Available online at www.sciencedirect.

com

Innovative Food Science and Emerging Technologies 9 (2008) 280 – 289

www.elsevier.com/locate/ifset

Drying of foods using supercritical carbon

dioxide — Investigations with carrot
Z.K. Brown ⁎, P.J. Fryer, I.T. Norton, S. Bakalis, R.H. Bridson
Centre for Formulation Engineering, Department of Chemical Engineering, University of Birmingham, Edgbaston, Birmingham B15 2TT, UK
Received 19 January 2007; accepted 13 July 2007

Abstract

The use of supercritical carbon dioxide (scCO2) for the removal of moisture from cylindrical pieces of carrot has been investigated. The
experiments were carried out at 20 MPa pressure and the effects of temperature and co-solvent (ethanol) addition were examined. At the
investigated conditions, comparisons with air-drying indicated that drying kinetics and the associated drying mechanisms differed between the
techniques. The microstructural characteristics of carrot pieces that had been dried using the different techniques were compared using X-ray
microtomography and light microscopy. Carrots dried in the supercritical fluid environment were seen to retain their shape much better than air-
dried carrots which underwent shrinkage. Samples dried in ethanol-modified scCO2 possessed less dense structures and consequently displayed
more favourable rehydrated textural properties than the air-dried equivalents.
© 2007 Elsevier Ltd. All rights reserved.

Keywords: Carrot; Air-drying; Supercritical carbon dioxide; Drying kinetics; Microstructure

Industrial relevance: Drying is a common unit operation in food processing, but the rehydrated product is often of very poor quality.
Comparatively little work has been done on optimising drying for quality of the final, rehydrated product. Supercritical drying is a possible way of
maintaining product microstructure, and this paper describes experiments in which the supercritical drying of carrot is studied and compared with
conventional processes.

1. Introduction product's microstructure. This is unfortunate given that food

structure influences nutritional availability, chemical and
The main purpose of drying foods is to lower the moisture microbiological stability, texture, physical properties and
content in order to reduce water activity and prevent spoilage. transport properties (Aguilera, 2005; Aguilera, Chiralt, &
Water activity is a critical factor that determines shelf life, with Fito, 2003; Aguilera, Stanley, & Baker, 2000; De Roos, 2003).
most bacteria unable to grow below a value of 0.91 and moulds Conventional air-drying is the most commonly used dehydra-
ceasing to grow below 0.80 (Beuchat, 1983). Water activity also tion process in the food and chemical industry, producing products
plays a significant role in determining the activity of enzymes that are characterised by low porosity and high apparent density
and vitamins in foods and can have a major impact on their (Krokida & Maroulis, 1997). The high temperatures commonly
colour, taste and aroma. Additionally, moisture removal reduces used during industrial air-drying (typically 65–85 °C) cause
the weight and the bulk of food products to facilitate transport damage to the microstructure and may also have a negative
and storage. While imparting these benefits, loss of moisture influence on the colour, texture, taste, aroma and nutritional value
during drying may also inflict undesirable effects on the of the product thereby influencing the quality of both the dried and
subsequently rehydrated product. Although there are several
alternatives to air-drying (e.g. microwave-, freeze-, osmotic-and
⁎ Corresponding author. Tel.: +44 121 414 6965; fax: +44 121 414 5324. vacuum-drying), none is ideal in every respect (Vega-Mercado,
E-mail address: [email protected] (Z.K. Brown). Marcela Gongora-Nieto, & Barbosa-Canovas, 2001). For example,
1466-8564/$ - see front matter © 2007 Elsevier Ltd. All rights reserved.
doi:10.1016/j.ifset.2007.07.003
 Z.K. Brown et al. / Innovative Food Science and Emerging Technologies 9 (2008) 280–289 281

while microwave drying promotes quick drying, difficulties in Chimowitz, Kumaran, Shapir, & Fauchet, 1997). A particular
controlling the rapid mass transport may cause damage in the form advantage of supercritical fluid drying is that vapour-liquid
of “puffing”. Additionally, this technique is associated with high interfaces can be avoided. As such, capillary-induced tensile
start-up costs and ongoing costs relating to the regular replacement stresses seen during air-drying are not experienced by the
of expensive magnetrons (Nijhuis et al., 1998). In contrast, freeze- product, thus helping to preserve structure. Additionally, as CO2
drying is known to produce products of excellent quality, allowing has a low critical temperature (31.1 °C) it is possible to exploit
significant structural preservation (Sinesio, Moneta, Spataro, & this advantage at relatively low temperatures, much lower than
Quaglia, 1995). Indeed it has been reported that the products of in conventional drying. However, scCO2 is a non-polar solvent
freeze-drying are of higher porosity than those of air-drying and hence the solubility of water in this fluid is modest. From
(Karathanos, Kanellopoulos, & Belessiotis, 1996; Marabi, & the work of Sabirzyanov, Il'in, Akhunov, and Gumerov (2002)
Saguy, 2004, Rahman, Al-Amiri, & Al-Bulushi, 2002) and other water solubility in compressed CO2 can be estimated to be
drying methods such as microwave- (Tsami, Krokida, & Drouzas, approximately 4 mg/g at 50 °C/20 MPa and 2.5 mg/g at 40 °C/
1999), and vacuum-drying (Rahman et al., 2002). However, the 20 MPa (King, Mubarak, Kim, & Bott, 1992; Sabirzyanov
technique is generally considered to be expensive and energy- et al., 2002). A widely-used strategy employed to increase the
consuming. (Lombraña & Díaz, 1987; Ratti, 2001). Therefore, a solubility of polar substrates in scCO2 is to add small quantities
place exists for new drying techniques that could be used in the food of co-solvent to the system so that the CO2 is then said to be
industry. In this work the possibility of drying with scCO2 has been “modified”. Co-solvents are usually organic solvents, including
examined and the outcomes compared with a conventional air- for example ethanol, acetone and ethyl acetate and can have a
drying technique. large effect on the solubilities of many compounds in scCO2
Supercritical fluids (Fig. 1), in particular scCO2, are already (Kopcak & Mohamed, 2005; Tanaka & Nakanishi, 1994).
widely used in the food industry, primarily as alternatives to The aim of the work described here is to evaluate the
toxic organic solvents used in extraction processes (Brunner, potential of supercritical drying for foods using both pure
2005). One of the most widely cited examples of such a use scCO2 and ethanol-modified scCO2. Carrot has been used as a
relates to the removal of caffeine from coffee (McHugh & test material because its structure is well-characterised in
Krukonis, 1986), but many other flavours, fragrances and oils previous studies and it is commonly employed as a dried
have also been extracted using scCO2 (King & Bott, 1993). ingredient.
ScCO2 has also been used in conjunction with an extrusion
process to produce microcellular starch-based foams (Alavi, 2. Materials and methods
Rizvi, & Harriott, 2003a,b). Drying techniques using scCO2,
although not previously reported for food processing, have 2.1. Materials
attracted interest in other industries e.g. for drying gels
composed of silica (Takishima et al., 1997; Rangarajan & Fresh carrots were purchased from the local supermarket.
Lira 1991) or chromia (Abecassis-Wolfovich et al., 2003) to Prior to their use in the drying experiments, they were stored in
produce aerogels and also to create other inorganic porous a refrigerator at 4 °C within moisture-resistant packaging. CO2
structures (Namatsu, Yamazaki, & Kurihara, 1999; von Behren, (liquid withdrawal) was supplied by BOC (Guildford, UK).

Fig. 1. Phase diagram of a pure substance. The supercritical state is achieved above the critical point, which is the highest temperature and pressure at which a vapour/
liquid equilibrium can exist. Supercritical fluids are neither liquids nor gases, but do possess several advantageous characteristics of both these states of matter. The
liquid-like densities of supercritical fluids are high enough for them to act as solvents, while gas-like viscosities and self-diffusivities higher than in liquid solvents may
facilitate mass transfer and reaction rates.
282 Z.K. Brown et al. / Innovative Food Science and Emerging Technologies 9 (2008) 280–289

Absolute ethanol (99.9%) was supplied by Fisher Scientific and pressure using an air-driven liquid pump. The compressed CO2
was of Analar grade. was pre-heated to the desired experimental temperature (40, 50
or 60 °C) via a heat exchanger coil upstream of the pressure
2.2. Carrot preparation and initial characterisation vessel. Once the desired temperature and pressure (20 MPa) had
been reached, the exit micrometering valve was opened and a
A cork borer was used to cut cylindrical pieces (L = 2.5 cm constant flow rate of ∼ 2 l/min (at STP) achieved. The
and d = 0.4 cm) from the cortex of the carrot. As the core and micrometering valve was a component of the automated back
cortex have been shown to have different moisture transfer pressure regulator that was computer controlled (Thar Technol-
properties at different temperatures (Srikiatden & Roberts, ogies Inc., Pittsburgh, PA, USA). The flow rate was monitored
2006), only the cortex was used to ensure consistency. throughout the experiments using a wet test flow meter and
For each experiment seven carrot cylinders were cut from adjusted when necessary using the micrometering valve.
one whole carrot. Four were then used in a drying experiment Constant pressure was maintained via a back pressure regulator.
while the other three were used to determine the initial average For experiments carried out with ethanol-modified CO2, an
total moisture content of the carrot. The initial moisture content HPLC pump was employed to deliver the correct quantity of
of the carrot pieces was measured by drying the samples at ethanol. At the end of the required processing time for
70 °C in an oven to constant mass and comparing to the original experiments in which ethanol was added, an additional
mass. The initial moisture content was found to be typically 30 min phase in which pure, unmodified scCO2 was passed
∼ 90% w/w which is expected for vegetables of this kind. over the samples was also included and which accounted for
Both cooked and raw carrot samples were used in this work. 30 min of the total drying time reported. This ensured that all
Cooking was carried out by placing the carrot piece in boiling ethanol was removed from the system prior to depressurisation,
water for 10 min. thus preventing the carrots from being wetted by residual
solvent. At the end of every drying experiment the micrometer-
2.3. Supercritical drying ing exit valve was opened to allow depressurisation at a
constant rate of 0.4 MPa/min, controlled by the automated back
A schematic illustration of the supercritical drying rig is pressure regulator. As it was necessary to depressurise the
shown in Fig. 2. Four cylindrical carrot pieces were placed in a vessel before samples could be removed for analysis, individual
stainless steel mesh sample holder positioned in the centre of the experiments were carried out for each time point indicated in
50 ml pressure vessel. CO2 from the supply cylinder was passed Figs. 4, 5, and 6. The moisture content at each time point was
through a dessicator tube before it was cooled to a liquid state calculated from the total initial moisture content and the mass
(to allow efficient pumping) and compressed to the desired loss during drying. It was assumed that all mass loss was due to

Fig. 2. Schematic representation of the experimental apparatus. Key: V1 = carbon dioxide inlet valve; DT = dessicator tube; NRV1, 2, 3 = non-return valves; V2 = ball
valve; F1 = in-line filter; BPR = back pressure relief valve; HE1 = refrigerated heat exchanger (cooling); PUMP1 = pneumatic liquid CO2 pump; V3 = pneumatic pump
air valve; PRV = safety pressure relief valve; P1 = pressure gauge; HPLC = HPLC co-solvent pump; MV1, 2 = micrometering valves; HE2 = heat exchanger (heating);
T1, 2 = thermocouples; PV1 = reaction pressure vessel; ABPR = automated back pressure regulator; WTM = wet test flow meter.
 Z.K. Brown et al. / Innovative Food Science and Emerging Technologies 9 (2008) 280–289 283

water removal. All experiments (each involving 4 pieces of 2.7. Texture analysis
carrot) were carried out in triplicate. The mean values from each
of the three independent experiments were combined and an A texture analyser (TA-XT2i, Stable Micro System, UK)
overall mean ± one standard deviation of the mean calculated. with a cylinder probe (2 mm diameter) was used for puncture
The effects of experimental conditions on the overall mean compression test analysis. The probe was used to measure the
moisture contents (and other mean values discussed in this maximum force required to penetrate an individual rehydrated
contribution) were statistically examined. The Kruskal–Wallis cylindrical piece of carrot, to a depth of 2.5 mm, positioned
non-parametric statistical hypothesis test was employed to test horizontally over a heavy duty platform. The speed of approach
for differences between mean values where results relating to of the probe was 1 mm/s and a 5 kg load cell was used. Three
three or more conditions were under simultaneous scrutiny. The independent experiments (each comprising at least three carrot
Mann-Whitney U-test was used to examine for differences pieces) were performed for each drying method and rehydration
where experiments were paired for comparison. The chosen temperature, and the mean maximum force (N) recorded.
level of significance, α, for all tests was 0.05.
3. Results and discussion
2.4. Air-drying
3.1. Drying
A sample holder (the same as the one used in supercritical
drying) was used to hold four cylindrical carrot pieces in the As mentioned in the introduction, addition of a co-solvent
centre of a fan oven pre-heated to 40, 50 or 60 °C. The samples can favourably influence the solubility of polar substances in
were briefly removed at regular intervals and their mass scCO2. Their mechanism of action is thought to involve specific
determined over time until b 10% w/w moisture content was chemical and physical interactions that exist between the co-
reached. The moisture content at each time point was solvent and solute, including hydrogen bonding and dipole–
calculated from the total initial moisture content and the dipole attractions (Ekart et al., 1993; Ke, Jin, Han, Yan, & Shen,
mass loss during drying. It was assumed that all mass loss was 1997). Therefore the results shown in Fig. 3 are as expected i.e.
due to water removal. All drying experiments were carried out the extent of moisture removal over a 90 min drying period
in triplicate. increased with increasing quantities of ethanol. It should be
noted however that carrots processed in an environment of pure,
2.5. X-ray microtomography unmodified scCO2 also experienced considerable moisture loss.
Over half of the original moisture content of the carrot was lost
This technique was used to study the microstructure of the over this time indicating that drying in CO2 alone is possible.
dried carrot pieces immediately after drying, allowing for non- Drying profiles for carrots dried in pure scCO2, scCO2
invasive analysis of the samples' internal structure. The samples modified with 6 mol% ethanol and by air-drying are presented
were scanned using a high resolution desktop X-ray micro-CT in Fig. 4. Once again, the inclusion of ethanol can be seen to
system (Skyscan 1072, Skyscan, Belgium) consisting of a
microfocus sealed X-ray tube with a spot size of 5 μm operating
at a voltage of 100 kV and current of 96 μA. A more detailed
explanation of the technique is provided by Lim and Barigou
(2004), and its application to the quantification of food
microstructure described by Mousavi, Miri, Cox, and Fryer
(2007). The radiograph image captured was reconstructed using
a filtered back-projection algorithm allowing the internal
structure to be visualised.

2.6. Rehydration

Dehydrated carrot pieces (6–10% w/w moisture content)

were immersed in a glass beaker containing 100 ml distilled
water, maintained at 50 ± 3 °C or 80 ± 5 °C. The samples were
removed at regular time intervals (∼ every 5 min), blotted on
tissue paper, and their mass determined over a total period of
40 min. Each experiment was carried out at least six times (on Fig. 3. The moisture content of carrot pieces dried for 90 min using modified
two pieces of carrot from each of three independent drying carbon dioxide over a range of ethanol concentrations. All measurements were
experiments) at both of the rehydration temperatures. Results carried out at 20 MPa and 50 °C. Each point represents the overall mean from
are reported in terms of normalized moisture content which is three independent drying experiments ± one standard deviation of the mean
(vertical error bars). Horizontal error bars indicate variability in ethanol
defined as moisture content of rehydrated product (kg water/kg concentration ± one standard deviation of the mean. ⁎Normalised moisture
dry matter) ÷ the initial moisture content, before drying (kg content is the moisture content following the experiment (kg water/kg dry
water/kg dry matter). matter) ÷ the initial moisture content (kg water/kg dry matter).
284 Z.K. Brown et al. / Innovative Food Science and Emerging Technologies 9 (2008) 280–289

In summary it appears that, at the investigated conditions,

supercritical drying with pure CO2 proceeds via a different
mechanism to that occurring in air-drying. To investigate this
hypothesis further, the effect of pre-cooking the carrots before
drying was investigated (Fig. 5). Cooking destroys the cellular
structure of carrots, thereby decreasing the tortuosity and
increasing the effective diffusivity through the carrot during the
drying process. For air-drying, pre-cooking was shown to have
a statistically significant effect on the moisture content achieved
at all comparable time points and the overall time required to
dry to low moisture content was reduced by pre-cooking. In
contrast, such effects were not seen when drying with scCO2;
Fig. 5 shows no difference in the moisture content-time curves
of cooked and uncooked carrot. This suggests that in contrast to
air-drying the internal structure of the carrot does not control the
Fig. 4. Drying profiles for raw carrot pieces dried in air (▲), pure scCO2 (■) and
scCO2 modified with 6 mol% ethanol⁎ (○). All experiments were carried out at rate of moisture removal during supercritical drying. At these
20 MPa and 50 °C. Each point represents the overall mean from at least three conditions therefore, it is hypothesised that supercritical drying
independent sets of measurements ± one standard deviation of the mean. ⁎30 min with pure CO2 is instead controlled by the rate of mass transfer
of each time point includes drying without ethanol to purge the system of this from the surface. Future work will include a detailed
solvent before depressurisation (see text).
investigation of the heat and mass mechanisms during
supercritical drying. However, the limitations of scale in the
equipment are such that only 0.4 cm sizes were used;
increase the overall rate of drying in the supercritical systems. investigation of larger particles would be useful to identify
The differences in moisture content between carrots dried in controlling resistances.
pure or modified CO2 become statistically significant at and Fig. 6 illustrates the effect of temperature on moisture
beyond 90 min. Before this time, the effects of ethanol addition removal, over the range 40–60 °C. Greater moisture removal at
are less obvious for two main reasons. Firstly, the times comparable time points was achieved by increasing the
indicated include the 30 min purge time, during which pure CO2 temperature for both air- and supercritical drying. For air-
alone is passed over the sample to remove ethanol from the drying, this is as expected and is in line with the results of
system before depressurisation. Thus a 40 min experiment, for others. For example, Krokida et al. (2003) showed that air
example, involves exposing the samples to modified CO2 for temperature had a greater effect than air humidity or air velocity
only 10 min. Secondly, it takes time for the ethanol on the drying rates for several different types of vegetables. In
concentration within the pressure vessel to become fully the case of supercritical drying, such a relationship is supported
established as modified CO2 replaces the pure CO2 present in by the work of Sabirzyanov et al. (2002) and King et al. (1992)
the vessel at the start of an experiment. who showed that the solubility of water in compressed CO2
At the conditions employed, drying to any particular increases with temperature at pressures down to 5 MPa.
moisture content was seen to take longer for supercritical Solubility below this pressure was not reported.
drying than for air-drying, although comparisons of this nature
must be treated with caution due to the different flow regimes
employed for each. The shape of the drying profiles in Fig. 4
can also be seen to differ between air- and supercritical drying
with pure CO2. A plot (not shown) of ln (moisture content)
versus time for air-drying is a straight line indicating that air-
drying kinetics follow a 1st order relationship. This is consistent
with the falling rate period commonly associated with the latter
part of air-drying, where the rate of transport of moisture
through the food matrix, determined by the mechanism of
internal moisture transport, becomes limiting (Krokida, Kar-
athanos, Maroulis, & Marinos-Kouris, 2003). A preceding
constant drying rate period may be present in the earliest stages
of the drying experiment. In the case of supercritical drying with
pure CO2, the straight line relationship between moisture
content and time indicates zero order kinetics. It is more
difficult to examine the kinetics for the data obtained when Fig. 5. Comparison of carrot drying profiles for cooked (△) and raw samples
(▲) using air-drying, or cooked (□) and raw (■) samples dried with pure scCO2.
drying with modified CO2. As discussed above, early time-point All experiments were carried out at 20 MPa and 50 °C. Each point represents the
experiments carried out with this fluid include a substantial overall mean from at least three independent sets of measurements ± one
contribution from pure CO2 alone. standard deviation of the mean.
 Z.K. Brown et al. / Innovative Food Science and Emerging Technologies 9 (2008) 280–289 285

those used by Krokida and Marinos-Kouris), and possibly

alternative imbibition mechanisms brought about by differences
in porosity and chemical composition of the dried material
(Witrowa-Rajchert & Lewicki, 2006). The first and last of these
explanations may be linked i.e. drying temperature may affect
the porosity of the dried product. Additionally, a temperature as
high as 80 °C has a detrimental impact on sample texture during
rehydration and therefore comparison with rehydration at 50 °C
requires consideration of not only the temperature change but
also of the structural changes occurring.
Images from analysis with X-ray microtomography and
light-microscopy help to explain some of the observations noted
above as they provide information about the internal micro-
structure of the carrots. Fig. 8 shows representative 2-D
Fig. 6. Comparison of carrot drying profiles at 40 °C (▲), 50 °C ( ) and 60 °C
(△) using air-drying or drying with pure scCO2 (40 °C (■), 50 °C ( ),60 °C reconstructed images of the internal structure of carrot pieces
(□)). Each point represents the overall mean from at least three independent sets dried by each technique. Firstly the shrinkage behaviour is
of measurements ± one standard deviation of the mean. illustrated by the images. The general shape of the slice through
the sample indicates the degree of shrinkage and the darkness of
3.2. Rehydration the image gives an indication of the density of the dried
pieces — the darker the image, the more dense the sample as the
Fig. 7 shows normalized moisture content at 5 min (Fig. 7a)
and 40 min (Fig. 7b) against the initial normalized moisture
content (before rehydration) for cooked and raw carrot pieces
dried by air, pure scCO2 or modified scCO2 (6 mol% ethanol).
All results were obtained following rehydration at 50 °C. For
rehydration after 5 min, no conclusive differences between
samples dried through different techniques can be seen.
However, the observed trends suggest that the extent of
rehydration for carrot pieces dried with modified scCO2 may
be greater than for the others. A potential explanation for this is
discussed below. However, after 40 min of rehydration these
differences in moisture content had disappeared, suggesting
samples dried using modified scCO2 may have a faster ‘instant’
rehydration rate than those samples dried using the alternative
techniques. The temperature employed during supercritical
drying had no effect on the amount of moisture gained after
5 min or 40 min of rehydration, and therefore the results at
60 °C are not shown in this figure. Furthermore, although pre-
cooking increased the rate of air-drying, its effects did not
persist with respect to rehydration i.e. no difference was seen
between the normalized moisture contents obtained for air-dried
cooked and uncooked carrot at either 5 or 40 min rehydration
time. In contrast, cooking was seen to affect the outcome of
supercritical drying, with higher moisture contents being
generally observed with pre-cooked sample but only after
40 min of rehydration.
Interestingly, results for equivalent rehydration experiments
carried out at 80 °C showed very similar trends to those
conducted at 50 °C and as such are not shown separately. This is
surprising, as it contradicts the results of Krokida and Marinos-
Kouris (2003) who showed that temperature has a significant
effect on both the rate of rehydration and the equilibrium
moisture content achieved. We suggest that this difference Fig. 7. Normalized moisture content during rehydration plotted against initial
between our work and that of these authors may be due to normalized moisture content before rehydration at, a) 5 min and b) 40 min.
Samples were dried using: air-drying at 50 °C (▲), pure scCO2 at 50 °C (■),
several factors including different drying temperatures (50 °C ethanol-modified scCO2 at 50 °C (×), air-drying (after cooking) at 50 °C (△)
versus 70 °C respectively), different sample sizes (the volume and pure scCO2 (after cooking) at 50 °C (□). Each point is a measurement from
of carrot pieces before drying in our experiments was a tenth of a single piece of carrot.
286 Z.K. Brown et al. / Innovative Food Science and Emerging Technologies 9 (2008) 280–289

X-rays find it harder to penetrate. In the case of the image for the distinction between cellular structures and the drying-induced
raw, undried carrot (Fig. 8a) the darkness is also due to ∼90% pores to be made; i.e. while high-temperature rehydration or
water in the sample. During air-drying, shrinkage and collapse cooking destroys the cellular structure they do not destroy the
of the product was observed, producing a dense closely-packed pores. The pores were also visible using cryo-scanning electron
cellular structure with limited intercellular spaces, expected to microscopy (not shown) which indicated that they were not
restrict intercellular diffusion and hence rehydration (Fig. 8e). simply an artefact of the light-microscopy technique. The
The sample processed in pure scCO2 at 50 °C (Fig. 8b) appears presence of the pores provides a further explanation for the
to have a comparable density to the air-dried sample, suggesting difference in normalized moisture content seen during rehydra-
a reason why the rehydration properties were similar. Note tion — the highest after 5 and 40 min of rehydration, being
however that the deviation from cross-sectional circularity is associated with samples containing these new features. It is
less pronounced after the supercritical drying, indicating that the proposed that these pores may facilitate the movement of water
original structure was maintained to a greater extent using this into the samples' internal structure.
technique. In addition to the preservation of cross-sectional
geometry, the overall volume of the carrot pieces dried in the 3.3. Texture
supercritical fluid was qualitatively observed to be closer to that
of the original. Texture is one of the most important quality criteria for
Drying with modified scCO2 (6 mol%) (Fig. 8d) ensured sensory consumer acceptance of freshly prepared and processed
even greater retention of the original volume and shape. The fruit and vegetable products. Fig. 10 shows the results of the
low density, porous structure with increased intercellular space puncture test used as an indicator of textural properties of the
may explain the increased moisture gain seen for these samples dehydrated–rehydrated samples. Fresh raw carrot and cooked
in the early stages of rehydration, and also for the fact that they carrot were used as controls.
floated during rehydration while all others sank. However, the The structural and textural properties of fruits and vegetables
volume was not retained indefinitely. Some shrinkage, resulting relate highly to the integrity of cell wall components. For
in a volume reduction of the structures dried by modified scCO2 example, heat-induced texture degradation is attributed to pectin
was seen after leaving the dried samples for a few hours at room solubilisation and pectin depolymerisation via β-elimination
temperature and pressure. Similarly, highly porous materials (Sila, Smout, Elliot, Van Loey, & Hendrikx, 2006). Rehydrating
prepared by freeze-drying are susceptible to post-drying at 80 °C was seen to have an equivalent effect on the required
collapse when optimal conditions for storage are not maintained puncture force as pre-cooking, presumably as both lead to
(Levi & Karel, 1995). Therefore, optimisation of storage cellular degradation. As such, trends relating process conditions
conditions following drying may prevent the collapse seen. to puncture force are best seen for samples rehydrated at 50 °C.
The collapse may be due to the small amounts of water still At this rehydration temperature, the maximum average puncture
present within the open porous structure. A possible solution to force values decreased in the following order: air-drying N dry-
this would be to introduce a rapid heating step to remove any ing with pure scCO2 N drying with modified scCO2. Application
final moisture. of the Kruskal–Wallis test to these data sets indicated that any
Supercritical drying with pure CO2 at 60 °C produced difference between the three was right on the limit of statistical
structures (Fig. 8c) with less shrinkage and lower density than significance at α = 0.05. However, the general trends relate well
the equivalent process at 50 °C. Despite these structural to the density and porosity of the various dried samples
differences, no differences were observed between the rehydra- suggesting that the more porous, low density dried samples
tion properties. (i.e. those dried with modified scCO2) have a softer texture when
Light-microscopy images captured following rehydration rehydrated than those with low porosity and high density (i.e.
(Fig. 9) showed visible “pores” within those samples that had those subjected to air-drying).
been dried using modified scCO2. These were also evident in Puncture forces in this work were similar to those reported by
cooked samples dried with pure scCO2 but not in raw carrot that Lin, Durance, and Scaman (1998). Air-dried slices produced by
had been dried in the same manner. Images taken after these authors required a force of 18.1 N to puncture them, a force
rehydration at 80 °C or of pre-cooked samples, allowed greater comparable to that required to puncture air-dried carrot cylinders

Fig. 8. Reconstructed 2D cross-sectional X-ray images of (a) raw carrot and uncooked carrot pieces dried in: (b) pure scCO2 at 50 °C; (c) pure scCO2 at 60 °C;
(d) modified scCO2 at 50 °C; (e) air at 50 °C. Before drying each carrot piece was circular in cross-sectional shape.
 Z.K. Brown et al. / Innovative Food Science and Emerging Technologies 9 (2008) 280–289 287

Fig. 9. Light-microscopy images showing: (a) raw carrot (b) raw carrot dried in pure scCO2 at 50 °C and rehydrated at 50 °C; (c) raw carrot dried in pure scCO2 at
50 °C and rehydrated at 80 °C; (d) raw carrot dried in modified scCO2 and rehydrated at 50 °C; (e) raw carrot dried in modified scCO2 and rehydrated at 80 °C;
(f) cooked carrot dried in pure scCO2 at 50 °C and rehydrated 50 °C.

rehydrated at 50 °C in this work (16.32 ± 1.17 N). The force dissolved solutes that contribute to case hardening. Lin et al.
required in this work to puncture rehydrated carrots that had also related their findings to sensory analysis, with the VMD
been dried using modified scCO2,11.27 ± 2.38 N, is comparable samples receiving the better sensory ratings for texture (in both
with the value of 11 N obtained by Lin et al. (1998) for samples dried and rehydrated states). Given the similarities between the
dried using vacuum microwave drying (VMD). puncture forces associated with VM-dried and samples dried
“Case hardening” during drying was stated to be the reason with modified scCO2, it is anticipated that the texture of the
for the differences seen between air-dried and VM-dried latter may also be preferable to the air-dried equivalents,
samples by Lin et al. (1998). This phenomenon is associated although sensory evaluations are required to confirm this.
with systems where water removal from the surface is faster While case hardening can be used as a possible explanation
than the rate at which water migrates from the interior. The for the hard texture associated with the air-drying, that relating
surface therefore dries to form a hard layer containing to drying in pure scCO2 can not yet be fully explained.
previously-dissolved solutes. In VMD, heat is generated within
the product resulting in in situ vaporisation of water which is 3.4. Colour
able to rapidly diffuse out of the tissue without carrying
Carrots that had been dried with both pure and modified
scCO2 appeared to be paler in colour than raw carrot. Initially it
was hypothesised that this was due primarily to the extraction of
carotenes (α- and β-carotenes) as these substances are known to
be soluble in scCO2 (Cygnarowicz, Maxwell, & Seider, 1990;
Reverchon, 1997). Upon rehydration however, the colour
returned significantly so that the difference between fresh and
dried carrot became almost indistinguishable to the naked eye.
This suggests carotene extraction had not been as extensive as
expected. Instead, it is suggested that supercritical drying leads
to a decrease in the transparency of the cell walls that is
reversible on rehydration.
In contrast, air-dried samples were darker in colour than
fresh raw carrot. The collapse of the cellular structure, shrinkage
and therefore increased carotene concentration within a specific
area provides an explanation for this.

Fig. 10. Puncture testing of carrot dried using different techniques and 4. Conclusions
rehydrated at 50 °C (dark bars) and 80 °C (diagonal dark stripes). Non-dried raw
carrot (spots) and cooked carrot (checks) are shown for comparison. Each value
represents the overall mean from at least three independent sets of It has been demonstrated that carrots can be dried in pure and
measurements ± one standard deviation of the mean. modified scCO2, promoting better retention of original structure
288 Z.K. Brown et al. / Innovative Food Science and Emerging Technologies 9 (2008) 280–289

than air-drying. This novel technique has also been shown to King, M. B., & Bott, T. R. (Eds.). (1993). Extraction of natural products using
produce less dense microstructural characteristics in samples near-critical solvents. Glasgow: Blackie Academic & Professional.
King, M. B., Mubarak, A., Kim, J. D., & Bott, T. R. (1992). The mutual
dried using modified scCO2. The additional creation of pores solubilities of water with supercritical and liquid carbon dioxide. Journal of
during drying with modified scCO2 and drying pre-cooked Supercritical Fluids, 5, 296−302.
carrot with pure scCO2 may increase the total moisture gained Kopcak, U., & Mohamed, R. S. (2005). Caffeine solubility in supercritical
during rehydration, at 5 min and 40 min, respectively. This may carbon dioxide/co-solvent mixtures. Journal of Supercritical Fluids, 34(2),
209−214.
enable manipulation of rehydration rates in both the early and
Krokida, M. K., Karathanos, V. T., Maroulis, Z. B., & Marinos-Kouris, D.
latter stages. The subsequent texture of rehydrated samples (2003). Drying kinetics of some vegetables. Journal of Food Engineering,
dried with modified scCO2, was closer to the original texture of 59(4), 391−403.
raw carrot than that of samples dried via air-drying or using Krokida, M. K., & Marinos-Kouris, D. (2003). Rehydration kinetics of
pure scCO2, suggesting the discussed structural changes have a dehydrated products. Journal of Food Engineering, 57(1), 1−7.
positive impact on the texture following rehydration. It is Krokida, M. K., & Maroulis, Z. B. (1997). Effect of drying method on shrinkage
and porosity. Drying Technology, 15(10), 2441−2458.
anticipated that such microstructural differences may favour- Levi, G., & Karel, M. (1995). Volumetric shrinkage (collapse) in freeze-dried
ably alter sensorial properties of dehydrated–rehydrated foods. carbohydrates above their glass transition temperature. Food Research
Although some colour loss was observed following drying International, 28(2), 145−151.
with the supercritical systems this was found to be largely Lim, K. S., & Barigou, M. (2004). X-ray micro-computed tomography of
reversible on rehydration. cellular food products. Food Research International, 37(10), 1001−1012.
Lin, T. M., Durance, D., & Scaman, C. H. (1998). Characterization of vacuum
microwave, air and freeze dried carrot slices. Food Research International,
Acknowledgements 31(2), 111−117.
Lombraña, J. I., & Díaz, J. M. (1987). Heat programming to improve efficiency
Funding from the EPRSC and Unilever under the Engineer- in a batch freeze-drier. Chemical Engineering Journal, 35(3), B23−B30.
ing Doctorate scheme is gratefully acknowledged. Marabi, A., & Saguy, I. S. (2004). Effect of porosity on rehydration of dry food
particulates. Journal of the Science of Food and Agriculture, 84(10),
1105−1110.
References McHugh, M. A., & Krukonis, V. J. (1986). Supercritical fluid extraction:
principles and practice (pp. 181−198). London: Butterworths.
Abecassis-Wolfovich, M., Rotter, H., Landau, M. V., Korin, E., Erenburg, A. I., Mousavi, R., Miri, T., Cox, P. W., & Fryer, P. J. (2007). Imaging food freezing
Mogilyansky, D., et al. (2003). Texture and nanostructure of chromia using X-ray microtomography. International Journal of Food Science &
aerogels prepared by urea-assisted homogeneous precipitation and low- Technology, 42(6), 714−727.
temperature supercritical drying. Journal of Non-Crystalline Solids, 318 Namatsu, H., Yamazaki, K., & Kurihara, K. (1999). Supercritical drying for
(1–2), 95−111. nanostructure fabrication without pattern collapse. Microelectronic Engi-
Aguilera, J. M. (2005). Why food microstructure? Journal of Food Engineering, neering, 46(1–4), 129−132.
67(1–2), 3−11. Nijhuis, H. H., Torringa, H. M., Muresan, S., Yuksel, D., Leguijt, C., & Kloek,
Aguilera, J. M., Chiralt, A., & Fito, P. (2003). Food dehydration and product W. (1998). Approaches to improving the quality of dried fruit and
structure. Trends in Food Science & Technology, 14(10), 432−437. vegetables. Trends in Food Science & Technology, 9(1), 13−20.
Aguilera, J. M., Stanley, D. W., & Baker, K. W. (2000). New dimensions in Rahman, M. S., Al-Amri, O. S., & Al-Bulushi, I. M. (2002). Pores and physio-
microstructure of food products. Trends in Food Science & Technology, 11 chemical characteristics of dried tuna produced by different methods of
(1), 3−9. drying. Journal of Food Engineering, 53(4), 301−313.
Alavi, S. H., Rizvi, S. H. H., & Harriott, P. (2003). Process dynamics of starch- Rangarajan, B., & Lira, C. T. (1991). Production of aerogels. Journal of
based microcellular foams produced by supercritical extrusion I: Model Supercritical Fluids, 4(1), 1−6.
development. Food Research International, 36(4), 309−319. Ratti, C. (2001). Hot air and freeze-drying of high-value foods: A review.
Alavi, S. H., Rizvi, S. H. H., & Harriott, P. (2003). Process dynamics of starch- Journal of Food Engineering, 49(4), 311−319.
based microcellular foams produced by supercritical extrusion II: Numerical Reverchon, E. (1997). Supercritical fluid extraction and fractionation of essential
simulation and experimental evaluation. Food Research International, 36 oils and related products. Journal of Supercritical Fluids, 10(1), 1−37.
(4), 321−330. Sabirzyanov, A. N., Il'in, A. P., Akhunov, A. R., & Gumerov, F. M. (2002).
Beuchat, L. R. (1983). Influence of water activity on growth, metabolic Solubility of water in supercritical carbon dioxide. High Temperature, 40
activities, and survival of yeasts and molds. Journal of Food Protection, 46, (2), 203−206.
135−141. Sila, D. N., Smout, C., Elliot, F., Van Loey, A., & Hendrikx, M. (2006). Non-
Brunner, G. (2005). Supercritical fluids: technology and application to food enzymatic depolymerization of carrot pectin: Toward a better understanding
processing. Journal of Food Engineering, 67(1–2), 21−33. of carrot texture during thermal processing. Journal of Food Science, 71(1),
Cygnarowicz, M. L., Maxwell, R. J., & Seider, W. D. (1990). Equilibrium E1−E9.
solubilities of [beta]-carotene in supercritical carbon dioxide. Fluid Phase Sinesio, F., Moneta, E., Spataro, P., & Quaglia, G. B. (1995). Determination of
Equilibria, 59(1), 57−71. sensory quality of dehydrated vegetables with profiling. Italian Journal of
De Roos, K. B. (2003). Effect of texture and microstructure on flavour retention Food Science, 1, 11−18.
and release. International Dairy Journal, 13(8), 593−605. Srikiatden, J., & Roberts, J. S. (2006). Measuring moisture diffusivity of potato
Ekart, M. P., Bennett, K. L., Ekart, S. M., Gurdial, G. S., Liotta, C. L., & and carrot (core and cortex) during convective hot air and isothermal drying.
Eckert, C. A. (1993). Cosolvent interactions in supercritical fluid solutions. Journal of Food Engineering, 74(1), 143−152.
AIChE Journal, 39(2), 235−248. Takishima, S., Suzuki, Y., Konishi, Y., Ohmori, T., Sato, Y., & Masuoka, H.
Karathanos, V. T., Kanellopoulos, N. K., & Belessiotis, V. G. (1996). (1997). Production of silica aerogels by supercritical CO2 drying. Pro-
Development of porous structure during air drying of agricultural plant ceedings of the 4th International Symposium on Supercritical Fluids,
products. Journal of Food Engineering, 29(2), 167−183. Sendai, Japan, May 1997.
Ke, J., Jin, S. Z., Han, B. X., Yan, H. K., & Shen, D. Y. (1997). Hydrogen Tanaka, H., & Nakanishi, K. (1994). Solubility in supercritical fluid mixtures
bonding of some organic acids in supercritical CO2 with polar solvents. with co-solvents: An integral equation approach. Fluid Phase Equilibria,
Journal of Supercritical Fluids, 11(1–2), 53−60. 102(2), 107−120.
 Z.K. Brown et al. / Innovative Food Science and Emerging Technologies 9 (2008) 280–289 289

Tsami, E., Krokida, M. K., & Drouzas, A. E. (1999). Effect of drying method on von Behren, J., Chimowitz, E. H., Kumaran, V., Shapir, Y., & Fauchet, P. M.
the sorption characteristics of model fruit powders. Journal of Food (1997). Critical behaviour in confined fluids: Application to supercritical
Engineering, 38(4), 381−392. drying of porous silicon. Proceedings of the 4th International Symposium on
Vega-Mercado, H., Marcela Gongora-Nieto, M., & Barbosa-Canovas, G. V. Supercritical Fluids, Sendai, Japan, May 1997.
(2001). Advances in dehydration of foods. Journal of Food Engineering, 49 Witrowa-Rajchert, D., & Lewicki, P. P. (2006). Rehydration properties of dried
(4), 271−289. plant tissue. Journal of Food Science and Technology, 41(9), 1040−1046.