Dense phase carbon dioxide (DPCD) offers advantages of enhanced physical and nutritional qualities during the processing of juices. Here, freshly squeezed orange juice was treated with DPCD, and changes of physical properties and volatile components were investigated and compared with the original untreated and thermally treated samples. The correlations among physiochemical properties were also examined based on Pearson correlation, cluster analysis (CA), and principal component analysis (PCA). Significant correlations were found among the particle size, color parameters, and volatile compounds in the DPCD-treated samples. The 12 parameters were clustered into three groups using CA and PCA, and the eight volatile compounds were separated within the three groups. Nonanal and citronellol were clustered in group I, and they increased for a longer duration of more than 40 min with higher levels than the control. Parameters in group II included D (4,3), L, a, ethyl butyrate, and trans-2-hexenol, and they linearly decreased after 10–60 min DPCD treatment. The parameters of b and monoterpenes were clustered in group III, and they decreased within 40 min of DPCD treatment and then increased to an intermediate level. In addition, PCA clearly showed that the orange juice samples under DPCD for 10–60 min formed a “U” shape on the two-dimensional plot and that the samples treated by DPCD for 10 min and 20 min were closer to freshly squeezed orange juice than the heat-treated orange juice. This indicated that the nonthermal DPCD process offers the potential to be used more extensively in juice products.

1. Introduction

As a nonthermal process, the use of dense phase carbon dioxide (DPCD) is attractive because it causes less degradation to the quality of juices and beverages than thermal processes. The effects of DPCD on the nutrient content, stability, color, and sensory quality of foods have been widely studied. Many results have shown that DPCD does not modify the chemical, physical, and organoleptic qualities such as pH, °Brix, and turbidity of fresh liquid foods [13]. However, most results on the volatile components have shown negative conclusions: in most cases, the content of volatile components was reduced significantly. Gasperi et al. [4] reported that the contents of esters and aldehydes in fresh apple juice were reduced by more than half after DPCD treatment. Plaza et al. [5] reported that DPCD treatment reduced the total MS ion chromatogram peak area by 35% in guava purée. It is possible that the volatile components were removed by CO2 during depressurization. However, a few reports showed no significant change [6] or even an increase of some volatile components [7] after DPCD treatment. It is well known that the cloud juice is a liquid-solid system consisting of the serum and a heterogeneous water-insoluble phase consisting of small particles [8, 9]. In orange juice, there is a clear partitioning of the volatile components between the insoluble pulp and the aqueous serum [10]. Moreover, when juice is immersed in dense CO2, the particle size, solubility of volatile substances, and the distribution of pigments could change under certain pressure and temperature conditions [1113]. For example, in carrot juice [14] and peach juice [15], the particle size increased significantly. In contrast, the b value (yellowness/blueness) decreased in grapefruit juice [1] and mandarin juice [16] after DPCD treatment.

To date, little information is available on the relationship between changes in different properties under DPCD treatment, and determination of the quality changes with prolonged DPCD duration remains confusing. The present study focuses on the changes of particle size, color, and volatile components of freshly squeezed orange juice after DPCD treatment. Cluster analysis (CA) and principal component analysis (PCA) were used to investigate the correlations of the parameters after DPCD treatment; furthermore, the quality of the juices after DPCD treatment was compared with the original freshly squeezed and the heat-treated juice.

2. Materials and Methods

2.1. Orange Juice Preparation

Fresh Hamlin oranges were peeled manually and then mashed using a domestic blender. The pulp was then packed into a nylon cloth (80 mesh) and squeezed to obtain the juice. Two liters of juice were collected and mixed well. The juice was then divided into three groups. One was the untreated fresh samples, while the other two were immediately subjected to DPCD treatment and thermal treatment (90°C, 60 s), respectively. After treatment, all of the samples were stored at 4°C and all parameters were determined within 24 hours.

2.2. DPCD Processing

The DPCD system in this study was the same as that previously described by Zhou et al. [14]. A high-pressure vessel was heated to 55°C and then evacuated. Under vacuum conditions, 200 mL juice was drawn into the pressure vessel by opening the sample inlet valve for each treatment, and then the valve was closed immediately. Meanwhile, the CO2 inlet valve was opened, and the pressure was increased to 40 MPa at the rate of 10 MPa/min. The high pressure was applied for 10, 20, 30, 40, 50, and 60 min, and six DPCD-treated samples were thus collected.

2.3. Particle Size and Color Analysis

The particle size distribution was determined using an LS 230 laser diffraction particle analyzer (Beckman Coulter Inc., Brea, CA, USA) as described by Zhou et al. [15]. The volume mean diameter D (4, 3) was determined.

The color parameters (L, a, b) were measured using an automatic colorimeter (WSC-S, Shanghai Jingmi Instrument Co., Shanghai, China). The instrument was calibrated using a black cuvette for attenuated reflection and a white ceramic tile as the reference standard (National Standard Research Center, Beijing, China).

2.4. Determination of Volatile Compounds

The volatile components were extracted using a solid phase microextraction (SPME) system. An 8 mL juice sample was placed in a 15 mL vial containing a stirring bar and 2.56 g NaCl, and then the vial was placed into a water bath at 40°C and stirred at 200 rpm. After allowing 10 min for temperature equilibration, a 100 µm polydimethylsiloxane-coated fiber (Supelco Ltd., Bellefonte, PA, USA) was inserted and allowed to extract the volatile components for 20 min.

The volatile compounds were separated and detected using a 7890A/5975C MS system equipped with an HP-5 column (30 m × 0.25 mm × 0.25 µm thick film) (Agilent Technologies Inc., Santa Clara, CA, USA). The GC injection port was set at 220°C, and the MS detector was operated in the scan mode with 70 eV electron impact, scanning throughout the 30–450 m/z range. The multiplier was set at 1.0 kV, and the source temperature and transfer lines were maintained at 230 and 280°C, respectively. The oven program was as follows: initial 50°C for 2 min, increased by 4°C/min to 160°C, increased by 10°C/min to 220°C, and then maintained for 3 min at 220°C.

The C8-C24 standard alkane and standard flavor compounds of ethyl butyrate, trans-2-hexenol, α-pinene, phellandrene, limonene, linalool, nonanal, and citronellol (Sigma-Aldrich, Shanghai, China) were injected under the same chromatographic conditions. The volatile flavor compounds were identified according to the NIST 2013 database for standard compounds, and then the retention indices were calculated from the alkane series.

2.5. Statistical Analysis

Each sample was tested in triplicate, and the data were expressed as the mean ± standard deviation. All data were expressed as percentage change relative to the corresponding control with the average of the control sample as 100%. ANOVA, multiple correlation, and PCA were performed using JMP Pro 10 software (SAS Institute Inc., Cary, NC, USA). CA was performed and visualized using MultiExperiment Viewer software (http://www.tm4.org/#/welcome).

3. Results

3.1. Changes of Particle and Color Parameters

The effects of DPCD treatment on the physical parameters of orange juice are shown in Figure 1. The particle size of samples treated by DPCD for 10 min was not significantly different () from the untreated (control) sample but was reduced significantly—approximately 20%—by the 20 min treatment. The changes in particle size of the orange juice remained insignificant until the DPCD treatment time reached 50 or 60 min. In other words, the D (4, 3) values of samples treated for 30 or 40 min were not significantly different from those of the sample treated for 20 min. In addition, there were no statistically significant differences in particle size between control and thermally treated samples (99.34% that of the fresh sample), and the smaller values of DPCD processing correspondingly resulted (71.83–81.13%).

No significant differences were found () among the L values of all the treated and control samples, while a values of all the DPCD-treated samples were lower than those of the control sample (). The b value reached the lowest magnitude after 30 and 40 min of DPCD treatment and then slightly increased (Figure 1). All of the b values were higher than those of the control sample, while b value of the heat-treated sample was close to () the lowest value of DPCD-treated samples, which occurred after DPCD processing for 20 and 30 min.

3.2. Changes of Volatile Compounds

The retention of volatile compounds in orange juice is shown in Figure 2. A gap in the vicinity of 70% can be seen in the figure, which divided the 8 compounds into 2 classes. Ethyl butyrate, trans-2-hexenol, α-pinene, phellandrene, and limonene were retained to a lesser extent, while linalool, nonanal, and citronellol were retained more effectively; furthermore, nonanal and citronellol were obtained in higher content than the untreated sample (100%) with DPCD duration times beyond 30 min. In contrast, the contents of the monocyclic terpenes exhibited the lowest values after 30 or 40 min of DPCD treatment. Compared with the DPCD-treated samples, the contents of the volatile compounds in the heat-treated sample (Supplemental data Figure S1) were near the values of the DPCD samples after 60 min duration (), except those of trans-2-hexenol and citronellol, which showed lower levels ().

3.3. Linear Regression Analysis

The changes of the parameters with extending DPCD treatment time were linearly fitted, and results are shown in Table 1. D (4, 3), L, a, ethyl butyrate, and trans-2-hexenol decreased linearly (), while b, α-pinene, phellandrene, limonene, linalool, and citronellol were not linearly changed. Nonanal was the only volatile component which increased linearly. It is worth noting that the L values decreased linearly, although these were not significantly different when using ANOVA.

3.4. Correlation Analysis

The interdependence of the physical and volatile variables was investigated by analyzing their correlation coefficients (Table 2). The particle size D (4, 3) was significantly correlated () with the ethyl butyrate () and citronellol (), indicating that a high particle size value was always associated with a high ethyl butyrate concentration but with a low content of citronellol. L showed no significant correlations with any other parameters. The a value showed a significant positive correlation with the ethyl butyrate (, ) and trans-2-hexenol (, ) but a significant negative correlation with the content of nonanal (, ). The value of b was also positively correlated with the content of α-pinene (, ), phellandrene (, ), and limonene (, ). There were also significant positive correlations between the contents of some volatile compounds. Linalool was significantly correlated () with the content of α-pinene, phellandrene, and limonene, and the three monoterpenenes were significantly positively correlated with each other.

The associations and differences in the parameters can be clarified by a simple inspection of the heat map (Figure 3): the levels of some parameters increased during processing (group I, nonanal and citronellol), while others decreased (group II, D (4, 3), L, trans-2-hexenol, ethyl butyrate, and a). The levels of the parameters in group III (linalool, phellandrene, limonene, α-pinene, and b) were also closely related, decreasing within 40 min of DPCD processing and then increasing after 50 and 60 min treatment to intermediate values between those for 10–20 min and 30–40 min treatments.

The results can be summarized by PCA (Figure 4). The two major components accounted for 90.5% of the cumulative variance. The PC1 (horizontal) explains 61.8% of the variance; on this axis, the samples were clearly arranged from right to left according to their DPCD duration time on the score plot (Figure 4(a)). PC2 (28.7%) separated the samples with medium treatment time (30 and 40 min) from the other four samples and all of the samples forming a “U” shape on the dimension plot. The PCA clustered the parameters in the same manner as the CA results in Figure 1, but the PCA plot showed opposite directions of group I and group II, with group III settled between them. When the data from the original fresh sample and the heat-treated sample were involved in PCA plots, limonene, linalool, α-pinene, phellandrene, and a were close to the fresh sample, which indicated that the higher values of these parameters represented the characteristics of fresh orange juice, while the changes in the nonanal and citronellol contents, as well as the b values, were correlated with the longer DPCD treatments. In addition, the samples with DPCD treatment time below 20 min were closer to the original fresh sample than the heat-treated sample on PC1 (Figure 5), and the increasing trend of L values was generally correlated with the heat-treated sample.

4. Discussion

Multivariate analysis is a powerful method to extract hidden information from a large amount of data. It can generate visual graphs or plots, and it has been used widely to elucidate the inner relationships/differences between food structures, quality, processing procedures, etc. [17]. It is well known that the heat treatment of orange juice generally causes significant changes, while DPCD-treated samples usually present some advantages in quality retention [3, 18]. CA and PCA made it possible to illuminate the differences of DPCD-treated samples from the untreated and heat-treated samples. In the present study, CA and PCA were used to reveal the association among the parameters relevant to the quality of orange juice processed by DPCD. Indeed, CA and PCA helped to group these parameters according to the three different ways in which they changed with lengthening DPCD duration.

Most researchers considered that DPCD treatment could lead to smaller particles in the juice colloid, probably by means of the homogenization effect, which facilitates the splitting of larger particles into smaller ones [14, 19, 20], while Zhou et al. [15] found that the particle size increased in carrot juice and peach juice as a result of acid-induced protein coagulation. However, the particle size in this study was not successively decreased. Most probably, this change was the combined outcome of both CO2-induced coagulation and DPCD-induced homogenization, among which the homogenization effect should be the major effect, although it required some time to be obviously manifested. Otherwise, the overall reduction of particle size by DPCD treatment might indicate the stability enhancement and quality improvement of the juice [9].

According to the reports on color changes in citrus juices after DPCD treatment, DPCD increased the yellowness values of carrot juice [14], orange juice [19], and mandarin juice [16] while decreasing their redness values. Our results in the present study confirmed these changes. Some other treatments such as heat treatment could also result in similar changes in color. For example, Lee and Coates [21] found the same color changes in orange juice after pasteurization when a value decreased, while b increased significantly. They considered that the isomerization of 5,6-epoxide carotenoids to 5,8-epoxide carotenoids was probably responsible for the color changes. However, Arena et al. [22] provided the insight that the varying distribution of carotenoids between the serum and the pulp, as well as the modification of particles, could be responsible for the change of the orange juice color; Kraska et al. [11] reported that the solubility of β-carotene, a low-solubility pigment in compressed CO2, depended on the density of the supercritical fluid which was an outcome of pressure and temperature [12]. In the present study, the increased yellowness and the decreased redness might be attributed to the comprehensive effects of the distribution change and isomerization of carotenoids as well as the modification of particles.

The volatile components of orange juice are partitioned between the insoluble pulp and the aqueous serum. The insoluble components in orange juice were classified as pulp particles (>2 µm) and finer particles (<2 µm) [19, 23]. The monoterpene, sesquiterpene, and long chain aliphatic aldehydes existed primarily in the orange juice pulp, while esters and monoterpene alcohols were mainly in the serum [24]. In this study, the mild increase of three monoterpenes (limonene, phellandrene, and α-pinene) and nonanal in the DPCD-treated samples with duration times longer than 40 min might be due to the splitting effect of the coarsest particles (>2 µm), since they were primarily in pulp. However, ethyl butyrate and trans-2-hexenol decreased linearly with time extension of DPCD treatment, indicating that they were mainly present in the serum and unable to be enhanced by the DPCD effect. Considering the lower level of the volatile compounds after either the heat treatment or the DPCD treatment compared with the control, the declining trend of some volatile components might be attributed to their volatility, the effect of heat, and their release during depressurization [25, 26].

The juice quality also depends on the volatile compounds. In fresh orange juice, fewer than 25 volatile components are odor-active [10]. trans-2-Hexenol arises from 3-hexenal, which is an important contributor to the green, grassy top-note of freshly prepared orange juices, which is nearly always reduced significantly in processed juices [27]. The DPCD-treated samples in this study retained significantly more trans-2-hexenol than the heat-treated one, possibly indicating that a higher intensity of fresh odor remained. Ethyl butanoate is one of the most potent esters in orange products and is described as a fruity flavor [10]. Linalool is the most aroma intensive alcohol in orange juice and possesses a distinctive floral, sweet odor [10]. Both ethyl butanoate and linalool are important contributors to the desirable aromas. Linalool is less sensitive to the DPCD and heat treatment without significant change. Additionally, ethyl butanoate was reduced linearly with DPCD treatment time extension, and the samples with the longer treatment times (50 and 60 min) showed similar content to the heat-treated one, suggesting a decrease of fruit aroma intensity.

Limonene is the most abundant component, usually with a peak area over 90% of the total area when determined using GC [28] but is readily converted into α-terpineol and carvone which are indicators of the aroma degradation of orange juice during processing and storage [29]. Limonene and α-pinene contribute to the minty, lemon, citrus-like and resinous, pine tree, ethereal odor of orange juice, respectively [10]. The aliphatic saturated aldehyde nonanal contributes a significant odor component, described as metallic, to orange essence oils [30], but a citrus-like, soapy, floral character in juice [10]. Citronellol and α-pinene are aroma components with minor content in orange juice [30], contributing floral, sweet, metallic, and minty odors [31].

Nonanal and the monoterpenes nearly always demonstrate higher content in mechanically extracted juice than in hand-squeezed samples, since they are correlated with pulp content [24] and peel oil [10]. The reduction of ethyl butanoate and the monoterpenes, as well as the increase of nonanal, was reported in high hydrostatic pressure-processed orange juice, while the aroma characteristics of HHP-treated juice were unchanged [27]. Therefore, the changes of the flavor component concentrations in orange juice possibly do not significantly modify the sensory aroma; nevertheless, extended treatment should be avoided in order to retain the fresh flavor.

Most previous studies only reported the reduction of volatile compounds after DPCD [35]; in this work, increases of some volatile compounds were detected, and the increase of the volatile components was found to be due to the decreasing changes of particles. In addition, the high correlations of the volatile components and color parameters might reflect that the volatile components and pigments changed simultaneously in the juice system.

5. Conclusion

A major challenge for the juice processing industry is to produce a juice with a long shelf life and a flavor close to freshly squeezed juice. Here, the DPCD-treated samples with duration times below 20 min could be closer to the freshly squeezed than the heat-treated sample. The CA and PCA were powerful tools to help assess the changes in the orange juice samples. The changes in particle size, color, and volatile components of orange juice during DPCD processing were significantly correlated. The changes in the orange juice indicated that DPCD offers the potential to be used to produce juice products of excellent quality.

Data Availability

The data used to support the findings of this study are available from the corresponding author upon request.

Conflicts of Interest

The authors declare that they have no conflicts of interest.


The authors are grateful to the Key Laboratory of Fruit and Vegetable Processing, China Agricultural University, for providing the DPCD equipment. This work was supported by the Natural Science Foundation Program of Jiangsu Province (BK20161376).

Supplementary Materials

Figure S1. Effect of DPCD processing on volatile components of orange juice: (A) ethyl butyrate; (B) trans-2-Hexenol; (C) α-pinene; (D) phellandrene; (E) limonnene; (F) linalool; (G) nonanal; (H) citronellol. Different letters on the columns indicate significant differences (). F, freshly squeezed juice; 1–6, DPCD-treated samples with time of 10, 20, 30, 40, 50, and 60 min, respectively; H, heat-treated sample. (Supplementary Materials)