. Part 2: raw fish fillets. J Food
Eng 63:451–8. Wilke CR, Chang P. 1955. Correlation of diffusion coefficients in dilute
solutions. AICHE J 1:264–70.
Smiddy M, Fitzgerald M, Kerry JP, Papkovsky DB, O’ Sullivan CK,
Guilbault GG. 2002a. Use of oxygen sensors to non-destructively measure Wise DL, Houghton G. 1969. Solubilities and diffusivities of oxygen in
the oxygen content in modified atmosphere and vacuum packed beef: hemolyzed human blood solutions. Biophys J 9:36–53.
impact of oxygen content on lipid oxidation. Meat Sci 61: Wong B, Zhang Z, Handa YP. 1998. High-precision gravimetric technique
285–90. for determining the solubility and diffusivity of gases in polymers. J Polym
Smiddy M, Papkovskaia N, Papkovsky DB, Kerry JP. 2002b. Use of oxygen 2025–32.
sensors for the non-destructive measurement of the oxygen content in Yasuda T, Yonemura S, Tani A. 2012. Comparison of the characteristics of
modified atmosphere and vacuum packs of cooked chicken patties; small commercial NDIR CO2 sensor models and development of a portable
impact of oxygen content on lipid oxidation. Food Res Int 35: CO2 measurement device. Sensors 12:3641–55.
577–84. Yokozeki A, Shiflett MB. 2011. The solubility of CO2 and N2O in olive oil.
Soccol MCH. 2003. Use of modified atmosphere in seafood preservation. Fluid Phase Equilib 305:127–31.
Brazilian Arch Biol 46:569–80. Zaritzky NE, Bevilacqua AE. 1988. Oxygen diffusion in meat tissues. Int J
Solomos T. 1987. Principles of gas exchange in bulky plant tissues. Heat Mass Transf 31:923–30.
HortScience 22:766–71. Zeng AP. 1995. Effect of CO2 absorption on the measurement of CO2
Tammam J, Williams A, Banks J, Cowie G, Lloyd D. 2001. Membrane inlet evolution rate in aerobic and anaerobic continuous cultures. Appl Microbiol
mass spectrometric measurement of O2 and CO2 gradients in cultures of Biotechnol 42:688–91.
Lactobacillus paracasei and a developing Cheddar cheese ecosystem. Int J Zhao P, Cai WJ. 1997. An Improved Potentiometric p CO2 Microelectrode.
Food Microbiol 65:11–22. Anal Chem 69:5052–8.
Thiele T, Kamphoff M, Kunz B. 2006. Modeling the respiration of Zhao Y, Wells JH, Mcmillin KW. 1994. Applications of dynamic modified
Pseudomonas fluorescens on solid-state lettuce-juice agar. J Food Eng atmosphere packaging systems for fresh red meats: review. J Muscle Foods
77:853–7. 5:299–328.

286 Comprehensive Reviews in Food Science and Food Safety r Vol. 13, 2014 
C 2014 Institute of Food Technologists®

You might also like